12:51
Robin
no it doesnโt make sense. the device either consumes 3kw (which results in 72kwh per day at full load, at $0.3 per kwh thats $21.6/day) or it consumes 3kwh in a certain timeframe, which does not make sense if you account for 24 hours and also does not fit his calculation
๐
12:52
so dan you should have written โ3kwโ because thats power, kwh is energy ;)
12:52
Dan "Fuserleer" Hughes
ok ok maybe 3kw was a little high, you get the point
12:53
Lennart
Oops my bad
12:53
Mike (stringhandler)
Yeah I think about 1kw is high but probably a decent maximum for estimates
12:53
Robin
all good, just want to avoid confusion ;)
12:53
2x192 thread epyc consumes 2x400 watt alone
12:53
Mike (stringhandler)
I think data transfer will be quite a high cost, especially if using AWS or similar
12:53
Dan "Fuserleer" Hughes
flashbacks to my HP blade box that draws 10kw at full everything lol
โค
12:54
Robin
plus ram plus fans is easily 1,5k kw
Deleted invited Deleted Account
12:54
Dan "Fuserleer" Hughes
yeah pure electricity cost is perhaps 25-40% dpeending on setup, location, etc
12:55
which you also need to account for ofc
if the load is 0, you still have immutable costs (rent, internet etc)
12:56
so yeah, efficiency of the protocol can potentially reduce the fee, maybe quite a bit, but the other things are just static costs you cant really do anything about
12:59
Deleted Account
I'm just wondering why did you focus around volume in your comparison tweet instead on liquidity. Because rush explicitly said liquidity/marketcap ratio is the issue. Which makes sense because liquidity is much harder to game than volume
13:00
Dan "Fuserleer" Hughes
I couldnt get historic liquidity scores from CMC, or depth
Volume forms part of the liquidity metric. Higher volume = thicker order books = better 2% depths = better liquidity
13:01
Deleted Account
Do you have any idea what is tier 1 exchange feedback when radix applies for listing?
13:02
Dan "Fuserleer" Hughes
Exchanges want volume because volume = fees
๐
13:02
Deleted Account
Thanks
13:03
Dan "Fuserleer" Hughes
Is this an acceptable answer for you?
padre
13:05
Lennart
That's seems like an impressive amount, so this doesn't vary very much with the amount of shards that are involved in the transaction?
13:07
Oh wait, the amount of validators does increase linearly with the amount of shards involved right
13:08
Dan "Fuserleer" Hughes
Ive tested Cassie in its current level of implementation and optimizations to about 5k tps per group with simple transactions.
Transactions which touch more shards obviously eat into that baseline budget
the 1-2k tps is a mix of transactions, from the simple, to the more complex (touching 8-10) distributed as a longtail ...as in more complex ones are more rare so there are less of them
13:09
so if you have 10 groups, but ONLY the simple baseline type transactions then the network can do 50k total
13:09
Lennart
So that would result in a net reduction of tps, as multiple validator sets will handle the same transaction
13:09
Dan "Fuserleer" Hughes
if you have a mix like some of my tests, then the network can do 10-20k total
13:09
Explore the Universe
For Xian a simple all in one solution might be good for average users.
A box that you plug into your network with a cellular fallback so the network doesn't slow down because people can't keep their servers up at their home networks.
13:10
Dan "Fuserleer" Hughes
if you pile more validators into a group (say 500 rather than 250) thoughtput doesn't really change, but latency/finality goes up a little
13:11
Explore the Universe
That's what I meant with slowing down when suddenly lots of unskilled users run nodes.
13:11
Dan "Fuserleer" Hughes
heh, well the network decides where your validator goes
13:12
and you'd need 1000s of validators per group to make a noticable difference as the latency increase is logarithmic over the baseline
13:13
Explore the Universe
So no all in one box solution with cellular backup required?
13:13
Dan "Fuserleer" Hughes
I mean you could do that but I dont think we want to get into hardware too ๐
13:17
Lennart
Great stuff, I am curious to see to what level of transaction complexity we will go in the upcoming years. Thanks for your time and insights!
๐
14:22
Haumikura
@danhughes, to accomplish something of this magnitude we can easily find people willing to help with the organization, all it takes is a leader to go through the details of what is needed. Radix asks for community support to grow radix, so let's put the community to work and use it.
This is your orchestra, define your musicians and let them do the rest.
๐
6
22:12
Martin
I met him in Amsterdam this week. He said he is editing ๐
๐
6
23:41
Guilherme Vieira - Dont DM - Wont check
Guilherme Vieira - Dont DM - Wont check 05.11.2023 23:38:34
In couple minutes Iwill be Live streaming with one of the majors YouTube channel in Brazil, talking about radix lending protocols...
(Portuguese language)
Any support with engagement is welcome!!
https://www.youtube.com/live/FlASfZyOOGE?si=0jwGH07dsrn5ARc-
๐ฅ
20
๐
1
๐
1
12:50
XRDWhale.xrd |๐ฆ๐บ Radix Ambassador
So there is a great marketing metric that just gets everyone tickled in all the right places, even if sometimes its all BS. People tend to get excited from it and it draws a crowd, entices them to have a taste.
It's no other than TPS
Now
@danhughes you did huge TPS on Tempo of~1.4m TPS and that was a hit from a marketing perspective and got a lot of attention. We now have TON getting all the attending on ~100k TPS and being classed as the world record for any blockchain.
Would you consider going all out with
#Cassie by breaking the record for TPS?
Make a spectacle of the whole thing. Don't just do a little better than the 100k but blow it out the water even blow out the water the Tempo test if you can. Now that would get heads turning every which way.
14:25
Dan "Fuserleer" Hughes
If we were going to do a tps test with Cassie I'd be shooting to smash tempo's record let alone TON
14:26
That would require a lot of participants tho with decent machines
14:26
But doable with enough planning
14:26
Anima
More raspberry Pi's you say?
๐คฉ
14:27
Dan "Fuserleer" Hughes
Cassie's database vamos is easy faster than the dB we used in tempo. Immediately that would give us a 30+% increase in throughput with the same network topology
๐ฅ
13
โค
2
๐
2
14:31
Haumikura
Dan, give us some numbers, so we can see if this is possible to achieve, how many devices are we talking about?
14:32
Inspector Crypto.xrd
TPS porn ๐ฅ
๐ฅ
7
โค
๐
1
14:34
RadixDeFiMatrix
If you want people to know that it's incredible, then you have to show it to them and bring in the media to broadcast it. You win over the public and investors at the same time, and it gives you a strong marketing argument ๐
14:35
Haumikura
My mind can't imagine the size of the noise and black hole we would create if you managed to beat the tempo record on a community supported network.
10 listings on binance wouldn't make the noise this would.
๐ฅ
14:55
h0ll0wstick ๊ฎ Old account
Tempo was like +1k nodes
14:56
How can we do 2k? How realistic is this ? How many people would be willing to participate ?
15:08
Matheus Piva
See this sponsored ad ๐
๐
๐
๐คฏ
15:13
Haumikura
wtf kkkkk
๐
15:15
Dan "Fuserleer" Hughes
100k is amateurs
If you look at the Cerberus white paper we smashed 1m there with a pretty small network and crude implementation.
Granted that was particularly decentralised, but it could have been. Just no need to do so for the paper
๐
15:16
We could probably rip tempo these days with 1k ... Maybe even less. Everything is much more efficient in the stack, from dB, network Io all the way to Cassie and consensus
15:19
Mike (stringhandler)
How do you get to the 1m in the Cerberus paper? From the graphs I see you need 61000 VNs and 2 objs per shard
15:19
Am I reading that correctly
15:19
Haumikura
@danhughes i would say that up to 500 would be possible to reach with a LOT of organization, above that I believe it would be very difficult to find enough people in the communities, including that some would need to use more than 1 machine. The rest would have to be complemented with cloud. And if it were possible to use Windows it would also make it much easier to reach these people.
๐
15:20
RadixDeFiMatrix
I'm saying that a small official oklm world record could give us a nice little spotlight. Maybe Visa and MasterCard will even want to come and take a look at Radix. ๐
๐
15:20
Inspector Crypto.xrd
How would we convey to ppl that the txns types are a mix of simple and complex? Just show transaction history afterwards?
15:20
h0ll0wstick ๊ฎ Old account
It s java so windows is no problem
15:20
Inspector Crypto.xrd
Coz thats an important point to make
๐
15:22
Lennart
This might actually be one of the better opportunities for marketing that we have as a community, don't know how difficult it would be to organize though.
15:24
h0ll0wstick ๊ฎ Old account
Ok so lets say 1k, would it be from your side (rdx works) or with everyone?
15:24
Last time we were like 15 + dan ๐
15:25
Lennart
Even though it might not represent current network capabilities, it might still bring in devs because of the prospect of scalability
15:25
Dan "Fuserleer" Hughes
No, 1m+ is 2 shards per tx with 1024 shard groups as per the Cerberus test.
Tho my bad a little, I was thinking Cassandra when I said "small network" and the results in that paper which is being finalized
15:27
Something to think about or we look if tons was just simple tx and do the same.
๐
15:27
Mike (stringhandler)
But how many vns per shard group
15:27
Dan "Fuserleer" Hughes
I'd like it to be much more community and less me/radix
We got beef for that last time with tempo even though it WAS a fair test
๐
15:28
Inspector Crypto.xrd
I bet theirs is just simple
15:28
Dan "Fuserleer" Hughes
For a pure tps test like this we could reduce the number but still show sufficient safety bounds.
So instead of 100 perhaps 20-25
๐
15:28
Depends what our validator budget is I guess
15:29
h0ll0wstick ๊ฎ Old account
Makes sense
15:30
Dan "Fuserleer" Hughes
Whoever had the idea of distributing a Cassie token for participating is a good idea tho
โค
6
15:30
About time they were put to use
15:35
h0ll0wstick ๊ฎ Old account
Indeed
15:35
Explore the Universe
Is there a way to get a network of universities involved? That could give a lot of additional machines.
๐
9
15:36
h0ll0wstick ๊ฎ Old account
So like an airdrop to everyone participating? + an nft to register
๐
15:40
Dan "Fuserleer" Hughes
maybe, would that silence the trolls and their decentralization fud though?
15:40
I R
Pretty sure there'll be a lot of people from the community willing to join, IF we get an "idiot-guide" of how to set up a validator.
I myself am one of the people who'd need this to help :)
โค
15:41
Dan "Fuserleer" Hughes
its pretty "idiot friendly" anyway tbh ๐
basically, install java, run provided script, done
15:45
h0ll0wstick ๊ฎ Old account
Also Allocate at least 8gb of ram when you install java
15:45
I R
You continue breaking it down to a few smaller and more specific steps and I'll tell you when it's "idiot-enough" for the rest of us ๐
๐คฉ
4
15:45
Dan "Fuserleer" Hughes
hehe ok deal
15:49
Haumikura
Perhaps if a foundation created a "marketing call" inviting academia to participate, we could get some good additional numbers.
15:51
oh, you can be sure we will be able to, I made some housewifes manage to use a python bot to create accounts and stake them there in olympia.
๐
15:56
Buddhini
is there a github to cassie or smth?
16:00
Inspector Crypto.xrd
Get on a Mario crypto spaces and put the call out for anyone who wants to participate in a world record attempt
16:01
h0ll0wstick ๊ฎ Old account
Not sure mario's space is the best way to find people dedicated enough but is sure a great niche to get people to watch
16:02
Inspector Crypto.xrd
yeh not sure either to your first point
16:02
Buddhini
how is our sharding better/worse than NEAR sharding for example? is it the atomic composability that sets us apart?
16:20
h0ll0wstick ๊ฎ Old account
Whats the tg for cassie token?
16:21
All of them yeah it s how you handle AC on cross shard, last time i checked nightshade from near does async composability meaning lock state involve cross shard
16:21
It s blockchain so...
16:23
And finality on near is high with very low demand so far
16:24
They are doing 5tps atm not much happening on near
16:24
Buddhini
now that i think about not being a blockchain - do we have any way of proving that a transaction took place in the past after it has been pruned? (like you can do e.g. with merkle trees in blockchains)
16:55
Timan | Astrolescent
With a bit of preparation we could probably create an AMI image on Amazon AWS which would make it rather easy for everyone to start up nodes. Even starting 10, 20, 30 would be easy depending on the configuration needed for every single machine
โค
5
๐
16:56
And running 600 machines for an hour is as expensive as running one for a month
16:57
I would be happy to help
โค
17:12
Emmoglu Leo Magal - I don't DM first
That's what I thought as well. With scripts to spin several nodes at once
17:18
Slammer
Ooft...things you love to hear ๐
17:19
Haumikura
The best thing would be to use as little cloud as possible, just what is necessary to complete the numbers that are necessary. This would show how robust the algorithms are, capable of dealing with hardware and Internet inconsistencies.
17:20
Matus โ
been lurking in the radix main channel in the last 3 months but here now after the rumors that 1.4mil tps is expected to be broken with cassandra soon ๐
๐ฅ
5
17:23
Zoran
Ha ha, same here. Active on twitter and in several other groups (excluding main) but was not member here.
Saw
@luke55 posting about 1.4M tos so realized it's about time to join ๐
17:27
lukas.xrd | Ociswap
๐
8
17:32
Emmoglu Leo Magal - I don't DM first
Let's keep this cooking... We can cover this in several YouTube channels live simultaneously at the same time ๐๐ฅ
๐
7
17:32
Zoran
Do not worry, I'm shilling in the name of Radix almost everyday ๐ Should change my nick here to Zoddo too, that's how most of you know me ๐
๐
5
17:34
Loccoco
We could maybe do the same with other providers (azure, gcp...) to distribute network and be closer to real time conditions
๐ฅ
17:35
Would be happy to spin a few nodes for a day too if there is a tutorial at hand
17:36
Emmoglu Leo Magal - I don't DM first
@danhughes we're going to make a poll at the brazilian community to gather how many are willing to join.
Could you please remember us what would be the minimum system recommend?
Iirc 4 core, 8gb, nvme SSD, 100mb bandwidth!?
17:38
h0ll0wstick ๊ฎ Old account
Let s ask in every country who would be interested?
๐ฅ
6
17:40
Haumikura
It's important for Dan to talk about whether a low-performance SSD would be capable, as NVMe in Brazil only applies to very new devices!!!
๐
17:40
h0ll0wstick ๊ฎ Old account
Everyone needs to have java installed allocate 8gb minimum of ram and have a ssd hd
17:40
Haumikura
It would be good to spread it among local community groups.
17:41
h0ll0wstick ๊ฎ Old account
The more it spreads geographically the better
โค
17:42
Japan
/korea /china ambassadors, ask c9 if they would have a few machines to spare, oci team im sure are up for it
Battlefield6 invited Battlefield6
17:44
h0ll0wstick ๊ฎ Old account
Does community council help that would be great for communication as well
[PSB] PSILOBLOX | Boris invited [PSB] PSILOBLOX | Boris
17:55
[PSB] PSILOBLOX | Boris
Whatโs happening ๐
17:55
Kansuler
Would've been nice with that amount of likes ๐๐
๐ฅฒ
17:56
projectShift
So, we're making a community run of Cassie?
๐
17:56
lukas.xrd | Ociswap
"World's Fastest Blockchain" well played... :D
they can always say that they never claimed to be the fastest DLT lol
17:57
projectShift
Last version i run could do multie instances, so if that's still avaliable, we can have high numbers, albeit clustered per actual site
17:57
Explore the Universe
Just to clarify, someone said the TON demo was with smart contract calls, not just simple TPS. Is that accurate?
17:58
projectShift
Do we have a time for this already? On phone, didn't read previous msg
18:00
Haumikura
Not yet, we are looking at feasibility and waiting for the boss's green light.
๐
18:03
lukas.xrd | Ociswap
๐
18:09
Guilherme Vieira - Dont DM - Wont check
Thanks for calling my 6 year old notebook "very new" ๐
18:10
Haumikura
did it come from the factory with nvme?
18:10
Guilherme Vieira - Dont DM - Wont check
18:12
Dan "Fuserleer" Hughes
Challenge accepted. What were the machine specs tho? And how was this test conducted.
If someone can find that while I do family time would be greatful otherwise I'll have a look later
๐ฅ
8
๐
3
18:22
Blind5ight (Watch out for scammers)
Time to line up the Howitzers?
18:25
lukas.xrd | Ociswap
Blog article and video is really lacking more insights about this test. Only nodes from TON foundation without community interaction? No clue. It really feels like everything is orchestrated from the team and all nodes from the same provider... meh. Also we need to trust numbers they are showing
on a dashboard and that's it.
105k peak TPS (complex not simple), 6s finality with 512 shardchains, 256 validator nodes running on servers of Alibaba cloud is basically the only information I'm getting out of this.
18:27
so there is 2x the amount of blockchains "shardchains" for the amount of validators, how's this even possible lol?
๐
18:29
Blind5ight (Watch out for scammers)
18:38
[PSB] PSILOBLOX | Boris
๐
๐
18:40
Haumikura
Ohhhhhh yeahhhhh, letssss dooooo iiittttt
18:42
Dan = thanos
Glove = community
๐ฅ
18:57
Explore the Universe
Good detective work. Would not try to attack their narrative though, just trump it.
19:08
Slammer
Getting targeted...Time to show them who's boss ๐
@danhughes
19:11
Dan "Fuserleer" Hughes
๐
19:40
h0ll0wstick ๊ฎ Old account
Maybe it wasn t permissionless
20:47
Cooper .xrd
IMO, doing a tps test with something that is far from production is a waste of time. But if you can make Cassie run Scrypto, or otherwise make it close to Babylon, then itโs a winner. There are too many reasons at this time for people to dismiss the results.
20:48
Explore the Universe
Even people dismissing results is good for attention and engagement.
Radix needs more eyes.
20:54
Al Bunderino
2 machines here.
4 core i5, 8 core Ryzen SSD and 64 gigs of ram each.
Ready when you are.
๐
20:55
projectShift
please someone ping in here when we have enough details for that Cassie run.
I can run from home and from the cloud, several locations, especially if it's still now in Nov :-P
@danhughes does Cassie still have the godix option so we can spawn locally several instances?
I can use two machines here in the home lab and two of the servers
21:13
SeF
Ryzen 9 5950x 16 cores 32 threads, 128GB RAM, NVMe Datacenter SSDs 2x2TB and 2x4TB. Could provision a few VMs if there are specs for Cassie.
@danhughes
21:31
A test in this size would be impressive.
21:31
The radix community is not the only one asking questions on how TON did that๐
21:32
Dan "Fuserleer" Hughes
Cassie has a very similar statemodel to Babylon (on purpose obviously)
There's also an execution engine that shares a lot of the same fundamentals and concepts to ensure that Cassie research is compatible with the Radix engine as close as possible without bolting it on (lot of work)
That said something I plan to do is to bolt on the Radix engine and scrypto but that's probably a couple of months to do while staying on top of current workload.
๐
21:33
It's always been my intention with Cassie to rip apart all doubt of the tempo tests at some point.
Taken longer than expected to get there but this kinda stuff really is up my street
๐ฅ
11
21:34
Jelthebest
Has the babylon tps been tested already? I know itโs not going to be impressive, but nice to know the number.
21:34
Lennart
I'd love to see it, and be part of it๐ฅ Never had a chance to be part of a cassandra network test
21:35
Dan "Fuserleer" Hughes
Yeah we did tests internally with Olympia and came out at about 50
Babylon had some improvements and optimization that maybe increases that a little bit we haven't done a direct test of it
My guess would be ~60-70 at best without more work to optimize certain areas
๐
21:36
If we did that work, maybe pushes it to 100
Speculation ofc
21:37
Loccoco
Any chance to perform a test with current configuration to silence the plebs (like me ๐) or it doesn't make sense ?
21:38
Dan "Fuserleer" Hughes
Effort Vs an addition 10 teepees doesn't really seem worth pulling someone off current tasks for a couple days to do it
21:38
Loccoco
I think a little web3 twitter twitch would be refreshing for many
21:40
Got it, i'll wait till your out of the batcave. Ping us when you'll need community nodes, i'm sure all will provide
21:51
Cooper .xrd
I get it, but the more disclaimers you have, like these, the less credibility the results will have, for observers who are not already in the Radix camp.
22:04
Dan "Fuserleer" Hughes
Yup get that, learned that from tempo tests
22:06
Markus
But they would talk about us. If there is nothing they can doubt, they will ignore us.
22:06
XRDWhale.xrd |๐ฆ๐บ Radix Ambassador
XRDWhale.xrd |๐ฆ๐บ Radix Ambassador 06.11.2023 22:06:30
22:09
Blind5ight (Watch out for scammers)
Is it a def gonna happen or is the community getting ahead of itself? ๐
22:14
XRDWhale.xrd |๐ฆ๐บ Radix Ambassador
It will happen if it's popular enough and add value enough to do. Dan seems pretty keen too as this is right up his alley. Everything goes hand in hand, community wanted marketing, Dan wants some excitement, RDX Works wants positive sentiment, we all want more engagement. This ticks all the boxs
๐
7
โค
3
22:16
Haumikura
We are waiting for Dan to confirm the minimum configurations he needs so we can see how many people would participate in our Brazilian community.
โค
22:16
BassPhil
I think it is a great Idea!
โค
22:17
Blind5ight (Watch out for scammers)
Yeah but let's not paint Dan into a corner he doesn't necessarily want to commit to yet
22:21
XRDWhale.xrd |๐ฆ๐บ Radix Ambassador
I'll let Dan answer if he feels this is something he wants to do, or feels like it is valuable to do and how much budget can be spent getting this set up. What i'm trying to do is show Dan and RDX Works that this is a popular enough idea to get behind and you can see by the early engagement this is what people are wanting. Some grass roots rough and ready excitement, mixed with a little Cyrpto competition to our brothers over on other chains.
๐
8
๐ฅ
6
โค
2
22:21
So get involved, offer your services, spread the word, yell out that you support this
๐ฅ
โค
22:39
Marlon
I have a laptop, 10th generation core i5, 250g SSD, 8GB memory Ram, 500 megabytes of fiber internet. It will be an honor to participate in this
โค
22:41
Andri.xrd
Noptera team members would participate with personal and cloud machines. Count us in!
๐ฅ
10
โค
3
BearosSnap invited BearosSnap
22:51
AfterWave
Just reached out to a friend who's company holds several Guinness World Records. Will see if we can get some contacts over at Guinness.
๐ฅ
15
22:52
Mrx
I have a desktop, and maybe a laptop if it meets required specs to participate. Count me in!
โค
22:54
Haumikura
whow, this is another level
Deleted invited Deleted Account
22:55
Guilherme Vieira - Dont DM - Wont check
If they send a representant to check a possible record and i buy a couple millions because of it before any announcement... Is it insider trading? ๐
๐คฉ
22:57
Slysmik
my machine is shit but free to make a banging meme to bring attention
โค
3
๐ฅ
22:59
maybe a breaking bad theme one this time. "let him cook"
22:59
oh and available to help prettify any front end ๐
23:18
Blind5ight (Watch out for scammers)
Dan talked about this during a twitter spaces
Let me try and find the spaces and a timestamp
โค
23:20
Buddhini
the twitter space is sufficient m8, don't need to go through the entire thing to find the timestamp :-)
23:23
Blind5ight (Watch out for scammers)
23:26
Might be more in the vicinity of what you're asking, instead of straight on target
But yeah, I think you'll want to hear this discussion around availability
Paul Sangreal invited Paul Sangreal
01:30
Paul Sangreal
Yeah if we can contribute to the record breaking test what are the requirements ?
โค
Matheson invited Matheson
Jason | Cassievision | PegacornCo invited Jason | Cassievision | PegacornCo
03:11
AfterWave
While waiting for a reply, fished around on their website regarding application process. Current application time running 16 weeks, of course reduced with a fee.
There would appear to be no existing category , so it would be applying to create a new one. They only have "most expensive digital art" which was an nft; only thing coming up on DLT or Blockchain search.
โค
03:13
Question; if wanting to apply to Guinness for an official record, who should do application? RDXworks, Dan, the Community?
๐ฅ
03:19
TaserFace | ๏ธRadixStake
Community would be a good way to show that Radix is not just fast but decentralized as well
03:21
AfterWave
That's what I thought when asking the question.
โค
03:37
AfterWave
So we can start defining reasonable standard metrics to measure by, as this would be part of the application process. Need help on this.
For a faster review time and pay for GWR consultancy and Lisence fee to use their name, set up a Backeum for funding that fee (no idea how much that is yet as I have not dug deeply enough into application process, but started to).
๐
โmystic invited โmystic
Paracloud invited Paracloud
Don Dude invited Don Dude
07:43
Don Dude
This gets me excited again ๐๐ผ
๐ฅ
marco michelino invited marco michelino
Jim โ Tsengar invited Jim โ Tsengar
10:51
Andri.xrd
What could be the most viral subject for the community test? Replaying the whole ETH history in 'x' hours?
๐
10:56
โmystic
Is there any strategy to market this testing. The TON testing was pretty viral throught the crypto sphere
10:59
Ed | RadUp.io
More than happy to spin a machine up for a TPS record attempt. ๐ซก
11:01
Dan "Fuserleer" Hughes
Specs I test with are 4 core, 8gb, SATA ssd, 100Mbit connection
100Mbit connection is overkill though unless we're really pushing the tps per validator set. IIRC I did some profiling and ~30Mbit was enough for 1000tps. Upload matters though, 1:1 connections are best if possible
๐ฅ
17
11:03
Due to the variance of hardware we're gonna see I'd rather run more shard groups with less tps per group if we have enough participation
Those with beefy machines could run multiples instances on the same box so the resources are used as mush as possible
๐
5
11:03
Mik3ology
This could be a good benchmark for a Guinness book of records, as long as it also includes all tx failures.
๐
6
11:05
Dan "Fuserleer" Hughes
I'll make a tweet with some thoughts later and expected timelines we could aim for
๐
9
โค
6
11:05
ArtQ.xrd
Also more than happy to help out with this - I've a 500Mb FTTP line so it would just be a case of putting building the hardware to suit as required ๐
11:06
Dan "Fuserleer" Hughes
I have a hunk of work to wrap up and cassie needs a bit of housekeeping as shes mainly been used to support the paper recently and theres a bunch of bugs and things I need to fix in various places which I havent got to
11:07
depending how quick I can get through all that we're looking at mid-December best case, min-Jan worst case
๐
13
11:08
WIll take probably a few weeks to orchestrate all of this anyway, so was never going to be a quick experiment anyway
๐
9
โค
4
๐ฅ
2
11:10
XRDWhale.xrd |๐ฆ๐บ Radix Ambassador
I think everyone would be happy if we could get it by Jan, that's still reasonably quick.
Not only do you need to get Cassie up to scratch but a marketing campaign would be great too by the team and all community members involved etc
๐
10
๐ฅ
4
11:10
AfterWave
For Guinness I have started an application just to find out thier timelines . It not like you call them and order a world record and they deliver a plaque by amazon๐.
They do have consultancy that help market the record. As I find out more, will share here for sure. That is why I had suggested a backeum would be good for this cause too.
๐ฅ
15
11:14
XRDWhale.xrd |๐ฆ๐บ Radix Ambassador
Make sure everyone who's influential and all who need to know should be contacted and made aware to watch.
Also make sure that all technical specs are known and it would definitely pass anyone's critique for a passed TPS test, so it is properly recognised.
Get those academics to recognise it too as the world record!!!
๐
11
๐ฅ
3
11:27
AfterWave
We might want to pin the appropriate starting point message for those being invited to participate and start showing up.
๐ฅ
๐
1
11:29
Durante Ali
Should we invite Adam_XRD here?
11:31
XRDWhale.xrd |๐ฆ๐บ Radix Ambassador
Adam should know, the ambassadors channel is made aware of this which Adam is apart of
๐
alexmachado invited alexmachado
11:56
oz.xrd
I personally think he is being disingenuous about his motive. His primary motive is for building his internal knowledge or company research and tapping all the innovative chains and techies for his or his companyโs benefit.
11:57
Explore the Universe
Before that we need to setup a google form where everyone who wants to participate can put in their Twitter handle or email.
That way we catch people outside Radix who are interested to participate early on.
๐
11
โค
1
12:02
flightofthefox (proven.network)
Nothing wrong with having multiple reasons for doing a thing. Only a bit iffy if theyโre not ever released at all but I see no reason why thatโd be the case. Someone in his position can get meetings without any pretense ๐
12:05
Haumikura
Man, if we got placed in Guinness, my god...
12:15
Explore the Universe
My Occam's razor reason, he is simply overloaded and can't get stuff done.
12:15
We've all been there.
12:16
h0ll0wstick ๊ฎ Old account
Can we use backeum to give everyone participating a cassie nft as an incentive to help out with this project? Or use the cassie token somehow?
๐
8
โค
12:58
โMr. Peanutbutter ๐ซ๐ท
0.1 cassie not 1 ๐ donโt kill ze tokenomic
๐
13:07
SeF
The Cassie tokens were for the testers. No tokenomics should be relevant here
13:10
โMr. Peanutbutter ๐ซ๐ท
Of course. Still, donโt dump it, think of future testing and stuff
13:26
Emmoglu Leo Magal - I don't DM first
well, those that are going to run on cloud need some kind of incentive to cover costs
๐
14:49
Buddhini
it didn't quite touch it.
maybe i can ping
@danhughes :D
so the question again - if my transaction is pruned in the future (i just assume pruning will have to take place at some point, especially because you need to sync with new validators on next epoch rotation), how can i proof later that my transaction existed?
Usually you'd have some merkle tree on your blockchain and i can create and store a proof together with my transaction and in 20 years i can say "look here is my tx and my proof and there is this merkle root on epoch xxx", but since we aren't a blockchain, do we have somethign different?
For example i could imagine our shardgroups creating those merkle trees on epoch boundary and all shardgroups together then forming a bigger merkle tree which gets assigned to the epoch or smth?
15:01
flightofthefox (proven.network)
also interested in this question
๐
15:14
Inspector Crypto.xrd
Justin said his storage drive got corrupted but managed to retrieve data now, so is now getting on with editing.
๐
MP (I will never DM you first) invited MP (I will never DM you first)
17:36
AfterWave
So digging in to GWR application. We are applying for a title that, as far as results my searching their data base show, does not yet exist.
What are suggestions for a New World Record Title?
1st suggestion:
"Highest Transactions Per Second for a Distributed Ledger Technology"
Other suggestions?
17:40
Explore the Universe
Higherst transaction count per second on a DLT network.
๐
17:41
AfterWave
Yes "On a DLT" as opposed to "for".๐
17:43
Explore the Universe
Think Dan answered that question in the MultiversX Twitter space, or in the JSpeak space.
Even if one shard group disappeard you could get everything back from all the other shards that interacted with it or something along those lines.
17:43
Buddhini
yes i heard that, it doesn't answer the question tho :-)
๐
17:45
AfterWave
We may want to close the door at future attempts for BC by including "...including Blockchains" in the title. That way future applications will see it's covered under DLT.
17:53
woooptydoo
How about in the crypto sphere instead of just DLT ?
17:53
Put us all in the same bubble and show that we are better
17:55
Fastest TPS for a cryptocurrency
18:01
I'm just thinking about it from the average person point of view. People won't really know what a DLT network is, maybe not care either and just fly by it. "Fastest crypto" sounds much more intriguing and will probably attract more people. Once again, thinking about it from the average persons point of view. Depends what cards you want to play.
๐
18:05
SeF
Fastest (TPS) DLT consensus protocol
18:05
woooptydoo
How many other DLTs are there ?
18:06
What is the leading competitor
18:08
Inspector Crypto.xrd
I keep thinking where are GWR gonna find the expert to adjudicate our attempt? Even 'experts' in crypto don't think its possible lol
18:18
woooptydoo
Ok that's basically most of crypto, if I'm understanding correctly. So why not use that as a marketing tactic?
18:28
Libertant | RADIX.wiki
I wouldn't assume that transactions will need to be pruned. The state isn't fully replicated across validators, meaning there's no global merkel root and far less need for pruning than if every validator had to hold every tx. Shard groups occur where validators overlap on substates so new validators would presumably only need to sync on the substates they're covering.
18:37
Buddhini
Either shardgroups need to prune or their existence is unjustified in the first place :-)
We have sharding because we want to be able to handle more than our current max 50 TPS, so we got to assume such a load in the first place when lookign at our sharding solution.
ETH has to prune with very low throughput already, so i expect a shardgroup (with hihher throughput) has to prune at some point aswell
๐
19:01
Frosties
Not sure if it was on the J Speak chats with Dan or the EGLD one - but I recall Dan saying thereโs no need for total history of the state if we can ensure correctness of the state. So he was struggling / undecided on whether we should provide for merkle trees (which add a degree of complexity) or to prune.
19:04
Buddhini
I don't think it is "merkle trees or prune", rather a "merkle trees plus prune".
With that you have the best of both worlds - you don't store the entire history, but everyone who is willing to store some data is able to proove its inclusion later on
19:07
but i agree with the stance of assuming / ensuring correctness of the state instead of saving the entire history, it's a different topic tho :-)
Teixeira Pawtato invited Teixeira Pawtato
00:56
Kansuler
Just reach out and I'll help any way I can! ๐
Deleted invited Deleted Account
Francis โฮฃ invited Francis โฮฃ
Pierre Wolf invited Pierre Wolf
Alain Vettor invited Alain Vettor
Titi Ham invited Titi Ham
19:32
Mike (stringhandler)
uh oh someone mentioned an airdrop
21:12
Loccoco
Nop, many legit accounts interested in the Cassie stress test
21:12
TON claim made us furious
๐ฅ
๐
1
21:13
Haumikura
"Cassie guiness book TPS test"
๐ฅ
9
Deleted invited Deleted Account
Cometa Chase invited Cometa Chase
shuzy.xrd invited shuzy.xrd
07:12
shah
bro i am ready to participate
09:35
yosh moryno
If guiness can participate with cassie i believe it would be the best marketing. RDX works should look into it even a little chance the money not spent not paying CEX cab be use.. I think this is how Kaspa gathered so much hype even its just a simple coin without smart contract because of the participation of the miners.
๐
8
๐ฅ
3
๐
1
Matti โ invited Matti โ
20:43
Emmoglu Leo Magal - I don't DM first
20:44
second sign of troll account... but maybe he can just answer you
20:56
Buddhini
cerberus is designed for a network with multiple shardgroups, why would they have that in babylon? do you actually have a real question or do you just wanna be annoying?
๐
20:57
so you just wanna be annoying, i see
20:58
Emmoglu Leo Magal - I don't DM first
I spotted it from the first sentence
20:58
Taylor Kauรฃ
When Xian testnet?
20:58
Buddhini
i did. have a great weekend sir
21:00
Emmoglu Leo Magal - I don't DM first
yes, I am, but not in the mood of answering you, although I know the answer, and Im sure you know it too, but just want to be annoying. Have a good day, sir. And
@Chen_RL please ban the clown
21:03
Markus
That, ladies and gentlemen, is the favorite activity of some on a Friday (evening). You can all make up your own minds about the sadness of such people.
๐
21:04
flightofthefox (proven.network)
Go for a walk Jimmi ๐
05:53
Yessir
guiness book is actually a cool idea but I doubt they would accept as it would be viewed as hidden advertising
๐
12:12
Dan "Fuserleer" Hughes
๐ฅ
45
12:14
Im probably going to regret the long hours imma have to do to pull this off but screw it!
12:14
needs to be done and cassie is so ready
๐ฅ
28
โค
1
12:15
flightofthefox (proven.network)
pressure makes diamonds
12:15
Explore the Universe
Let us know what we can do to support.
12:15
Dan "Fuserleer" Hughes
everyone here please fill in the form with the hardware you have
๐
9
12:15
this will help me to configure the network on boot in such a way to make the most of the hardware we have
๐
4
12:16
Kansuler
Do anyone want to start a Cassandra 1M TPS campaign on Backeum? I think it could be a fun community event for everyone to support the node runners by handing in some XRD for a good cause, we can close the ability to mint the Trophy after the test has been run.
Just an idea from
@h0ll0wstick for engagement :)
12:17
Dan "Fuserleer" Hughes
Great idea! Whos good with artwork in here? ๐
12:22
Explore the Universe
Macbook Pro also possible?
12:22
J C
I see the upload requirement was lowered, but guess Iโm still out with my blazing 2 Mbit upload speed ๐
12:23
Dan "Fuserleer" Hughes
if it can run Java, ofc
IIRC some folks used macs before
12:23
Explore the Universe
And just checking, the Tweet and the form say TPS. Does this test also demonstrate smart contract calls per second or only simple TPS?
12:23
Dan "Fuserleer" Hughes
yeah 1M+ TPS even with a lot of shards needs decent upload capabilities
๐
12:24
it'll be a mix
so long as we perform 100,001 smart contract calls per second we beat TON and everyone else ๐
๐
9
12:24
Kansuler
Let's aim for 10x that then ๐
12:25
Dan "Fuserleer" Hughes
I need to dig into that TON test and determine exacly what SC calls they made
I would imagine they were the cheapest they could do and still get away with calling them "smart contracts"
12:26
Buddhini
i don't know if that's really worth your time
12:27
i don't even think anyone outside gives a crap about some performance of some research testnet
12:28
TON used a test net, plenty of folks cared
Cassie is a test net, plenty of folks will care
Our test will be much more fair like real world, more folks will caare
12:28
"so the time-to-finality of TON is 6 seconds." - Cassie should be a bit better than this with the latest improvements but I'm yet to test it very hard
๐ฅ
13
12:30
Buddhini
u sure anyone cared? to me it feels it's just our community desperate for attention / token pump. did anyone care about the Solana 1.25m tps with firedancer?
12:31
Dan "Fuserleer" Hughes
Thats kinda moot tbh
"Radix Tempo test was centralized, Radix can't do 1M TPS, Radix yada yada"
Once we've done this, point and STFU
๐ฅ
8
12:32
Will also stop all the internal doubt about Xian
๐
15
โค
1
12:32
Inspector Crypto.xrd
Firedancer test was all over my twitter feed for days
12:32
Explore the Universe
It's a win win anyway. Brings the community together and allows people from outside to also take part. There is a lot of interest in such an open test in the space.
12:32
Buddhini
better focus on the tech and not on pleasing degens or fudders, but just my 2 cents
12:32
Blind5ight (Watch out for scammers)
I admire your boldness
I'm both excited and anxious
Excited because of the potential, anxious because of the blow back into our faces potential
But we'll stand by you
โค
12:33
Dan "Fuserleer" Hughes
Its something I've wanted to do for a while anyway, so it's just killing many birds with one stone
๐
14
๐ฅ
2
12:34
I've done 50+k before with Cassie, we've done 1.4M before with Tempo, its not really a "risk"
Providing we have the hardware a 20x of my peaks is a walk tbh
๐ฅ
14
12:36
Blind5ight (Watch out for scammers)
Which streamers/youtubers do we know?
Would be awesome to have them join and livestream it
An example I can think of is the Brazilian professor that
@magal36 got into contact with and also typed in this channel in the past
12:37
Explore the Universe
This needs participation of many universities.
12:37
Everyone with contacts please activate them.
12:49
BassPhil
Hi Dan
I have a MacBook Air M1 Chip.
Is it possible to participate on the Test?
12:50
Where can I find a link to information about cross shards times
12:52
Dan "Fuserleer" Hughes
He doesn't even understand how our Cerberus cross shard works, so timings aint going to help
12:53
and this is AFTER a 1:1 for an hour of me explaining it
๐
12:53
Explore the Universe
It doesn't take minutes, it's not asynchronous. He is talking about MultiversX and pretend it's Radix.
12:54
Jelthebest
I think you are the best at explaining it xd, Iโm not expert enough to tell him how it works. But I do remember that it doesnโt take minutes.
12:54
Explore the Universe
Either strong confirmation bias or malicious behavior.
12:54
Dan "Fuserleer" Hughes
I tried, on a Twitter spaces, remember
๐
12:54
Explore the Universe
I heard it. Even I understood it at that moment.
12:55
Emmoglu Leo Magal - I don't DM first
in brazil, we have a few that would be interested, some small, some medium sized channels, and we could even tthink of going after the big ones (since our community has been loud on them before, there are aware of Radix already)
๐ฅ
6
12:55
Jelthebest
Haha iknow, itโs not really to convince him,but others that might be misinformed by his tweet
12:56
Explore the Universe
We need to bring universities on board for this test. Everyone with connections and contacts, please get them involved.
๐
6
R๐t๐rstar ๐ฆโ๏ธ invited R๐t๐rstar ๐ฆโ๏ธ
13:14
Markus
A text about Radix couldn't contain more misinformation
MultiversX folks are almost the PoS version of Kadena folks
"Atomicity is unimportant!"
*Spends time denying Radix it's atomicity and designing MvX's VM to mimic cross-shard atomicity*
๐
๐
ฑ๏ธS โ invited ๐
ฑ๏ธS โ
13:37
Paul Sangreal
Focusing on the tech for years is daily business. Making some waves keep the sentiment up is necessary. Do good and talk about it.
13:39
Haumikura
What was the most hardware you could get?
13:45
XRDWhale.xrd |๐ฆ๐บ Radix Ambassador
Exactly Dan and extremally happy you've run with the idea.
If there's anything you need from us Ambassadors just say the word and i'm sure we'll all be happy to chip in and make this a successful campaign.
Let's get this officially recognised and silence the naysayers, the FUD Brigade and stick it up 'em ๐ช
๐
7
13:47
Haumikura
Dan, I couldn't be happier to see you moving forward with this idea. You can be sure we will do our best to make it big!
Just tell us what you need.
Deleted invited Deleted Account
13:52
Deleted Account
Damn, my work device is more powerful than my personal device, but I'm not gonna use that for this test ๐
13:52
Lennart
What are you supposed to fill in here? All the specs that are also given a minimum requirement of in the description above?
13:55
Haumikura
and
@danhughes, you need to help us with how to run on windows/linux/mac, so we can put together easy tutorials for the community.
13:58
XRDWhale.xrd |๐ฆ๐บ Radix Ambassador
@Radstakes Is this something maybe you could help with?
13:59
Haumikura
And we need know if can run in VM, to run more than 1 node per machine.
CharlieCrypto invited CharlieCrypto
14:24
Faraz.xrd | Radstakes | XRD.Domains
Absolutely, I'll gladly help with any docs, guides and support to get as many of us running as possible ๐
โค
24
๐
2
14:26
XRDWhale.xrd |๐ฆ๐บ Radix Ambassador
You're a Superstar Faraz ๐
I love this community!
โค
9
14:29
Emmoglu Leo Magal - I don't DM first
I wish my DevOps and scripting skills were a bit more evolved, but we could provide scripts to spin up several cloud instances for those with $$ to pay for them
๐ฅ
14:31
avaunt | ShardSpace | Notix
๐ฅ
4
14:33
flightofthefox (proven.network)
Appreciate the work you do in the node runner community. Even from the outside, it shows
โค
9
14:36
Slammer
Honestly, people can try to discredit it all they like as long as they're talking about Radix. We're in the attention economy and the more of that the better - good or bad
๐
9
๐ฅ
1
14:37
Dan "Fuserleer" Hughes
this too ofc
14:37
Slammer
Is this something a non-tech person could participate in?
14:37
Dan "Fuserleer" Hughes
yups
๐ฅ
9
Nikola | SRWA invited Nikola | SRWA
14:38
Nikola | SRWA
1M+ TPS ๐ฅถ
14:45
Chris | CaviarNine.com
how many boxes do you need Dan? We can spool a few up
๐ฅ
6
14:48
Dan "Fuserleer" Hughes
the more we have the more TPS we can push
Imma do some baseline benchmarking this week on Cassie so I can gauge a minimum number at the lowest spec we'd need to hit 1M
๐ฅ
17
ollydigi invited ollydigi
14:53
Chris | CaviarNine.com
'more' is always the correct answer
๐ฅฐ
6
14:56
Haumikura
@danhughes , can you be a little more specific about the performance required in terms of processor?
Unfortunately "4 cores" is generic and can lead to error.
Let me explain: a processor used in servers (like a Xeon like you use in cloud tests) has lower per-core performance than a desktop processor like an i7. Often an inferior desktop processor, such as an i3 with 2 physical cores with 4 threads, outperforms a Xeon with 4 physical cores.
And as I know the target audience for this test, this can cause a lot of problems for us in the tests, because if you say that there are 4 powerful cores, people who have inferior notebooks that have processors with 2 physical cores and 4 threads will cause problems for the test, because for them their computer is 4 cores.
But if you give us more technical details, we will be able to translate this for people who don't understand, I'll give you an example:
If you tell me that the nodes you use have Xeon E5-1607 processors, then we will know that processors like an i7-7600u or even an i3-8145u, which are 2-core processors (with 4 threads), will be able to participate in the test.
It's just technical information so that those of us who organize communities know how to tell people whether or not they will be able to participate, we don't want anyone with computers that could get in the way.
๐
14:57
Brendan
Hey Dan, I am curious as to what specs you would like to see if one were to host two nodes on the same computer
14:58
AfterWave
Worth an effort, and as for marketing it is part of their business model where they have a paid for consultancy to assist in marketing campaign around a record.
๐
14:58
Emmoglu Leo Magal - I don't DM first
Or if he states he just means 4 threads than the inferior cores would suffice
14:59
When asking for technical information from people who don't understand, we need to be as specific as possible.
Heisenberg invited Heisenberg
15:21
r3animation โ | Never DMs first
My steam decks, rog allies and legion go are on standby!
๐ฅ
15:46
Blind5ight (Watch out for scammers)
Linear scaling is pretty cool huh ๐ค
16:05
Explore the Universe
Nicolai | Sentura 11.11.2023 13:35:43
how to be shure that only radix supporters are taking part and no one who likes to see bad performance ?
16:06
Explore the Universe
My guess is we simply outnumber them in combination with monitoring bad node behavior?
16:09
Aleksander
Any given shard group should be able to work at the speed 66 % of the nodes in the shard greoup can support I guess?
16:12
๐
ฑ๏ธS โ
@danhughes how many have already registered so far?
Andrรฉs Leรณn invited Andrรฉs Leรณn
17:08
Emmoglu Leo Magal - I don't DM first
Actually that would be actually good for the test, if Dan is confident enough on it.
17:08
[PSB] PSILOBLOX | Boris
Looking forward to see what specs are needed for this test ๐
๐ฅ
17:11
Explore the Universe
You can see them when you click the "sign up form" on the Tweet.
17:12
[PSB] PSILOBLOX | Boris
Yea i just noticed that hahahah
17:12
okay specs are not bad at all.
17:17
Have to beat TON. I mean... it's TON ๐
17:18
Haumikura
It's too early to say, I haven't even signed up yet, in the Brazilian community we haven't announced it yet...
17:18
Dan "Fuserleer" Hughes
Well so long as friendlies are 2f+1 let me try. That's a real test of bft then
17:18
Perhaps we boost the vote power of the friendlies ๐
๐คฉ
17:19
Just to be sure James doesn't spend his "hard earned" millions firing up lots of nodes
โค
17:20
Dan "Fuserleer" Hughes
He's sick, be nice โค๏ธ
๐
17:20
Lennart
@danhughes are we going to do a specific set of transactions like the history of btc? Or how do we decide what transactions are going to be submitted to the network
17:21
Dan "Fuserleer" Hughes
No imma probably create a bunch of accounts with lots of flex in. Then spam them around calling smart contracts, token transfers etc efc
๐ฅ
17:21
I want the test to run for a few hours at 1m tps and nothing in this space has enough transaction history to replicate for that long ๐
๐ฅ
15
17:22
Durante Ali
I will wait for Black Friday deals to upgrade up to specs to join you guys.
๐
17:22
Dan "Fuserleer" Hughes
Remember bitcoin only took like 15 mins on tempo
17:23
Haumikura
What in the world have 3.6bi in operations? Maybe all time sp500 with HFT...
17:24
Emmoglu Leo Magal - I don't DM first
Just create a pattern of smart contract calls that resembles reality by a far approximation, and rinse and repeat them ad infinitum. Itt doesn't really matter, since they are not real users, earning or loosing money lol
๐
17:26
Haumikura
What if you do all the transactions on all the networks that have already crashed due to overloads?
๐
17:27
Emmoglu Leo Magal - I don't DM first
anyway it wouldn't be enough. Needs to make them up
17:27
Haumikura
Marketing ๐๐๐
17:28
Emmoglu Leo Magal - I don't DM first
you could say that the number of tx outnumbers all of those ๐
๐
17:29
Haumikura
All of these x10 ๐
17:29
TyPo.XRD
When a node tries to join the network isnโt there a spec check of the system trying to join and if below requirements gets rejected?
17:32
Dan "Fuserleer" Hughes
Not really no and I think that would be bad form anyway
If the spec is too bad it'll just never actually sync and always be catching up
๐
Deleted invited Deleted Account
17:32
Deleted Account
Testing with "friendlies" only is equivalent to a permissioned system. Change my mind.
17:34
[PSB] PSILOBLOX | Boris
thanks, i'm out of it a bit for sure, and i did not take offence. It's not like he went and said "Look at sign up form, you dumb ****" ๐
๐คฉ
17:35
flightofthefox (proven.network)
TON did purely synthetic load. I think actually generated from within by their "bomb" SCs. Seems like a bit of a shortcut to bypass all ingress overhead but W/E
17:42
Dan "Fuserleer" Hughes
Yeah the "bomb" is not the bomb
โค
4
17:42
We'll at least match it, but I want more complex stuff
โค
6
17:45
flightofthefox (proven.network)
how many nodes did you run for max tilt in the last paper?
17:49
Slammer
Rerun Solana for the flex?
๐ฅ
17:50
flightofthefox (proven.network)
i'm gonna need more storage ๐
17:51
๐
ฑ๏ธS โ
@danhughes If we have enough participants, would 2 million TPS or more be possible?
17:51
Slammer
But we're sharded 'member?
17:54
TyPo.XRD
X is now plastered with 1M+ anything less will now be scoffed at no matter how much it still beats TON. The vast majority of people just see the number and donโt care about the specifics. Justin Bon will care mind and you never know he might be forced to actually mention Radix.
17:57
Chen (โ, โพ๏ธโ๏ธ)
We can do a trillion TPS.
Just need enough nodes and shard groups.
Simple smart contract txs with many smart contracts across all shard groups and get as close to 2 shards per tx.
Pushing TPS to the max
๐ฅ
17:58
Lennart
For now it's just important to get as many people running a node as possible for the test. Maybe it's possible to get the official radix account to make a mention of it?
๐
18:01
TyPo.XRD
Maybe pin a message at the top with spec requirementโs or a form to fill in, I definitely would like to be part this test.
18:07
Haumikura
How many did this using the community's own computers around the world? This is true marketing
18:09
John โข
Let's see how many fill the form, i think we should push as high as possible if we have the resources
๐ฅ
18:13
TyPo.XRD
Apologies, what I meant is radix fans have been posting that Radix is going to do a community test to smash TON record, by setting its own new world record with 1M+ TPS. I shouldnโt have mentioned it since this group isnโt a chitchat group.
18:15
Dan "Fuserleer" Hughes
It's fine, excitement is allowed ๐
๐
4
18:42
Ahmed Sedeek
Hello
@danhughes Do you require this to be on the cloud or on our personal computers?
18:44
[PSB] PSILOBLOX | Boris
I think unless you are doing this on some crappy internet line, it will be fine. But we need good latency i think for this ๐ my 2 cents.
๐
18:45
i recommend NVMe SSDs, too.
18:50
rotane.xrd
Will we have an explorer to verifiy the 1m TPS? I am thinking of something like this: every transaction should have a text message stating a unique ID (kind of unsername or so) and an increment counter and after the test I can query the network asking โhow many entries did I make into the network?โ in a verifiable way
โค
Deleted invited Deleted Account
18:56
I R
Yes, and the explorer needs to have an option to show data in the simplest of ways for regular people and ofc a view for the tech folks
18:59
cactus.xrd๐ต
Saw the tweet, I'm in for the cassie test in Jan!
๐
6
19:29
Charlie
Always 2.56 X more than radix right now โฆ. So youโre telling that Radix is centralized ?
BTW TON did this with a marketing plan behind and itโs worked, we see the result everywhere on X, medias, โฆ If radix doesn't have a marketing plan to get the word out and promote it no one will care about the result.
A bit like Babylon ๐
๐
19:35
marco michelino
Anyone can join
19:36
Deleted Account
No one will seriously try to break it unless there is money to be made
19:36
That is why such testing fails imo
19:42
Emmoglu Leo Magal - I don't DM first
it didn't fail for TON marketing
19:45
Charlie
because the know how to market and promote it
19:45
Emmoglu Leo Magal - I don't DM first
but now RDX won't bee promoting by themselves. We need to make our noise as a community too
๐ฅ
19:47
Charlie
Yes community + Radix team need to do noise
๐
19:50
Deleted Account
sure, its a nice project
Jamoleite.xrd invited Jamoleite.xrd
19:54
Jamoleite.xrd
Hi everyone ๐โโ๏ธglad to be here
20:22
Durante Ali
Please, pay attention to this comment. We have had cassi tests before and we may extrapolate. No need to step on same rake over again. Marketing suport is imperative.
20:35
J100
Yes, the test is important. The technology has to prove itself and these tests are important not only for marketing, but also for the improvement of the technology, as it allows the developer, in this case Dan, to make observations, and perhaps to bring more confidence to investors until Xian. This is closer to the real world (computers, networks, with different capabilities) in which Cerberus technology will work, so I don't see it as a waste of time.
๐
8
20:55
h0ll0wstick ๊ฎ Old account
It s perfect, let s roll out the plan clear for everyone and make the DLTs world shake, being part of a world record that could set the standard in 10 to 20 years is something to tell the grand kids about
20:57
Java runs on linux/windows/mac
21:01
@danhughes is cassie still hybrid on POW/POS for the test?
21:13
Emmoglu Leo Magal - I don't DM first
If running mon AWS... what would be the desired instance type? c5.xlarge?
m7a.xlarge? i4g.xlarge?
21:14
Loccoco
If Cloud is allowed I can manage to package something with Terraform and Ansible on aws, azure, gcp if networking and install requirements are given 1&2 week earlier
โค
21:15
Emmoglu Leo Magal - I don't DM first
yup, cloud is allowed, and this is a gret idea!!
21:32
Ftt Fff
someone here must have mega access to cloud compute resources. I know myself that i could possibly spin up 5 instances for this.
๐
21:35
Haumikura
@danhughes, the specifications you gave some account do not fit. If you need 30mbps, the maximum it could occupy on disk in a 3-hour test would be around 40gb.
Sorry to be the annoying hardware guy, I know how people who don't understand this work, and Internet (which is a very low requirement, great) and SSD disk space are the hardest to get.
๐
21:36
Loccoco
Probably a crude github action pipeline to clone with api credentials in clear but mangeable for everyone
21:36
I'll check if there are any gift cards for clouders also, so we can lower costs
๐
21:37
We'll need some academics though
21:38
Cerberus reviewers would be good for starters I think
๐
5
22:05
Spangular Directions
I'm not sure that's true, if Alice sends a tx to a new account Bob, and the validator is storing both the tx and the state, then the stored data is growing by tx + change to state with every transaction, but the only data actually sent over the internet is the tx
๐
22:13
Nikola | SRWA
Just realized the whole BTC ledger has less than 1M transactions in total :)
And we're going to print that in second.
๐
22:17
Dan "Fuserleer" Hughes
I'm assuming we get enough for 1000 shard groups.
30mbps is enough to support approx 2000 tps per shard group with 8 outgoing connection per node connections
That's just wire io, there's other overheads in local storage such as proposals, indexes etc so disc IO becomes perhaps 50mbps
200GB is enough for days and days usage at 2000 tps per group. Which would be nice if we can do it.
Going back to my assumption, 1000 shard groups capable of 2000 tps (mixed) is 2m per second peak
Hence my spec
๐ฅ
13
๐
7
22:21
Octavio
Is around 7-9 Mbit upload enough ?
22:21
Loccoco
Does having bigger size for instance help with throughput ?
22:22
I may have capability to push for bigger instances if my company likes the project
22:22
Dan "Fuserleer" Hughes
Not really. We need at least 1000 tps per shard group and that requires at least 15mbps upload for even the simplest transactions
22:22
Loccoco
Allocating a 1kโฌ for a day might give us 10 nice instances
22:23
Dan "Fuserleer" Hughes
You can only go as fast as the slowest majority. So huge instances won't help.
What you can do is spawn multiple smaller spec instances in VMS and run multiple validators on the same machine if you have the resources available
๐
22:24
So if you had 32 cores and 64 GB machine you could do 4-8 validators on that machine.
You need a beefy SSD and internet too though
22:24
Enough to handle 8x the io
22:25
Emmoglu Leo Magal - I don't DM first
22:30
Dan "Fuserleer" Hughes
Ooh I forget the spec of these weirdly notarised aws instances. I'll have a look
๐
Space Sizzler invited Space Sizzler
00:01
Haumikura
and how long you imagine running the test. Because if we are going to do it with the maximum number of participants, it cannot take more than a few hours.
00:02
Blind5ight (Watch out for scammers)
Dan "Fuserleer" Hughes 11.11.2023 17:21:56
I want the test to run for a few hours at 1m tps and nothing in this space has enough transaction history to replicate for that long ๐
00:04
Haumikura
So, why 200gb? If 200gb is enough for days...
00:04
Blind5ight (Watch out for scammers)
What's your calculation?
00:05
Dan say 200gb is enough for days
00:06
Explore the Universe
Glad we understand each other โค๏ธ๏ธ๏ธ๏ธ๏ธ๏ธ๏ธ was meant in a purly informative way.
00:06
Blind5ight (Watch out for scammers)
Hmmm so he's playing with the idea of maybe running it beyond hours into days territory?
So wants a buffer for the machines?
00:07
Haumikura
If Dan says 50 GB is enough, then we will get a lot of participants
00:07
Tommy
Even 1tb is literally nothing ๐คท๐ผโโ
๐
00:08
Blind5ight (Watch out for scammers)
Let's see what he says tomorrow or later
โค
00:08
Haumikura
not for the test. Most notebooks here are sold with 128GB and 256GB SSDs
00:09
Tommy
Sorry, this group was popping up right now. I flew over the last posts -- short summary: we can help with setting up a node to get the new record?
00:16
Ok, flew over some more messages ๐
Please Pin HW requirements, the installation Skript and WEN ๐!
โค
00:20
TaserFace | ๏ธRadixStake
00:22
TaserFace | ๏ธRadixStake
Katan Logical Moon ๐ 12.11.2023 00:22:22
๐
00:24
rotane.xrd
@danhughes hopefully Adam is aware? I see ads for TON 100kTPS everywhere. Can we change that?
00:25
The nice side effect of this community test is that it gives the community purpose.
๐
00:26
If we make it into Guiness book of WR every participant can claim his share of the laurels
00:26
This will become a meme
๐
00:26
james
Guiness book?
00:27
rotane.xrd
someone raised the idea to make an entry if we succeed
00:27
you can basically pay for an entry
00:27
nice way of marketing and very efficient
00:28
james
Is this officially endorsed or...?
00:28
rotane.xrd
Lets chase Adam
๐ฅ
00:29
[PSB] PSILOBLOX | Boris
of course. Mostly, itโs only Germans I have a problem understanding ๐ I had gotten into some arguments because every time I see a reply I think Iโm being trolled. Then after I confronted some, we realized that I need to learn German ๐ best thing to do is to confront the person when you feel you are being trolled, solves all misunderstanding ๐
โค
00:30
btw I donโt mean to offend any Germans here Iโm just offering up my personal experience ๐
00:31
Explore the Universe
This is hilarious. I'm German btw ๐๏ธ๏ธ๏ธ๏ธ๏ธ๏ธ
We tend to say what we mean, no hidden message.
00:31
Tommy
Why so low requirements?
I don't want to complain, I just want to understand. Would an NVME be more efficient or would more cores be more efficient?
00:32
Markus
So that as many people as possible can take part, I assume
00:32
Ftt Fff
because then they will say only powerful computers can be validators and that not everyone can participate
00:32
[PSB] PSILOBLOX | Boris
thatโs funny ๐ okay Iโm going to be honest, I did question some of your replies on Twitter I just havenโt confronted you yet ๐
00:32
XRDWhale.xrd |๐ฆ๐บ Radix Ambassador
I was wondering where i get that from lol
00:34
Tommy
Hmm... Ok, get it. As long as we see 5m TPS, I'm happy. ๐
00:36
Under 5m is fud. Achievable @Dan? If so, how much power could we possibly need?
Let me get this straight. I need to see radix on the f*cking top with this live test. No laboratory under perfect condition "fuel milage test". ๐
00:39
Emmoglu Leo Magal - I don't DM first
Nmves are recommend. More cores shouldn't make a difference. If you have double or more of cores and RAM, run two or more instances
00:39
[PSB] PSILOBLOX | Boris
When did we go from 1M TPS to 5m ๐
00:39
flightofthefox (proven.network)
happy typo?
00:40
Radsoy
Will anyone else from the RDX works team be involved during this process? Will they be part of the code โcleaningโ before the test happens?
00:40
Emmoglu Leo Magal - I don't DM first
00:42
Tommy
Ok, so we need fastest access time. ๐
๐
00:43
rotane.xrd
One of the pitfalls of POW blockchains is that the race for more hashpower raises the bar for participations and this is bad for decentralisation. The lower the specs the more decentralised the network can get. โDefi for everyoneโ is a good claim.
00:44
[PSB] PSILOBLOX | Boris
Yes, we should try to keep node requirements low. Otherwise, we start excluding people from participating.
โค
00:44
flightofthefox (proven.network)
did anyone answer
@Buddhinii's question about storage compression?
00:45
[PSB] PSILOBLOX | Boris
where ?
00:46
flightofthefox (proven.network)
this
00:49
[PSB] PSILOBLOX | Boris
oh, this is a good one. I'm going to let
@danhughes answer that as my understanding of the design is limited.
00:52
Tommy
Yeah, I hear the "everyone will be able to run a node on your router"-thing from 2021 still in my ears. ๐ This is planned for Xian afaik.
I see huge marketing potential in this test, that's why I'm pretty excited
00:54
flightofthefox (proven.network)
i think storage onsiderations might not agree
00:55
[PSB] PSILOBLOX | Boris
Depends on the router, my current router can run the current validator node if i wanted ๐ I just have a powerful one so i can do IPS/IDS on all my traffic ๐
00:58
flightofthefox (proven.network)
what kinda chonky router do you have?
00:59
[PSB] PSILOBLOX | Boris
00:59
running pfsense plus
00:59
and suricata for ids/ips
01:01
flightofthefox (proven.network)
what are pfsense and suricata?
01:02
[PSB] PSILOBLOX | Boris
pfsense is router software. Suricata is an addon to inspect traffic. it actually saved me one day, i forgot to update my unifi appliance and it had that nasty java logging bug
01:03
Some hackers were trying to exploit it ๐
01:04
flightofthefox (proven.network)
i see, not in my realm
01:04
[PSB] PSILOBLOX | Boris
Rafaล Moล invited Rafaล Moล
01:45
TaserFace | ๏ธRadixStake
01:50
[PSB] PSILOBLOX | Boris
8MB RAM ๐
01:56
TaserFace | ๏ธRadixStake
thanks good catch
๐
Lucas Luize invited Lucas Luize
Nate Incognateo invited Nate Incognateo
Christiaan wowo.xrd invited Christiaan wowo.xrd
02:27
Emmoglu Leo Magal - I don't DM first
02:35
octo.xrd
Does the 8 gbs of ram mean 8 gbs of ram free at all times or just a PC running (for instance) Windows with 8 gbs of RAM in total.
02:38
Octavio
Just donโt open a single firefox/chrome tab
02:40
XRDWhale.xrd |๐ฆ๐บ Radix Ambassador
@r3animation @Chen_RL @danhughesSome advice to
pin the information for specs and signup to become a validator for the world record tps test. Maybe create a special post with all information needed and a FAQ section
I've see a few comments around this and will make it easier for people to find rather than scrolling or searching messages
๐
04:14
avaunt | ShardSpace | Notix
8GB means a PC that has 8GB of RAM to begin with
๐
Deleted invited Deleted Account
Cerkoryn invited Cerkoryn
05:45
Cerkoryn
Hello World!
05:45
Just signed up for the 1M TPS test. I'm about done building a dedicated NAS, so I'll have tons of extra compute laying around.
๐ฅ
5
๐
1
RADKam | HotOnRadix.xrd invited RADKam | HotOnRadix.xrd
xLukex.xrd invited xLukex.xrd
07:27
xLukex.xrd
signed up , i can probly run 3 instances on my laptop ๐
๐ฅ
CryptoGravi invited CryptoGravi
09:23
Deleted Account
I heard we're going for 15M tps. Are the rumors true?
๐ฅ
09:27
Arabald
How is it with SMT? E.g I got a cpu with 6cores / 12 threads. But very fast internet, nvme,.. can i run 1 or 3 instances?
09:45
h0ll0wstick ๊ฎ Old account
09:50
Dan "Fuserleer" Hughes
Oh wow got busy in here. Just grabbing breakfast then I'm here for the day ๐
โค
6
๐ฅ
2
09:56
Speakers
Great! rumours are we're shooting for 1 billion teepees
๐
09:56
Lennart
Yes we are going for 150M tps
09:57
SeF
The question of hardware requirements has already been asked several times.
Is it 4 physical cores = 8 threads (8 VM cores) or
4 VM cores (virtualized server) = 2 physical cores (4 threads)
So
4 CPUs (8 threads) or 2 CPUs (4 threads)
09:59
BassPhil
Is a MacAir Laptop with M1 Chip 8GB enough?
10:01
Deleted Account
1M TPS minimum
10:02
Carl ๐ฑ โพ๏ธ โ๏ธ ๐ ๐ฉ
Can we get the 1M TPS test, minimum computer Specs pinned please?
10:02
Deleted Account
#1 rank minimum
10:09
SeF
Dan "Fuserleer" Hughes 11.11.2023 22:17:42
I'm assuming we get enough for 1000 shard groups.
30mbps is enough to support approx 2000 tps per shard group with 8 outgoing connection per node connections
That's just wire io, there's other overheads in local storage such as proposals, indexes etc so disc IO becomes perhaps 50mbps
200GB is enough for days and days usage at 2000 tps per group. Which would be nice if we can do it.
Going back to my assumption, 1000 shard groups capable of 2000 tps (mixed) is 2m per second peak
Hence my spec
๐
10:11
SeF
The requirements to operate a Cassandra validator are low, the minimum specification we are looking to use for this test is as follows:
4 core CPU
8 GB RAM
200GB SSD (SATA is fine)ย
30Mbit download / 10Mbit upload
Those in possession of more powerful hardware are able to run multiple instances of Cassandra during the test should they wish.ย
10:13
Dan "Fuserleer" Hughes
Performance of each individual core isn't particularly important, it's more to do with threading and context switching efficiently
๐
10:36
๐
ฑ๏ธS โ
2m TPS ๐?
10:56
h0ll0wstick ๊ฎ Old account
Your upload mbps will be important and allocating 8gb ram to java will make your laptop dedicated to cassie
10:59
Most recent machines will do, 8gb of ram and 4 core cpu is common nowadays, we need good bandwidth nice upload, if you use home network make sure you don t have everyone else on your network streaming videos or downloading torrents
๐
11:01
Just make some popcorn and say to your families/friends tonight we are breaking a world record watch this ๐
๐
4
๐ฅ
2
11:21
JM_Borg
Likewise, I'm using an M1 Chip with 8GB RAM.
I'll be testing the network this morning. The network requirements seem easily achievable.
Does anyone know the expected duration of the test?
โค
11:27
Spangular Directions
I think the days thing could be cool, but personally I'm not willing to dedicate my machine to days of use, and i suspect that's going to be the same for everyone using a home machine.
A really good compromise could be 1mil tps for, say 1 hour, plus 150k tps (or whatever, some other big number) sustained for a few days
๐
4
11:44
JM_Borg
@danhughes can we participate in someway if we only have available 150 GB
๐
11:45
Yeah, thatโs my case, 256 GB SSD
12:36
Paul Sangreal
200GB free space on the SDD that is ?
12:44
Mike (stringhandler)
1 mil TPS generates a lot of data
13:11
Explore the Universe
Does it even make sense? The crypto space seems decades from requiring such throughput sustained over a long period. Demonstrating that the network can handle such peaks even over 1h seems already sufficient.
13:31
Dan "Fuserleer" Hughes
Yeah that will be plenty, especially if we only do a short run of a few hours
๐ฅ
13:33
A sprint is one thing, but I would like the network to be up longer for "interrogation"
Fud = did radix really do this?
Us = the network is still up at idle. Here is an explorer, feel free to count the transactions one by one
๐
13:33
SeF
There is nothing to be said against keeping the network running for longer. Gradually, more and more nodes will withdraw anyway. For the record, 1 hour will be a very good value. 24 hours would also be very nice but not a must. You can also test a few bad actor things later. The peak TPS depends on the number of shardgroups... So open end. The question is whether you can fuel the high load over a longer period of time. With BTC History the test would be over after a few seconds.
๐
13:37
Dan "Fuserleer" Hughes
Some machines (the more powerful ones) will take the role of "clients" who will flood the network with various types of transactions
13:40
Another reason Im asking for specs so I can determine how many we have which can do that and how many I might have to spin up to consume the deficit (if any)
I think it's ok for me/radix to have about 20% position in the network. That's well below the 33% threshold you could shout "controlled" and in any real public network the probability that a few actors had 20% of influence is realistic.
๐
10
13:43
๐
ฑ๏ธS โ
Are XRD needed for this? When Yes, how much?
13:44
Win_Q
Is there a chance that someone will want to sabotage the test by pretending to be a node validator? I don't understand the technicalities, but I'm worried that there are competitors who want to take us down.
๐
13:46
marco michelino
Testnet tokens, not real XRDs
13:46
Radsoy
Not sure if you saw my message Dan but this is something some people including myself are wondering about
13:47
Dan "Fuserleer" Hughes
Of course there will be, but so long as they don't control 33%+ of the network they can't really do shit other than cause a little unresponsiveness
๐
13:47
We can combat that by giving some trusted community actors a bit more vote power than those who simply join anonymous
๐
7
13:48
You'd have a long tail looking distribution in a live network anyway so the result is a network that looks more "real"
13:51
I'll probably grab one of the Dev ops guys for the duration of the test to help us spin up and run the test etc
Iirc shambu did a lot of that in the tempo test so he's got some experience with this stuff so I'll probably grab him if possible for the day.
As for everything else I'd like to leave our developers out of it, they have enough to do.
Coding wise I'm sure me + community can cover everything , especially peripheral stuff like explorers and wallets etc
Can use the stuff I already built but polish it a little.
Most of the work is in Cassie core and there's only really me who knows that codebase and what needs to be done. So some long night incoming for that ๐
๐
7
13:52
Radsoy
Thanks for the reply!
14:05
rotane.xrd
Stupid question. How do you gauge the risk that competitors reverse engineer the Cassie code and run away with it?
14:10
Radsoy
I might be risk averse but having additional help from the radix team would be great as this will have a LOTS of eyes on it and it needs to be 10000% smooth
๐
14:13
Dan "Fuserleer" Hughes
Because there will be some components removed that are critical for a main net but we don't need for a short test.
So if they DO take the code and launch a network with it, it will be fragile against some forms of coordinated attacks
14:13
Also I won't be mentioning what those components are ofc ๐
๐
โค
14:14
Dan "Fuserleer" Hughes
To know what those components are you'd have to go very deep and rediscover/ reinvent things that we haven't published yet
14:15
That'd take a long time to do, or you risk running a fragile network
14:17
But from a general theory pov anyone looking at Cassie code would be able to verify that all the pieces are there and valid to support the claims
14:19
rotane.xrd
I think scammers are ok with running a fragile as long as they can claim to do the same TPS as the Cassie test had achieved. They can just leverage Cassieโs reputation and in case they get hacked/attacked the damage would backfire on you and Radix. Are you using code obfuscation?
Deleted invited Deleted Account
14:22
Dan "Fuserleer" Hughes
Not sure how someone taking our open source code and using it for scams backfires on us.
Plenty of scams cloned other codebase and those projects are still around and doing fine
14:23
rotane.xrd
Makes sense
15:46
Inspector Crypto.xrd
Dan setting booby traps heh
17:25
projectShift
damn, too many msg, TL;DR
But already filled the form.
Anything else for the moment?
17:26
Yessir
@danhughes great to see you posting on X more, this brings back confidence and hype (TVL + TPS tests etc)
๐ฅ
๐
17:47
Dan "Fuserleer" Hughes
Just making sure we're first to market ๐
๐
4
๐ฅฐ
1
18:57
Blind5ight (Watch out for scammers)
It seems the winter season does something to Dan: Urge to explore unexplored frontiers from the cozy Coding Cave
Start of build Flexathon/Cassandra was also in November
18:59
Dan "Fuserleer" Hughes
Heh true!
19:00
Don't commit to that project when flexathon turned serious for the reasons rotane raised earlier
19:00
We are first to market with a cross shard atomic network
Not some skank leeching my GitHub lol
David Moreno invited David Moreno
19:03
Markus
Wouldn't the better description be "cross shard synchronous (atomic) network" before one or the other asynchronous maxi claim otherwise?
๐
19:04
Dan "Fuserleer" Hughes
Noted ๐
19:06
SeF
In any case, you have to include synchronous. This is often left out and the asynchronous calls... we can also
19:08
Cofind
We should also add quantum where possible, it adds to seriousness or makes avenger like project
๐คฉ
21:40
Wilson | Radical Staking
For the 1M TPS test, do we need to make any configuration changes or installations that could potentially conflict with an existing Babylon node v1.0.4 installation? Or is it better to use a fresh Ubuntu 22.04 machine?
22:14
Dan "Fuserleer" Hughes
Shouldn't but probably best to use a new machine or at least a separate VM to be sure
๐
6
01:58
TaserFace | ๏ธRadixStake
๐
โค
08:30
Mike (stringhandler)
there's been so much chatter, so maybe I missed it, but what data is going to be used to test the 1mil TPS
10:08
Don Dude
What are the marketing plan around this?
Do we make sure that every Crypto news channels, and major influencers is aware of this stesstest of the Network is being established?
11:31
Dan "Fuserleer" Hughes
Part of the design spec for vamos is efficient reads, which means Cassandra networks don't need things like gateway nodes etc
Flexathon twitter was a nice example of processing high load / data but still able to query (streaming video etc) big requests.
Whether it makes sense for Xian to obscelete gateway nodes depends on a lot of things but for Cassandra I wanted to be sure it could at least be done and done well
11:32
Nothing that complex
Async and parallel execution, with total order agreement on overlapping transaction dependency
Works just fine
11:32
A mix of simple transfers, smart contract calls, and heavy payloads (something no one else has done)
11:38
Mike (stringhandler)
how many transactions in total?
11:51
Dan "Fuserleer" Hughes
Heh I dunno. Depends how long we run for.
Say an hour, that'd be something ~3.6B
๐คฏ
7
11:58
Mrx
Let's say I could put 2 machines to the test, do I need to have at least 2x the minimal upload rate (= 20 Mbit) ?
12:00
Emmoglu Leo Magal - I don't DM first
yes, otherwise there's a risk you wouldn't be able to keep them both in sync
12:01
Mrx
ok, what I thought. So I can't, I'm already at 9 something in practice (10 theoretically)
13:33
Win_Q
I think this is worth pinning on your X profile ser. So that it doesn't get drowned out by other posts and anyone who visits your profile immediately sees this. Imho
๐
14:11
Dan "Fuserleer" Hughes
Ahh yes I meant to do that
16:10
Space Sizzler
Dan, donโt want to draw the wrong assumption with your comment. Did you mean that some of the Radix team donโt believe in Xian? Or did you mean something else?
16:14
Inspector Crypto.xrd
Nah think he means within the community
๐
16:27
Emmoglu Leo Magal - I don't DM first
he means within the community
๐ฅ
16:51
projectShift
Its nice we are seen as "internal" ... you don't get that everywhere ๐
Doubts are good, fud is not
โค
18:10
Dan "Fuserleer" Hughes
Yeah community sorry
๐
6
๐ฅฐ
19:11
olit
If the test is in the first half of January I'd like to participate as my PC just meets the requirements. Will need a guide though ๐
๐
12:17
Jamoleite.xrd
is it possible that actually we can run more ... if we have 2TB 64gb m2 max, 12 cores( 8 perf. 4 efficy ?
12:19
flightofthefox (proven.network)
Can we just work everything backwards from whatever we need to do 6.9m teepees?
๐คฉ
12:21
Emmoglu Leo Magal - I don't DM first
Yes
๐
12:27
h0ll0wstick ๊ฎ Old account
You ll go at the speed of slowest majority in the shardgroup, yes you can run mulitple instances but what matters most with bigger config like yours is your bandwidth availability
๐
12:31
XRDWhale.xrd |๐ฆ๐บ Radix Ambassador
Specifically upload. I think it's 10Mbits per instance, but don't quote me on that
๐
12:32
Jamoleite.xrd
mine was 500mbps downlaod 25 upload
12:35
XRDWhale.xrd |๐ฆ๐บ Radix Ambassador
4 core CPU
8 GB RAM
200GB SSD (SATA is fine)
30Mbit download / 10Mbit upload
12:35
Those are the minimum specs
12:35
Emmoglu Leo Magal - I don't DM first
Iirc it's minimum 30 download
12:36
Jamoleite.xrd
okay so far my on bright side ๐
12:37
Emmoglu Leo Magal - I don't DM first
Multiply that for the number of instances ๐
๐ฅ
12:43
Jamoleite.xrd
is there any documentation that would help to gain a little bit more knowladge before we do this ?
12:45
Emmoglu Leo Magal - I don't DM first
๐
12:46
But that's an old one, let's wait to see if it needs updating
๐
Deleted invited Deleted Account
ClickEnter invited ClickEnter
Luiz Candido invited Luiz Candido
Pyro - WILL NEVER DM YOU invited Pyro - WILL NEVER DM YOU
18:55
Pyro - WILL NEVER DM YOU
8 core 8 thread 4.5ghz 9770k, 32gib, nvme ssds, 500mbps down 50mbps up
At your service
๐ฅ
18:55
Responded to the form
18:59
Haumikura
you can run multiples instances
๐
19:07
flightofthefox (proven.network)
One 5 year old 36โ Samsung smart TV. Unknown mbps up/down (hacked neighbours WiFi)
At your service
๐ฅ
๐
๐คฉ
19:09
Deleted Account
Gameboy color with pink link cable and Gameboy camera
At your service
๐ฅ
๐ฅฐ
19:11
woooptydoo
No gameboy printer though ?
19:11
Deleted Account
Unfortunately no
๐ฅฐ
19:12
Can I run an extra node on this light master?
๐คฉ
19:15
Cpt. Charles
Gameboy camera?! I gotta google this ๐
19:15
Deleted Account
I want to play wario land 3 again on my game boy, but I kind of forgot how shitty it actually is to have a tiny screen, no backlight and bad colors
19:15
Cpt. Charles
Woah. I have never seen that camera thingy
19:15
woooptydoo
Now obsolete
19:16
Cpt. Charles
I do remember the โlightsโ you could add onto the gameboy, because the thing wasnโt properly backlit lol
๐
19:16
woooptydoo
You got full blown back-lit lcd mods now
19:17
woooptydoo
I had one of those shitty lamps as a kid too
19:17
Deleted Account
This, but in the backseat of the car during the night on the way to your holiday destination
๐
๐
19:18
We even have a proper music making game lol
19:19
Still don't know my way around it though, probably never will.
19:21
Cpt. Charles
Did you mod that one yourself or where does one order this? ๐
19:24
woooptydoo
I got this one off a modder from Taiwan on etsy but I did also mod one for my kid. If you are comfortable with basic electronic repair/modding and know how to solder it should be relatively simple but the time it takes, might as well just buy it from a professional maybe. There's a few stores out there that sell mod kits and blank shells/buttons of all colors.
19:32
Cpt. Charles
Yea, I don't have any tools to pull that off myself ๐
I'll have a look around, nice for nostalgia purposes!
๐
19:50
[PSB] PSILOBLOX | Boris
alright bros, i just got a lot more firepower for that test, soon ๐
19:50
64 cores... i can probably run a bunch of nodes on there for sure ๐
๐
19:58
Mike
Cool, I'm friends with the guy who programmed LSDJ
๐ฅ
19:58
my favorite tracker ever, or at least the only one I use nowaday (used to be a FT2/Renoise guy)
๐
20:06
[PSB] PSILOBLOX | Boris
@danhughes what do you think about changing the garbage collector to an alternative low-latency one instead of the default? When i started running validator back before babylon, i found that changing to ZGC helped me with missed slots, etc. Since this is high-performance test maybe it would work better with different GC ?
20:06
Deleted Account
pump it
20:07
better have a good internet connection, lol
20:07
[PSB] PSILOBLOX | Boris
gigabit fibre
20:08
woooptydoo
Nice, from what I understand it's a very complete tracker indeed. Very cool that it all runs on the gameboy.
20:08
[PSB] PSILOBLOX | Boris
This is what i used to start node, however, i had to hack the code and remove the memory leak detector as its not compatible with the new GC:
20:08
export JAVA_OPTS="--enable-preview -server -Xms18g -Xmx18g -XX:+UnlockExperimentalVMOptions -XX:+UseZGC -XX:+UseDynamicNumberOfGCThreads -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch -XX:MaxDirectMemorySize=2048m -XX:+PerfDisableSharedMem -Djavax.net.ssl.trustStore=/etc/ssl/certs/adoptium/cacerts -Djavax.net.ssl.trustStoreType=jks -Djava.security.egd=file:/dev/urandom -DLog4jContextSelector=org.apache.logging.log4j.core.async.AsyncLoggerContextSelector"
20:10
Just an idea, not sure if it will help or not... but it could.
20:12
Mike
Still being updated as well, which is crazy. He refactored the whole thing ,changed some basic implementations of previous features + fixed some bugs just last winter iirc
๐
20:13
Pyro - WILL NEVER DM YOU
20:13
[PSB] PSILOBLOX | Boris
However, I was asking around some Java Pros at Ergo community for example, and one said "ZGC is more deterministic at small pauses than Shenandoah, but lacks specific optimisation flags like CompressedOOPS, which are very helpful in reducing memory usage. And ZGC has coloured pointers for its GC algo, so in practice its off-heap memory usage is through the roof."
20:14
So i guess Shenandoah and ZGC could be some good candidates, potentially?
๐
20:14
Pyro - WILL NEVER DM YOU
Half java flags are overrated, the only real good ones are increased memory and use new generation gc
20:15
[PSB] PSILOBLOX | Boris
well, like i said, i was able to get better performance with a different GC algo. i think this high-performance test will matter even more.
20:16
Pyro - WILL NEVER DM YOU
XX:+UseG1GC
20:16
[PSB] PSILOBLOX | Boris
that's the default man lol
20:16
Pyro - WILL NEVER DM YOU
If you're on Linux though zgc is better
20:16
But windows can't use it AFAIK
20:17
Is the node to be run in a vm?
20:17
[PSB] PSILOBLOX | Boris
well, what sux is ZGC not supported on FreeBSD yet.
20:17
well yea linux VM most likely or freebsd if we going with a GC that works there lol
๐
20:18
Pyro - WILL NEVER DM YOU
I hope we later rewrite it to be in rust and not java ๐คฃ๐คฃ
20:20
[PSB] PSILOBLOX | Boris
well, for now... it is what it is ๐
so gotta find some ways to improve performance if we can, no? if want to go for that big D TPS ๐
20:20
Deleted Account
I expect Xi'an node software to be written in Rust ๐ I believe the team have stated that there are, or will be efforts to rewrite the node software in Rust.
๐
20:21
Dan just knows his way around Java best and there's probably not much point in him mastering Rust for his experiments
20:21
(but I don't want to speak for Dan of course)
20:22
Dan "Fuserleer" Hughes
Cassie is coded to reduce GC pressure as much as possible but that's a good idea ๐
20:23
[PSB] PSILOBLOX | Boris
I haven't done a real heap profile or anything, but this is my anecdotal evidence :P
20:23
Pyro - WILL NEVER DM YOU
I mean java works and is used by many large projects, mainly enterprises, but it's a dog on memory
20:25
Dan "Fuserleer" Hughes
But 1m tps in java is an extra middle finger no?
I did consider a C++ rewrite a while back as that's my true native language but seemed poor in effort Vs progress at the time
๐คฉ
20:25
Pyro - WILL NEVER DM YOU
๐
20:27
Dan "Fuserleer" Hughes
Same with rust. Marginally better than java, especially for prototypes but again, time sunk Vs gain seems lowish
20:27
If I was gonna shoot for outright bang for buck I'd go c
20:34
[PSB] PSILOBLOX | Boris
a fellow c++ bro ๐ค mind you, i coded in that back in Borland C++ dayz ๐
20:36
Dan "Fuserleer" Hughes
Borland lol were old as fuck ๐
20:36
[PSB] PSILOBLOX | Boris
okay maybe i'm not that old, I used the c++ visual builder too ๐
20:37
But then i stopped programming and went into infrastructure. now i'm tired of infra and getting back into programming ๐
๐ฅ
20:38
Dan "Fuserleer" Hughes
Welcome back
๐
20:40
Mike
Refuse to work in anything but turbo pascal
โค
20:43
Dan "Fuserleer" Hughes
Goat
20:45
[PSB] PSILOBLOX | Boris
u mean, Delphi right ๐
20:45
or u running DOSBOX ๐
20:59
Sebรถรถรถl
My first coding was in a C64 Basic, followed later by a DOS Basic 3.3 or so... with an huge 4M Hard drive an 20MHz CPU ๐คฃ
โค
21:05
Carsten ๐ฒ
Similar to me. C64 Basic, QBasic, Turbo Pascal, Java, Python and latest some glimpses into Rust. But all as a hobby.
โค
21:05
Pyro - WILL NEVER DM YOU
Speaking of delphi, I'm working with a company to migrate their delphi environment to python
โค
๐ฅ
21:07
Deleted Account
Long time since I've heard of Delphi
21:08
I played around with it a while as a kid, because I used a program for some game I played and it was written in Delphi, so obviously I was inspired ๐
๐ฅฐ
21:18
oz.xrd
I have an iMAC desktop with the fllowing specs:
Model Name: iMac
Model Identifier: iMac20,1
Processor Name: 8-Core Intel Core i7
Processor Speed: 3.8 GHz
Number of Processors: 1
Total Number of Cores: 8
L2 Cache (per Core): 256 KB
L3 Cache: 16 MB
Hyper-Threading Technology: Enabled
Memory: 64 GB
Disk: 3.28 Tb SSD available
Will I be able to run multiple nodes in separate virtual Terminals?
21:25
Sebรถรถรถl
From Maschine Specs Yes, depends on your Internet Speed.
21:27
oz.xrd
wiFi download 210 Mbps upload 220 Mbps
21:32
Perfect, 2-4 instances should be no problem.
I'm sure we will do some per tests before the final test.
๐
21:34
Mike
Where you involved in the demoscene?
21:37
Sebรถรถรถl
No we progged Text-based SiFy Games for ourself as a 10 year old together with my cousin and brother.
21:38
Emmoglu Leo Magal - I don't DM first
Two instances ๐
22:11
Octavio
I still remember counting Ram in Mb .. low number Mb
โค
22:38
Tobias | vido.info
640kb is more than enough ! ๐
23:27
Wekkel
It was expensive stuff
01:49
Rod
Lets go guys! Cant wait to test Cassandra!
Daniel_Elohinn invited Daniel_Elohinn
05:26
Maybe we can convince Ivan to cover the test? ๐
๐
09:25
Pyro - WILL NEVER DM YOU
Ideally we should do similar to what ton did if
@danhughes feels confident we can top their tps figures. Advertising over telegram would embarrass their campaign
09:31
flightofthefox (proven.network)
TON basically is Telegram though ๐
assume thatโs the only reason they could do those ads
If anyone could do it, weโd probably be seeing ads in project announcement channels all the time
09:32
Pyro - WILL NEVER DM YOU
Anyone is free to advertise on telegram
09:32
Not sure of the cost tho
09:38
flightofthefox (proven.network)
But can you target your competitors announcement channels? Neat if so, just surprised I havenโt seen people doing it all the time if thatโs the case
๐
09:42
Actually I think I see the reason. Minimum commits are expensive as hell:
However, ad campaigns are available only after you top up the balance with 2 million euros. Where 1 million is spent on advertising, and the second part is frozen as a deposit. If the advertiser decides to terminate the contract with Telegram before spending 10 million euros on advertising in the last 12 months, Telegram takes the deposit for itself. If it succeeds, 1 million euros will be returned to the advertiser.
09:46
Deleted Account
jesus
09:48
Is this what the community means when they say they want marketing? ๐
09:48
flightofthefox (proven.network)
2m euros could fund a lot of grants. you can bet TON did not have to pay that though
09:49
Deleted Account
One day Adam will jsut get pissed and say "Fuck it, you'll get your TG ads. It costs 10 million euros, but you'll friggin' get it"
09:49
And then everyone starts complaining that it's too expensive
๐คฉ
09:49
ooooh, TON = Telegram
๐
09:54
flightofthefox (proven.network)
dunno, something must be a filter - otherwise thereโd be all out fomo/fud wars in announcement channels ๐
10:24
Pyro - WILL NEVER DM YOU
Hory sheet. Yup no wonder it was up for a while, free ad. Monopoly
10:38
avaunt | ShardSpace | Notix
great way to boostrap your token distribution to 800M users.
TON will be the exclusive payment token on Telegram
10:44
Pyro - WILL NEVER DM YOU
Really need to buy those stickers, such value!
13:34
avaunt | ShardSpace | Notix
@danhughes @Chen_RL are you able to pin a relevant 1M TPS Cassie Test message in here please. Something we can use to point people to. ๐
๐
13:41
Faraz.xrd | Radstakes | XRD.Domains
I'd also suggest that some more admins might be beneficial when the time comes. This place is going to be lit when the time comes. with wen cassie token support questions. Happy to volunteer if I can be of use ๐
๐
13:44
A
@danhughes have you considered asking Ivan on Tech to participate in the TPS test?
๐
13:46
I could imagine that he would live stream the community test if he was interested.
๐
14:13
Chen (โ, โพ๏ธโ๏ธ)
I'm not high-ranked enough to pin things :D just banning evil ghosts.
14:33
avaunt | ShardSpace | Notix
15:15
ClickEnter
Let's Go 1M TPS
๐ฅ
7
15:31
h0ll0wstick ๊ฎ Old account
Good idea ๐
15:34
Jazzer | DefiPlaza.net
Played this from age 15-50 or something like that.
15:36
rotane.xrd
once we get the entry in Guiness book of World Records can we legally enforce TON to stop their ad campaign?
15:37
Deleted Account
Cool! I really liked it and Utopia Angel was an inspiration for me in programming ๐ I actually made my own tool at some point to summarize and format thief and spell ops (or whatever they were called)
15:38
Jazzer | DefiPlaza.net
Actually toyed with the idea of bringing something like that to Radix as I feel a game with the whole state on ledger may be quite a great thing to have on Radix and for this you could do that.
Use delegated fees, make an app and the users basically wouldn't even know they're using crypto..
๐คฏ
15:39
Yeah I tried to reverse engineer the Crystal Ball encoding to build stuff on top but failed. And they wouldn't say how it worked.
15:48
Deleted Account
That'd be really cool
15:50
I see my tool is listed on the "outdated tools" section in a forum post from 2010 ๐
15:51
pretty cool seeing some recognition haha
15:51
(the code was exceptionally shitty, lol, because young me just did things until they worked)
15:55
Just checked and I was 13/14 when I made it
15:55
I'll forgive myself for shitty code
๐คฉ
19:16
Emilia
Radix sounds very interesting. I read about the new consensus protocol that supports atomic updates between shards. However, I still have a few questions:
1) Are shards actually shards or are they threads?
2) Is there a limit to the number of shards? Can the number of shards grow and shrink dynamically depending on the network load?
3) What is the finality time of a transaction in both intra-shard and cross-shard cases?
4) What is the minimum and maximum number of validators expected? I have read that the first 100 validators validate blocks. The others?
5) How does the number of shards scale with the number of validators?
6) What is the actual TPS and when will the testnet version that is supposed to beat the TPS record reach the mainnet?
Thank you
๐
19:20
Deleted Account
Cool, someone with good questions!
19:25
I can answer 2, 4 and 5 I think:
2) With the Xi'an upgrade then network will be pre-sharded with 2^256 shards, which means that the number of shards is technically finite, but practically infinite. It is fixed, though, so no shrinking and growing.
4) Again with Xi'an: the network can scale linearly, meaning there is theoretically no max amount of validators. Minimum I don't know tbh.
5) Answered by 2, but I will add that validators will serve a set of shards and as more validator sets exist, they will serve smaller groups of shards if i'm correct
๐
19:25
TaserFace | ๏ธRadixStake
๐
19:33
Emilia
Thank you for your answers.
2) 2^256 is a monstrous number. If the network load is low, does it make sense to use resources for such a large number of shards?
4) The total number of validators is not important, what is important is how many of them participate in the consensus and especially those who allow it to be stopped. The famous Nakamoto coefficient that is 33% of validators for BFT blockchain PoS.
How many validators participate in the consensus? I read the top 100 based on their stake. Are the others divided into groups for each shard?
19:38
Deleted Account
I'm not knowledgable enough to answer your questions, so I won't try to ๐ The infographic series above is a good start, though, because it lays out the concepts that will probably already make some of your questions irrelevant, is my guess
19:39
But it's nice to see someone digging again. People in this group love getting into the technicals, so I'm sure someone will take up the challenge ๐
๐
19:51
Ed | RadUp.io
On 2) - itโs presharded, but nodes cover a shard space which is dynamic. In times of low demand, nodes might cover more shard space (eg more of the shards) than in times of high demand.
19:54
Emilia
OK, but if I am not mistaken, each node still has to allocate resources to maintain the status of the shards.
19:58
Ed | RadUp.io
Only those for the shard space they cover.
20:00
Emilia
Correct. How many validators are needed to cover 2^256 shards, an impressive number? Do only the first 100 validators cover them or also the others?
20:09
Emmoglu Leo Magal - I don't DM first
They don't actually cover shards, but shard groups. Shard groups are dynamic, shards are fixed. There can be only one shard group, or there can be many. Each shard group is at any time covered by enough validators (ideal > 100)
๐
20:28
projectShift
in Xi'an that 100 number is not the goal and certainly not fixed like currently.
Dan has previously referred to higher counts but he also said there was practical optimization that wouldn't allow it to go too high, since horizontal scaling is always best when there's linear scalability (Xi'an, again)
So the design goes towards less shards per shard group and more shard groups, iirc.
In essence, validator number shouldn't vary too wildly per shard-group served ... but more shard groups will mean more validator sets.
At a limit, a node might only be able to sustain belonging to only one validator set and thus serve only the shards that belong to the shard group served by that set, but on average, nodes are expected to belong to several validator sets and thus serve several shard groups.
Since this is all adjusted dynamically, there's a constant optimization, *per node*, on the workload it's able to handle versus the sets it belongs to.
iirc, that dynamic adjustment is to be take place on a per epoch basis, that will probably (but unclear and so far never saw it discussed publicly, but I could be wrong) have its own dynamic adjustment to network load conditions in order to maintain a targeted epoch temporal size (like currently on Babylon).
As you can also see, we keep referring to Xi'an, the fully sharded future, as we all have it pretty clear that the current single-shard design is the main factor driving the static values you can see and mentioned for validator set size.
21:33
TyPo.XRD
Is Flexathon down for future Cassie test?
22:01
Octavio
They claim to be the fastest blockchain
Deleted invited Deleted Account
Nick Dorian invited Nick Dorian
06:07
Inspector Crypto.xrd
In addition to the other replies, this conversation is pretty good for better grasping Radix's scaling solution. And with regards to 1).. right at the end (28:40mins) they mention that internally they've discussed not calling it sharding but threading.
https://www.youtube.com/watch?v=WfrHsupphOQ
07:38
avaunt | ShardSpace | Notix
๐
09:22
flightofthefox (proven.network)
Actually wouldn't be a bad name. it is quite different from any other sharded system
Too bad Hyperthreading is trademarked by Intel because that would sound cool
09:54
Inspector Crypto.xrd
damn that does sound good
10:27
avaunt | ShardSpace | Notix
SuperDuperThreading is not trademarked it seems
๐ฅฐ
10:34
h0ll0wstick ๊ฎ Old account
"Masterthreading" if people ask: top level in pokemon
10:36
That s what comes after "hyper"
12:09
TyPo.XRD
Wormholethreading because itโs on a completely different space time continuum, just blows peoples minds
12:17
h0ll0wstick ๊ฎ Old account
SonicELONvoldomoreInuthreading then..if people ask: don t.
๐คฉ
12:37
marco michelino
DanThreading beats them all
๐คฉ
13:30
avaunt | ShardSpace | Notix
A message has been pinned
19:30
AfterWave
Apogee-threading
19:30
TheWoodsman
Uber-threading
19:32
AfterWave
Grabbing an Uber via Radix Uber-threading๐
19:32
TheWoodsman
Soon ๐
02:28
Guilherme
Guys, If Dan were to pass away, would Xian continue to be developed normally?
02:38
Is it currently possible for someone else to fully take over Dan's work if he becomes incapacitated?
03:19
marco michelino
Dan can't die: his soul is sharded in 2^256 parts, like Voldemort
๐ฅ
6
03:27
[PSB] PSILOBLOX | Boris
03:27
okay now i have the firepower ๐
03:28
Loaded up FreeBSD14/ARM64 on this 64 core beast
๐ฅ
4
07:32
Pezburger
Be fun to see ervery Solana txn on testnet๐๏ธ๏ธ๏ธ๏ธ๏ธ๏ธ
๐
07:35
yosh moryno
if possible not only solana every top coin and benchmark how long it takes, imagine the hype
๐ฅ
09:09
Deleted Account
It has definitely been asked before and iirc the answer is: yes
๐
18:14
Christiaan wowo.xrd
I will also join
๐ฅ
18:15
Have a decent desktop and 1gb up and down fiber
๐ฅ
18:48
Christiaan wowo.xrd
12 threads and 32gb ram
18:49
Might also use my new work laptop
18:53
Whats most important for multiple instancances
19:20
AfterWave
I believe it was commented on connection throughput becoming bottleneck with too many instances.
StakedByRadix ๐ | ๐ทXSEED STAKING invited StakedByRadix ๐ | ๐ทXSEED STAKING
21:59
RetiredUncle
Dan has answered this question before. Does anyone have cop yto post/pin here.
TLDR he is in constant touch with others who knows exactly why the cassie source code does what it does.. Despite the re-assurance, it will certainly be easier if the team can consult with Dan as they build a productized version.
22:35
Christiaan wowo.xrd
Alright I have enough
Silvio Caixeirinho invited Silvio Caixeirinho
Azerdina invited Azerdina
12:47
Don Dude
Get i touch with elon musk, and get him to put his โDojoโ super computer into the test ๐๐ผ
13:00
AfterWave
It is interesting to read how this was done. Essentially spun up 256 servers in more of a lab like run.
This is where I find Dan's proposed test far more interesting as there will be a variety of nodes as a public test on community run nodes which is far more realistic.
๐
13
14:03
Win_Q
How many participants already submit the run test?
๐
15:29
Faraz.xrd | Radstakes | XRD.Domains
๐
BLW๐งis EARLY โก๏ธโก๏ธ invited BLW๐งis EARLY โก๏ธโก๏ธ
23:18
Speakers
Would it be worth changing the name from Cassie to something that more directly ties back to Radix? Just in case (when) this blows up and gets a lot of attention?
๐
4
01:49
Deleted Account
I like cassie :,(
๐
4
01:50
Speakers
Me too, but if it's going to be printed in the Guinness World records book, do we want it to say Cassandra network, or Radix Testnet or something?
01:50
Deleted Account
Most tech projects have "code names" for major updates or r&d projects
01:51
Not sure why cassandra would be better or worse than any other
01:51
Can call it Radix Cassandra Research Network if you want
01:52
It's pretty explicitly not the radix testnet. The public testnet is stokenet
02:07
Speakers
Yeah I don't mind what it's called, but better if it's somehow linked back to Radix.
If this blows up and thousands of people start google searching for Cassandra network, they're not immediately going to find their way back to Radix
02:16
Deleted Account
Whatโs the status of the Guinness World Record entry?
03:26
AfterWave
Very early stages. Until we have an arrangement figured out, best to speak of simply World Record so as not to cross liscening lines.
Once we get there, there will be "sharables" pre and post record attempt.
04:16
michelepsiu
Me too.
05:41
Win_Q
I like Kassandra
12:09
Deleted Account
Maybe the Maguires World Records. Aldi knockoff of Guinness
Nick Turk invited Nick Turk
Deleted invited Deleted Account
Deleted invited Deleted Account
Deku uzumaki invited Deku uzumaki
NightOcean invited NightOcean
MoRpHoZz invited MoRpHoZz
17:27
Tibor P.
Hey, I would love to take part with my comp in January, but what's the technical knowledge req? Like pro node runner knowledge and rust expert?
17:36
TaserFace | ๏ธRadixStake
just basic linux user knowledge, if you have any issues we are here to help
17:38
Tibor P.
That one I have. Main system is in Windows, but got a finance Linux backup system, do I need to use that?
17:39
TaserFace | ๏ธRadixStake
you could run it bare metal on the Linux box or as a VM on Windows using something like VirtualBox.
17:42
Tibor P.
Is there a guide? Also, really 300gb needed?
17:42
Emmoglu Leo Magal - I don't DM first
You can also run it on windows, it's just Java. But probably there's some messing with configurations so that windows doesn't get in the way of the test (doesn't do anything unimportant on background)
๐
17:43
Markus
From https://docs.google.com/forms/d/e/1FAIpQLSfj7m4Hr4c5vVADqqGKqRb0_BGQt04j6d-dPLPooM9U0sI_jg/viewformWe would like this test to be fair and unquestionable, therefore we are looking to the community to assist us by providing the majority of compute resources to act as validators during the test. The requirements to operate a Cassandra validator are low, the minimum specification we are looking to use for this test is as follows:
- 4 core CPU
- 8 GB RAM
- 200GB SSD (SATA is fine)
- 30Mbit download / 10Mbit upload
17:43
TaserFace | ๏ธRadixStake
No guide yet, just have the specs above
17:45
You could also spin up some "free" VMs in the cloud. Many providers offer an initial credit to use their service. For example, I believe Linode has a $100 credit with no contract which should be enough to run a node for a while for free.
โค
Raves.xrd invited Raves.xrd
22:51
THis
Dan mentions the Vamos database.
Do I have that right?
I can't find it on the internet, I thought I should just see what its approach is.
Does anyone have a link?
23:11
avaunt | ShardSpace | Notix
23:37
Sorry that was meant to be a reply to
Tibor
10:31
Tibor P.
hey, thx, so just run the Cassie testing and it should be fine, right?
12:30
projectShift
Why is Windows getting fud guys? C'mon ๐
It's just as good and fit as system D ... it's your limitation, not the systems ๐
Anyways, to those that may wonder - running it on windows is just as easy and straightforward as a system D setup on linux
12:32
Pyro - WILL NEVER DM YOU
Windows is good for games because Microsoft own a monopoly on direct x, that's about it
12:33
In every other area it is garbage
Fatih Sezer invited Fatih Sezer
21:20
Loccoco
I'm considering packaging deployment for aws and azure to make it easier for non tech.
About 10$ an instance for 2 days from first calculations using AWS Calculator
But curious about which types of cloud instances would be better. Seems as it would be java based, memory optimised ressources would be the best option ?
๐ฅ
21:56
THis
I read through,
#vamos, let's go!
Great backgrounder, thanks.
22:00
I am too, but don't want to commit just yet as life is busy anyway.
I alread have the AWS account and am familiar enough with how it works.
Yes, we need to know what instance type to specify and then make sure it has OpenJDK on it.
And know what ports should be open.
I guess for this we do not need a failover, or to monitor it otherwise, although monitoring could be useful for test results?
22:01
Loccoco
Yes, to me Terraform+github action+Ansible would do the job to provision security groups, IP, middleware and such
22:02
We could make a git repo to clone by users
22:02
But same as you, not engaging yet and need precise requirements
23:52
Haumikura
I was looking at things about star citizen, a game in development for years, and I came across a problem they had a while ago that reminded me of Cassandra, the problem was "server meshing", they needed interaction between several servers, which must be able to support the very high demand from players, at the same time that they save the position state of items/players dynamically, at one point they divide the servers into "shards" so that they divide the bigger problem, into minor problems (just like xian will do).
This brought me to the question, could Cassandra's test simulate Star Citizen multiplayer gameplay?
The marketing behind this would be very large, as Star Citizen, like Radix, is the longest game in development in history, having a gigantic fan base: "Radix, the ecosystem capable of supporting Star Citizen Multiplayer with infinite players".
@danhughes
23:54
Lennart
As far as I've seen Starfield has been kind of a flop. Apparently concurrent playerbase after 2 months has dropped below skyrim, which is a game from the same devs, but around 12 years older
23:56
Haumikura
Star Citizen is a game that has been in development for over a decade, several technologies needed to be created for the game to be able to be created, it has over 650 million in crowdfunding.
The active player base is not large because it is still in the alpha phase, nor is it in beta yet.
00:21
AfterWave
Hell of a motivation seeing this ad in the anouncement channel. Sweet and sour!
00:30
Lennart
Oh wait, my bad. I confused the 2 games๐
00:32
Haumikura
and I didn't even notice, now that I saw it too lol ๐๐
08:38
oz.xrd
TON has thrown the performance gauntlet to the L1 community based on 106.5k TPS using complex transactions, 258 nodes, 256 dynamic sharding side chains, 6 sec time to finality & audited by Certik running their main net code in testnet. Very professionally executed.
In light of this audited benchmark we need to dial down our planned stress testing as a preliminary internal validation exercise.
https://www.youtube.com/live/jWWl1sLGY7s?feature=shared
Deleted invited Deleted Account
14:54
THis
I bought up the topic of the
#vamos database.
So my understanding is that it has been built by Dan?
So the
#Cassandra #2024
#test -- soimeone needs to invent a single hashtag for this -- will also be a test of the DB?
It is important to be clear about what is being tested.
What of the final design, even though not fully coded as release candidate, is being tested.
Can the test be designed so as to embrace different possible scenarios?
There are complex SC dependency interactions.
What about large sums being shifted to arbitrage such as what I quote here from the Kaiko Newsletter:
"
For the months of September and October it registered very few transactions. However, it awoke at the beginning of November, registering more than $1.5bn in outflows to a variety of lending and borrowing protocols, CEXes, and DEXes in just a single day. "
https://t.me/RadixUnofficial/28571The point I am trying to understand here is to do with the vast difference in fees that such actions accrue on BTC / ETH compared to XRD.
This seems to me to be a whole, large, topic in itself which would touch on potential arbitrage possibilities and MEV issues.
People with the test results would have to investigate further and draw their own conclusions.
14:57
Just staying with
#vamos, I think this question belongs here too?
Does the ledger guaruntee the complete history of all transactions in perpetuity?
Where is *all* of this data going to be stored?
How is it *all* going to be accessed?
15:08
AfterWave
Although to a certain extent I agree but I would say "dial it over" not down and here is why...
The TON test was done with a copy of their network and although yes, this is a copy of the launched mainnet, it was run under a "laboratory " set up where they spun nodes up with alibbaba which I presume were uniform in nature; not as real world as what Dan is proposing IMO.
One of the prudent approaches I appreciate about Radix is test and design before launching. So yes, this is the research network and good job pushing the limits before taking a leap with other people's money.
As for complexe transactions, Dan has stated that these will figure into the test design.
I am curious what is meant by complexe txs in the TON test. Were there elements that worked across shards? Seems like the test bomb was a self replicating scheme, so wouldn't that mean it starts as one shard and as their dynamic sharding spins up, the test txs copy themselves as contained to single shards, but does it get into atomicity with truly complex txs? Would be great if they did, but have not seen this sort of granular detail yet, so if known, please share.
15:14
Spangular Directions
Xian validators only need to store current state. Historic state and tx info is not required for consensus
15:15
Buddhini
still trying to get an answer on that
@danhughes :-)
15:16
Emmoglu Leo Magal - I don't DM first
are you sure of that?: Because that would mean stateless, right? Iirc That is Dan goal, but not quite there yet... Curretly sharding does divide the storage load accross nodes, though
15:19
Spangular Directions
I'm pretty sure, not 100%, but lots of stuff would get very very heavy if it wasn't the case.
Eg, when a validator gets reshuffled, do they then have to import every transaction that has made contact with a shard in their grouping? That could be many many gb of data
15:20
Emmoglu Leo Magal - I don't DM first
๐คท๐ปโโ๏ธ let's wait for Dan to shed the light
15:20
Deleted Account
I thought they kept part of the historic state. Buuuut I could've also made that up
17:03
flightofthefox (proven.network)
still interested in this question ๐
17:13
that can be fine depending how often nodes are shuffled
17:20
THis
But historic state will be required for many use cases.
Verifiable transaction detail, history and provenance.
It may be necessary to store an arbitrary large slice of history, I don't know.
Is this being addressed?
17:21
h0ll0wstick ๊ฎ Old account
Every second, on average, around 6,000 tweets are tweeted on Twitter