<eddie:oblak.be> Fucking qubic still annoying

<eddie:oblak.be> lol decrypt.co/338223/monero-51-attacke…model-but-it-cant-do-basic-math-yet

<eddie:oblak.be> it's shown that giving correct answer to 1+1 requires 100 pages of thoughts. What a moron honestly

<eddie:oblak.be> Humans learn by copying not by computing everything from scratch every time. Because it's not efficient nor useful.

<syntheticbird> @eddie:oblak.be: step 1. copy a 40M LLM model from hugging face

<syntheticbird> step 2. rename it and expand it

<syntheticbird> step 3. claim it's thanks to Qubic miners

<syntheticbird> step 4. AGI IS NEAR!!!!

<eddie:oblak.be> step 5. sprout nonsense that sound smart but it still nonsense

<eddie:oblak.be> true AI lfmao

<eddie:oblak.be> The only thing he's actually capable of is disrupting the monero network probably at a loss

<eddie:oblak.be> * Ivancheglo warned that mocking the model now may have bad future consequences and AIGarth may be hiding its intelligence.* 🤣

<sbt:nope.chat> New gift card site: xmr.cards

<sbt:nope.chat> Bought a few, love the no js and blazing fast.

<eddie:oblak.be> looks nice!

<basses:matrix.org> @sbt:nope.chat: lol, trust on first sight?

<basses:matrix.org> they arrived?

<rbrunner7> Somebody has to go first, otherwise new things are never able to get off the ground ...

<syntheticbird> @sbt:nope.chat: US only unfortunately

<syntheticbird> can't they support all the other 384 countries out there

<eddie:oblak.be> don't they?

<eddie:oblak.be> I see giftcards for other countries

<syntheticbird> @eddie:oblak.be: which category?

<syntheticbird> oh wait there is a country selector

<syntheticbird> my bad

<syntheticbird> Finally

<syntheticbird> I'll be able to buy my Robux in XMR

<syntheticbird> or my "EA FC 25" coins

<syntheticbird> or whatever the hell that is

<eddie:oblak.be> Prepaid phone credit

<alexandre:uii.pt> wouldn't that be the one shown on cake wallet? i recall it had built in gift card buying option

gift card enjoyers see the same api with a different front end from unknown persons

@alexandre:uii.pt i believe the cake wallet one was called "cake pay" and is built-in, yes

<alexandre:uii.pt> yeah, they call it cake pay, has more options, at least, it seems

<syntheticbird> plowsof can you stop revealing the civilization-old proxy scheme

xD

are there any known tricks to accelerate monero node synchronisation? as soon as i reach blocks from ~2020, sync times explode a bit, and it doesn't feel like that's because of CPU performance

cuprate sync times are faster for me on the same machine, by the way (but cuprate sadly isn't quite usable yet), and it syncs in around 1 day whereas monerod takes 3-4 days

<ofrnxmr:xmr.mx> Has a few countries > <@syntheticbird> US only unfortunately

<eddie:oblak.be> Anyone here can give me tips on inspecting an lmdb similar to sqlite shell ?

<syntheticbird> helene: ~2020? what height range exactly?

<eddie:oblak.be> @eddie:oblak.be: nvm gottit

<syntheticbird> @eddie:oblak.be: There are no tools for such. I think there is only mdb_tools for having some stats on the internal b+tree but no cli to read/write data

<syntheticbird> bad timing

<eddie:oblak.be> you can do it in python apparently

syntheticbird: i can try re-creating it, but i don't believe it will give an exact range still (but at least a better approximation)

<syntheticbird> no need for precision, you don't remember what kind of height it started slowing down

<syntheticbird> you can approximate it to the 250k

i've stared at it for a while now so i don't have the numbers anymore (and the scrollback/log buffer got too full), but because i've been able to replicate it on every machine i have it shouldn't be a problem for me to replicate it

<syntheticbird> alright

<syntheticbird> You can refer to this chart of cuprate syncing:

<syntheticbird> mrelay.p2pool.observer/m/monero.social/kQObHnUEUeLXQFcLnGzArkjy.png (syncing20241217.png)

<syntheticbird> monerod follow approximately the same bottlenecks

it might check out, but i'll confirm with you once i have the proper info :) thanks!

it seems to scale very strongly at around 2.2m block height; LMDB performance seems very relevant there but also note that this is on a aarch64 machine, so I wonder if some crypto might be less optimized

it also seems that the more blocks there are since that point, the slower it gets

<citizenfour4:matrix.org> hey

<syntheticbird> helene: At this height the high transaction volume makes blocks very big and batches decreases. This is more often a network bandwith bottleneck at this point than a CPU or IO one

<syntheticbird> are you on a gigabit connection?

network bandwidth being a bottle neck is really odd since i have 10gbit symmetric

<syntheticbird> bare-fiber connections? no tunnel?

<syntheticbird> i just made the word bare-fiber, but basically no vpn?

it's on a 2.5gbit ethernet link to the router with the 10gbit symmetric link, it could be buggy Realtek drivers on Linux/this device though

there is no extra layer like a VPN no

2020?

That's the conversion to RandomX

<syntheticbird> alr, we can safely assume that the 2.5g is saturated

<syntheticbird> or at least can

The entire chain is only 250Gb, it's not network sync related, you can sync perfectly fine on a 1gbit link with no slowdown, there's rarely enough high speed peers to auctually cause you to saturate a 1gbit link.

Unlikely to be a network bottleneck unless really shitty network.

A slowdown in 2020 is likely the change to RandomX given that you said it was an ARM chip.

download link is barely doing 1MiB/s currently, and I know my network to be fairly reliable

Run sync_info in monerod, if it starts with m with a few o's afterwards, it's 99% not network.

<syntheticbird> I say that because when sync speedrun any% benchmaxxing Cuprate network bandwith quickly became the primary issue once we hit ringCT

that would be my theory indeed, i don't presume randomx would be very well optimised for aarch64 right?

<syntheticbird> snipa is probably right

I'm sure it's been optimized at least partially, but it's not likely to be anywhere near an amd64.

another x86 machine (Ryzen 7940HS) doing a sync via another 2.5gbit link on a XFS filesystem is managing to achieve it under 7h

(2.5gbit link to the same router)

<syntheticbird> lmao well that eject the network theory out

I'm pretty sure my node sync (Over starlink of all things) is only 5-6 hours.

<syntheticbird> Space node

Joys of not being able to get anyone to pull you fiber for less than a quarter million bux.

<syntheticbird> feel your pain, took a whiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiile to get fiber at me

i'm just lucky to be in a country where ISPs for home use provide such links via FTTH :)

<syntheticbird> there was 2 trials actually between the city and the ISP because they fucked up the installation twice

Eh, I'm a little far from the closest node. :)

If I lived down in the city, I could get fiber no problem, I just prefer not living in cities at this point in my life. :D

moneromooo: [mooooooooooooooooooooooooooooooo.oo.ooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo..oooooo.....]

<syntheticbird> you could always try the autistic space dish method, where you pay your neighbor to install the gigantic not suspicious antenna in their roof and you just receive it where you are

<syntheticbird> Snipa: fair

helene - What arch chip are you playing with? I'm curious, about to spin up some Amperes

<syntheticbird> TOTAL AMPERE SUPREMACY

<syntheticbird> sorry my inner x86 hater just woke up

It was an excellent excuse to pick up some to play with, because I'm not a fan of one of my current AMD boxes.

And I needed something that could host 24 u.2 NVMe's.

Snipa: It's a pretty terrible SoC that I've been using to also run a NAS, it's a RK3588 (4x Cortex-A55, 4x Cortex-A76)

How much memory does it have?

RX is memory hungry, and if you're having to run it in limited mode, it's gonna be /real/ slow.

8 GiB, LPDDR4X 2133MHz soldered (the 16Gib version was sold out at the time I wanted it, lol)

<syntheticbird> is monerod the only thing running ?

pretty much

<syntheticbird> also yeah terrible soc

<syntheticbird> 8GiB system with only monerod is more than acceptable

Eh, my monerod uses 6Gb.

So yeah, it's marginal imo.

Not as bad as tari's 10Gb. xD

<syntheticbird> POWERED BY RUST

tari is also heavier on CPUs (and disk space usage) i've noticed lol

<basses:matrix.org> sync when sleeping? > <Snipa> I'm pretty sure my node sync (Over starlink of all things) is only 5-6 hours.

<syntheticbird> @basses:matrix.org: A: Question?

<syntheticbird> B: Why ?

Tari is super heavy on iops.

The wallet in particular because of it's architecture.

<basses:matrix.org> @syntheticbird: don't want to be working while it is syncing taking all of their bandwith

Cryptocurrency is part of my work. :)

<syntheticbird> @basses:matrix.org: nice thought but tbf they claim to have 10gb symmetric with 2.5gb nic on each machine

<syntheticbird> so that shouldn't cause any issue

sounds like you're having fun at work then :)

I mean, I have 100gbit on anything important internally. :)

ah, i am very not bothered about monerod's bandwidth usage

<syntheticbird> biggest flex in this chat since its creation

<syntheticbird> hands down

torrenting is much heavier on bandwidth haha

Lemme see if I have my syncing stats for the wallet still.

There we go, that was the syncing of the tari hatchling pool wallet, with /nothing/ else running on the entire VM.

what filesystem was it running?

EXT4.

even for ext4 that is indeed pretty heavy

Yeah, 4Tb of writes for a 4gb final size. xD

200Gb of ram.

It's impressive how leaky that wallet can be.

It's extremely difficult to load either the SXMR or hatchling/jagtech legacy wallets because of how they do txn loading.

M5M and I thought the Monero SXMR wallet was bad. xD

i don't really like the tari wallet codebase but then again, maybe it's because i'm not used to their programming style

(and i'm not a fan of gRPC either)

So, having talked w/ the devs /alot/ about it, the wallet was written a long time ago, and it kinda got dropped because they never really perfomance tested it on the scale that a pool would.

They did finally drop the GRPC and P2P layers, and are mostly http at this point for wallet <-> node comms.

Though I think there's still some extant p2p stuff left.

i can see why they're rewriting it from scratch, hopefully to a point where it would be easier to play around with scripting :)

It's auctually quite scriptable right now. :)

oh, you've managed to play around with TariScript?

I do alot of work with automating tari as the hatchling/jagtech op xD

ah, automating! i meant the coin scripting (for atomic swaps and such) :)

Oh, that insanity. Yeah, haven't gotten there yet.

I know a few people have finally gotten grpcurl working nicely, though I just have a repo of go scripts I use. xD

I need to update that for 5.0.1 soon(tm) though.

anyway, i'll look if i can do a bit of profiling to confirm that it's randomx (which would explain why i have similar struggles with cuprate without its fast sync) :) thanks for helping!

No problem! Good luck with it, and if you really just want it to go fast, export it from your other node and pull it in. :D

<jairbolsonaro:nope.chat> Good morning everyone.

<frogmindset:matrix.org> hi, question. is cake wallet good for holding my monero?

<frogmindset:matrix.org> im new to all of this so i would appricate the advice

<basses:matrix.org> @frogmindset:matrix.org: yes

<basses:matrix.org> getmonero.org/downloads/#mobilelight

<frogmindset:matrix.org> @basses:matrix.org: why did you link this? cake wallet isnt on there

<frogmindset:matrix.org> oh, monero.com wallet links to it, weird

<alexandre:uii.pt> monero.com is cake wallet's site, not endorsed by monero itself

<basses:matrix.org> @frogmindset:matrix.org: yeah same software stack by same dev but focused on only one coin (wallet)

<basses:matrix.org> im surprised their faq doesn't include any differences

<basses:matrix.org> include the differences*

<basses:matrix.org> they don't link directly because of seo backlinks making them rank higher than getmonero.org

<basses:matrix.org> dumb to choose monero.com as domain for their wallet

<ofrnxmr:xmr.mx> Monero.com is cake's monero-only wallet

<ofrnxmr:xmr.mx> @basses:matrix.org: The only difference is that monero.com doesnt ship any other coins code

<hopeful24:matrix.org> So dumb question how do you buy monero anonymosly?

<ofrnxmr:xmr.mx> anonymously? Using cash is the only way

<ofrnxmr:xmr.mx> If you use a wire transfer etc, its no longer anonymous. if you are swapping other crypto for it, did you obtain the other crypto anonymously?

<ofrnxmr:xmr.mx> If cash in person, its not anonymous

<ofrnxmr:xmr.mx> So i guess youll need to define what is acceptable to you

<usb:envs.net> how many people here have done cash-by-mail

<usb:envs.net> i wanna know your experience

<usb:envs.net> like wont the cash be stolen or what

<frogmindset:matrix.org> How do you even go about buying it with cash?? > <@ofrnxmr:xmr.mx> anonymously? Using cash is the only way

<ofrnxmr:xmr.mx> Cash by mail

<ofrnxmr:xmr.mx> Xmrbazaar or retoswap

<basses:matrix.org> still got the exchanges stuff? > <@ofrnxmr:xmr.mx> The only difference is that monero.com doesnt ship any other coins code

<basses:matrix.org> @hopeful24:matrix.org: retoswap/haveno

<basses:matrix.org> @ofrnxmr:xmr.mx: find it on the street

<ofrnxmr:xmr.mx> @basses:matrix.org: Yeah, exchanges are present

<ofrnxmr:xmr.mx> @basses:matrix.org: Only some payment methods allow for anonymity

<ofrnxmr:xmr.mx> Like, using a mule or fake return address for cash by mail, or wearing a mask and walking or taking public transportation to meet in a dark alley