<r​ecanman:agoradesk.com> Woow, that is a very beautiful machine

<r​ecanman:agoradesk.com> Woaw, that is a very beautiful machine

a pretty wasteful design. 64 core chip, 64MB cache, so you can only use 32 cores per chip

sg2042 analysis published a couple weeks ago arxiv.org/abs/2309.00381

Wait, 18 CPUs per machine? lol

And SODIMMs, lol

It's $20k+ per machine in componnents

SG2042R is not the regular SG2042 though, so it probably has AES

And one memory stick per CPU? 212/18 = 11.78 kh/s per CPU, so they probably reduced CPU speed and voltage as much as possible to make it hash at 11-12 kh/s

One memory stick can do 11-12 kh/s

Yes, it's DDR-3200, it can do this

DDR4-3200

lol youtu.be/RkWGbkJjXa8?t=242

If SG2042R is a version of SG2042 with 32 cores/64 MB cache and AES, it makes the most sense

11778/32 = 368 h/s per thread, that's low. They don't need 2 GHz for this.

Since each thread is so slow, the balance between RandomX program execution latency and DDR4 latency is shifted to program execution - so any increase in program size will decrease their hashrate proportionally. Our RandomX tweaks (256 -> 288 instructions + 16 AES instructions in the main loop) will slow it down 15%, from 212 kh/s to 185 kh/s

And Ryzen CPUs will get a small 2% hashrate increase while keeping power usage the same (I tested it yesterday)

So the tweaks won't brick X5, but it will reduce it's efficiency quite a bit, it will be on par with Ryzen rigs after this

hm, sg2042r, the R could stand for "reduced" - reducing core count to 32 could make sense

half as many cores per NUMA region, but same number of regions

I'm sure they also added AES instructions

No way they get this efficiency with software AES

Hardware AES, 32 cores with 64 MB cache, clocked very low and with low voltage

And a single DDR4-3200 SODIMM, it's probably also very energy efficient because it's a notebook memory

8 GB stick = only a few memory chips, and they can run it at < 1.1V with loose timings

368 h/s per thread means they need memory with < 150 ns latency which is 2 times more than a regular DDR4 latency

DDR4 latency is 40-80 ns depending on board and CPU design

Found the memory sticks they use: jm.pl/en/8gbddr43200mhz260p-memory-…-2133mhz-8gb-512mx8/9208147/produkt

M4SE-8GSSOC0M-F

DD=VDDQ= 1.2 Volt (1.14V~1.26V)

They definitely set it to 1.14V or even lower

"Low-Power auto self-refresh (LPASR)"

This is Innodisk M4SE modules: rosch-computer.de/produkt/speicher-…ded/so-dimm/ddr4/innodisk/DDR4-M4SE

$107.5 per stick: wiredzone.com/shop/product/10022826…x8-sodimm-mem-dr480l-il02-so32-9241

I can't even find what timings these modules have :D

they must have found a source of memories like used corporate laptops to source it cheap

They've been mining since December 2021, and if my math is correct, each device mined 20-30 XMR since then. Even if they sell them for $3k, they got no more than $8k from each device, and I doubt Bitmain even got their money back, lol

Of course if you order 3000*18 = 54k DDR4 sticks, they will be cheaper than $100 per piece

and I doubt that they tore down 27k laptops :D

It was just a bulk order

I wonder if they utilize this design for data centers etc. looks very much like very "general purpose" ai board

They are selling them for 3k and just the memory costs 2k? lol

maybe they are pulling amd athlon and disabling bad cores on the 64 core soc to utilize bad batches

Judging by the performance, the SG2042R chip would have a market price of at least $500 a piece.

If they sold the miner at the market price of the components, it should be at least $12k. So there is no real cost advantage compared to a dual Epyc setup.

tevador unrelated to the current discussion: did you check how many more infinities we get with 288 program size?

I also just realized that since the AES tweak mixes F and E registers better, infinities don't hurt scratchpad entropy anymore

btw I finished my RandomX v2 aarch64 implementeation

elucidator it can be an early batch of SG2042 with added hardware AES. Yes, they can have 64 cores on chip, but only 32 cores active - putting a bad batch to use, as you say

I will check the effect of the 256->288 change later. I first need to integrate the VM/JIT changes.

RandomX v2 will not brick Bitmain X5, based on what I know so far

It must have hardware AES to get this level of performance

368 h/s per threads = 150 ns per one loop iteration, so they are limited by CPU loop execution, not by memory latency. Increasing program size to 288 will automatically slow it down by 12.5%

And adding AES in the main loop will also slow it down by 1-2%

So I expect RandomX v2 will slow down X5 by ~15%

now the only hardware aes is left to wonder

I wonder why the board has this chip: en.sophgo.com/product/introduce/cv1835.html

perhaps for AES?

Ah, it's the controller board.

elucidator About hardware aes - the only way to find it out is to disassemble their risc-v firmware code

they must have bought or had manufactured all of their hardware in huge bulks... or they've "falled off the back of a production line" somewhere

In other news, Raspberry Pi 5 finally has AES: hackster.imgix.net/uploads/attachme…press%2Cformat&w=1280&h=960&fit=max

X5 costs $20k if you sell individual components. In other words, Bitmain could get $20k if they sold components of X5, not X5

It definitely costs them less than $20k to produce

Since RPi5 finally has AES, RandomX v2 AES tweak is just in time :D

RPi5 is probably going to sell at around the same price as a Lychee 4 I'd imagine, if not higher :/

sech1: for the life of me, I can't find the source of the rV binary anywhere on that firmware image.

xmrig gen binary does some spi operations but can't see anything related to the actual operations. just fetch run fetch run

paulio_uk: Raspberry Pi 5, coming at the end of October. Priced at $60 for the 4GB variant, and $80 for its 8GB sibling

RPi5 looks juicy for Monero + p2pool + xmrig combo

Even 4GB variant can run all three, if p2pool runs with "--no-cache --no-randomx"

and it has AES, so it should have decent hashrate

<h​bs:matrix.org> \o/

qu.ax/XEp.jpg m2 hat looks like this, slap a 1 TB and long term local node

I'll order 8GB variant. monerod will run with 2 GB dataset for fast block verification, p2pool will run with "--no-cache --no-randomx", and there will still be 6 GB left for xmrig

Question is, does it support 256 GB microSD cards?

It can also boot from external SSD connected to USB3

And it's fast enough: youtu.be/9hYfQ7bRgZg?t=794

sech1: all the raspi models can boot from usb hdd

I do it with rpi4 at home rn

Is microSD card fast enough for Monero node?

Or external SSD is the way to go?

never tried, I don't use big SD cards. don't trust their reliability. either only use it for redirecting the boot or don't use it at all

<r​ecanman:agoradesk.com> afaik microsds are pretty slow and degrade pretty quickly

$80? That surprises me. Guess they'll do the "We're out of stock" thing 10seconds after it launches

definitely ^

5 minutes later... RPi5 for $400 on ebay :D

why break with traditions huh? :D

<r​ecanman:agoradesk.com> I still cannot comprehend why they are so expensive

<r​ecanman:agoradesk.com> I got a 3b+ in 2018 for $30

and for the first 6 months, you can't order a board but you can order a kit with extra 10 LEDs and resistors and a crappy how to blink led book for $999.99

it's a "kit" so parts add value ofc

<r​ecanman:agoradesk.com> matrix.monero.social/_matrix/media/…radesk.com/XOQnHbXegiAjVUGLJCozKtoF

<r​ecanman:agoradesk.com> matrix.monero.social/_matrix/media/…radesk.com/EqMjskpfrSawujoyYJXXULDR

<r​ecanman:agoradesk.com> Source: hardforum.com/threads/please-help-t…icrosd-card-crystaldiskmark.1999752

I guess microSD is fine for running an already synchronized node (do sync on desktop, then just copy over the blockchain)

yeah probably, but still you better install log2ram and enable zram instead of swap for longer SD card life

actually it's better if you just enable overlayfs on SD card if OS is running on it, then keep the blockchain and node files on external drive

lol, log2ram reminded me of the old days where I put Windows temp folders on the ramdrive to speed up the system (it was running on an HDD, SSDs weren't a thing yet)

Oh lol, I used to do that with DJGPP on DOS before I switched to linux :D Flood of memories...

All system includes on a ramdisk. I loved DOS, you could understand the OS back then...

I even remember experimenting with WinPE (Windows version that could boot from CD and run entirely on the ramdrive)

Only 150 kh/s? cdn.discordapp.com/attachments/7899…30/1156703452362182696/IMG_5347.png

Theoretical hashrate 71.94 kh/s per board, 11.99 kh/s per chip

But it does only half that

36-37 kh/s per board

110 kh/s based on "Real Time Hashrate" numbers

ok, "Chain's rate" graph shows that it started at 70 kh/s, but then slowed down to 36. Overheating probably

Discord messages say that it only happens on Zephyr. So non-Monero coins make it blow up :D

Need to wait for Monero tests - actual sustained hashrates and power usage numbers

hmm yeah total of "real" is more like ~110k

That would be fun if it can theoretically do 212 kh/s, but overheats and drops to 110k :D

turns out managing 18 cpus is difficult

Cat Processing Unit

based on the images of the hardware, does it look repurposable? capable of being repurposed

i don't see any sata ports or even USB headers

i guess if it was cracked, though, you could feed it data to crunch over the network. I mean, if its a modded CPU at 32 threads, thats still 576 threads per unit.

or if they are original spec, 1152 cores in 22620 cm^3

erg, threads

You could at least rip out DDR4 sticks and sell them for $2k :D

Their control board runs some flavor of Linux, but I don't think they run any OS on SG2042 boards. They just straight up upload RandomX code there and run it

If someone hack the firmware and finds way to modify it, it could be a nice number cruncher with 576 threads

<e​ndor00:matrix.org> Compile your workload with the rx VM as target and send it as a "mining" job :D

On the other hand, if you need 576 threads to compute something, you probably need a GPU

RandomX is unique because it's not SIMD, and 99.9% heavy threaded workloads are SIMD

GPUs will crush this potential "hacked X5" on these workloads

you'd be surprised how useful pure thread count can be for scientific workloads. genetic sequence alignment, for example.

I'm sure it can be ported to GPU

X5 only has 32 (possibly 64) threads and 8 GB RAM per chip, so it's limited in what it can do

yeap bmcbioinformatics.biomedcentral.com…/articles/10.1186/s12859-019-3086-9

"Overall on Geforce GTX 1080 Ti GPU, GASAL2 is up to 21x faster than Parasail on a dual socket hyper-threaded Intel Xeon system with 28 cores"

I can't come up with a workload that requires CPU specifically. Maybe code compilation or databases, but 8 GB per chip is not enough

<e​ndor00:matrix.org> There are some numerical algorithms that cannot be sped up through parallelization, so they require faster single-thread performance rather than lots of threads

<e​ndor00:matrix.org> Though I'm not sure that the SG2042 could compete even in that

It's known that nothing is known about possible parallelization :D en.wikipedia.org/wiki/P-complete#Motivation

Even if some algorithm is sequential, you can run 576 different instances of it on X5

lol, RIP twitter.com/MiningRabid/status/1707237819373293758

<e​ndor00:matrix.org> en.m.wikipedia.org/wiki/Parallel_algorithm#Parallelizability

Yes, it's classic "dependency chain" issue in parallelization

But the other wiki page says "it is not known whether there are any tractable problems that are inherently sequential"

so even Newton's method can be theoretically parallelized (or at least it's not disproven yet)

Sure, but "let's buy GPUs to run thus because it's not proven it will always be slow" isb't a great pitch to the boss :)

What's of salience here is practical speed now (or at least in the near future) no ?

yes

I just didn't hear much about people buying i9-13900ks in bulk, LN2 cooling and overclocking them to speed up their sequential algorithms :D

But GPUs in datacenters are everywhere

yup yup, our place has 4 12u racks full of the nvidia non-gpu gpu processing cards

damn things sit idle a lot of the time

<e​ndor00:matrix.org> AWS and Azure (and I'm pretty sure GCP too) have a "compute optimized" tier of VMs, offering fewer cores but with higher clock speeds ("up to" 3.5-3.6 GHz on AWS) compared to the other tiers

<e​ndor00:matrix.org> (Although "fewer" is a relative term, given that the biggest amd ones still reach 192C/384T)

could still host a decent network doom tournament on them

<e​ndor00:matrix.org> Though I guess if you actually need that much speed, you're probably better off looking into some dedicated fpga/asic to optimize that specific compute step?

pretty sure my place only bought the kit outright because we started looking into AI automation, and someone high up had a friend whos neighbour's dogwalker's aunt's first cousin's son runs a hardware firm that deals in them, and they probably had shares in it, so why not spend the public money on it?

moneromooo, cat processing unit <3

The more watts, the better for cat

yes, I think if some sequential algorithm is very important, there will be some specialized chips for it

RIP youtube.com/watch?v=WRLZitXASaM

dayum

It's locked to Monero only :D

such enterprise much wow

it makes sense, other coins use different RandomX configs

most, but Zephyr seems to use an unmodified rx, unfortunately

No, it can't even mine Zephyr which uses stock RandomX

They are either very bad at writing software and hard-coding everything for Monero, or they deliberately locked it to Monero

maybe they hard locked it to XMR so they can mine Zeph with less competition on their next system =p

But they advertised it as multi-coin miner. False advertising? Selling used hardware? Classic Bitmain

tru

if zephyr is the same PoW as monero, I don't see why it would not work unless it's intentionally only working with XMR addresses

The theory is that their firmware can't process address formats other than XMR

Or they intentionally locked it

Mining job header has fork version and block height, so it's easy to lock it to XMR

I suggested him to try to connect to xmrig-proxy with fake XMR address, but it "didn't work"

most HPC/scientific computing tasks are easily parallelizable because it's all matrix math

I saw N-body problem as an example of "hard to parallelize" algorithm

Because each body interacts with all other bodies

I see why

all parallel threads will have to access and modify the same data structure representing the N bodies

or do two phases, with redundant copies of all state. 2nd phase to coalesce the multiple views back into a single one.

but now we're tactually talking rocket science, of course it's not straightforward or easy

even solving the original rocket equation isn't easy.

One phase (where you calculate next positions) is easy to parallelize, but the other phase where you have to sync all threads, is mostly sequential

yes

There's no escaping Amdahl's Law

btw, bitmain > apple, RAM not soldered to the board!

lol

At least buyers can recoup some costs after it gets useless :D

18 DDR4 sticks :D

heh but ddr4-3200 is already obsolete

if someone jailbreaks it, at least they can upgrade the RAM to a usable amount for some number crunching

That would require not only jailbreak, but also a custom BIOS. They hardcoded their firmware for these specific modules (timings, voltage, frequencies etc)

It's a shame Bitmain doesn't opensource their firmware

And a breach of GPLv3 :D

They use some "xmrig-mango" binary

"3 people now with dead hashboards" from discord

live now youtube.com/watch?v=VdcK1QMDA1E

It takes 1 minute before it starts mining... Do they generate the dataset on control board and then upload it to risc-v boards? lol

They probably didn't bother implementing JIT compiler for SuperscalarHash. My quad core risc-v board takes 9 minutes to initialize the dataset.

oh, so they run it without JIT for dataset

Quality! p2pool.io/u/ef6a78845da8a060/image.png

well they'd already beat on these things for 2 years. thermal stress was bound to destroy them

I read in discord that people set fans to 50% and then VRMs overheated

so they did it to themselves? should have left the fans alone?

Probably. But consumer products shouldn't just die after poking around in the official web gui :D

why is it every unboxing video/notes I've seen all complain about it only being able to mine XMR? I mean thats what it's sold as :P

now if BitMain threw in a p2pool node and remote monero node into the X5, that would have been pretty damn nice

They advertised it as "multi currency miner"

ah well thats worth a damn slap then :|

sech1: given the excellent ROI for Bitmain, is it even worth it to hamstring Rx on the X5?

Todd doubling down on his weird takes "it's a significant step towards the creation of a Monero ASIC"

s/weird/dumb/

I have to say I'm pretty impressed how accurately sech1, hyc, and tevador guessed at what was in these things before seeing a teardown

well the SG2042R is a decent step towards creating a processor nicely suited to RandomX, but I still think if ARM was a bit more open source, it could be a better route to take

the X5 however, is more of a significant step towards bankruptcy than a Monero ASIC

<p​olar9669:matrix.org> It’s just a step towards next gen cpus

well given they (the subsidiary) are developing the chips for wider use cases, it's really a case of how much did Bitmain spend on all the other bits (e.g. RAM)

but yeah, even 2 years mining on them then offloadig at 3k per unit, they surely must have lost money on this endeavor

<p​olar9669:matrix.org> Won’t be surprised if it’s stated funded r&d for cpus

that's a good point

^^ hpcwire.com/2023/07/19/how-china-is…en-national-chip-plan-around-risc-v

<p​olar9669:matrix.org> Yup x6 might already be on the network, they need to stress test their cpus

just as rx should form part of the common CPU benchmarks, chip manufacturers should design around rx ;)

it was always hyc's retort, to build a rx asic is basically building a better cpu

<p​olar9669:matrix.org> Yup better cpus that is and small rx tweaks are fine