<pauliouk> "yup theres a connection that..." <- raffle was shot, added some sanity checks, testing continues 😉

It has probably been mentioned here before, but both CryptonightR and RX are included in Techpowerup benchmarks: techpowerup.com/review/amd-ryzen-th…e-asus-rog-zenith-ii-extreme/8.html

I lobbied with several review sites to add these to their benchmark suites

heh

Well, RandomX is legit a thorough test of CPU+RAM performance

it is!

and its something that quite a lot of people would like to see benchmarks of before making purchases

there's sites like betterhash that'll show the various processors and their submitted speeds, but there's no guarantee someone hasn't just photoshopped a screenshot before submission

I'd quite like to see a more confirmed way of hashrates being provided - such as mining to a side chain until enough information on the hashrate has been collected

xmrig benchmarks are reliable

reliable, but can be edited

even so pauliouk mining to a side chain can be done using multiple miners

yup I thought about that after clicking submit heh

I guess it would take a custom version of xmrig, setup to mine to say p2pool, but then xmrig would submit the findings to the remote host with some sort of validation

it's all client side pauliouk

it can be changed

I suppose with a digitally signed app you can prove that the client hasn't been tampered with

that is to trust the signature

you can still run whatever, and change the results

for it to be trusted you would need to run it from start to end in secure mode, from bootloader to OS/application

no.... not if the client generates the signed result

the client can then reverse engineer such result, and change it to something else and sign

but sure, you could run it inside a VM with a slowed down clock

to get massive ops per econd

second

oh yeah didn't even think there is no certified time, you could at least get around that with the centralized clock signatures

well... benchmarking sites have been around for ages, nobody is doing that

well, multiple users provide their own :)

xmrig validates benchmarks by running always online - it checks the time passed between sending a RandomX seed and receiving the final hash to confirm hashrate

vs one single tester that has a prototype

couldn't that be faked by just throwing external hashrate at it?

though indeed, effort

sech1: what do you mean, it checks the time taken for every individual hash iteration?

or for a batch of N iterations, same thing

no, to calculate 1M hashes

ok

so it confirms the hashrate, but it's possible to submit wrong number of threads or CPU name

*to report

yeah, you could tweak the source to report whatever you want

and I guess there have been some instances of people "leaking" faked benchmark results

claiming to be engineering samples of processors etc. but totally bogus

I'm a bit sad nobody has beat my xmrig score yet

for a specific processor, or the AMD sample Inge ?

I'm fairly sure I have a record on xmrig for the lowest hashrate from a AMD processor :)

poor old athlon didn't stand a chance.

I still hold a record for Ryzen 5 5600X hashrate there

8921.88 h/s. That machine can do 8830 h/s 24/7 with my daily overclock

DataHoarder: Threadripper 3970x yeah

<@hyc> I lobbied with several review sites to add these to their benchmark suites

<@hyc> Well, RandomX is legit a thorough test of CPU+RAM performance

you should try contacting hwbot again

now that have a guy doing maintenance on it, not like november last year

sech1: I see quite a few benchamrks include the motherboard in the benchmark, but about as many do not...

sech1: is that different versions, or some motherboards can't be ID'ed or what?

looks like it changed quite a bit. for mine it includes which dimm slots are populated and stuff

differing versions

there was a lot of benches already before they started grabbing smbios info

also, non root runs

also, non root runs on linux i think wont include smbios data

dmidecode thingy blah blah

on closer inspection, looks like someone managed to get in right below my top spot

I thought I had like 8 of the top 10 but there IS some competition there

i cant even grab top spot on my cpus

but hey, atleast i still probably has the highest freq memory ran there

got 3 new outputs since yesterday

heh, all 3 this morning, btw 3-4am

it's the morning rush :)

3 days in a row with block-heavy mornings

interesting. maybe someone else's hashrate goes offline then

well if anything we get more then, but we were at 34MH/s for a while now

<\x> "i cant even grab top spot on..." <- bad cas?

also can someone explain why sometimes my difficulty is like 1/800th the network hashrate which I think is default, and sometimes its enough to solve shares every minute? kinda weird

It is a random process :)

although if you restart it might mine at lower difficulty while it syncs