The Scottish Boinc Team

I am looking for a project that will give an old AMD card something to do - it is an R7-350 2Gb (Cape Verde)

Collatz will run but have tried various optimisations which just crash so it jugs along at the default config (37K points in approx 2.5 hrs)
Moo Wrapper doesn't seem to be dishing out work for AMD

None of the other projects in the menu have the ATI logo so I wondered if there was anything else I could try for fun. If it helps with an FB Marathon then so much the better

einsiein or milkyway?

Okay thanks - I'll try them. I realise Collatz probably scores higher but I'm getting big figures on that from other machines

I'd definitely try MW on the AMD. They pay better on that than Nvidia of the same vintage.

Einstein pays better on AMD but for a GPU project it pays poorly.

So far ...

• Einstein completely crashes the system (twice)
  main circuit breaker tripped - probably unrelated
  Milkyway gets Computation errors
  Moo Wrapper isn't handing out AMD GPU WU

Off to bed leaving it running Collatz for an easy life

Enigma also favors AMD I think

I haven't run Enigma since I hit my goal. Aren't there optimized apps for it or is that "old" information?

Ha hah ha your old card is far newer than my new cards oh well shows how times move on.

For sure Milkyway is a good option as the extra double floating point units on the ATi benefits over nVidia as mentioned.

You mentioned trying to tweak the config. I managed to double my output last night on mine using: -

kernels_per_reduction=48
threads=8
lut_size=17
sieve_size=30
sleep=0
cache_sieve=1
reduce_cpu=0

Maybe those will work on your 370 as well?

Oh and Primegrid does GPU on AMD but not sure what the scores are like.

Thanks Jock - will try that later

Milkyway also gives computation errors. I've got something badly wrong in this machine : fresh lubuntu install, default amdgpu drivers, boinc for amd/opencl. It was all so very easy for Collatz (but even that doesn't like the usual optimisations)

If you want to run new Moo! WUs:

• Detach
• Re-attach
• enable CPU in preferences on account page
• enable test applications

Bryan wrote: ↑Tue Nov 27, 2018 3:37 am I haven't run Enigma since I hit my goal. Aren't there optimized apps for it or is that "old" information?

No optimized app, just speaking on my experience by comparing my 1070ti VS my RX 580. Not 100% sure on this, but if you look at the top computers, most of them have AMD cards.
Collatz favors Nvidia ... about 2.5x faster with optimized app. This is why I put my AMD cards to E@H and Enigma.
My 1070ti is producing almost the same as bluestang's Vega on Collatz. 6.4m VS 6.7m

No optimized app, just speaking on my experience by comparing my 1070ti VS my RX 580. Not 100% sure on this, but if you look at the top computers, most of them have AMD cards.
Collatz favors Nvidia ... about 2.5x faster with optimized app. This is why I put my AMD cards to E@H and Enigma.
My 1070ti is producing almost the same as bluestang's Vega on Collatz. 6.4m VS 6.7m

So for AMDs are you saying it is more lucrative that I move them from Collatz to E@H and Enigma?

Of course milage may vary :)

I guess I will have to fire the second pair of identical cards up and compare...

I could certainly be mistaken, but I don't think any GPU project is going to compete with Collatz as far as credits go. Especially not E@H.

... and I just came on to post that Collatz is the only project which doesn't give Computation errors on AMD GPU

different mobo
different GPU (R7 350 and Rx 580)
different OS (Ubuntu 18.04 and Lubuntu 18.10)

If anyone is crunching Einstein or Milky Way etc on Linux AMD GPU then please do tell me which driver. I'd like to fix it so that I can help on sprints next season

davidBAM wrote: ↑Tue Nov 27, 2018 8:57 pm ... and I just came on to post that Collatz is the only project which doesn't give Computation errors on AMD GPU

different mobo
different GPU (R7 350 and Rx 580)
different OS (Ubuntu 18.04 and Lubuntu 18.10)

If anyone is crunching Einstein or Milky Way etc on Linux AMD GPU then please do tell me which driver. I'd like to fix it so that I can help on sprints next season

Problem with AMD cards/Intel IGPs is that the Open Source driver (all proprietary Intel and AMD drivers are deprecated in the latest Linux builds) is not complete and that in a lot of cases the card will not be detected at all, or only partially. Linux Mint 19.0 sees the Intel UHD 605 of my Pentium J5005 as 'Intel Corporation Device 3184', and BOINC sees no usable GPU.
The IGP of my AMD A12-9800E wasn't seen either by Linux Mint 19.0
When you have AMDs latest and finest the GPU is recognized by the last Linux kernels -but the OpenCL drivers may not be complete, thus preventing proper working in BOINC.
You can try your luck with the AMDGPU PRO driver

Thank you.

Oddly enough, I was just reading this link about an hour ago ... https://www.phoronix.com/forums/forum/l ... ctionality

Is there a recognised repository for amdgpu-pro or does that link look promising?

davidBAM wrote: ↑Tue Nov 27, 2018 11:52 pm Thank you.

Oddly enough, I was just reading this link about an hour ago ... https://www.phoronix.com/forums/forum/l ... ctionality

Is there a recognised repository for amdgpu-pro or does that link look promising?

That is the latest AMPGPU pro driver.

jockmacmad2 wrote: ↑Tue Nov 27, 2018 8:40 am Ha hah ha your old card is far newer than my new cards

Have you many many of them or have you been out shopping before you can turn in 20M credits / day

Yes I have lots of them like boxes of them and nearly all dual-gpu

For the 20 mil though I borrowed some off a man I know for about 6-7 hours.

Seems like a useful sort of a chap to know. It has done wonders for your WCG returns too

As to your question on AMD. Has any one looked at https://rocm.github.io/ROCmInstall.html ??

Seems like this is being used on the 'official' Docker BOINC images for AMD GPU crunching.

I am going to take a look at the whole BOINC in Docker setup, https://github.com/BOINC/boinc-client-docker, when I get home although I think all my nVidia & AMD cards are simply too old to test anything GPU wise.

I think my mate was looking at that but didn't get anywhere with it. Dirk might be in the know.

I've just put a pair of RX580 into a machine and I see one of them is showing 81C - is that too hot (not rising above that). The other one is 20C cooler but I suspect the hot one is the one without much gap to feed the fans. Case is open of course

When do they start missing cycles to prevent overheating?

My AMD usually run around 78-80C with 38% fan. I use a custom fan profile on Windows so that if they go up to around 82C the fan goes up to 50% and at 85C goes to 100%. They were bought in 2010 and ran flat out for 4 years at that and are running again now. So I would say from experience that is a normal temperature for them running Collatz.

Off the top of my head there are certain things that cause Milkyway failures.
Firstly the GPU must be Double Precision, some older GPU's and mobile GPU's are not.
OpenCL must work.
A lot of older chipped cards can not run the nbody tasks. I remember having to run all my 69xx series cards under old drivers or they don't work. Seperation tasks should work though.
Is collatz using openCl or Cuda ?