log in |
21)
Message boards :
Number crunching :
FMA application for windows_x86_64
(Message 2190)
Posted 2 Feb 2021 by Mr P Hucker It happens that the performances of the various versions are almost the same. Why would it need to? Unless your computer has changed, eg you upgraded it. Which should cause a re-evaluation, not sure how Boinc or the server would know you changed it though. |
22)
Message boards :
Number crunching :
FMA application for windows_x86_64
(Message 2187)
Posted 1 Feb 2021 by Mr P Hucker In Linux, my 3900X, 3950X and Epyc 7402P settled on FMA exclusively as fastest app. Takes about two weeks trying out AVX, SSE2 and FMA before declaring a winner. So the server actually calculates which is the most efficient version to send to each machine? That's cool :-) So this "two weeks" being discussed is the server seeing which works best? |
23)
Message boards :
Number crunching :
FMA application for windows_x86_64
(Message 2184)
Posted 1 Feb 2021 by Mr P Hucker In Linux, my 3900X, 3950X and Epyc 7402P settled on FMA exclusively as fastest app. Takes about two weeks trying out AVX, SSE2 and FMA before declaring a winner. How come my two fast machines (see links in last post) are almost identical? I can't see how that would change in two weeks, the timings are indistinguishable. Unless it's an OS thing, I only use Windows 10. |
24)
Message boards :
Number crunching :
FMA application for windows_x86_64
(Message 2181)
Posted 31 Jan 2021 by Mr P Hucker I just deployed the FMA optimized version of the application for Windows x64. It had problems, some time ago, with the early releases of the AMD Ryzen cpus (although, if I remember correctly, the problem were solved with BIOS upgrades). I notice FMA and AVX running on my Ryzen 9 3900XT (brand new, with brand new motherboard with latest BIOS) and my i5 8600K. The FMA ones don't seem to be any faster than the AVX tasks. Are they doing more or different calculations? Also, why with both AVX and FMA, is the Ryzen so slow compared to the i5? It should be just over 3/4s of the speed per core, but it's only 1/2 the speed. I don't know much about FMA and AVX, but from what I've read, FMA is a subset of AVX, which has me even more confused. Then there's FMA3 and FMA4, and Intel and AMD choosing one then the other, then going back again.... My Ryzen's completed tasks: http://gene.disi.unitn.it/test/results.php?hostid=63548&offset=0&show_names=0&state=4&appid= My I5's completed tasks: http://gene.disi.unitn.it/test/results.php?hostid=65175&offset=0&show_names=0&state=4&appid= |
25)
Message boards :
Science :
AMD COVID-19 HPC Fund
(Message 2038)
Posted 23 Oct 2020 by Mr P Hucker Server can't mostly support the crunching power we have now with few cpus, it is "useless" to put any effort in a GPU version.What for? It seems to manage ok now. But yes they might need a bigger one (or just upgrade it) for GPUs. It can't cost that much for some RAM or a disk or something. And if work can be completed many times faster, well that's the whole point of this science. To get the answers quickly and save lives! |
26)
Message boards :
Science :
AMD COVID-19 HPC Fund
(Message 2032)
Posted 22 Oct 2020 by Mr P Hucker Must get a better server before releasing GPU WUs. OpenCL please. I can't afford overpriced Nvidia cards. |
27)
Message boards :
Science :
AMD COVID-19 HPC Fund
(Message 2028)
Posted 22 Oct 2020 by Mr P Hucker How much do they need? How many of us are there? I'm sure a lot of us would crowd fund it. It wouldn't be very expensive between 1000 people. SETI raised quite a bit of cash by asking folk. If they can't accept charitable donations, the bureaucrats need sacked right now. That is absolutely beyond a joke. |
28)
Message boards :
Science :
AMD COVID-19 HPC Fund
(Message 1998)
Posted 11 Oct 2020 by Mr P Hucker They submitted a grant request for a new server several months. Not sure how that went. How much do they need? How many of us are there? I'm sure a lot of us would crowd fund it. It wouldn't be very expensive between 1000 people. SETI raised quite a bit of cash by asking folk. Looks like the same on other BOINC forums: Ah, thanks. I guess there's a manual somewhere. I thought it maybe wasn't a function here, since on (I think) Einstein, you get a box asking for the address and the wording. But they have a non standard setup. |
29)
Message boards :
Science :
AMD COVID-19 HPC Fund
(Message 1996)
Posted 11 Oct 2020 by Mr P Hucker They dont want to. This http://gene.disi.unitn.it/test/gene_science.php implies there's about 5 years work to do on the current project, not to mention the on hold ones. I'm sure they'd love more computing power. By the way, how do I make the URL get named in here (i.e. some of the words of this sentence become the link), is that possible with BB tags or whatever they're called? I've seen people do it in other forums but maybe they use different code. |
30)
Message boards :
Science :
AMD COVID-19 HPC Fund
(Message 1995)
Posted 11 Oct 2020 by Mr P Hucker There is no GPU app available at TN-Grid. Looking at the specs of those AMD things, they don't seem like much more than some of the volunteers I know crunching with several cards for Milkyway and Einstein. If you can get OpenCL working nicely, collectively we'll be able to do far more than that AMD thing. Have fun playing with it (or tearing your hair out trying to get the program converted! Folding@home managed, WCGrid and Rosetta didn't.) I have 19 TFlops of GPU sat here.... |
31)
Message boards :
Number crunching :
Lost credits?
(Message 1994)
Posted 11 Oct 2020 by Mr P Hucker This is very strange, we did not suffer any db crash or similar nasty things (you are also the first, if I remember correctly, that has this problem). Did you start crunching within the gridcoin group? Anyway, if you can tell me (with some proof) the number of missing credits I can add them to you manually. I've no idea how to find out how many missing credits I had. But I made posts 3 years ago in the forums, stating I ran 4 computers, so I'm sure I got some credits. It's no big deal, I don't mind if they're irrecoverable. There seems to be no evidence of my computers from back then in my account either, although I don't know how long those records stay. I can only see the current 6 I just attached. With more powerful machines now I'll get the credits back faster than I generated them in the first place! 66 cores available. At least my account's still here. Climate Prediction deleted it altogether! |
32)
Message boards :
Science :
AMD COVID-19 HPC Fund
(Message 1990)
Posted 10 Oct 2020 by Mr P Hucker Considering the original announcement of the experiment back own 2018, if I am not mistaken the dataset is a floating point matrix. Not knowing exactly what precision is needed for those computations (F16/32/64 precision), with 8 x Radeon Instinct MI 50 accelerator compute units, the peak compute capability at F16 would be 212 TFLOPs, the min at F64 at 52.8 TFLOPs. (Data according to https://www.amd.com/en/products/professional-graphics/instinct-mi50-32gb) TN Grid only has 683 users with recent credit. If it was publicised more, there would be more users. It's not in the list of projects to choose within Boinc, or here: https://boinc.berkeley.edu/projects.php I only heard about this project through a mention on another project's forum. |
33)
Message boards :
Number crunching :
Lost credits?
(Message 1988)
Posted 10 Oct 2020 by Mr P Hucker I was here 3 years ago. I just came back and found I have 0 credits? Who ate them? |
34)
Message boards :
Science :
AMD COVID-19 HPC Fund
(Message 1987)
Posted 10 Oct 2020 by Mr P Hucker There is no GPU app available at TN-Grid. So what is running on the Radeons AMD lent you? |
35)
Message boards :
Number crunching :
Error: disk limit exceeded
(Message 1121)
Posted 1 Oct 2017 by Mr P Hucker 0% must be ok, that's the default on all my machines and I've never changed it. 0 must mean infinite. Where do I find the log? P.S. Uninstalling Malwarebytes fixed the problem with all three projects. Strange as Malwarebytes didn't inform me it was blocking anything. |
36)
Message boards :
Number crunching :
Error: disk limit exceeded
(Message 1119)
Posted 30 Sep 2017 by Mr P Hucker The machine has about 20GB free (not 12GB as I misremembered above). Set to: "Use at most 0GB disk space" (which means no limit). "Leave at most 0GB disk space free" "Use at most 0% of total disk space" "Write to disk at most every 30 seconds" "Use at most 0% of page file" These are the settings it's always had, and the settings my other 3 machines have. And the problems only started on this one machine a few days ago. It's doing the same with the two other projects (SRBase and Yafu). P.S. I wasn't emailed when you replied, yet I'm subscribed to the thread. |
37)
Message boards :
Number crunching :
Error: disk limit exceeded
(Message 1117)
Posted 29 Sep 2017 by Mr P Hucker I'm getting an error "disk limit exceeded" on quite a few tasks (about 25% of them) from one computer only out of my 4. There is no disk space shortage (I have 12GB free), and I haven't specified a limit for disk space in BOINC. Any ideas what's going on? http://gene.disi.unitn.it/test/result.php?resultid=19495327 Just started in the last few days, changes made to the machine that might be the cause are: 1) Some windows updates (including a major one). 2) A different (very old 4MB) graphics card was inserted into the machine (just for display purposes, not crunching). 3) Malwarebytes changed from the free version (run on demand) to the trial full version (realtime). |