Message boards :
News :
GPU app - beta version for linux nvidia
Message board moderation
Author | Message |
---|---|
Send message Joined: 8 Jul 11 Posts: 1344 Credit: 532,710,280 RAC: 546,005 |
After all these years, we finally have our first GPU app. It's only a beta version for 64bit linux with Nvidia GPUs. Support for other platforms and GPUs will be coming soon. If you'd like to help test this app, you will need to check the "run test applications" box on the project preferences page. I generated a special batch of work for this app from some older WUs that I have truth for. This will help to find any potential bugs that are still there. A few potential issues: 1. This was built with the Cuda SDK version 10.1, so it uses a relatively new Nvidia driver version and only supports compute capability 3.0 and up. If this affects too many users out there, I will need to rebuild with on older SDK. 2. I was not able to build a fully static executable, but I did statically link the ones most likely to be a problem (i.e. pari, gmp, std c++) Please report any problems. I am still relatively new to the whole GPU app process, so I am sure there will be issues of some kind. Also, feel free to leave comments regarding what platform, GPU, etc I should concentrate on next. I was thinking I would attack linux OpenCL (i.e. ATI/AMD) next as that should be a quick port of what I did with Nvidia. I think the windows port will take much longer, since I normally use mingw to cross-compile but I don't think that's compatible with the nvidia compiler. |
Send message Joined: 25 Oct 18 Posts: 15 Credit: 112,744,248 RAC: 313 |
Not getting any WU for GTX1080 on Linux Mint 19 : "NVIDIA GeForce GTX 1080 (4095MB) driver: 390.11 OpenCL: 1.2" Perhaps the driver is too old? Ditto on Ubuntu 18.10 - same hardware, same driver |
Send message Joined: 8 Jul 11 Posts: 1344 Credit: 532,710,280 RAC: 546,005 |
Not getting any WU for GTX1080 on Linux Mint 19 : "NVIDIA GeForce GTX 1080 (4095MB) driver: 390.11 OpenCL: 1.2" Yeah, I set the minimum driver to 418.39 in the plan_class, as that is what got installed with the SDK. I'm not sure if these are backward compatible or not, but thought I would error on the safe side. If enough people have older drivers I might just have to build from an older SDK. I'll wait to see what the concensus is. |
Send message Joined: 1 Feb 17 Posts: 23 Credit: 62,059,567 RAC: 5,617 |
No wonder I couldn't get anything with 396.51. I didn't even know NV drivers were up to 418. |
Send message Joined: 25 Oct 18 Posts: 15 Credit: 112,744,248 RAC: 313 |
Not getting anything with "NVIDIA GeForce GTX 1080 (4095MB) driver: 418.43 OpenCL: 1.2" on Ubuntu 18.10 either Are there definitely WU available? |
Send message Joined: 13 Mar 19 Posts: 2 Credit: 220,432,972 RAC: 2,009 |
I got one and it ran fine. |
Send message Joined: 8 Jul 11 Posts: 1344 Credit: 532,710,280 RAC: 546,005 |
Not getting anything with "NVIDIA GeForce GTX 1080 (4095MB) driver: 418.43 OpenCL: 1.2" on Ubuntu 18.10 either Sorry, it looks like they were gobbled up pretty quickly. I just put a bunch more out. So far no compute errors... |
Send message Joined: 20 Dec 14 Posts: 17 Credit: 12,153,123 RAC: 0 |
My recommendation would be to work on OpenCL for Nvidia, AMD, and Intel. When Nvidia's Pascal GPUs came out, CUDA was not supported for several months until CUDA 8.0 shipped. However, OpenCL was supported from launch on those GPUs. |
Send message Joined: 8 Jul 17 Posts: 10 Credit: 40,060,092 RAC: 1,171 |
Thanks Eric. From looking at your GTX 1050 results, the credit looks super-awesome. I think you ran roughly do like 690K PPD. Keep that credit score high :) |
Send message Joined: 23 Aug 11 Posts: 7 Credit: 35,760,148 RAC: 390 |
Not getting any WU for GTX1080 on Linux Mint 19 : "NVIDIA GeForce GTX 1080 (4095MB) driver: 390.11 OpenCL: 1.2" Please build for older drivers also. I'm sure there are many, like myself, with older cards that would love to help but the 418.xx drivers aren't supported for our cards. |
Send message Joined: 25 Oct 18 Posts: 15 Credit: 112,744,248 RAC: 313 |
Thank you Eric - got some overnight so please do keep them coming |
Send message Joined: 8 Jul 11 Posts: 1344 Credit: 532,710,280 RAC: 546,005 |
How old are we talking? After compute capability 3.0? What version OpenCL does your card support? |
Send message Joined: 30 Sep 18 Posts: 1 Credit: 235,682 RAC: 0 |
First of all: Great work Eric! Here's another challenge for you: 2 x AMD Radeon HD 7800 Series - doesn't seem strong, but I play in FullHD mostly with 60 FPS in high quality ;-) Windows 10 Intel Processor I would be happy to test for you if you dare to try Windows and AMD. |
Send message Joined: 7 Apr 15 Posts: 4 Credit: 18,772,724 RAC: 0 |
Hello :) I waited a long time for GPU app. Great job! I don't getting any GPU tasks. I have drivers version 418.39. Linux 4.13.16-hiveos. My machine: https://numberfields.asu.edu/NumberFields/show_host_detail.php?hostid=1465179 5: 23-Mar-2019 10:04:14 (low) [] CUDA: NVIDIA GPU 0: GeForce GTX 1060 3GB (driver version 418.39, CUDA version 10.1, compute capability 6.1, 3019MB, 2943MB available, 4053 GFLOPS peak) 6: 23-Mar-2019 10:04:14 (low) [] CUDA: NVIDIA GPU 1: GeForce GTX 1060 3GB (driver version 418.39, CUDA version 10.1, compute capability 6.1, 3019MB, 2943MB available, 4053 GFLOPS peak) 7: 23-Mar-2019 10:04:14 (low) [] CUDA: NVIDIA GPU 2: GeForce GTX 1060 3GB (driver version 418.39, CUDA version 10.1, compute capability 6.1, 3019MB, 2943MB available, 4053 GFLOPS peak) 8: 23-Mar-2019 10:04:14 (low) [] CUDA: NVIDIA GPU 3: GeForce GTX 1060 3GB (driver version 418.39, CUDA version 10.1, compute capability 6.1, 3019MB, 2943MB available, 4053 GFLOPS peak) 9: 23-Mar-2019 10:04:14 (low) [] CUDA: NVIDIA GPU 4: GeForce GTX 1060 3GB (driver version 418.39, CUDA version 10.1, compute capability 6.1, 3019MB, 2943MB available, 4053 GFLOPS peak) 10: 23-Mar-2019 10:04:14 (low) [] CUDA: NVIDIA GPU 5: GeForce GTX 1060 3GB (driver version 418.39, CUDA version 10.1, compute capability 6.1, 3019MB, 2943MB available, 4053 GFLOPS peak) 11: 23-Mar-2019 10:04:14 (low) [] CUDA: NVIDIA GPU 6: GeForce GTX 1060 3GB (driver version 418.39, CUDA version 10.1, compute capability 6.1, 3019MB, 2943MB available, 4053 GFLOPS peak) 12: 23-Mar-2019 10:04:14 (low) [] CUDA: NVIDIA GPU 7: GeForce GTX 1060 3GB (driver version 418.39, CUDA version 10.1, compute capability 6.1, 3019MB, 2943MB available, 4053 GFLOPS peak) And I getting this: 74: 23-Mar-2019 10:17:03 (low) [NumberFields@home] Sending scheduler request: Requested by user. 75: 23-Mar-2019 10:17:03 (low) [NumberFields@home] Requesting new tasks for NVIDIA GPU 76: 23-Mar-2019 10:17:05 (low) [NumberFields@home] Scheduler request completed: got 0 new tasks 77: 23-Mar-2019 10:17:05 (user notification) [NumberFields@home] This project is using an old URL. When convenient, remove the project, then add http://numberfields.asu.edu/NumberFields/ |
Send message Joined: 14 Sep 15 Posts: 2 Credit: 35,763,335 RAC: 0 |
First, YEAH!!!!! A new GPU project! Cool and thanks!!!! Driver version 4xx.xx (at least for winblows) is the first driver to include support for the 2000 series RTX GPUs, and at least initially, it severely hampered the performance of the 10-series cards. I do not think many of your 10-series users will be willing to try a 4xx.xx driver version anytime soon. EDIT: I decided to do my homework (AFTER I have already run my mouth :D ), and perhaps just the first few 4xx.xx drivers hurt the 10-series, I guess I will be updating my driver now to find out for myself. :) |
Send message Joined: 1 Feb 17 Posts: 1 Credit: 17,259,519 RAC: 0 |
If you want to target users that are running the "standard" drivers from popular distros like Ubuntu (LTS), Mint, and Debian, then you will probably need to build the app with the CUDA 9.1 Toolkit. Apps built with the older toolkit will still be compatible with newer GPUs. Below is a table from the CUDA Toolkit docs that lists Toolkit version with minimum driver requirements. Not sure which libraries you are using for your application, so using the older toolkit may or may not affect performance. CUDA Toolkit and Compatible Linux Driver Versions CUDA 10.1.105 >= 418.39 CUDA 10.0.130 >= 410.48 CUDA 9.2 (9.2.148 Update 1) >= 396.37 CUDA 9.2 (9.2.88) >= 396.26 CUDA 9.1 (9.1.85) >= 390.46 CUDA 9.0 (9.0.76) >= 384.81 CUDA 8.0 (8.0.61 GA2) >= 375.26 CUDA 8.0 (8.0.44) >= 367.48 CUDA 7.5 (7.5.16) >= 352.31 CUDA 7.0 (7.0.28) >= 346.46 |
Send message Joined: 23 Aug 11 Posts: 7 Credit: 35,760,148 RAC: 390 |
Compute capability 2.0 OpenCL 1.1 If there isn't enough interest, no problem. Thanks for all you do. |
Send message Joined: 21 Sep 11 Posts: 9 Credit: 14,403,447 RAC: 3,772 |
For those needing a list of generations and compute capability https://en.wikipedia.org/wiki/CUDA Supported CUDA level of GPU and card. See also at Nvidia: CUDA SDK 1.0 support for compute capability 1.0 – 1.1 (Tesla)[20] CUDA SDK 1.1 support for compute capability 1.0 – 1.1+x (Tesla) CUDA SDK 2.0 support for compute capability 1.0 – 1.1+x (Tesla) CUDA SDK 2.1 – 2.3.1 support for compute capability 1.0 – 1.3 (Tesla)[21][22][23][24] CUDA SDK 3.0 – 3.1 support for compute capability 1.0 – 2.0 (Tesla, Fermi)[25][26] CUDA SDK 3.2 support for compute capability 1.0 – 2.1 (Tesla, Fermi)[27] CUDA SDK 4.0 – 4.2 support for compute capability 1.0 – 2.1+x (Tesla, Fermi, more?) CUDA SDK 5.0/5.5 support for compute capability 1.0 – 2.1+x (Tesla, Fermi, more?) CUDA SDK 6.0 support for compute capability 1.0 – 3.5 (Tesla, Fermi, Kepler) CUDA SDK 6.5 support for compute capability 1.1 – 5.x (Tesla, Fermi, Kepler, Maxwell). Last version with support for compute capability 1.x (Tesla) CUDA SDK 7.0/7.5 support for compute capability 2.0 – 5.x (Fermi, Kepler, Maxwell) CUDA SDK 8.0 support for compute capability 2.0 – 6.x (Fermi, Kepler, Maxwell, Pascal). Last version with support for compute capability 2.x (Fermi) CUDA SDK 9.0 – 9.2 support for compute capability 3.0 – 7.2 (Kepler, Maxwell, Pascal, Volta) CUDA SDK 10.0/10.1 support for compute capability 3.0 – 7.5 (Kepler, Maxwell, Pascal, Volta, Turing) I for one have mostly older cards and would like to see at least Fermi and newer support. Realistically, I would like to see all CUDA capable cards supported as there are a lot of systems out there with entry level cards to give servers video like the 8400 GS and 210's. However, many wont logically choose to run on those cards if efficiency is brought up. |
Send message Joined: 1 Feb 17 Posts: 23 Credit: 62,059,567 RAC: 5,617 |
First, YEAH!!!!! A new GPU project! Cool and thanks!!!! I recall hearing that somewhere as well and a reason I installed just 396 on my 1080Ti. Pls report back if it's still the case. Attempting to install 418 broke my Ubuntu install and I don't recall the exact OC clocks on the Ti but it's performing just as well on E@H with 418 vs 396 after the Ubuntu reinstall. The E@H frequency is constantly changing which changes the compute time slightly so if there is a driver difference it seems small with OpenCL. Comparing the 11 tasks I received here at NF and with a random Ti at Collatz, the RAC is about 1/3 at Collatz which is known to be a very high RAC project. Once people realize the RAC possibilities here, expect more members to flock here. Especially with the limited # of GPU projects. |
Send message Joined: 8 Jul 17 Posts: 10 Credit: 40,060,092 RAC: 1,171 |
Once people realize the RAC possibilities here, expect more members to flock here. Especially with the limited # of GPU projects. Agree. This will attract more crunchers, some who may love the high RAC and also good for the project admin (project completed quicker), so win-win for both parties. Further, I think the project admin (Eric) did a fantastic job in last year pentathlon and Formula Bonic sprints. Looking at BOINCstats, NumberFields currently has 390 active crunchers versus 1897 in Collatz. Imagine if the number of active users doubled or perhaps more if the apps can support both Linux and Windows plus older AMD/Nvidia GPU cards. I know there will be lots of work ahead but this is definitely a good start. |