skulstation Posted January 19, 2018 Posted January 19, 2018 (edited) A small update. I did some more test whit the new gpupi3.1 , and til now i am not abel to brake it or come close to my e5645 1B record I traid it whit almost all my xeon cpu's. fsb down / up whit eleet in windosw , boot whit low / high fsb , stabel / not stabel voltages ( forcing it to bsod ) My conclusion is , i whil deleet my 1B score whit the e5645 cpu's Edited January 19, 2018 by skulstation Quote
_mat_ Posted January 20, 2018 Author Posted January 20, 2018 That's good to hear. Thank you for your testing! Quote
Administrators websmile Posted January 21, 2018 Administrators Posted January 21, 2018 GPUPI 3,1 mandatory from february 1st http://forum.hwbot.org/showthread.php?p=504138#post504138 Please all, delete your sr2 results and rebench with 3.1, we will check and delete the results still on made with older versions Quote
Crew Strunkenbold Posted January 21, 2018 Crew Posted January 21, 2018 Thx mat and skulstation for testing and fixing this! Quote
Administrators websmile Posted January 21, 2018 Administrators Posted January 21, 2018 +1 again on the thanks for this good and helpful job you both did - to all users, to make it more clear, we do not have mainboard info on all submissions, so of course we will delete scores that are doubrful/out of line at efficiency. II ask for some understanding on this because this will be no easy task - thank you in advance Quote
kmac20 Posted February 4, 2018 Posted February 4, 2018 (edited) Hello, I am trying to submit GPUPI for 100M CPU, but I cannot find my CPU to select in the calculate settings. All that I see listed are CUDA devices (GTX 1080) and OPENCL GPU DEVICES (GTX 1080). I see no option to select my Ryzen 1700. Any ideas? If someone could get back to me I'd GREATLY APPRECIATE IT. This is the only benechmark I have remaining in the current, cup, and depending on how high I score I could essentially be taking 1st-3rd place. Thank you very much in advance. I"ve searched around, seen others with the same problem. I saw that installing a certain AMD SDK could fix it? But AMD no longer offers the SDK, or rather, never offered it for Windows 10. So I'm at a bit of a loss here. Far as I know my catalyst drivers are 17.30. Not the VERY MOST recent, but those only cam out a couple days ago and are from what I understand mostly aimed at supporting the new Ryzen chips coming out soon (the ones using essentially binned dies from the first Ryzen series, but different CPUs because of different steppings/batch/etc). I'd greatly appreciate if someone could either let me know what I need to do, or at the very least point me in the right direction. Thank you all very much. Edit: and just so you know, here's what it says in GPUPI, so essentially nothing: "API: OpenCL CPU with 0 devices" Edited February 5, 2018 by kmac20 Quote
_mat_ Posted February 5, 2018 Author Posted February 5, 2018 If your CPU is not shown, there is no OpenCL driver installed, that's compatible with your Ryzen. The Catalyst driver for an AMD graphics card would normally install the OpenCL driver for CPU and GPU, but you'll need to plug-in an AMD gpu temporarily to install it. The alternative is the AMD APP SDK, as you already mentioned. It IS available for Windows 10 and installs without a problem for me. Only check the driver, uncheck the rest of SDK: https://developer.amd.com/amd-accelerated-parallel-processing-app-sdk/ Quote
kmac20 Posted February 5, 2018 Posted February 5, 2018 (edited) Unfortunately the only AMD graphics card I have is a 1900XTX which is about....oh i dont know 12 years old by now? Doubt it would do anything. One first question: when I'm picking the install features, which one is the driver? They're not labeled very clearly. Several are listed as sample source code, several are listed as pre built samples. Which would I be selecting here? Thank you for the link to the APP SDK, I couldn't find that page searching for it. I just have one more question though: will this driver mess up any other OpenCL reliant programs/etc? For example, if I install this and run cinebench in OPENCL to test a GPU, will it still run on the GPU or will it default to a CPU from now on? I'm assuming this will be fine, and I greatly appreciate your help. Thanks a bunch. Edited February 5, 2018 by kmac20 Quote
_mat_ Posted February 5, 2018 Author Posted February 5, 2018 The file you need to download is listed as: AMD-SDK-InstallManager-v1.4.87 (1).exe When the installer starts, select "Custom" and check only the "OpenCL Runtime libraries". Follow through with the install. Regarding multiple OpenCL drivers you don't need to worry. The install right next to each other, each runtime will add one ore more OpenCL platforms that you will be able to select in the benchmarks/programs that uses them. But not CineBench, that's OpenGL. Quote
havli Posted February 5, 2018 Posted February 5, 2018 I was trying the new version 3.1 today and unfortunately it is not working for me. The current PC is i5 2500k, GTX 465 (378.78 drv), windows 7 x64 SP1 and AMD OCL driver (for CPU). While old 2.3.4 version is working perfectly, the 3.1 just stops after I start the calculation. Same behavior for both OCL @ CPU and OCL @ GPU. Here is the log file: LOG START at 2018-02-05 23:31:51 ---------------------- Could not parse version string successfully: OpenCL C 1.1 OPENCL PCI address [0000:01:00.0] Could not parse version string successfully: OpenCL C 1.2 Invalid topology output type: 0 OPENCL PCI address [undefined] CUDA driver version is insufficient for CUDA runtime version Maybe it could be somehow related to Windows locale settings? I'm using english win7 with regional setting set to czech. Also does the last line means Fermi will be no longer supported (using CUDA at least)? It is sad to see the support for older HW is slowly disappearing. First GTX 200, then HD 4800, now all 32-bit SSE2-capable CPUs and perhaps even Fermi series GPUs. Quote
kmac20 Posted February 5, 2018 Posted February 5, 2018 Ah yes, I stupidly thought Cinebench was OpenCL, when its obviously OpenGL. Guess I'm so used to dealing with OpenGL that I just got it all mixed up. Thanks a bunch, I had the installer downloaded correctly, just wanted to know which options to pick under custom install so I just got what I needed. Thanks again! Quote
bolc Posted February 6, 2018 Posted February 6, 2018 given 3.1 is only useful for preventing bugs 1366/20xx sockets xeon bi xeon, why not keeping 2.3.4 alive for others....? Quote
_mat_ Posted February 6, 2018 Author Posted February 6, 2018 I was trying the new version 3.1 today and unfortunately it is not working for me. The current PC is i5 2500k, GTX 465 (378.78 drv), windows 7 x64 SP1 and AMD OCL driver (for CPU). Maybe it could be somehow related to Windows locale settings? I'm using english win7 with regional setting set to czech. Also does the last line means Fermi will be no longer supported (using CUDA at least)? It looks like an issue with HWiNFO. I will add a button to disable hardware detection to circumvent problems like this. Things like that happen with low level hardware detection and although I anticipated that, I didn't want to add the possibility for deactivation until a real problem arises. It is sad to see the support for older HW is slowly disappearing. First GTX 200, then HD 4800, now all 32-bit SSE2-capable CPUs and perhaps even Fermi series GPUs. I have a lot of love for old hardware, but it takes some time to trim a new version down to make it working again. I am currently working on GPUPI on a daily basis, that's why there has not been an official annoucement of GPUPI 3. But it's coming and and will be much better than it's predecessors. Until then please grab the hottest GPUPI versions here: GPUPI 3.1.1 | GPUPI 3.1.1 - Legacy Version The new Legacy Version works with Windows XP, supports 32 bit OS and is compatible with old OpenCL 1.1 devices. It was successfully tested with GTX 280, which is the oldest NVIDIA graphics card that supports double precision. Please be aware that you will always need to install the newest OpenCL drivers possible for these old devices. Back then CUDA and OpenCL was horribly bugged and our testing showed that there are situations where you won't be able to produce a valid result due to precision problems in old runtime libraries. The next step will be a native OpenMP API path so you will not need to install OpenCL drivers to benchmark CPUs with GPUPI. It's coming along quite nicely and will support AVX as well. My latest implementation shows that it's also faster than the OpenCL CPU path. On top of that it supports every CPU out there. Quote
_mat_ Posted February 6, 2018 Author Posted February 6, 2018 given 3.1 is only useful for preventing bugs 1366/20xx sockets xeon bi xeon, why not keeping 2.3.4 alive for others....?That's the current change log for GPUPI 3.1.1: Mixed multi GPU calculation CUDA 9.1 support with improved performance for newer graphics cards that support Warp Shuffle Supports Titan V, which is not working with GPUPI 2.3.4. Supports Xeon Phi accelerator cards Hardware detection of clock frequencies and more, that will conveniently be submitted to HWBOT Improved timer detection (EVGA SR-2 fixed, HPET not necessary and approved platforms like Skylake and above) Command line mode via GPUPI-CLI.exe Improved Legacy version for old hardware Encrypted result transaction to HWBOT via https Many many bugfixes and smaller improvements like better device name detection for AMD, HWBOT submission fixes, result file validation fixes, ... So there is much more to it than the EVGA SR-2 timer fix. GPUPI 3 is faster, more convenient for competitive benchmarking and the Legacy version should be more compatible as well. Plus it's more time consuming to maintain two versions of a product. Not only for me but for the moderators as well. So the target is to concentrate on GPUPI 3 and make it better in every aspect. If there are drawbacks for you, just let me know, because there shouldn't be. Quote
havli Posted February 6, 2018 Posted February 6, 2018 Wow, that is some great news. I wasn't aware legacy version 3.x already exists. I'm really looking forward to test GTX 200 series and other older hardware too. I like the idea of CPU computing without OCL driver, sounds really interesting. This should make the CPU version more popular and easier to use... while it isn't that hard to install OCL driver, many systems don't have it by default and some people might have problem with it. I really appreciate your work and effort to make GPUPI compatible with as wide hardware base as possible. My favorite kind of benchmark is the one which can run on old CPU but also is capable of utilizing new instruction sets and scale with modern architectures. There aren't that many unfortunately... at least that I know of - only GPUPI, Y-Cruncher and x265. I understand it takes extra time to support this much different HW, especially when you are developing the whole application. I try to keep x265 oldschool-HW ready as well, and while it is "only" a sophisticated wrapper to the encoder executable I did spend a ton of time doing pre-release testing on many different platforms. Quote
bolc Posted February 6, 2018 Posted February 6, 2018 That's the current change log for GPUPI 3.1.1: Mixed multi GPU calculation CUDA 9.1 support with improved performance for newer graphics cards that support Warp Shuffle Supports Titan V, which is not working with GPUPI 2.3.4. Supports Xeon Phi accelerator cards Hardware detection of clock frequencies and more, that will conveniently be submitted to HWBOT Improved timer detection (EVGA SR-2 fixed, HPET not necessary and approved platforms like Skylake and above) Command line mode via GPUPI-CLI.exe Improved Legacy version for old hardware Encrypted result transaction to HWBOT via https Many many bugfixes and smaller improvements like better device name detection for AMD, HWBOT submission fixes, result file validation fixes, ... So there is much more to it than the EVGA SR-2 timer fix. GPUPI 3 is faster, more convenient for competitive benchmarking and the Legacy version should be more compatible as well. Plus it's more time consuming to maintain two versions of a product. Not only for me but for the moderators as well. So the target is to concentrate on GPUPI 3 and make it better in every aspect. If there are drawbacks for you, just let me know, because there shouldn't be. oh wow that sums it all, amazing job ! will try 3.1.1 on older hd4xxx again Quote
bolc Posted February 7, 2018 Posted February 7, 2018 Hi in this example I run the regular version but it says legacy in the bench windows. Does the legacy mode auto loads? any advantages to run 3.1.1 legacy executable over the standard, if detection goes fine ? http://hwbot.org/submission/3779554_bolc_gpupi_for_cpu___100m_pentium_4_650_5min_5sec_741ms Quote
yosarianilives Posted February 7, 2018 Posted February 7, 2018 That's the current change log for GPUPI 3.1.1: [*]Supports Xeon Phi accelerator cards Will be interesting to see the results of this one, I've always been interested in xeon phi as kind of this wacky thing from intel, as a sort of solution without a problem. The newest one was the only one that I could find a real use for, and that's only because it finally has "real" cpu cores so would work good for a vm server or something. Quote
QuantumX Posted February 8, 2018 Posted February 8, 2018 (edited) Is 980 Ti also legacy? The normal version does not detect it as a CUDA device only OpenCL. In the legacy version it shows up as a CUDA device but the calculation is 5 seconds slower than in 2.3.4 Edited February 8, 2018 by QuantumX Quote
_mat_ Posted February 8, 2018 Author Posted February 8, 2018 (edited) No, the GTX 980 Ti is not legacy. It only uses the reduction type "Legacy", that's why it is written in the same line as the reduction size. If your device is capable of "Warp Shuffle" for reductions, the type will state this. Just to clarify: There is no distinct seperation between legacy hardware and modern hardware. GPUPI takes what it gets and uses it to perform the best with it. The Legacy version exists because it is compiled with CUDA 6.5, which is the last CUDA toolkit to support the GTX 200 series cards (Compute Capability 1.3). The normal version is compiled with the latest CUDA toolkit (currently 9.1) which supports Compute Capability 3.0 onwards. So the legacy version has to be used for all GeForce cards between GTX 200 and GTX 500 series. All other cards should use the normal version. As for the GTX 980 Ti not being listed with GPUPI 3, I will look into it. Edited February 8, 2018 by _mat_ Quote
bolc Posted February 8, 2018 Posted February 8, 2018 No, the GTX 980 Ti is not legacy. It only uses the reduction type "Legacy", that's why it is written in the same line as the reduction size. If your device is capable of "Warp Shuffle" for reductions, the type will state this. Just to clarify: There is no distinct seperation between legacy hardware and modern hardware. GPUPI takes what it gets and uses it to perform the best with it. The Legacy version exists because it is compiled with CUDA 6.5, which is the last CUDA toolkit to support the GTX 200 series cards (Compute Capability 1.3). The normal version is compiled with the latest CUDA toolkit (currently 9.1) which supports Compute Capability 3.0 onwards. So the legacy version has to be used for all GeForce cards between GTX 200 and GTX 500 series. All other cards should use the normal version. As for the GTX 980 Ti not being listed with GPUPI 3, I will look into it. Thanks for the clarification For AMD, any rules as to which version to use for HD 4xxx 5xxx 6xxx etc ? Quote
_mat_ Posted February 8, 2018 Author Posted February 8, 2018 The solution for the GTX 980 Ti and similiar problems is easy: Just install the newest GeForce drivers. Old ones are not compatible with the new CUDA toolkit. You will very soon have an extensive debug log on your side to aid you with internal problems of device detection. It will come with GPUPI 3.2. As for old Radeon cards: That depends heavily on your driver situation. If you have an OpenCL 1.2 and higher driver installed on your system, you can use the normal GPUPI version. Otherwise it will give you an error like "cl... function not found in OpenCL.dll". If you have old OpenCL 1.1 drivers, you will need the Legacy version. Quote
bigblock990 Posted February 12, 2018 Posted February 12, 2018 Hi, seems to be a cpu frequency bug in gpupi 3.1 See here: gpupi reporting 5472mhz, bench ran at 5740mhz http://hwbot.org/submission/3783579 https://d1ebmxcfh8bf9c.cloudfront.net/u117988/image_id_1978470.jpg Thanks Quote
_mat_ Posted February 12, 2018 Author Posted February 12, 2018 Thanks guys, I will look into it! Quote
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.