Posts by Wailing Angus Beef

21) Message boards : Number crunching : Amount of CPU time on GPU WUs (Message 71582) Posted 3 Jan 2022 by Wailing Angus Beef Post: Engineering sample CPUs are cheaper but there are usually different versions of the samples. They don't have the same clocking specs as production CPUs (usually lower), there may be some bugs and they don't work in all motherboards. Research well before buying any.
22) Message boards : Number crunching : Amount of CPU time on GPU WUs (Message 71580) Posted 3 Jan 2022 by Wailing Angus Beef Post: If it does have multiple devices, it's not like a 7990 which actually appeared as 2 GPUs to your computer and BOINC. I think the K80 tesla is like that. I think the P100 and later teslas can be partitioned in a way so they can be shared in a data center environment but I'm not sure how that's done.
23) Message boards : Number crunching : Amount of CPU time on GPU WUs (Message 71578) Posted 3 Jan 2022 by Wailing Angus Beef Post: System has dual Xeon V4 engineering sample CPUs in it, 28 threads per CPU, 56 threads total for system. Don't know why it reports 12GB and 4GB. As for 12GB instead of 16GB, there are both 12GB and 16GB model P100s. Something else I don't understand, if you look at the task info it says it found 2 CL devices. On WUs for all other devices I've only seen 1 CL device reported.
24) Message boards : Number crunching : Amount of CPU time on GPU WUs (Message 71574) Posted 2 Jan 2022 by Wailing Angus Beef Post: What controls the amount of CPU time used for the GPU WUs? I have a P100 in system running linux Mint 20.1 using the 470.86 driver and it uses practically a full CPU thread for each WU. Here's the host tasks...https://milkyway.cs.rpi.edu/milkyway/results.php?hostid=906244 I also have a Titan V running in a nearly identical system also linux Mint 20.1 using the 470.86 driver but only uses about 1/3 of a CPU thread. Why? I understand the differences in the 2 GPUs but what controls how much CPU is used? I've also noticed the Titan V uses up to 1266MiB GPU memory per WU while the P100 only uses up to 1066MiB. Both GPUs have 12GB of memory. Thanks, scole of TSBT
25) Message boards : Number crunching : Nvidia Tesla P100 (Message 71568) Posted 1 Jan 2022 by Wailing Angus Beef Post: I got it all worked out a while back. I only had a couple different systems to try the P100 in. I could not get it to work in a Z77 mobo with a i7-3770 CPU. Under windows the device manager said not enough resources. Under linux it never would recognize the GPU. Installed it in a C612 chipset mobo with Xeon V4 CPU and it works fine under windows and linux. As for cooling, I had originally planned to take the cover off the heatsink and zip tie a couple 120mm fans to it but the cover is all sealed up with no way to get the heatsink open to zip tie fans to. That was easy to deal with though. I just cobbled together fan shroud to hold a 120mm fan which blows air through the cooler. I used a Noctua fan which moved over 100 cfm with high static pressure. Keeps GPU temps 65C or less.
26) Message boards : Number crunching : Nvidia Tesla P100 (Message 71279) Posted 26 Oct 2021 by Wailing Angus Beef Post: Do you know what mobo/cpu/os combo was used?
27) Message boards : Number crunching : Nvidia Tesla P100 (Message 71268) Posted 25 Oct 2021 by Wailing Angus Beef Post: Does anyone own or have ever installed a Nvidia P100 PCIe GPU in a system and got it working? I realize it doesn't have an video outputs and it only has a passive cooler but I'm curious what hardware/OS requires there are to install and use one these. Thanks, Steve
28) Message boards : Number crunching : Updated GPU Requirements (Currently not supporting GPU tasks) (Message 68133) Posted 10 Feb 2019 by Wailing Angus Beef Post: Thanks for that explanation. And I will compare those keeping in mind the results you found. Regarding the GPUs with passive cooling, I saw where people remove the shroud and zip tie a couple case fans on the heat sink.
29) Message boards : Number crunching : Updated GPU Requirements (Currently not supporting GPU tasks) (Message 67992) Posted 8 Jan 2019 by Wailing Angus Beef Post: For Milkyway@Home, GPUs with high DP rule. Einstein@Home uses some DP also but not as much as this project. Cooling is going to be the challenge with a Tesla. As the GPU temps rise above the 60C-70C range it will begin to downclock a little and if they keep going up it will keep downclocking and you will never get the performance you expected. That was my experience messing around with an AMD FirePro S9100 which had passive cooling. Also keep in mind the K80 is really 2xK40 GPUs on a single board so the system and Boinc will see it as two separate GPUs. This may not matter on all projects but those where there's a lot of CPU to GPU communication, like Einstein, the number of PCIe lanes available to each GPU makes a little bit of a difference. A K80 will be 2 GPUs using just 16 PCIe lanes. Another option to get the same processing power as a K80 would be to try to snag a couple Titan Black GPUs. They are the same Kepler chip family as the K40/K80, the Titan Black has a cooler with a fan and if you put them in a LGA 2011 or new system, each GPU should have 16 PCIe lanes. Just food for thought. Of course, if money were no obstacle, get a Titan V!
30) Message boards : Number crunching : Updated GPU Requirements (Currently not supporting GPU tasks) (Message 67990) Posted 7 Jan 2019 by Wailing Angus Beef Post: Sorry for being difficult. But it looks like DP FPU is more useful here. I plan to buy something to go with my RTX 2070, But from my reading a K80 has 4 times more DP. Will it work for your work? I can't verify a K80 will work but I wonder how do you keep it cool? It has passive cooling and you need to force air through it. What do you plan on spending? For the money there may be better options.
31) Message boards : Number crunching : AMD Firepro GPUs (Message 66572) Posted 26 Aug 2017 by Wailing Angus Beef Post: Upon further research, it appears there were issues with getting the Hawaii GPUs to crunch. Did that ever get resolved are they still a no-go? This is the R9 390/R9 390X and Firepro S9100 series GPUs.
32) Message boards : Number crunching : AMD Firepro GPUs (Message 66571) Posted 26 Aug 2017 by Wailing Angus Beef Post: Will the AMD Firepro server type GPUs, like Firepro S9100, run Boinc/Milkyway apps?
33) Message boards : Number crunching : Android/ARM WUs all failing with computation errors (Message 65593) Posted 4 Nov 2016 by Wailing Angus Beef Post: http://milkyway.cs.rpi.edu/milkyway/results.php?hostid=606245

Previous 20