Posts by vseven

1) Message boards : Number crunching : GPU Utilization (Message 1112) Posted 3 Apr 2019 by vseven Post: No, no errors., Just slow and pegs my CPU. Right now I'm running 44 WU across 4 GPU's on a 48 CPU core machine. No errors.
2) Message boards : Number crunching : GPU Utilization (Message 1092) Posted 20 Feb 2019 by vseven Post: The problem is you then need more CPU cores. I'm running graphics cards that can do 16 WU at a time at a 22 kernel but I don't have the CPU cores to support it. So I have it set to 23 and can run 11 on a 12 CPU core machine.
3) Message boards : Number crunching : ETA of (search up to 10^20) (Message 1088) Posted 17 Feb 2019 by vseven Post: No, not at all. As long as there doing good work then they should be kept. I just know with how long CPU tasks are taking that a large amount going to those computers are probably timing out. If they are still completing over eight thousand tasks a day then that's great for the project. I would worry though as time goes on if they aren't completing all of those but they will move to other projects. And with that said maybe the deadline on work units needs to be increased not only so that doesn't happen with them but also for more casual users they won't get things timed out.
4) Message boards : Number crunching : ETA of (search up to 10^20) (Message 1084) Posted 16 Feb 2019 by vseven Post: Also as a side note if you look at the top participants of this program you're going to see a bunch for charity engine. This is a program, running on a modified boinc client, that runs on people's computers with the intention of generating gridcoin and then they give out prizes to the participants randomly. Feel free to look them up. Now I personally also crunch for gridcoin but the issue is almost every single computer under the charity engine umbrella is going to be a CPU only machine. And if CPU work units are not finishing in the allocated time that means the literally thousands and thousands of work units being sent to those people's computers are simply timing out. then once they time out they become available again. I believe this is why there are so many work units in process, the majority are never going to be completed. GRCPool, another in the top participants, I can guarantee has a lot of users that have gpus and won't have this problem. Just something else to throw in here. Would also be nice to hear from an admin from this project about these issues or see if anything can be done to get things moving a little quicker. From how it looks, if things keep going the way they are, the search at this level will never be completed. As a side note I'm going to bring a machine online next week that can do about 22 WU every 2.5 hours at the current difficulty and I should be able to leave it running for 7-8 days. So that should help.
5) Message boards : Number crunching : ETA of (search up to 10^20) (Message 1080) Posted 15 Feb 2019 by vseven Post: I'm wondering if people are leaving because its taking so long and CPU's are kinda useless at this point. Maybe increase the RAC given since its taking a lot more work to help retain people? If your estimates are right and people are leaving things are going to get worse.
6) Message boards : Number crunching : Work unit errors (Message 947) Posted 10 Oct 2018 by vseven Post: I've seen the same issue, a WU failing upon startup. Not just in this project but in others also. But with 3 failed out of 900+ I don't know if its worth trying to figure out.
7) Message boards : Number crunching : GPU version: kernel size tuning and less UI lags (Message 938) Posted 1 Oct 2018 by vseven Post: They probably have a 2080 and a second older card. BOINC only reports back how many cards you have installed and only the model of one, usually the fastest one. So the 133 is coming from the 2080 and the 600 is from whatever other card they have.
8) Message boards : Number crunching : GPU version: kernel size tuning and less UI lags (Message 936) Posted 29 Sep 2018 by vseven Post: No, I only got a 2080, no Ti. I'm actually really surprised the performance was that good. My 1080 only eked out 300 second WU's but this was very surprising. Have to do more testing with other Boinc projects and see if these results are similar or something that is just benefiting this project. Soooo.....about that 24 or 25 kernel size. lol. :)
9) Message boards : Number crunching : GPU version: kernel size tuning and less UI lags (Message 933) Posted 29 Sep 2018 by vseven Post: dont expect much increase for RTX cards, nearly half of gpu core take that "tensor".. Just installed my RTX 2080.....averaging 134 seconds a task. :)
10) Message boards : Number crunching : GPU version: kernel size tuning and less UI lags (Message 932) Posted 27 Sep 2018 by vseven Post: Yeah, I ran a Titan V for about a month and compared to the Tesla v100, which spec wise is very similar, in some projects it was close (within 10%) but in others it was way off (more then double the time). I ended up selling the Titan as for the price it wasn't worth the added processing over a 1080 Ti. But I agree, maybe a 24 or 25 setting just for testing.
11) Message boards : Number crunching : GPU version: kernel size tuning and less UI lags (Message 930) Posted 26 Sep 2018 by vseven Post: Are there any plans on increasing the kernel size allowed? Running a Tesla v100 SXM2 16Gb: Kernel at 18 = 160 sec Kernel at 20 = 92 sec Kernel at 21 = 85 sec Kernel at 22 = 81 sec Kernel at 23 = 79 sec Not sure if 24 would be faster but logic would say yes, maybe slightly. Also the new RTX 2080 just got released so they might be more capable then current cards.
12) Message boards : Number crunching : Massive Improvement to Runtimes With New Nvidia Driver (Message 857) Posted 15 Jun 2018 by vseven Post: The CUDA runtimes were updated from 9.1 to 9.2 and that is where the increase is coming from. I've seen a drop from 105 sec to 90 sec on a Tesla v100 which is a huge difference for just a software change.