Message boards :
Number crunching :
CPU usage, clock speed, RAM speed - what do you do?
Message board moderation
Previous · 1 · 2 · 3 · 4 · 5
Author | Message |
---|---|
![]() Send message Joined: 29 May 19 Posts: 86 Credit: 6,688,090 RAC: 32 ![]() ![]() |
Thanks for the responses, yeah I guess I'll stick with stock for now! Maybe upgrade GPUs in the near future, thanks once again! |
![]() Send message Joined: 28 Nov 02 Posts: 5126 Credit: 276,046,078 RAC: 462 ![]() |
Just for "baseline" purposes, I ran 100% cpu (all cores) for a couple hours on my Ryzen 5 2400G and the iGPU came to a screeching halt. Just as soon as I dropped it back to 75%, the next time I checked the gpu tasks were processing in a range near 40 minutes again. The cpu tasks continue to run 2+ hours (mostly). Depending on who you read/believe it SOUNDS like the Zen+ version of the 2400G aka: 3400G will be available on July 7. Given that I have two other Zen+ cpus which regularly process cpu tasks in under an hour. And I want to keep running an APU that has some reasonable processing speed... Plus the fact that it is not real expensive... I expect to be ordering one. The other rumor is that late in the Fall (2019) both the newest Threadripper (Amd confirmed Threadripper is NOT going away) and a 7nm version of the 2400G/3400G will ship. I believe that the Threadripper will come out in 7 nm form. I hope it will come out with the memory issues fixed. I am not convinced the 2400G/3400G will come out in the 7 nm form because AMD has "always" had it running one generation behind the mainline Ryzen cpus. It would be interesting though :) Tom A proud member of the OFA (Old Farts Association). |
![]() ![]() ![]() Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 ![]() ![]() |
I am not convinced the 2400G/3400G will come out in the 7 nm form because AMD has "always" had it running one generation behind the mainline Ryzen cpus. The new APU's will be on the 7nm process node. But they will still be the normal monolithic die. No chiplets or separate I/O die. Seti@Home classic workunits:20,676 CPU time:74,226 hours ![]() ![]() A proud member of the OFA (Old Farts Association) |
![]() ![]() ![]() Send message Joined: 30 Nov 05 Posts: 282 Credit: 6,916,194 RAC: 60 ![]() ![]() |
Just for "baseline" purposes, I ran 100% cpu (all cores) for a couple hours on my Ryzen 5 2400G and the iGPU came to a screeching halt.What do you mean exactly by screeching halt? Seti@home classic: 1,456 results, 1.613 years CPU time |
![]() Send message Joined: 28 Nov 02 Posts: 5126 Credit: 276,046,078 RAC: 462 ![]() |
Just for "baseline" purposes, I ran 100% cpu (all cores) for a couple hours on my Ryzen 5 2400G and the iGPU came to a screeching halt.What do you mean exactly by screeching halt? Estimated processing time went from 40 odd minutes to 5+ hours. Tom A proud member of the OFA (Old Farts Association). |
catavalon21 Send message Joined: 2 Nov 01 Posts: 13 Credit: 7,238,152 RAC: 48 ![]() ![]() |
Curious - why avoid iGPU tasks? I have a PC that is integrated GPU only (i7 6700K), and have just started running a combo of CPU / iGPU tasks. Will omitting the iGPU let CPU tasks (currently running 4 or 5 concurrent CPU tasks, depending on what else I am doing) finish more quickly? |
![]() ![]() ![]() Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 ![]() ![]() |
Curious - why avoid iGPU tasks? I have a PC that is integrated GPU only (i7 6700K), and have just started running a combo of CPU / iGPU tasks. Will omitting the iGPU let CPU tasks (currently running 4 or 5 concurrent CPU tasks, depending on what else I am doing) finish more quickly? Normal convention says yes. Seti@Home classic workunits:20,676 CPU time:74,226 hours ![]() ![]() A proud member of the OFA (Old Farts Association) |
![]() ![]() ![]() Send message Joined: 30 Nov 05 Posts: 282 Credit: 6,916,194 RAC: 60 ![]() ![]() |
Normal convention is yes, but lately I feel my experience has been the opposite. I would experiment for yourself to see what works best. Try running with it and running without it. Also try running running iGPU tasks but with one less CPU (7/8 = 87.5%). I'm not familiar with your CPU's architecture but you may have to run with one less core (in that case, 75%), instead of one less thread.Curious - why avoid iGPU tasks? I have a PC that is integrated GPU only (i7 6700K), and have just started running a combo of CPU / iGPU tasks. Will omitting the iGPU let CPU tasks (currently running 4 or 5 concurrent CPU tasks, depending on what else I am doing) finish more quickly? The TL;DR for my situation is I'm finding I crunch more tasks with 3 cores working on CPU tasks and the integrated GPU (coupled with that last CPU core) working on a GPU task. The GPU technically would work faster if one CPU core was not crunching, but overall production was not as great. However, that is with an AMD APU, so that could be completely different from your situation. On another computer I'm running Intel HD 630 graphics. The performance isn't great relative to other GPUs, but I think the times a little faster than the CPU. The jury is still out, though, I've only experimented with it for a few days. One other thing to keep an eye on is if Boinc suspends itself because the CPU is busy. If you are not actively using the computer and you find this is happening then something is slowing down your performance. Usually disabling the iGPU at that point helps out. Seti@home classic: 1,456 results, 1.613 years CPU time |
![]() Send message Joined: 28 Nov 02 Posts: 5126 Credit: 276,046,078 RAC: 462 ![]() |
Curious - why avoid iGPU tasks? I have a PC that is integrated GPU only (i7 6700K), and have just started running a combo of CPU / iGPU tasks. Will omitting the iGPU let CPU tasks (currently running 4 or 5 concurrent CPU tasks, depending on what else I am doing) finish more quickly? So far, for most of the tests, the faster crunching on the Intel iGPU has slowed the entire rest of the system down. However, I have NOT tried crunching with 75% of cores available and the iGPU driven by one of those 75% of cores/threads. So Bill has really brought up a very good point. We need some experimentation using the same "tricks" that we have been using on the Amd APU type systems (A4, A6, A8, A10, Ryzen 3 2200G, Ryzen 5 2400G) to see if that make a difference. As far as I can tell when the Intel iGPU is running it slows the entire system down so that "total" production is less than running simply the cpu tasks. But I am uncertain if anyone has re-tested with the above settings. If I have to, I will take my one Intel with iGPU down, take out the discrete gpu and re-test. I have never been able to get the Intel iGPU to run if I have a discrete video card in there. (Spent hours on that). Tom A proud member of the OFA (Old Farts Association). |
![]() ![]() ![]() Send message Joined: 30 Nov 05 Posts: 282 Credit: 6,916,194 RAC: 60 ![]() ![]() |
I kindof imply in my last post that an iGPU can crunch task efficiently. I base that on only the fact that my iGPU tasks complete around 5,000-,6000 seconds, whereas the normal CPU tasks for that computer complete around 8,000. That might not be a strong enough argument. Additionally, that computer is my daily driver, and I am sharing its resources with other programs, pausing computation throughout the day (either the CPU and/or GPUs), and it also runs two other projects. I am definitely not running a controlled environment with that computer. Still, I encourage to experiment. Seti@home classic: 1,456 results, 1.613 years CPU time |
![]() ![]() ![]() Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 ![]() ![]() |
If you want the TL;DR analysis of crunching with an APU, whether Intel or AMD, read Raistmer's posts at Lunatics where he has graphs and everything. Pretty much lays out what to expect with crunching times for both cpu and gpu tasks. http://lunatics.kwsn.info/index.php/topic,1735.msg58456.html#msg58456 Seti@Home classic workunits:20,676 CPU time:74,226 hours ![]() ![]() A proud member of the OFA (Old Farts Association) |
©2025 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.