Another card of advanced age is GTX 750ti. One task takes about 1550 sec. Still in working-age? :-)
Well, it's always a compromise isn't it.
Found some data saying the GTX 750ti consumes ~60W @ gaming, The RTX 3070ti takes ~270W. This makes the 3070 TI more efficient, which is no surprise :) (4x the power, but 12x faster in crunching) But I think 1550 seconds is still a lot of crunching.
The RTX 3060ti LHR 8GB can use a little update on OC mode.
OC it can crunch 157s per WU average since a couple days now, the few errors it had were induced by myself when I tried to suspend one or more units while being crunched. It won't take suspending active units.
Suspend all non-active units and let it run dry wil do the trick to stop crunching without errors.
Also this good 'ol X79 machine made it to the top 50 machines on this project Today with a RTX 3060ti.
Some additional data points for older NVIDIA GPUs running 3012L01_892:
GTX1060 6GB 661s 1874MHz (+0) - Win10
GTX1660ti 600s 1980MHz (+50) Ubuntu 18.04.6 LTS
GTX1070ti 495s 1987MHz (+160) Ubuntu 18.04.6 LTS
RTX2060 329s 1980MHz (+50) Ubuntu 18.04.6 LTS
RTX2070 295s 1935MHz (+50) Ubuntu 18.04.6 LTS
I also have a RTX2060super in the mix but there is something wrong there as it is taking 452s @1995MHz (+50) under Ubuntu when it should be somewhere between the 2060 and 2070.
I also have a RTX2060super in the mix but there is something wrong there as it is taking 452s @1995MHz (+50) under Ubuntu when it should be somewhere between the 2060 and 2070.
Maybe you should use older drivers (not the 500 series), this worked for me. I also reset Einstein project in BOINC
Maybe you should use older drivers (not the 500 series), this worked for me. I also reset Einstein project in BOINC
I might give that a try but I'd rather see if I can find the root cause as I like having all my rigs built and configured the same for my sanity's sake ;-)
I'm still on Ubuntu 18.04 but I just reached a milestone in Folding at Home I've been waiting for so I'm going to start to re-build them using 20.04 shortly so it will likely have to wait for then to get resolved. I primarily do F@H on my newer (Turing) GPUs and use my excess CPU Threads and older GPUs for OpenPandemics on WCG and Einstein at weights of 100 and 0 respectively but with WCG paused while the Server Infrastructure is transitioned from IBM to Krembil I've moved all my GPUs except 5 2070 Supers to Einstein GRPS (don't have enough Memory and/or Memory Bandwidth for GWS)
Hello, does anyone have benchmarking numbers with Windows 10 hardware accelerated GPU scheduling on vs. off? I unfortunately do not have any Windows systems running capable GPUs at the moment to test if this has any impact on E@H numbers and wasn't able to find existing posts in the forum that explored this.
Thanks for you data! *Updated
)
Another card of advanced age
)
Another card of advanced age is GTX 750ti. One task takes about 1550 sec. Still in working-age? :-)
solling2 wrote: Another card
)
That is still faster than the iGpu of a Ryzen 3 2200g that is processing in the ballpark of 34 minutes per GR#1 task.
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor)
solling2 schrieb: Another
)
Well, it's always a compromise isn't it.
Found some data saying the GTX 750ti consumes ~60W @ gaming, The RTX 3070ti takes ~270W. This makes the 3070 TI more efficient, which is no surprise :) (4x the power, but 12x faster in crunching) But I think 1550 seconds is still a lot of crunching.
The RTX 3060ti LHR 8GB can
)
The RTX 3060ti LHR 8GB can use a little update on OC mode.
OC it can crunch 157s per WU average since a couple days now, the few errors it had were induced by myself when I tried to suspend one or more units while being crunched. It won't take suspending active units.
Suspend all non-active units and let it run dry wil do the trick to stop crunching without errors.
Also this good 'ol X79 machine made it to the top 50 machines on this project Today with a RTX 3060ti.
Yay :P
Hurray for the Rtx 3060 ti!
)
Hurray for the Rtx 3060 ti!
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor)
Some additional data points
)
Some additional data points for older NVIDIA GPUs running 3012L01_892:
GTX1060 6GB 661s 1874MHz (+0) - Win10
GTX1660ti 600s 1980MHz (+50) Ubuntu 18.04.6 LTS
GTX1070ti 495s 1987MHz (+160) Ubuntu 18.04.6 LTS
RTX2060 329s 1980MHz (+50) Ubuntu 18.04.6 LTS
RTX2070 295s 1935MHz (+50) Ubuntu 18.04.6 LTS
I also have a RTX2060super in the mix but there is something wrong there as it is taking 452s @1995MHz (+50) under Ubuntu when it should be somewhere between the 2060 and 2070.
gordonbb schrieb:I also
)
Maybe you should use older drivers (not the 500 series), this worked for me. I also reset Einstein project in BOINC
Markus Windisch wrote:Maybe
)
I might give that a try but I'd rather see if I can find the root cause as I like having all my rigs built and configured the same for my sanity's sake ;-)
I'm still on Ubuntu 18.04 but I just reached a milestone in Folding at Home I've been waiting for so I'm going to start to re-build them using 20.04 shortly so it will likely have to wait for then to get resolved. I primarily do F@H on my newer (Turing) GPUs and use my excess CPU Threads and older GPUs for OpenPandemics on WCG and Einstein at weights of 100 and 0 respectively but with WCG paused while the Server Infrastructure is transitioned from IBM to Krembil I've moved all my GPUs except 5 2070 Supers to Einstein GRPS (don't have enough Memory and/or Memory Bandwidth for GWS)
Hello, does anyone have
)
Hello, does anyone have benchmarking numbers with Windows 10 hardware accelerated GPU scheduling on vs. off? I unfortunately do not have any Windows systems running capable GPUs at the moment to test if this has any impact on E@H numbers and wasn't able to find existing posts in the forum that explored this.
Non Microsoft article: https://www.windowslatest.com/2020/07/07/windows-10-hardware-accelerated-gpu-scheduling-explained/