Success. Here is what I did for those that might be looking at Arcs and use Windows. I think it was just the spoof file name I had incorrect.
- Change the coproc_info file:
<name> to "<name>Intel(R) HD Graphics 5500 pretending to be HD Graphics 599</name> TWICE in the file (you will see the first <name> close to the top of the file and then again seven lines down from the first <name>).
- Save file
- Then, change the file to "read only" in "properties".
Yeah, agreed. I think it is more of just "proof of function" right now and then I will start looking at BRP7. Additionally, I want to see if these validate.
Nice! Assuming those MeerKat tasks were both done at 1X for both the A4500 and B580, the 580 is twice as fast.
Yep, they were 1x. I am collecting data overnight running 1x to come up with some means and then will give 2x a try. The GPU is running 100% stock speeds/frequencies and basically seems to sit at ~68w, which is pretty impressive to me when comparing to the A4500 on the same system.
Nice! Assuming those MeerKat tasks were both done at 1X for both the A4500 and B580, the 580 is twice as fast.
Yep, they were 1x. I am collecting data overnight running 1x to come up with some means and then will give 2x a try. The GPU is running 100% stock speeds/frequencies and basically seems to sit at ~68w, which is pretty impressive to me when comparing to the A4500 on the same system.
your older A4500 tasks from that same host must have been running at 2x.
your other A4500 Windows host is doing the same ~1250s runtimes and your host reports that it's running 2x in the scheduler log:
Quote:
2025-01-15 00:49:22.6031 [PID=2973779] [version] Checking plan class 'BRP7-cuda55'
2025-01-15 00:49:22.6040 [PID=2973779] [version] reading plan classes from file '/BOINC/projects/EinsteinAtHome/plan_class_spec.xml'
2025-01-15 00:49:22.6040 [PID=2973779] [version] parsed project prefs setting 'gpu_util_brp': 0.500000
A4500 still about 5-10% faster than the B580 at your current settings.
(but of course the A4500 could be like 2x as fast on linux ;) )
Nice! Assuming those MeerKat tasks were both done at 1X for both the A4500 and B580, the 580 is twice as fast.
Yep, they were 1x. I am collecting data overnight running 1x to come up with some means and then will give 2x a try. The GPU is running 100% stock speeds/frequencies and basically seems to sit at ~68w, which is pretty impressive to me when comparing to the A4500 on the same system.
your older A4500 tasks from that same host must have been running at 2x.
your other A4500 Windows host is doing the same ~1250s runtimes and your host reports that it's running 2x in the scheduler log:
Quote:
2025-01-15 00:49:22.6031 [PID=2973779] [version] Checking plan class 'BRP7-cuda55'
2025-01-15 00:49:22.6040 [PID=2973779] [version] reading plan classes from file '/BOINC/projects/EinsteinAtHome/plan_class_spec.xml'
2025-01-15 00:49:22.6040 [PID=2973779] [version] parsed project prefs setting 'gpu_util_brp': 0.500000
A4500 still about 5-10% faster than the B580 at your current settings.
Correct, running 2x on the A4500 gpus. Will be interesting tomorrow to see what happens when I switch to 2x.
Edit: I misspoke in my earlier post due to reading too quickly. B580 is running 1x, A4500 gpus running 2x.
Success. Here is what I did
)
Success. Here is what I did for those that might be looking at Arcs and use Windows. I think it was just the spoof file name I had incorrect.
- Change the coproc_info file:
<name> to "<name>Intel(R) HD Graphics 5500 pretending to be HD Graphics 599</name> TWICE in the file (you will see the first <name> close to the top of the file and then again seven lines down from the first <name>).
- Save file
- Then, change the file to "read only" in "properties".
i see you're running the BRP4
)
i see you're running the BRP4 tasks on that B580.
it will probably be more worthwhile to run BRP7 instead.
_________________________________________________________________________
Yeah, agreed. I think it is
)
Yeah, agreed. I think it is more of just "proof of function" right now and then I will start looking at BRP7. Additionally, I want to see if these validate.
Glad to see you figured the
)
Glad to see you figured the conundrum out.
Thank you to all that
)
Thank you to all that helped!
Right now, we are running MeerKAT 1x successfully, it seems.
Congratulations. I think I
)
Congratulations. I think I see the first two validations on BRP7 tasks.
(hint: look for MeerKAT tasks sent January 14 or later).
Nice! Assuming those MeerKat
)
Nice! Assuming those MeerKat tasks were both done at 1X for both the A4500 and B580, the 580 is twice as fast.
Keith Myers wrote: Nice!
)
Yep, they were 1x. I am collecting data overnight running 1x to come up with some means and then will give 2x a try. The GPU is running 100% stock speeds/frequencies and basically seems to sit at ~68w, which is pretty impressive to me when comparing to the A4500 on the same system.
Boca Raton Community HS
)
your older A4500 tasks from that same host must have been running at 2x.
your other A4500 Windows host is doing the same ~1250s runtimes and your host reports that it's running 2x in the scheduler log:
A4500 still about 5-10% faster than the B580 at your current settings.
(but of course the A4500 could be like 2x as fast on linux ;) )
_________________________________________________________________________
Ian&Steve C. wrote:Boca
)
Correct, running 2x on the A4500 gpus. Will be interesting tomorrow to see what happens when I switch to 2x.
Edit: I misspoke in my earlier post due to reading too quickly. B580 is running 1x, A4500 gpus running 2x.
I need to slow down when reading...