I keep getting errors about 1 out of 10 work units. Any suggestions how to eliminate this? Or this is the new normal? Errors were very rare until about a week ago.
I keep getting errors about 1 out of 10 work units. Any suggestions how to eliminate this? Or this is the new normal? Errors were very rare until about a week ago.
I bought an Asus rtx 3080 ti Liquid Cooling on the grounds (from a review) that it runs an graphics clock of near 1965 Mhz (and it does) and the allegation that it was as "fast" as the slower rtx 3090's.
So far I haven't been able to persuade to run much past 1.5M RAC. I was hoping I could get it to run 1.6+
I have reduced the MTM to 1500 since 1700 didn't seem to be making much of a difference.
The PL is running the default of 400 watts. I have a late model GPU driver.
With the liquid cooler the gpu is averaging 53C.
The pcie is 3rd gen (motherboard limit)
What should I tinker with to get the average processing speed to its lowest total (and RAC to the maximum)?
Does the speed of the CPU make that much difference (I have it throttled at the moment rather than at it's maximum)?
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor)
Nvidia orin nano is the
)
Nvidia orin nano is the future. It is not cheap with about 550$ per unit.
Stefan
Stefan wrote: Nvidia orin
)
What kind of performance for boinc GPU crunching are they likely to provide?
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor)
Tom M wrote: Stefan
)
Take a look at my Nvidia Jetson TX2-NX host running BRP4/4G tasks.
Nvidia Jetson TX2-NX host on BRP4 and Guarav Khanna app
Then quadruple the output. The TX2-NX is two generations behind Orin.
I keep getting errors about 1
)
I keep getting errors about 1 out of 10 work units. Any suggestions how to eliminate this? Or this is the new normal? Errors were very rare until about a week ago.
Thank you.
Einstein@Home 1.28 Gamma-ray pulsar binary search #1 on GPUs (FGRPopencl2Pup-nvidia) LATeah4021L12_1068.0_0_0.0_15388485_0 00:14:04 (00:13:49) 98.294952607 100.000000000 7/4/2023 7:34:21 AM - 1C + 0.5 NVIDIA GPUs Computation error 20t.2
Lowering sclk, GPU clock
)
Lowering sclk, GPU clock speed, fixed mine.
Miklos M. wrote: I keep
)
since your hosts are hidden, please link us to the host details page so the configuration can be inspected.
of particular interest is what drivers you have installed
_________________________________________________________________________
Thank you both.
)
Thank you both.
Miklos M. wrote: Thank you
)
does that mean you solved it?
_________________________________________________________________________
It seems to have faded away.
)
It seems to have faded away. Maybe those units were error prone on my Titans.
I bought an Asus rtx 3080 ti
)
I bought an Asus rtx 3080 ti Liquid Cooling on the grounds (from a review) that it runs an graphics clock of near 1965 Mhz (and it does) and the allegation that it was as "fast" as the slower rtx 3090's.
So far I haven't been able to persuade to run much past 1.5M RAC. I was hoping I could get it to run 1.6+
I have reduced the MTM to 1500 since 1700 didn't seem to be making much of a difference.
The PL is running the default of 400 watts. I have a late model GPU driver.
With the liquid cooler the gpu is averaging 53C.
The pcie is 3rd gen (motherboard limit)
What should I tinker with to get the average processing speed to its lowest total (and RAC to the maximum)?
Does the speed of the CPU make that much difference (I have it throttled at the moment rather than at it's maximum)?
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor)