Fixed the core clock to 4.2GHz paired with Tesla P100 for this host. No other cpu project is running.
With 3 tasks per gpu, average run time is about 630 secs when staggering the 3 tasks. Run to run variation is small (look for those between 1880 - 1890 secs range). About 1.37M PPD. GPU board power fluctuates from 80W to 130W.
Do you mean a gpu like this: NVIDIA NVIDIA P104-100 (8117MB) because that one, mine, is running them in 1700 seconds for each task running just 1 at a time. My core clock is nowhere near where yours is!!
Fixed the core clock to 4.2GHz paired with Tesla P100 for this host. No other cpu project is running.
With 3 tasks per gpu, average run time is about 630 secs when staggering the 3 tasks. Run to run variation is small (look for those between 1880 - 1890 secs range). About 1.37M PPD. GPU board power fluctuates from 80W to 130W.
Do you mean a gpu like this: NVIDIA NVIDIA P104-100 (8117MB) because that one, mine, is running them in 1700 seconds for each task running just 1 at a time. My core clock is nowhere near where yours is!!
It has 4096 bit memory bus, thus higher memory bandwidth (that helps a lot to speed up read/write access between the gpu chip and vram). Not certain if the higher FP64 matters much for this OA3S task but perhaps is does (just an observation) since I read here that the 4090 even with higher memory bandwidth can only do around 600 secs per task, similar to P100 which is already 3 generations behind. Maybe someone else here can shed a light if FP64 compute is used in O3AS.
No. A P100 is the Nvidia Tesla P100 data center card. Nvidia names all their generational data center flagships like this, P100/V100/A100/etc. A P100 is like a 1080Ti with better FP64 and HBM memory.
your P104-100 is a mining GPU, which is basically just a 1070 with GDDR5X memory.
Fixed the core clock to 4.2GHz paired with Tesla P100 for this host. No other cpu project is running.
With 3 tasks per gpu, average run time is about 630 secs when staggering the 3 tasks. Run to run variation is small (look for those between 1880 - 1890 secs range). About 1.37M PPD. GPU board power fluctuates from 80W to 130W.
Do you mean a gpu like this: NVIDIA NVIDIA P104-100 (8117MB) because that one, mine, is running them in 1700 seconds for each task running just 1 at a time. My core clock is nowhere near where yours is!!
It has 4096 bit memory bus, thus higher memory bandwidth (that helps a lot to speed up read/write access between the gpu chip and vram). Not certain if the higher FP64 matters much for this OA3S task but perhaps is does (just an observation) since I read here that the 4090 even with higher memory bandwidth can only do around 600 secs per task, similar to P100 which is already 3 generations behind. Maybe someone else here can shed a light if FP64 compute is used in O3AS.
No. A P100 is the Nvidia Tesla P100 data center card. Nvidia names all their generational data center flagships like this, P100/V100/A100/etc. A P100 is like a 1080Ti with better FP64 and HBM memory.
your P104-100 is a mining GPU, which is basically just a 1070 with GDDR5X memory.
Ahhh I hate when they do that stuff with the names, it's almost like they want to confuse people.
Fixed the core clock to 4.2GHz paired with Tesla P100 for this host. No other cpu project is running.
With 3 tasks per gpu, average run time is about 630 secs when staggering the 3 tasks. Run to run variation is small (look for those between 1880 - 1890 secs range). About 1.37M PPD. GPU board power fluctuates from 80W to 130W.
Do you mean a gpu like this: NVIDIA NVIDIA P104-100 (8117MB) because that one, mine, is running them in 1700 seconds for each task running just 1 at a time. My core clock is nowhere near where yours is!!
It has 4096 bit memory bus, thus higher memory bandwidth (that helps a lot to speed up read/write access between the gpu chip and vram). Not certain if the higher FP64 matters much for this OA3S task but perhaps is does (just an observation) since I read here that the 4090 even with higher memory bandwidth can only do around 600 secs per task, similar to P100 which is already 3 generations behind. Maybe someone else here can shed a light if FP64 compute is used in O3AS.
Ahhh that's a very nice card!!
The P100 has dropped a lot in price over the past year or so. You can get one in ebay for USD160ish. This card is only good for project requiring high memory bandwidth and FP64, so has limited use in BOINC. In the past, I used it in Milkyway and maybe will use it in gpugrid (quantum chemistry) when tasks are available. This version has no active cooling. Certainly good for O3AS but the high credit will eventually revert within two months or so.
Fixed the core clock to 4.2GHz paired with Tesla P100 for this host. No other cpu project is running.
With 3 tasks per gpu, average run time is about 630 secs when staggering the 3 tasks. Run to run variation is small (look for those between 1880 - 1890 secs range). About 1.37M PPD. GPU board power fluctuates from 80W to 130W.
Do you mean a gpu like this: NVIDIA NVIDIA P104-100 (8117MB) because that one, mine, is running them in 1700 seconds for each task running just 1 at a time. My core clock is nowhere near where yours is!!
It has 4096 bit memory bus, thus higher memory bandwidth (that helps a lot to speed up read/write access between the gpu chip and vram). Not certain if the higher FP64 matters much for this OA3S task but perhaps is does (just an observation) since I read here that the 4090 even with higher memory bandwidth can only do around 600 secs per task, similar to P100 which is already 3 generations behind. Maybe someone else here can shed a light if FP64 compute is used in O3AS.
Ahhh that's a very nice card!!
The P100 has dropped a lot in price over the past year or so. You can get one in ebay for USD160ish. This card is only good for project requiring high memory bandwidth and FP64, so has limited use in BOINC. In the past, I used it in Milkyway and maybe will use it in gpugrid (quantum chemistry) when tasks are available. This version has no active cooling. Certainly good for O3AS but the high credit will eventually revert within two months or so.
Thanks I will look but the passive cooling is not a good thing for me.
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
Fixed the core clock to 4.2GHz paired with Tesla P100 for this host. No other cpu project is running.
With 3 tasks per gpu, average run time is about 630 secs when staggering the 3 tasks. Run to run variation is small (look for those between 1880 - 1890 secs range). About 1.37M PPD. GPU board power fluctuates from 80W to 130W.
Do you mean a gpu like this: NVIDIA NVIDIA P104-100 (8117MB) because that one, mine, is running them in 1700 seconds for each task running just 1 at a time. My core clock is nowhere near where yours is!!
It has 4096 bit memory bus, thus higher memory bandwidth (that helps a lot to speed up read/write access between the gpu chip and vram). Not certain if the higher FP64 matters much for this OA3S task but perhaps is does (just an observation) since I read here that the 4090 even with higher memory bandwidth can only do around 600 secs per task, similar to P100 which is already 3 generations behind. Maybe someone else here can shed a light if FP64 compute is used in O3AS.
Ahhh that's a very nice card!!
The P100 has dropped a lot in price over the past year or so. You can get one in ebay for USD160ish. This card is only good for project requiring high memory bandwidth and FP64, so has limited use in BOINC. In the past, I used it in Milkyway and maybe will use it in gpugrid (quantum chemistry) when tasks are available. This version has no active cooling. Certainly good for O3AS but the high credit will eventually revert within two months or so.
Thanks I will look but the passive cooling is not a good thing for me.
I ran across some after-market GPU cooling for passively cooled cards. Maybe one fits the P100?
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
pututu wrote: Fixed the core
)
Do you mean a gpu like this: NVIDIA NVIDIA P104-100 (8117MB) because that one, mine, is running them in 1700 seconds for each task running just 1 at a time. My core clock is nowhere near where yours is!!
mikey wrote:pututu
)
Mine is this.
It has 4096 bit memory bus, thus higher memory bandwidth (that helps a lot to speed up read/write access between the gpu chip and vram). Not certain if the higher FP64 matters much for this OA3S task but perhaps is does (just an observation) since I read here that the 4090 even with higher memory bandwidth can only do around 600 secs per task, similar to P100 which is already 3 generations behind. Maybe someone else here can shed a light if FP64 compute is used in O3AS.
No. A P100 is the
)
No. A P100 is the Nvidia Tesla P100 data center card. Nvidia names all their generational data center flagships like this, P100/V100/A100/etc. A P100 is like a 1080Ti with better FP64 and HBM memory.
your P104-100 is a mining GPU, which is basically just a 1070 with GDDR5X memory.
_________________________________________________________________________
pututu wrote: mikey
)
Ahhh that's a very nice card!!
Ian&Steve C. wrote: No. A
)
Ahhh I hate when they do that stuff with the names, it's almost like they want to confuse people.
mikey wrote:pututu
)
The P100 has dropped a lot in price over the past year or so. You can get one in ebay for USD160ish. This card is only good for project requiring high memory bandwidth and FP64, so has limited use in BOINC. In the past, I used it in Milkyway and maybe will use it in gpugrid (quantum chemistry) when tasks are available. This version has no active cooling. Certainly good for O3AS but the high credit will eventually revert within two months or so.
After a short test yesterday
)
After a short test yesterday with about 270 GW-WUs, I can't report a single crash on my host. Linux, Nvidia 4090, driver 535.129.03. No problem here.
Two WUs at the same time results in more than double the runtime, a delayed start helps, but requires constant manual control.
i am also waiting for the Nvidia version.
Minor nitpick: I still have two orphaned wus (741768987 and 741769147) from the first test last july in my tasklist.
pututu wrote: mikey
)
Thanks I will look but the passive cooling is not a good thing for me.
mikey wrote:Do you mean a
)
Post deleted.
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
mikey wrote:e]pututu
)
I ran across some after-market GPU cooling for passively cooled cards. Maybe one fits the P100?
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!