Im useing a GeForce RTX 3060 (not TI). As the weather is warming up the GPU is sounding the temp alarm at 80 c. I think I remember a piece of software from Nvidia for control of fans, temps, OCing, etc. Can anybody tell me the name of such software? This is my daily driver rig, but I never play games... the GPU is used mainly for E@H...?
Im useing a GeForce RTX 3060 (not TI). As the weather is warming up the GPU is sounding the temp alarm at 80 c. I think I remember a piece of software from Nvidia for control of fans, temps, OCing, etc. Can anybody tell me the name of such software? This is my daily driver rig, but I never play games... the GPU is used mainly for E@H...?
Although not from Nvidia, I like MSI Afterburner. If you download it, make sure it's directly from the MSI website. It's clean, simple, and works well with Nvidia. Just my opinion though.
I am wondering ( as always ) what NVIDIA GPU works best at resolving Einstein@Home GPU tasks? I am just wondering because that's what I do.
I ask this because I do know that some NVIDIA GPUs work better at other tasks, like Milkyway@Home works best on a Titan V, so I am asking because I want to know.
Also, is there a water block for GPUs that would make the PCB of the GPU only one slot thick?
Also my intentions for the future are to use a 7 slot motherboard with 7 GPUs running at the same time.
Both the GRP and BRP tasks respond best to maximum CUDA core counts. So the 3080 Ti is best at the top of the RTX 3000 stack. The 3080 Ti mostly matches the 3090 and at much less cost and much less wattage.
I'd stay away from the RTX 4000 cards as evidenced by the current users having issues with high invalid counts.
EK has covered just about every Nvidia card with a single slot block.
Both the GRP and BRP tasks respond best to maximum CUDA core counts. So the 3080 Ti is best at the top of the RTX 3000 stack. The 3080 Ti mostly matches the 3090 and at much less cost and much less wattage.
I'd stay away from the RTX 4000 cards as evidenced by the current users having issues with high invalid counts.
EK has covered just about every Nvidia card with a single slot block.
Thanks Keith. You've always given me good advice, including this one.
For a good example of what I am telling you, look at my most productive GR#1 host with a single 3080 Ti and a single 3080. Both cards running the same amount of memory and type. Both running 2X work units per card.
The 3080 Ti is consistently 60-80 seconds faster than the 3080. That is what an extra 1280 CUDA cores gains you.
All fine and good for gaming stats comparison. Not much help for what the 4070 might do here crunching GR#1 or BPR7.
Plus look at the current posts and status of those that are already running 4080 or 4090 cards and the high amount of invalids and the barely perceptible decrease in crunching times compared to a 3080 Ti.
archae86 wrote:Thanks for
)
Don't forget to use the optimized versions for grp#1 and brp7.
++++++
Oops, I forgot your a windows only setup.
Sorry about the bum advice.
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor)
Im useing a GeForce RTX 3060
)
Im useing a GeForce RTX 3060 (not TI). As the weather is warming up the GPU is sounding the temp alarm at 80 c. I think I remember a piece of software from Nvidia for control of fans, temps, OCing, etc. Can anybody tell me the name of such software? This is my daily driver rig, but I never play games... the GPU is used mainly for E@H...?
Nvidia-settings is the gui
)
Nvidia-settings is the gui version for fan/clock control. nvidia-smi is the command line version.
Styx N Stones wrote: Im
)
Although not from Nvidia, I like MSI Afterburner. If you download it, make sure it's directly from the MSI website. It's clean, simple, and works well with Nvidia. Just my opinion though.
I am wondering ( as always )
)
I am wondering ( as always ) what NVIDIA GPU works best at resolving Einstein@Home GPU tasks? I am just wondering because that's what I do.
I ask this because I do know that some NVIDIA GPUs work better at other tasks, like Milkyway@Home works best on a Titan V, so I am asking because I want to know.
Also, is there a water block for GPUs that would make the PCB of the GPU only one slot thick?
Also my intentions for the future are to use a 7 slot motherboard with 7 GPUs running at the same time.
Thanks in advance!
Proud member of the Old Farts Association
Both the GRP and BRP tasks
)
Both the GRP and BRP tasks respond best to maximum CUDA core counts. So the 3080 Ti is best at the top of the RTX 3000 stack. The 3080 Ti mostly matches the 3090 and at much less cost and much less wattage.
I'd stay away from the RTX 4000 cards as evidenced by the current users having issues with high invalid counts.
EK has covered just about every Nvidia card with a single slot block.
Keith Myers wrote: Both the
)
Thanks Keith. You've always given me good advice, including this one.
Proud member of the Old Farts Association
For a good example of what I
)
For a good example of what I am telling you, look at my most productive GR#1 host with a single 3080 Ti and a single 3080. Both cards running the same amount of memory and type. Both running 2X work units per card.
The 3080 Ti is consistently 60-80 seconds faster than the 3080. That is what an extra 1280 CUDA cores gains you.
Darksider
https://www.techradar.com/ver
)
https://www.techradar.com/versus/rtx-4070-vs-3080
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor)
All fine and good for gaming
)
All fine and good for gaming stats comparison. Not much help for what the 4070 might do here crunching GR#1 or BPR7.
Plus look at the current posts and status of those that are already running 4080 or 4090 cards and the high amount of invalids and the barely perceptible decrease in crunching times compared to a 3080 Ti.