Page 1 of 1
Higher then average PPD?
Posted: Wed Nov 04, 2020 1:14 am
by Unibrowser
I just got a 2070 Super FE and started folding with it expecting to get around 1.6m PPD like all these charts I saw on here said. To my surprise I'm getting between 2m and 2.3m PPD. Is this normal? Is there a way I can do more? I went from 33,000 ppd folding with a GTX 560ti, so this is a whole new level for me. I also checked the logs to make sure It was using CUDA, which it is.
Mobo: Asus P6X58D-E
CPU: i7-960
GPU: 2070 Super FE (Driver 457.09)
Windows 10 64bit
Thanks in advance for any info
Re: Higher then average PPD?
Posted: Wed Nov 04, 2020 1:27 am
by bruce
Yes, CUDA does provide an extra bonus even though the Points are allocated based on what OpenCL is expected to do. This may change at some time in the future.
Re: Higher then average PPD?
Posted: Wed Nov 04, 2020 1:29 am
by Unibrowser
bruce wrote:Yes, CUDA does provide an extra bonus even though the Points are allocated based on what OpenCL is expected to do. This may change at some time in the future.
Oh ok so the PPD is currently artificially high?
Re: Higher then average PPD?
Posted: Wed Nov 04, 2020 5:23 am
by bruce
That's not quite what I said, but it's not wrong either. What is it that the stock-brokers say? Something like ...past performance is not a prediction of future performance....
Re: Higher then average PPD?
Posted: Wed Nov 04, 2020 7:38 am
by PantherX
Here's some comparison of CUDA VS OpenCL performance, Linux with Project 11765, time for 1% in seconds:
73s - 1080Ti OpenCL
57s - 1080Ti CUDA
49s - 2080Ti OpenCL
39s - 2080Ti CUDA
36s - 3080 OpenCL
31s - 3080 CUDA
30s - 3090 OpenCL
26s - 3090 CUDA
Thus, CUDA does speed up the rate at which Science is being done when compared to OpenCL. Thus, the boost in PPD is exponential, not linear due to the formula used.