Folding Forum

I am guessing the IDs of other Super GPUs based on what we already know (and some unidentified IDs I found) :

- 2080 Super should be a TU102 chip, so it will likely be ID 0x1e81 which is between the last known TU102 ID and the first TU104 ID.
- 2070 Super should be a TU104 chip, so it should be (and confirmed once) ID 0x1e84 which is between known IDs of the two chips that equipped the 2080. 0x1ec7 is also a possible ID (it's still in the TU104 chips range), but it would more likely be a professional GPU judging by the surrounding IDs ...
- 2060 Super should be a TU106 chip, so it will likely be ID 0x1f06 which is between known IDs of the two chips that equipped the 2070 and the 2060.

I added them to GPUs.txt with "TBC" as the chip. Please confirm those IDs as soon as you get one of these GPUs ...

Seems like the higher core count helps folding more than gaming.
Average % increase of 2070 to 2070 super was 16%.
At 1,4M PPD, the 2070 Super folds about 25% faster.

I'm eager to know how the 2060 Super will do.

Certainly something strange going on.

I still can't get the thing to appear as an option for folding so i guess i'll have to wait a few weeks and see if GPU-Z gets updated or something so i can get a better idea of what the card version actually is.

Thanks for the help and replies so far

nodezeon wrote:Certainly something strange going on.

I still can't get the thing to appear as an option for folding so i guess i'll have to wait a few weeks and see if GPU-Z gets updated or something so i can get a better idea of what the card version actually is.

Thanks for the help and replies so far

Please post the beginning section of your log file, that will help show what the F@h client detects for your GPU. Include enough to show the folding configuration information.

Just a quick update,

I've gave it another shot tonight, completely reinstalled F@H and the nvidia drivers and voila it popped up right away.

Not sure if it was the client or the drivers but all sorted now so thanks for the help everyone, much appreciated.

When a science fiction writer (or maybe God, herself) endows a character with "super powers" you'll probably notice that many subsequent stories involve the unforeseen aspects of that endowment.

MeeLee wrote:Seems like the higher core count helps folding more than gaming.
Average % increase of 2070 to 2070 super was 16%.
At 1,4M PPD, the 2070 Super folds about 25% faster.

I'm eager to know how the 2060 Super will do.

Following roughly 65 hours of folding we have seen it up as high as 1.51M and as low as 1.28M, but we do seem to be averaging a shade over 1.4M

Right now on PRCG 11719 (0, 2463, 143) it's sitting pretty at an estimated 1,435,324 ppd which is in line with most other projects it's worked on.

Very happy indeed, it might not be best bang per buck but within my budget it's exactly the right return on investment.

I've always contended that FAH has always been about GFLOPS which is equivalent to (#Shaders @ GHz) ... [with an unnecessary factor of 2]. Other features don't help FAH or only make only incremental differences.

Shirty wrote:

MeeLee wrote:Seems like the higher core count helps folding more than gaming.
Average % increase of 2070 to 2070 super was 16%.
At 1,4M PPD, the 2070 Super folds about 25% faster.

I'm eager to know how the 2060 Super will do.

Following roughly 65 hours of folding we have seen it up as high as 1.51M and as low as 1.28M, but we do seem to be averaging a shade over 1.4M

Right now on PRCG 11719 (0, 2463, 143) it's sitting pretty at an estimated 1,435,324 ppd which is in line with most other projects it's worked on.

Very happy indeed, it might not be best bang per buck but within my budget it's exactly the right return on investment.

It might change on other kind of WUs ...

I'll update the thread if there are any significant changes to the output (that I pick up on).

When gamer sites post gaming improvements of 10-16%, I guess the remaining 10-15% we can attribute to the return bonus?
Most points are made from the Quick return bonus anyway; it doesn't accurately represents the amount of work load done.
With RTX cards, I presume in the near future, they can lower the return bonus by a bit; as, if in the future GPUs will be even faster, and we'd have to start counting in billion points per day.
We might as well go with scientific notation (I gained 8e6 points today )

MeeLee wrote:When gamer sites post gaming improvements of 10-16%, I guess the remaining 10-15% we can attribute to the return bonus?
Most points are made from the Quick return bonus anyway; it doesn't accurately represents the amount of work load done.

This isn't quite the way FAH sees it.

If WU is slated to expire in N hours, it takes the same amount of gaming resources to complete it in N hours or to complete it in N/10 or N/100 but the scientific value of finishing it in 10% of the time is more than delaying the result until 100% of the allotted time and it's worth MUCH MORE to science if someone completes it in 1% of the allotted time.

Looking at it another way, if a project can be completed in 10 years, completing it in a year is a lot better for science and it's even better if someone can complete the project in 36.5 days.

As a result, the bonus points for a powerful GPU will be significantly higher than for a GPU that takes longer to complete exactly the same calculations -- quite unlike the measurements as seen by the gamer. FAH Points are awarded both for the calculations you do complete AND for how rapidly you complete them.

2080 Super confirmed as 0x1e81, 2070 Super as 0x1e84 and 2060 as 0x1f06.

I still don't know what 0x1ec7 is and why someone got a GPU with this ID ...

0x1ec7 is a nvidia rtx 2070 super variant.
0x1f47 is a nvidia rtx 2060 super variant.
0x1F42 is a nvidia rtx 2060 super variant.
https://build.opensuse.org/package/view ... w?expand=0