Page 9 of 11
Re: 9401 fails on 750ti [GM107] but not 750ti [GK106]
Posted: Fri May 30, 2014 5:48 pm
by sheroo
Guys wondering if you can tell me what the current status of the 750ti running under Linux is, as I'm eager to purchase one of these cards for folding. btw what PPD can I expect under Linux?
Re: 9401 fails on 750ti [GM107] but not 750ti [GK106]
Posted: Fri May 30, 2014 9:14 pm
by PantherX
Welcome to the F@H Forum sheroo,
Please note that currently, FahCore_17 isn't supported on Maxwell GPUs due to a Driver bug. Nvidia is aware of it and once a fix is released, development can test FahCore_17 on Maxwell GPUs. Do note that there isn't any ETA provided for this. Also note that you can fold FahCore_15 WUs on Maxwell GPUs only in Windows. Since you have a Linux system, you can't fold on that GPU for now.
Re: 9401 fails on 750ti [GM107] but not 750ti [GK106]
Posted: Fri May 30, 2014 9:18 pm
by sheroo
Thank you for your quick response and warm welcome.
Re: 9401 fails on 750ti [GM107] but not 750ti [GK106]
Posted: Sun Jun 29, 2014 9:53 pm
by james888
I have been folding on two 750ti's for a couple weeks. I get ppd levels of either ~14k ppd or ~30k ppd per gpu and rarely in between. I do not know why there is variance or if that is normal.
@PantherX
When core_17 wu's become available and the drivers are fixed, will I see more performance out of my 750ti's?
Re: 9401 fails on 750ti [GM107] but not 750ti [GK106]
Posted: Mon Jun 30, 2014 1:06 pm
by uddarts
i'm seeing the same sort ppd(11k & 29k).
ud
Re: 9401 fails on 750ti [GM107] but not 750ti [GK106]
Posted: Mon Jun 30, 2014 3:26 pm
by bruce
Welcome to foldingforum.org, james888.
james888 wrote:I have been folding on two 750ti's for a couple weeks. I get ppd levels of either ~14k ppd or ~30k ppd per gpu and rarely in between. I do not know why there is variance or if that is normal.
@PantherX
When core_17 wu's become available and the drivers are fixed, will I see more performance out of my 750ti's?
Updating the drivers is your responsibility (provided NVidia has released a version containing the correction). I'm not convinced whether the problem is (A) exclusively in the drivers or (B) exclusively in Core_17, or (C) a combination of the two. I have not tested it personally and I have not seen a
definitive report of it being tested by anyone else. Have you?
Re: 9401 fails on 750ti [GM107] but not 750ti [GK106]
Posted: Mon Jun 30, 2014 4:47 pm
by soa_rru
62k ppd (21.5k per wu) with my 750ti (GM107) on a 9201 using 337.88 drivers and Version 0.0.55 core.
Running perfectly stable just like it was before
Re: 9401 fails on 750ti [GM107] but not 750ti [GK106]
Posted: Mon Jun 30, 2014 5:08 pm
by PantherX
james888 wrote:...When core_17 wu's become available and the drivers are fixed, will I see more performance out of my 750ti's?
Welcome to the F@H Forum james888,
Regarding the potential increase in performance, please note the following:
1) Recently, AMD GPUs got a decent boost in their PPD by using the newer driver version. Thus, it might be possible that if the GPU vendor optimizes their code, the customers can gain an increased performance by using the latest driver version. Since OpenCL is industry standard while CUDA is Nvidia only, not sure if Nvidia will choose to fine tune their OpenCL performance.
2) If the current FahCore_17 is using a workaround to allow Maxwell GPUs to fold, they won't be folding at their full potential until the driver bug is resolved (viewtopic.php?p=262342#p262342).
3) OpenMM would be further optimized for Maxwell GPUs so hopefully, it would result in increased performance on Maxwell GPUs (viewtopic.php?p=259238#p259238).
Re: 9401 fails on 750ti [GM107] but not 750ti [GK106]
Posted: Wed Jul 02, 2014 12:55 am
by james888
bruce wrote:Welcome to foldingforum.org, james888.
james888 wrote:I have been folding on two 750ti's for a couple weeks. I get ppd levels of either ~14k ppd or ~30k ppd per gpu and rarely in between. I do not know why there is variance or if that is normal.
@PantherX
When core_17 wu's become available and the drivers are fixed, will I see more performance out of my 750ti's?
Updating the drivers is your responsibility (provided NVidia has released a version containing the correction). I'm not convinced whether the problem is (A) exclusively in the drivers or (B) exclusively in Core_17, or (C) a combination of the two. I have not tested it personally and I have not seen a
definitive report of it being tested by anyone else. Have you?
Thank you for welcome.
I am on 337.88, and at the moment am updating to the latest beta driver to see if it help. If A, B, or C is really why I am here asking questions here.
soa_rru wrote:62k ppd (21.5k per wu) with my 750ti (GM107) on a 9201 using 337.88 drivers and Version 0.0.55 core.
Running perfectly stable just like it was before
PantherX wrote:james888 wrote:...When core_17 wu's become available and the drivers are fixed, will I see more performance out of my 750ti's?
That is a lot higher than I am getting. I would love to get that kind of ppd as it would double my output.
Welcome to the F@H Forum james888,
Regarding the potential increase in performance, please note the following:
1) Recently, AMD GPUs got a decent boost in their PPD by using the newer driver version. Thus, it might be possible that if the GPU vendor optimizes their code, the customers can gain an increased performance by using the latest driver version. Since OpenCL is industry standard while CUDA is Nvidia only, not sure if Nvidia will choose to fine tune their OpenCL performance.
2) If the current FahCore_17 is using a workaround to allow Maxwell GPUs to fold, they won't be folding at their full potential until the driver bug is resolved (viewtopic.php?p=262342#p262342).
3) OpenMM would be further optimized for Maxwell GPUs so hopefully, it would result in increased performance on Maxwell GPUs (viewtopic.php?p=259238#p259238).
I am just glad it is being worked on. These maxwell gpu's are really cool. My 750ti's use about 30watts each at load, and I feel have a lot more potential than what I am getting out of them. As soa_rru just mentioned, maybe there is more I can get now. I am looking into this now.
EDIT:
A FAH member messaged me on what I need to change to get the ppd these cards are meant for. FAH now says I have estimated 76,000 ppd, but I suspect that is a bit high and will not show in real day to day numbers.
Re: 9401 fails on 750ti [GM107] but not 750ti [GK106]
Posted: Tue Jul 08, 2014 9:01 am
by muffinxy
EDIT:
A FAH member messaged me on what I need to change to get the ppd these cards are meant for. FAH now says I have estimated 76,000 ppd, but I suspect that is a bit high and will not show in real day to day numbers.
Could you tell me as well what I need to change to get the higher ppd on my 750ti Maxwell GPU? Thank you!
Re: 9401 fails on 750ti [GM107] but not 750ti [GK106]
Posted: Tue Jul 08, 2014 2:21 pm
by JimF
muffinxy wrote:Could you tell me as well what I need to change to get the higher ppd on my 750ti Maxwell GPU? Thank you!
Just set the Advanced flag, by setting
Name = client-type and
Value = advanced in the Configure => Slots tab.
This gives some guidance, though the pictures are no longer available.
viewtopic.php?f=67&t=18187
(I am getting 67k on each of my two Asus GTX 750 Ti's, not overclocked.)
Re: 9401 fails on 750ti [GM107] but not 750ti [GK106]
Posted: Tue Jul 08, 2014 5:52 pm
by PantherX
Please note that that post is superseded by this post (viewtopic.php?p=261088&f=24#p261088) which contains detailed instructions for the latest V7 version. There aren't any pictures in it since I found that the text is sufficient for the vast majority and is easy on the bandwidth (also, don't have to worry about hosting images).
Re: 9401 fails on 750ti [GM107] but not 750ti [GK106]
Posted: Thu Jul 10, 2014 6:59 am
by muffinxy
Thank you! It's working...Estimated PPD 63475
Re: 9401 fails on 750ti [GM107] but not 750ti [GK106]
Posted: Thu Aug 07, 2014 1:24 am
by james888
You don't need to change anything to get high points on 750ti's anymore. Just installed another 750ti and didn't set anything beyond passphrase and I am getting 60k.
Re: 9401 fails on GM107 but not GK106 {Hopefully fixed}
Posted: Sun Oct 05, 2014 1:27 am
by Freightanimal
I have not seen any activity on this thread for a while now. My 750ti had been purring right along. I have been busy but noticed that for about the last month and a half I have not received any core17 wu's. I had been getting them regularly and completing them each time. Today I got one that failed, then another that failed, then several more failures. Not sure what to make of it. Just looking for some feedback. Has this happened to anyone else?