Advanced search

Message boards : Graphics cards (GPUs) : Obsolete Video Cards

Author Message
Carl
Send message
Joined: 2 May 13
Posts: 8
Credit: 2,438,161,537
RAC: 267,620
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 47022 - Posted: 18 Apr 2017 | 20:28:47 UTC

With the rollout of the new Cuda drivers, it looks like the Fermi based video cards are now rendered obsolete.

http://docs.nvidia.com/cuda/cuda-toolkit-release-notes/index.html

Deprecated Features:

•Fermi Architecture Support. Fermi architecture support is being deprecated in the CUDA 8.0 Toolkit, which will be the last toolkit release to support it. Future versions of the CUDA Toolkit will not support the architecture and are not guaranteed to work on that platform. Note that support for Fermi is being deprecated in the CUDA Toolkit but not in the driver. Applications compiled with CUDA 8.0 or older will continue to work on Fermi with newer NVIDIA drivers.

If this is true, then it needs to be officially announced as it will affect many users. I knew this day would be coming, just hoping it would be delayed a little bit longer.

wiyosaya
Send message
Joined: 22 Nov 09
Posts: 114
Credit: 589,114,683
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 47137 - Posted: 28 Apr 2017 | 23:06:22 UTC - in response to Message 47022.

Have a look at this thread - especially this post
____________

Carl
Send message
Joined: 2 May 13
Posts: 8
Credit: 2,438,161,537
RAC: 267,620
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 47147 - Posted: 1 May 2017 | 4:16:55 UTC - in response to Message 47137.

Thanks for the reference. I have already retired my old GTX 570 with a new GTX 1070. Wow, What a difference! Much more efficient power wise and tasks finish in hours instead of days now!

Engagex BOINC-SETI
Send message
Joined: 7 Oct 16
Posts: 5
Credit: 949,150
RAC: 0
Level
Gly
Scientific publications
wat
Message 47484 - Posted: 20 Jun 2017 | 16:56:17 UTC

I run three dedicated machines. One of them has a 9600 GT and two GT 310s. They will only support up to the 340 drivers. Granted they are not very fast, but aside from the 9600 which was given to me, they only use 30 watts each and only cost $12 each.

So regardless of the task turnaround time they are very power efficient and the replacement cost is minimal. So unless you are crunching on the same machine you are gaming on there is no point in spending over $100 on a 10 series card.

3de64piB5uZAS6SUNt1GFDU9d...
Avatar
Send message
Joined: 20 Apr 15
Posts: 285
Credit: 1,102,216,607
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwat
Message 47485 - Posted: 20 Jun 2017 | 18:30:26 UTC - in response to Message 47484.

I run three dedicated machines. One of them has a 9600 GT and two GT 310s. They will only support up to the 340 drivers. Granted they are not very fast, but aside from the 9600 which was given to me, they only use 30 watts each and only cost $12 each.

So regardless of the task turnaround time they are very power efficient and the replacement cost is minimal. So unless you are crunching on the same machine you are gaming on there is no point in spending over $100 on a 10 series card.


It may be that you can barely see the energy cost of those little cards on your bill ... but I beg to differ in regard to efficiency ... compare the GT310 with the new GT1030, they have about the same TDP.

The new 1030 is 20x (!!) times faster in regard to SP GFLOPS and therefore 20 times more efficient. Put it another way, the 1030 requires 1,5 Watt for the same performance. Whereas the old GT310 pulls 30W.
____________
I would love to see HCF1 protein folding and interaction simulations to help my little boy... someday.

Erich56
Send message
Joined: 1 Jan 15
Posts: 1091
Credit: 6,603,906,926
RAC: 3,160,162
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwat
Message 47504 - Posted: 28 Jun 2017 | 16:52:56 UTC

Should a GTX750Ti be considered obsolete?

Why I am asking this?

As already explained in other threads, since the implementation of the new crunching software acemd 918.80 in April, my GTX750Ti in the Windows10 host (WDDM overhead!) has tremendous problems when crunching such demanding tasks like
..._crystal_ss_contacts_50_ubiquitin_... and ..._crystal_ss_contacts_100_ubiquitin_...
and sometimes also
...PABLO_all_data_goal_KIX_CMYB-...

What does the problem look like? All of a sudden, crunchings stops, and does not continue until the task is suspended and resumed in the BOINC manager. Even if not overclocked at all!
BTW, I have read about this behaviour also in other threads here in the forum, so it's definitely NOT a defect of the card itself.

If this happens some time during the night, or even during daytime when I am not at home and noticing what's going on, the task runs idle for several hours. Which, of course, makes no sense.
Recently, such a task stopped 7 or 8 times in the course of 3 days.

I had already proposed to the GPUGRID people, here in the forum, to introduce a third type of task for these ones mentioned above, called "very long runs" or similar (besides the current "long runs" and the "short runs" which don't seem to exist any longer). However, no reaction so far, and I doubt that there will be any.
Another way to fix this might of course be to repair the buggy crunching software. However, this doesn't happen either :-(

Whereas my other GTX750Ti which runs in the Windows XP host has none of these problems, although overclocked markedly (also, the crunching software is a different one: acemd 849.65).

So, in short: must a GTX750ti in a Windows10 system be considered obsolete? At least what concerns GPUGRID crunching?
Would be interesting what happens when crunching Folding@Home.

Qax
Send message
Joined: 29 May 11
Posts: 8
Credit: 67,402,347
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 47815 - Posted: 25 Aug 2017 | 5:10:10 UTC

I just ordered a computer with a AMD Radeon (TM) RX 580 with 8GB GDDR5 Graphics Memory. Will this work for GPUGrid?

Erich56
Send message
Joined: 1 Jan 15
Posts: 1091
Credit: 6,603,906,926
RAC: 3,160,162
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwat
Message 47816 - Posted: 25 Aug 2017 | 5:12:47 UTC - in response to Message 47815.
Last modified: 25 Aug 2017 | 5:13:06 UTC

AMD Radeon (TM) RX 580

to my knowledge, only NVIDIA works with GPUGRID

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 2343
Credit: 16,201,255,749
RAC: 1,038
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 47817 - Posted: 25 Aug 2017 | 23:05:16 UTC - in response to Message 47815.
Last modified: 25 Aug 2017 | 23:09:04 UTC

I just ordered a computer with a AMD Radeon (TM) RX 580 with 8GB GDDR5 Graphics Memory. Will this work for GPUGrid?
No, it won't work with GPUGrid.
Take a look at the apps page. There are only CUDA and CPU (multi-thread) clients present.
CUDA is NVidia's own proprietary GPU programming language.

Post to thread

Message boards : Graphics cards (GPUs) : Obsolete Video Cards

//