[QUOTE=Stargate38;291308]Use Mediafire or Sendspace to upload it.[/QUOTE]

Here you go:

[URL]http://www.sendspace.com/file/8vyd19[/URL]

I recommend using the -b switch to suppress the "report results" query.

memtestG80 -b [amount of RAM in MB to test] [number of iterations]

On a 1GB GTX460 I can get it to test ~825MB or a bit more when the card is driving the display. README suggests,
[CODE]If you suspect that your graphics card is having issues (for example, it fails
running Folding@home work units), we strongly recommend that you test as large
a memory region as is practical, and run thousands of test iterations. In our
testing, we have found that even "problematic" cards may only fail sporadically
(e.g., once every 50,000 test iterations). Like other stress testing tools,
to properly verify stability MemtestG80 should be run for an extended period of
time.[/CODE]

Unfortunately, MemtestG80 wont stress test a graphics card enough to reveal any instability issues with it.
I personally have overclocked my 480's memory 200 Mhz beyond stable, and after 500 iterations, no errors were found.
However, as soon as I launched FurMark, a shiny blue screen of death was looking at me.
Unfortunately, GPU manufacturers and designers consider FurMark/OCCT evil for their products, since GPUs with stock cooling cant safely handle that kind of stress for adequate amount of time. Protection for these kind of applications was created even!
That was the main reason I bought GTX 480, it has no BS protection and can be put to ultimate pressure.
It survived, and I know that it's stable.
But it has the best air cooling there is, and I a couple of extra fans to keep it really cool.

To summarize, apps which stress only certain parts of GPU at a time are not the best to get the whole picture.

P.S.
OCCT has a great test with error checking. If a "bad pixel" will be rendered, you will know about it straight away. My personal recommendation as a stability check along FurMark.

[QUOTE=msft;291152]BIts and pieces.
Thank you for your work![/QUOTE]

1.49 & 1.61 -> only 11 hours to go and still a match at 18950000 of 26026433.

[QUOTE=Karl M Johnson;291322]Unfortunately, MemtestG80 wont stress test a graphics card enough to reveal any instability issues with it.
[SNIP]
P.S.
OCCT has a great test with error checking. If a "bad pixel" will be rendered, you will know about it straight away. My personal recommendation as a stability check along FurMark.[/QUOTE]

Thanks for the info. I was unaware of OCCT. Still testing, but I've seen [B]lots[/B] of errors. I have to revise my estimates of reliability..... EDIT: of my current OC.

[QUOTE=frmky;291216]I also just started 26026433 using a Linux build of version 1.58 on a card that I trust. ETA is 31 hours:[/QUOTE]
And it finished successfully. The Linux build of 1.58 appears fine.
M( 26026433 )C, 0x457f73d49f90b822, n = 1572864, CUDALucas v1.58

Attached is the list of good residues.

[QUOTE=frmky;291373]And it finished successfully. The Linux build of 1.58 appears fine.
M( 26026433 )C, 0x457f73d49f90b822, n = 1572864, CUDALucas v1.58

Attached is the list of good residues.[/QUOTE]

Awesome! My check will be done in a few hours...

I had good double check with the version I have compiled (my changes on top of 1.55)

[CODE]M( 27786383 )C, 0x5768f3aaf1f2d798, n = 1572864, CUDALucas v1.55apsen[/CODE]

Currently running another double check using 1.58. ETA 48 hours (I have two copies of mfaktc running on the same card).

BTW residues from 27786383:

OCCT
 

@[URL="http://www.mersenneforum.org/member.php?u=10186"]Karl M Johnson[/URL]:

Again, thanks for mentioning this program (OCCT). I'm getting close to a revised OC level. I still have to run a longer test.

OCCT slams the GPU temp up faster and higher than even FurMark.

My previous testing had relied on fairly long runs of memtestG80 without errors, and on passing the long self-test in mfaktc (--st).

(Sorry for getting a bit OT for this forum, but this is where the information came up, rather than the mfaktc section.)

[QUOTE=kladner;291403]@[URL="http://www.mersenneforum.org/member.php?u=10186"]Karl M Johnson[/URL]:

Again, thanks for mentioning this program (OCCT). I'm getting close to a revised OC level. I still have to run a longer test.

OCCT slams the GPU temp up faster and higher than even FurMark.[/quote]

FYI - nV drivers (and AMD, too) detect FurMark and throttle the card back to prevent it from self-destructing.

[QUOTE=kjaget;291417]FYI - nV drivers (and AMD, too) detect FurMark and throttle the card back to prevent it from self-destructing.[/QUOTE]

Thanks for that. I had some suspicion that was happening, but had not pinned it down.

Still a perfect game
 

[CODE]Processing result: M( 28652753 )C, 0x99059620de9d6320, n = 1572864, CUDALucas v1.58
LL test successfully completes double-check of M28652753[/CODE]No mismatch in my GPU lifetime yet.
Btw, no overclocking here.