Looks like I just loaded up my first 4.2 app.
it's a Paola, which normally take around 12-15 hours on my 560ti @920.

I've dropped it to 880 for this one, and it's indicating a 5:45:00 min runtime.
If that proves correct, that's an amazing result!

whoa, using 96% GPU and 597 Mb of Ram. Best utilization I have seen!

WTG!!!!

hmmmm, maybe got ahead of myself.
after 1:15:00, the countdown to completion is still virtually the same at 15:42:32.

I'll be watching closely to see how long this actually takes.

Hi All,

If you are interested in the increase in computational times, I've pasted my Nvidia 470 results below.

http://www.gpugrid.net/host_app_versions.php?hostid=82476

The recursive acronym at it's best.

http://www.gpugrid.net/host_app_versions.php?hostid=122075
http://www.gpugrid.net/host_app_versions.php?hostid=125945 :)

However you also need to look at each task type individually to see the variation in improvement:
I4R101-NATHAN_RPS1120528-13-166-RND7789_0 3518586 24 Jun 2012 | 5:58:48 UTC 24 Jun 2012 | 18:53:21 UTC Completed and validated 38,683.76 1,187.91 60,900.00 Long runs (8-12 hours on fastest card) v6.16 (cuda31)

I5R92-NATHAN_RPS1120528-11-166-RND0162_2 3507092 25 Jun 2012 | 21:44:05 UTC 26 Jun 2012 | 6:38:11 UTC Completed and validated 24,334.46 517.02 60,900.00 Long runs (8-12 hours on fastest card) v6.16 (cuda42)

GTX470 on Ubuntu 12.04 (195.40), Nathan tasks - 4.2 is 58% faster than 3.1.

Anyone wishing to exclusively run CUDA 4.2 tasks (ie not 3.1 and 4.2)?
I suggest people select no new tasks, finish any tasks in progress and then reset the project. This will delete the 3.1 App. If you reset while you have a task, and are running it on 3.1, it's probably going to start running it again and fail.

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

I will let that run and see if I get any more cuda3.1 WU's.

Well, crap. After the "reset project" in BOINC, I did get through several long cuda4.2 WU's before getting another cuda3.1 WU. So, seems like it is luck of the draw right now.

This problem of receiving multiple applications is probably a bug of the server.
I am going to investigate it with Boinc people or maybe do a server upgrade.

For now, you have to live with it. Sorry.

gdf

Well, sporadic cuda4.2 is better than no cuda4.2 at all. :-) Please do look into the potential server issues sending out both types of WU's to a cuda4.2-eligible machine.

Anyone wishing to exclusively run CUDA 4.2 tasks (ie not 3.1 and 4.2)?
I suggest people select no new tasks, finish any tasks in progress and then reset the project. This will delete the 3.1 App. If you reset while you have a task, and are running it on 3.1, it's probably going to start running it again and fail.

Doesn't seem like that is a permanent solution. I did that and after receiving 2-3 cuda4.2 WU's, it went back to grabbing a cuda3.1 without intervention.

6/26/2012 9:00:34 AM | GPUGRID | Sending scheduler request: To fetch work.
6/26/2012 9:00:34 AM | GPUGRID | Requesting new tasks for NVIDIA
6/26/2012 9:00:37 AM | GPUGRID | Scheduler request completed: got 1 new tasks
6/26/2012 9:00:39 AM | GPUGRID | Started download of acemd.win.2352
6/26/2012 9:00:39 AM | GPUGRID | Started download of cudart32_31_9.dll
6/26/2012 9:00:42 AM | GPUGRID | Finished download of cudart32_31_9.dll
6/26/2012 9:00:42 AM | GPUGRID | Started download of cufft32_31_9.dll
6/26/2012 9:00:49 AM | GPUGRID | Finished download of acemd.win.2352
6/26/2012 9:00:49 AM | GPUGRID | Started download of I2R137-NATHAN_RPS1120528-12-LICENSE
6/26/2012 9:00:50 AM | GPUGRID | Finished download of I2R137-NATHAN_RPS1120528-12-LICENSE
6/26/2012 9:00:50 AM | GPUGRID | Started download of I2R137-NATHAN_RPS1120528-12-COPYRIGHT
6/26/2012 9:00:51 AM | GPUGRID | Finished download of I2R137-NATHAN_RPS1120528-12-COPYRIGHT
6/26/2012 9:00:51 AM | GPUGRID | Started download of I2R137-NATHAN_RPS1120528-12-I2R137-NATHAN_RPS1120528-11-166-RND7611_1
6/26/2012 9:00:59 AM | GPUGRID | Finished download of I2R137-NATHAN_RPS1120528-12-I2R137-NATHAN_RPS1120528-11-166-RND7611_1
6/26/2012 9:00:59 AM | GPUGRID | Started download of I2R137-NATHAN_RPS1120528-12-I2R137-NATHAN_RPS1120528-11-166-RND7611_2
6/26/2012 9:01:06 AM | GPUGRID | Finished download of I2R137-NATHAN_RPS1120528-12-I2R137-NATHAN_RPS1120528-11-166-RND7611_2
6/26/2012 9:01:06 AM | GPUGRID | Started download of I2R137-NATHAN_RPS1120528-12-I2R137-NATHAN_RPS1120528-11-166-RND7611_3
6/26/2012 9:01:11 AM | GPUGRID | Finished download of I2R137-NATHAN_RPS1120528-12-I2R137-NATHAN_RPS1120528-11-166-RND7611_3
6/26/2012 9:01:11 AM | GPUGRID | Started download of I2R137-NATHAN_RPS1120528-12-pdb_file
6/26/2012 9:01:33 AM | GPUGRID | Finished download of I2R137-NATHAN_RPS1120528-12-pdb_file
6/26/2012 9:01:33 AM | GPUGRID | Started download of I2R137-NATHAN_RPS1120528-12-psf_file
6/26/2012 9:01:54 AM | GPUGRID | Finished download of cufft32_31_9.dll
6/26/2012 9:01:54 AM | GPUGRID | Started download of I2R137-NATHAN_RPS1120528-12-par_file
6/26/2012 9:01:58 AM | GPUGRID | Finished download of I2R137-NATHAN_RPS1120528-12-par_file
6/26/2012 9:01:58 AM | GPUGRID | Started download of I2R137-NATHAN_RPS1120528-12-conf_file_enc
6/26/2012 9:01:59 AM | GPUGRID | Finished download of I2R137-NATHAN_RPS1120528-12-conf_file_enc
6/26/2012 9:01:59 AM | GPUGRID | Started download of I2R137-NATHAN_RPS1120528-12-metainp_file
6/26/2012 9:02:00 AM | GPUGRID | Finished download of I2R137-NATHAN_RPS1120528-12-metainp_file
6/26/2012 9:02:00 AM | GPUGRID | Started download of I2R137-NATHAN_RPS1120528-12-I2R137-NATHAN_RPS1120528-11-166-RND7611_7
6/26/2012 9:02:01 AM | GPUGRID | Finished download of I2R137-NATHAN_RPS1120528-12-I2R137-NATHAN_RPS1120528-11-166-RND7611_7
6/26/2012 9:02:20 AM | GPUGRID | Finished download of I2R137-NATHAN_RPS1120528-12-psf_file
6/26/2012 9:02:21 AM | GPUGRID | Starting task I2R137-NATHAN_RPS1120528-12-166-RND7611_0 using acemdlong version 616 (cuda31) in slot 3

Looks like I was wrong/it doesn't work that way/the way it's supposed to. So server bug it is.

You have 301.42, and did get a 4.2 task after resetting, so why the 3.1?
I see you have a GT240 (CC 1.2) in one system, but that would not explain the other getting 3.1 tasks (unless a GT440 is deemed as being too slow)?

Which card is the 3.1 task running on?

cufft32_42_9.dll looks like it's been written for both 3.1 and 4.2, actually 3.2 and 4.2.

Anyway, if it's not working maybe Paul Raney was right in asking for separate queues for 4.2 and 3.1, though that might not work in a mixed GPU generation environment (if that matters).

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

ok i have the same problem now, i got a cuda 42 app and as next i got a cuda 32 app. intersting enough, i got no speedup on the cuda 42 WU with the 285gtx :( Seems to be a only speedup on fermi (where i got a real speedup) & kepler.

---

DSKAG Austria Research Team: http://www.research.dskag.at

You have 301.42, and did get a 4.2 task after resetting, so why the 3.1?
I see you have a GT240 (CC 1.2) in one system, but that would not explain the other getting 3.1 tasks (unless a GT440 is deemed as being too slow)?

Which card is the 3.1 task running on?

I have two systems:

A: 1x GTX570 (GPUGrid only) and 1x GT440 (non-GPUGrid, only PrimeGrid, etc.)
B: 1x GTX570 (GPUGrid only) and 3x GT240 (non-GPUGrid, only PrimeGrid, etc.)


I use the cc_config.xml options to specify which projects get which cards. I devote the 570's to GPUGrid (since I started with you guys), and have merely demoted old GPUGrid cards to other projects that accept them.


I am OK getting a mix of cuda4.2 and cuda3.1 tasks, though I hope I don't get ONLY cuda3.1 tasks now, unless I do a reset project. I'd love the 30%+ credit bonus, and I'm sure GPUGrid would love the 30%+ throughput increase from Fermi/Kepler cards. If you guys can identify the server issue (or whatever it is), I think it'll be a win/win.

But, for now, I'm happy to continue with the cuda3.2's with the hopes a nice cuda4.2 strolls down my street. :-)

I would suggest making the long queue CUDA 4.2 only, but I'm not sure how many people have a suitable driver, how many don't, and of those that don't have compatible drivers for CUDA4.2 how many run normal length tasks or both normal and long?

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

I would suggest making the long queue CUDA 4.2 only, but I'm not sure how many people have a suitable driver, how many don't, and of those that don't have compatible drivers for CUDA4.2 how many run normal length tasks or both normal and long?

As was suggested before, maybe a solution is to have 2 long queues selectable in the preferences: 1 for cuda3.1 long and 1 for cuda4.2 long? Not sure if that is easy/possible, but I know if I could deselect the cuda3.1 long and only select the cuda4.2 long, that'd be great for me. For those that want both or only cuda3.1 long, it still allows full flexibility.

Wonder how long they will even keep 3.1 with these kinds of results though?

It's just needed because some people use older drivers, and don't read the forums.

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

Why not just post that as of such and such date, you will no longer be issuing cuda3.1 work units, that anybody who hasn't already done so, to updated your driver to 301.xx, and then do it. This will save a lot of aggravation, and will increase overall number crunching totals, even while losing a few crunchers.

I see that resetting the project when only Long runs are selected just means that 'acemdlong' is used. It does not specify that it's CUDA4.2 or CUDA3.1.

So basically there is just one 'acemdlong' app and two CUDA variants within & we can't exclude a CUDA3.1 app.

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

Given that everyone needs to supply an e-mail address to sign-up,
why just do a mass e-mail advised users to upgrade their drivers?

You might even get some people who have long stopped crunching come back
after getting the e-mail.

Having the choice of either 3.1 or 4.2 would be good for those people who
for whatever reason, cannot upgrade their drivers. Program incompatability or such.

Also, just completed


I3R197-NATHAN_RPS1120528-16-166-RND3620_0http://www.gpugrid.net/result.php?resultid=5538958

25669.37 675.84 60,900 points - compared to

I2R58-NATHAN_RPS1120528-6-166-RND3113_0http://www.gpugrid.net/result.php?resultid=5525951

46355.21 1526.58 60.900 points

very nice improvement. This is on a 560ti @880. Now running a Paola @900. See how stable it is. I was able to run the 3.1 units @920

I'm still receiving CUDA 4.2 and CUDA 3.1 by turns on the same host.
I never thought that I'll become one of those who have to abort tasks selectively one by one....

For Windows users it's a good time to upgrade their drivers to 301.42, otherwise they won't receive CUDA 4.2 tasks (if they use the last recommended driver v285.58, or earlier)

Just poking around as a comparison for tasks, and the last MJHarvy I did was around 11.9 hours on my GTX 580. I'm set to finish one in about 5.1 hours on the same card. That is quite an improvement. Kudos, GPU Grid researchers.

However, as an FYI, in my poking around I note that this WU http://www.gpugrid.net/workunit.php?wuid=3514448 which is a CUDA 4.2 app WU, was sent to a PC with 296.10 drivers - http://www.gpugrid.net/show_host_detail.php?hostid=89848

---

That card's not doing too well, http://www.gpugrid.net/results.php?hostid=89848
But it is using one of the known bad drivers. There might be more wrong drivers in use than than right!

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

I certainly agree that the card is not doing well. Every once in a while, it does do a WU right - the corollary to Murphy's Law in action. LOL

Retvari posted that the GPU Grid server would not distribute CUDA 4.2 tasks to PCs running drivers less than 301.42. Seems this exclusion is not functioning properly.

I don't know what is going on with that, however, I know that Milkyway or Einstein (I am not sure which) was successful at implementing a general exclusion on the distribution of WUs to PCs with improper drivers because for a while, I was at one of the 295 or 296 drivers and stopped getting WUs from one of those projects. I noted on that particular project's site that they were requiring 301.XX drivers (due to the known issues with 295 and 296), and when I updated to 301.28 (beta at the time, but I've been at 301.42 for a few weeks now) I started getting WUs from that project again.

I'm not complaining, but trying to be helpful.

If anyone is interested, here's a 3.1 MJHarvey task completed on my GTX 580 and the 4.2 MJHarvey task my 580 just completed.The 4.2 task completed in slightly less than half the time, according to the WU status, than the 3.1 task did. IMHO, fantastic results.

I wonder how much of an improvement, if any, my GTX 460 will exhibit. It would be great if it started doing Paolas, which now take about 23 - 25 hours, in 20 or 22 hours as right now, they sometimes finish and report just outside of the 24 hour 1.5X bonus window.

I'll be experimenting with tasks on the 460 this weekend. :)

---

it distributes cuda4.2 to drivers higher than 295.43 which is the linux version for cuda4.2.

gdf

Retvari posted that the GPU Grid server would not distribute CUDA 4.2 tasks to PCs running drivers less than 301.42.

What I was actually saying is users with the last recommended driver (v285.58 or earlier) won't receive CUDA4.2 tasks. The v295 and v296 drivers are CUDA4.2 capable but not recommended, because the monitor sleep bug. Because this bug, many users rolled back to the v285 drivers, that's why it is a good idea to update the drivers to the v301.42.

Seems this exclusion is not functioning properly.

This part is true....