Questions and Answers :
Windows :
Distributed.net has stopped working ..
Message board moderation
Author | Message |
---|---|
Send message Joined: 6 Dec 11 Posts: 60 Credit: 306,719,331 RAC: 0 |
Anyone else get this. It used to be manageable, but now with all 4 GPUs in one box it's 16 of them at once. Problem signature: Problem Event Name: APPCRASH Application Name: dnetc518-win32-x86-stream.exe Application Version: 2.9103.509.0 Application Timestamp: 4d0a6e2d Fault Module Name: aticaldd.dll Fault Module Version: 6.14.10.1589 Fault Module Timestamp: 4e95edbd Exception Code: c0000005 Exception Offset: 0017e048 OS Version: 6.1.7601.2.1.0.256.1 Locale ID: 1033 Additional Information 1: 0a9e Additional Information 2: 0a9e372d3b4ad19135b953a78882e789 Additional Information 3: 0a9e Additional Information 4: 0a9e372d3b4ad19135b953a78882e789 Read our privacy statement online: http://go.microsoft.com/fwlink/?linkid=104288&clcid=0x0409 If the online privacy statement is not available, please read our privacy statement offline: C:\Windows\system32\en-US\erofflps.txt |
Send message Joined: 20 Apr 11 Posts: 388 Credit: 822,356,221 RAC: 0 |
That seems to be a segmentation fault at the ATI driver level. I would concenrate troubleshooting efforts at the ATI installation on that host. Sometimes Windows 7 Data Execution Prevention (DEP) is responsible for these faults if it's enabled for all/certain applications. Of course, if applications/driver does something like this, it would indicate a bug.. -w |
Send message Joined: 22 Jun 11 Posts: 2080 Credit: 1,844,408,008 RAC: 2,455 |
Problem signature: Try renaming the app-info file so it doesn't load and see if it works then, if so then you can see what is wrong in there. |
Send message Joined: 22 Jun 11 Posts: 2080 Credit: 1,844,408,008 RAC: 2,455 |
That was the problem but that was the only way I knew to run one GPU per WU. Do you know of a better way? There is a small program, let me look for it and I will get back to you. |
Send message Joined: 6 Dec 11 Posts: 60 Credit: 306,719,331 RAC: 0 |
That was the problem but that was the only way I knew to run one GPU per WU. Do you know of a better way? Interesting... |
Send message Joined: 22 Jun 11 Posts: 2080 Credit: 1,844,408,008 RAC: 2,455 |
That was the problem but that was the only way I knew to run one GPU per WU. Do you know of a better way? Okay I found one thread that is similar to what you are doing now: http://moowrap.net/forum_thread.php?id=215 Do NOT go just by the thread subject, it kinda bounces around but ends up with an app_info.xml file. I am still looking for what I think I saw that is better!! |
Send message Joined: 6 Dec 11 Posts: 60 Credit: 306,719,331 RAC: 0 |
Yeah, I attribute the problem to having more then 2 GPU's in a system and running 2 WU's per GPU during the fragmentation period(s) and the main culprit being Windows and driver problems. I was having the issues when I had (8) GPU's in the system it was outta control on the APPCRASH's, if I left them to crunch 1 WU on the 8 GPU's together it was just too slow and the time from a WU finishing, and starting another one was almost 2.5 minutes, about half the time it took to complete a single "HUGE" WU with 8 5870s. This meant the system was getting only 1/2 of what it could and just wasn't going to work for me. I made almost $400 profit on the (4) 5970s I had so it allowed me to get (2) 7970s, which honestly are great cards, but for the price, on this project, you could do 300% more work with (4) 5970s in 2 different boxes. |
Send message Joined: 22 Jun 11 Posts: 2080 Credit: 1,844,408,008 RAC: 2,455 |
That was the problem but that was the only way I knew to run one GPU per WU. Do you know of a better way? I think I found it!!!: http://bitsum.com/ ps it is FREE!! |
Send message Joined: 6 Dec 11 Posts: 60 Credit: 306,719,331 RAC: 0 |
That doesn't spread the GPU work though, still need app_info for that. I've had PL for months. |
Send message Joined: 25 Jul 11 Posts: 5 Credit: 23,397,372 RAC: 0 |
Well, I discovered something interesting... I have an HD6990 and when I use the linked app_info file, it says it is running one task per GPU, but seems to be using both of them anyway! I suspended the pending tasks, then suspended ONE task and suddenly the GPU's go to 99% again... both of them. Try detaching, reinstall clean drivers, rebooting and all the tricks... it insisits on using BOTH GPU's even when the app_info and the status line say it is using only one... LOL! I was wondering why it was taking so long... :) |
Send message Joined: 22 Jun 11 Posts: 2080 Credit: 1,844,408,008 RAC: 2,455 |
Well, I discovered something interesting... I have an HD6990 and when I use the linked app_info file, it says it is running one task per GPU, but seems to be using both of them anyway! Since Moo is not designed to be Nvidia friendly, you might be better off just excluding the Nvidias here and using them on GpuGrid or PrimeGrid instead. Most projects work better with one platform or the other, Moo is better with AMD. That does NOT mean it won't work with Nvidia cards, they just won't be at their best. Since they are competitors, Nvidia and AMD I mean, they do things differently, they get the same answer in the end though. |
Send message Joined: 25 Jul 11 Posts: 5 Credit: 23,397,372 RAC: 0 |
Well, I discovered something interesting... I have an HD6990 and when I use the linked app_info file, it says it is running one task per GPU, but seems to be using both of them anyway! Umm, an HD6990 is an AMD/ATI card... like two 6970's on one card... and it happens to be in a box driven by a 3.75GHz 1100T CPU... Soo, not sure how your comment applies to my problem... :D |
Send message Joined: 22 Jun 11 Posts: 2080 Credit: 1,844,408,008 RAC: 2,455 |
Well, I discovered something interesting... I have an HD6990 and when I use the linked app_info file, it says it is running one task per GPU, but seems to be using both of them anyway! I thought you were talking about the machine that has these in it: [2] NVIDIA GeForce GTX 560 Ti (1024MB) driver: 290.53 |
Send message Joined: 25 Jul 11 Posts: 5 Credit: 23,397,372 RAC: 0 |
Ohh. Nope... talking about the Win7-955 machine... Used to have a 955 CPU... upgraded to 1100T a while back... :) |
Send message Joined: 22 Jun 11 Posts: 2080 Credit: 1,844,408,008 RAC: 2,455 |
Ohh. Nope... talking about the Win7-955 machine... Okay let's step back into your problem then, you said: "Well, I discovered something interesting... I have an HD6990 and when I use the linked app_info file, it says it is running one task per GPU, but seems to be using both of them anyway!" This is normal until you finish your first few batches of units then the software figure out you have multiple gpu's and starts running one workunit on all the gpu's. When I added a 2nd gpu to a machine a while back, it took 24 hours for it to figure it out. Now I do NOT use an app_info file, maybe you could disable yours for a day or so and see if this happens as a test, or if you are happy as it is just let it crunch. |
Send message Joined: 22 Jun 11 Posts: 2080 Credit: 1,844,408,008 RAC: 2,455 |
Faulting application name: dnetc518-win32-x86-stream.exe, version: 2.9103.509.0, time stamp: 0x4d0a6e2d I don't know but I see this for your pc: Operating System Microsoft Windows 7 Home Premium x64 Edition, Service Pack 1, (06.01.7601.00) Yet above in the log I see "dnetc518-win32-x86-stream.exe" could this be the problem? Are you running a 32 bit version of Boinc on a 64 bit machine? When I look at one of your valid units I see this.."dnetc v2.9109-518-CTR-10092920 for ATI Stream on Win32 (WindowsNT 6.1)". So maybe I am barking up the wrong tree, but I do know that Windows 64 bit and Windows 32 bit handle memory differently! |
Send message Joined: 22 Jun 11 Posts: 2080 Credit: 1,844,408,008 RAC: 2,455 |
This is the version of BOINC i'm using- I just checked one of my valid units, on a 64 bit pc, and see this: running dnetc518-win32-x86-stream.exe Yours says: running dnetc518-win32-x86-stream.exe So I guess it is right, hmmmm. One BIG difference is your version of the AMD software is newer than mine, you are using: driver: 1.4.1703 While mine says: driver: 1.4.1546 Meaning unless you are gaming you probably should downgrade to version 11.9 of the software, which is what I am using. I think the newest version is 12.1 or 12.3, but any release after 11.9 benefits gamers not crunchers. You can download the older drivers at the bottom of the page where you can download the latest drivers from AMD. |