Hi,

I joined about a week ago and have managed to complete 4 work units so far. I have an Intel Q6700 (core 2 quad), 2GB of ram and 4.23GB of virtual memory, an XFX NVIDIA GeForce 8600 GTS running under Vista Home Premium 32 bit, Boinc 6.3.21, CUDA 2.0., NVIDIA driver s 178.08

This evening I saw a GPU task had just started but was in danger of missing its deadline. I aborted that task and the task next in the queue as that was on a similar deadline. Boinc immediately got 2 replacements. However none of the four wu's in the queue would start processing. I reported the aborted tasks and still no GPU task would start. I even reset the project. In the end I shut down the Boinc client and waited for the auto-restart. That did the trick and a GPU task is now running. But what a hassle!

GPU shares the computer with 4 Cosmology tasks usually and these were running this evening. The resource share is 100 for both projects (50:50).

The GPU tasks take just over a day to complete and I'm wondering if I've got the preferred software mix right. Would I be better advised to try CUDA 2.1 Beta and associated drivers or a different version of BOINC?


Thanks in advance,

Phoneman1

This evening I saw a GPU task had just started but was in danger of missing its deadline. I aborted that task and the task next in the queue as that was on a similar deadline. Boinc immediately got 2 replacements. However none of the four wu's in the queue would start processing. I reported the aborted tasks and still no GPU task would start. I even reset the project. In the end I shut down the Boinc client and waited for the auto-restart. That did the trick and a GPU task is now running. But what a hassle!

Wasn't that a problem with 6.3 Clients that they don't started GPU tasks immediately? I still had a 6.3.x on a Linux machine that did that.

After an upgrade to 6.4.2 it works fine...
____________

pixelicious.at - my little photoblog

Almost the same here too with Q6700 + 4GB-ram , NVIDIA 8800GS & Vista HP-64, Boinc 6.3.21, CUDA 2.0, Nvidia driver 178.08

I did notice that High Priority running disappeared for few days ago and one issue of that is the lack of new WU-loading from the server due to the change of reference for the client. Before and as explain somewhere else on this forum, the resource allocation hadn't any impact on the GPU-activity. This seem to have disappeared. However and when the four or two WU (se below) are computed, I only detach and reattach to get new ones.

Those issues are also valid for a less performant system based on PD 915 with XP, NV8800, & so on.

Hi,

I joined about a week ago and have managed to complete 4 work units so far. I have an Intel Q6700 (core 2 quad), 2GB of ram and 4.23GB of virtual memory, an XFX NVIDIA GeForce 8600 GTS running under Vista Home Premium 32 bit, Boinc 6.3.21, CUDA 2.0., NVIDIA driver s 178.08

This evening I saw a GPU task had just started but was in danger of missing its deadline. I aborted that task and the task next in the queue as that was on a similar deadline. Boinc immediately got 2 replacements. However none of the four wu's in the queue would start processing. I reported the aborted tasks and still no GPU task would start. I even reset the project. In the end I shut down the Boinc client and waited for the auto-restart. That did the trick and a GPU task is now running. But what a hassle!

GPU shares the computer with 4 Cosmology tasks usually and these were running this evening. The resource share is 100 for both projects (50:50).

The GPU tasks take just over a day to complete and I'm wondering if I've got the preferred software mix right. Would I be better advised to try CUDA 2.1 Beta and associated drivers or a different version of BOINC?


Thanks in advance,

Phoneman1

I'm going to up upgrade the Nvidia drivers to 178.24 and Boinc 6.4.2 today.

I'll let the dust settle to decide on 6.4.3 - I see there are more issues with that elsewhere on this forum.

I forgot to mention in my earlier post I did suspend the other project
in an attempt to get a GPU task started. Either I didn't wait long enough (quite possible) or it didn't work for me!!!

Thanks to all who took the time and trouble to reply.

Incidentally my other machine has such a feable graphics card it could not complete a GPU task in 4 days - a solution to that problem is on order!

Phoneman1

I access one of my home crunching computers a lot by MS Terminal Services. I have a feeling that if I access the computer remotely from my work laptop, that when it finishes completing the current work unit on my CPU (8800GT) that the next one doesn't start correctly and errors out. IF it connect to the computer directly at its console again before the work unit completes, then the next work unit starts. This is a theory at this point only.

I can only guess that since I have connected from a laptop that doesn't have a valid GPU to process this type of work, and if I remain connected to it, that when the current WU completes and the next WU starts, it sees the GPU card on my laptop instead and errs out?

This is only a guess on my part, and I'm still trying to prove this theory. Does anybody else access their crunching home computer a lot via terminal services and see this behavior?

Does anybody know what should happen to a crunching computer if you access it remotely from a non-compatible laptop via terminal services?
____________
Crunching for the benefit of humanity and in memory of my dad and other family members.

I access one of my home crunching computers a lot by MS Terminal Services. I have a feeling that if I access the computer remotely from my work laptop, that when it finishes completing the current work unit on my CPU (8800GT) that the next one doesn't start correctly and errors out. IF it connect to the computer directly at its console again before the work unit completes, then the next work unit starts. This is a theory at this point only.

I can only guess that since I have connected from a laptop that doesn't have a valid GPU to process this type of work, and if I remain connected to it, that when the current WU completes and the next WU starts, it sees the GPU card on my laptop instead and errs out?

This is only a guess on my part, and I'm still trying to prove this theory. Does anybody else access their crunching home computer a lot via terminal services and see this behavior?

Does anybody know what should happen to a crunching computer if you access it remotely from a non-compatible laptop via terminal services?

I access constantly using remote desktop to a home premium vista 64 and have not seen this problem.

All the way back to the original post .... :)

All versions of BOINC Manager will not handle the queuing of work correctly ... the problem is that all versions assume that there is no such thing as a GPU ... and that all tasks run on the CPU ... as far as scheduling goes. So ...

THIS MEANS that if you have 50 / 50 share it is possible that the GPU will run out of work and BOINC Manager will be fine with that ... then it will later try to get work and sadly with the server issues that is still hit and miss ...

Though strangely I got two tasks into my queue almost as if it was working as it should ...

I guess what I am saying is that as far as keeping the GPU running what has been working for me on my XP Pro hosts is 180.84 video drivers, 6.5.0 BOINC Manager (yes it is beta and no it does not have any fixes for what ails us), resource share 500 (highest on the machine indicates 35% RS on the machine) ... the prime machine is running a mix of SIMAP, Cosmology, AI, LHC and Pirates (the last two only when I can get tasks obviously) with the remains of WCG tasks which should be done sometime tomorrow (if my calculations are right I should have enough work queued to get CEP Gold and as I will be still working WCG elsewhere I will still rack up earnings) ...

WIth all of that, I still babysit the machine and at times it will run dry but on the manual Update will pull work ... many times update before the queue is empty (like you have the last one in work and have a few minutes to go ...) you may or may not be successful getting new work ...

Early days ... were it now not a new goal and essentially "free" I would say a Pox on you (see my notes in Rosetta when they crashed half of my tasks after wasting upwards of 20 hours on a 6 hour task), but I digress ...

I am really of a mind that many of the manual interventions we have been doing have been nothing other than a pigeon test and none of them really affect the situation ... we only think they do in that we get work after performing the elaborate ritual ... Mine was up to 4 hours and involved, among other things ... wait a moment, why should I give away the secret of my magic?

:)

All the way back to the original post .... :)