Message boards :
Number crunching :
too short deadline
Message board moderation
| Author | Message |
|---|---|
|
Send message Joined: 10 Aug 08 Posts: 2 Credit: 11,906,694 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]()
|
Hi, I don't get too often some work for my GPU. Besides GPUgrid I get some work from primegrid. On my CPU's I run more than 30 other projects. My question for other users is do you also see that the deadline for GPU tasks from GPUgrid are too short and most of the time not beying even realistic? My GPU will not process that task before the deadline and I will get no credit for it. Maybe someone know how to extend the deadline? I can always cancel the task, but usually I only see it after it was completed more than 50-60%, so a lot of GPU time is already wasted... Thank you. Ionel |
Retvari ZoltanSend message Joined: 20 Jan 09 Posts: 2380 Credit: 16,897,957,044 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
There is no way to extend the deadline (which is 5 days, but after 2 days a WU will be sent to another host). GPUGrid needs fast WU turnarounds, because the next WU depends on the result of the currently processed one. Your current GPU (GeForce 9500 GT) is too slow for this deadline and therefore for this project too. See the GPU FAQ: Overview of cards that run the GPUGRID application thread for up to date information about recommended GPUs. |
|
Send message Joined: 10 Aug 08 Posts: 2 Credit: 11,906,694 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]()
|
The link is not available. I will not request new work from GPUgrid, then. I only have one computer with a GPU card which is even recignized by cuda. The rest either do not have a gpu card or is not nvidia/ati. Best regards, Ionel |
skgivenSend message Joined: 23 Apr 09 Posts: 3968 Credit: 1,995,359,260 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
|
|
Send message Joined: 6 Nov 07 Posts: 3 Credit: 0 RAC: 0 Level ![]() Scientific publications
|
I was upset about it too. I think GPUGRID.net needs to send only workunits that my computer can complete before of deadline. Boinc manager don`t send you the CPU benchmark? Today I access my account here and inside "GPUGRID Preferences" I set "No" to "ACEMD for long runs (8-12 hours on fastest GPU)" cause the last workunit was estimated to run for 140 hours (more than 5 days and increasing). Now I get a workunit that is estimated to run for 28 hours "only" (and increasing). My Sony Vaio have only a NVIDIA GEFORCE 310M (with only: 512MB DDR3; BUS WIDTH 64bit and 16 GPU cores) I don't know if the settings that I choose in my account are a solution...... I would like to help with this project too. Help me! (Sorry, but I speak portuguese :o) and I`m trying to write in english) |
Retvari ZoltanSend message Joined: 20 Jan 09 Posts: 2380 Credit: 16,897,957,044 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
Your GeForce 310M is also too slow for this project. Your current 33-KASHIF_HIVPR_cut_ba1-5-100-RND1811_0 workunit will be completed on your GPU in 3 days, while it will be reissued after 2 days to another (faster) host. You should crunch with this GPU for a different project with much shorter WUs. |
|
Send message Joined: 6 Nov 07 Posts: 3 Credit: 0 RAC: 0 Level ![]() Scientific publications
|
OK. I will (try to) finish this WU and think about it. :-) |
Retvari ZoltanSend message Joined: 20 Jan 09 Posts: 2380 Credit: 16,897,957,044 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
Just as I predicted, your current 33-KASHIF_HIVPR_cut_ba1-5-100-RND1811_0 workunit was reissued to another (faster) host. I guess it will complete this wu within 16 hours. However if your host finishes this wu before the other one starts processing it, it will be cancelled on the other host. |
Retvari ZoltanSend message Joined: 20 Jan 09 Posts: 2380 Credit: 16,897,957,044 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
The other host is finished processing of the workunit almost 2 hours ago. So now you can see: processing GPUGRID workunits with a GeForce 310M is only wasting your time and electricity (i.e. money). While you could make use of your GPU in another BOINC project. |
|
Send message Joined: 6 Nov 07 Posts: 3 Credit: 0 RAC: 0 Level ![]() Scientific publications
|
If remaining time was calculated and displayed in a more accurate way, it will be enough to know that I can't complete the task before the deadline with my hardware. The Work Unit was estimated to run for about 28 hours. And I`m seeing it here know: Progress: 54,279 Elapsed: 57:00:01 Remaing: 43:25:23 But I want to know another thing... This project was called PS3GRID too??? What can we do with a PS3 for the boinc community today? Thank You... (Now I`m in Buenos Aires, and I`m thinking if is a good choice buy a playstation at a Free Shop before return to Brazil ;-) |
Retvari ZoltanSend message Joined: 20 Jan 09 Posts: 2380 Credit: 16,897,957,044 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
If remaining time was calculated and displayed in a more accurate way, it will be enough to know that I can't complete the task before the deadline with my hardware. You are right about this. I've calculated your approximate runtime based on the runtime of a similar WU on a different GPU. But I want to know another thing... Yes. What can we do with a PS3 for the boinc community today? Nothing. In the meantime Sony made it illegal to run linux on PS3. Besides this, recent high end GPUs are much faster than PS3's Cell processor. (Now I`m in Buenos Aires, and I`m thinking if is a good choice buy a playstation at a Free Shop before return to Brazil ;-) To play games - yes. :) To use it for BOINC projects - no. :( |
nenymSend message Joined: 31 Mar 09 Posts: 137 Credit: 1,429,587,071 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
As I know, you can use PS3 for yoyo OGR and Primaboinca crunching yet. Both projects are mathematics.(Now I`m in Buenos Aires, and I`m thinking if is a good choice buy a playstation at a Free Shop before return to Brazil ;-) |
|
Send message Joined: 5 May 11 Posts: 1 Credit: 2,547,133 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]()
|
Hey guys, i finished my 77-KASHIF_HIVPR_cut_ba1-8-100-RND9442_0 today with a delay of 1,5 hours (it was not really my fault, usually it completes these tasks in time, but this time the BOINC manager started another CUDA-task while your task was still running and so it stopped it and it hadn't reached a checkpont, so that it had to start again from the last checkpoint and that's why it couldn't finish in time)...and i didn't get any credit, as far as i can see...i mean, come on, couldn't you just give me at least some credits for this task? My PC ran for about 5 days without a break to complete this one...i know i don't have the fastet GPU (NVidia 9500 GT), but still i give all the time i have for this project and this is my home PC! Best regards. |
|
Send message Joined: 10 Oct 08 Posts: 18 Credit: 39,100,916 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
Fanqua, it's simple. They want results fast, so the deadlines are tight. Moreover, if a task isn't completed in 2 days it gets sent to another host. I quite often see some of the tasks I get which are still running on another host but because their GPU is too slow it gets sent to me. This is a shame because it's just a waste of energy for the slow GPU to keep running it. Sometimes it does finish in time and get credit but old GPUs are better used in other projects like einstein or seti. |
|
Send message Joined: 11 Nov 09 Posts: 3 Credit: 558,331,226 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
When a work unit IS completed ON TIME but cannot upload due to network problems somewhere between my ISP and GPUGrid I waste my electricity. I spent several hours trying to get the upload going only to not get any credit. Enough is enough. Work done on time does not count on this project. My electric bill is high enough. If I spend money running your project on my GPU cycles done ON TIME and the work I donate to you is tossed back in my face saying it does not count. Nobody else but one person completed the WU on time and its marked as "Completed and validated" where mine is "Completed, too late to validate"? What other organization that should value voluntary participation treats volunteer participants like this? What incentive do I have to spend more of my money on your project? Not even a thank you. Nothing. Sheeesh. Excuse me. Guess its my fault for being stupid and expect reason. Time to lower expectations again. Bye. |
Carlesa25Send message Joined: 13 Nov 10 Posts: 328 Credit: 72,619,453 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
Hello: I understand your frustration, I happened to me something like, problems GPUGRID server I could upload some homework on time and I lost the rebate, yours is even worse. Not that PC has (hidden) but I guess if it takes several days to complete a task the GPU will be old or very little capacity. If your GPU is not in the list of those recommended in this project, I sincerely believe you'll get more satisfaction in another. Greetings. |
|
Send message Joined: 11 Nov 09 Posts: 3 Credit: 558,331,226 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
GPU is a GTX 460 so it should be okay. Turn arounds are not a problem with respect to computation time on that card so far. Upload was the problem. Fought that with SETI recently with their connectivity problems and not too long ago lost another 44K credits on this project like this time. Think they should have certificates for wasted credits too. Trying to upload one of my last WU on this project now so I get credit for what I am due then outta here. Upload is crawling right now like the other day when I wasted my time. Have fun. |
skgivenSend message Joined: 23 Apr 09 Posts: 3968 Credit: 1,995,359,260 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
You get credit for reporting a task, not completing it or uploading it. This is the same for all Boinc projects - don't report and you get no credit! Sounds like you may have needed to restart your router and then your system, but definitely an ISP issue. There is nothing more GPUGrid could have done about your Internet outages. GPUGrid already allows you to resume uploading a task from it's last percentage upload, unlike many other Boinc projects. One of the recommended GPUGrid configurations is to report tasks immediately. Are you using this? |
|
Send message Joined: 11 Nov 09 Posts: 3 Credit: 558,331,226 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
Router and modem rebooted mutiple times. I know how BOINC projects work. I am NOT a novice at this. |
|
Send message Joined: 16 Mar 11 Posts: 509 Credit: 179,005,236 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
Mole, I see very slow upload rates for GPUgrid as well. They are the slowest of any BOINC project I have ever seen. Unfortunately we have to live with that. They have to draw the line somewhere. They can't (or won't) give credits for results that report after the deadline and they won't extend the deadline because they need results back quickly. If your host can't meet GPUgrid's admittedly tough expectations then you need to withdraw and crunch for a different project. I'm sure they don't mind you lobbying for a longer deadline, some improvement to their upload speed or credits for results reported late but I think you'd be banging your head on the wall for nothing. |
©2025 Universitat Pompeu Fabra