1) Message boards : Number crunching : PCIe 2.0 and GTX 970 (Message 42787)
Posted 1 day ago by Profile Retvari Zoltan
In support of this setup I've been looking at my app_config and trying to figure out how to adjust it such that the top card (980Ti/970) will run two tasks at once and the bottom card (750Ti) only run one task. Is this possible? I'm not seeing any tags on the BOINC Client Config guide that would apply to this.
I can imagine only one way to achieve this: to install two separate BOINC managers in separate program & data folders, then set up the 1st to run two workunits on the GPUs and exclude the lesser GPU; then set up the 2nd to exclude the bigger GPU. I've never done it, but theoretically it could work.
2) Message boards : Number crunching : PCIe 2.0 and GTX 970 (Message 42771)
Posted 4 days ago by Profile Retvari Zoltan
So the CPU's 16 PCIe lanes don't divide up evenly between the 3.0 and 2.0 slots? I like your idea if that is the case.
No, the PCIe3.0x16 comes directly from the CPU, and divided between the two PCIe3.0 slots if necessary (when there is a GPU in the 2nd PCIe3.0 slot), while the PCIe2.0 comes from the "South Bridge" (Z97) chip.

3) Message boards : Number crunching : PCIe 2.0 and GTX 970 (Message 42769)
Posted 4 days ago by Profile Retvari Zoltan
I would put the GTX 750Ti to the PCIe 2.0 slot as well. Your hosts have OSes with WDDM, and the PCIe bandwidth limitation has lower impact on GPU performance than WDDM, so there's no point giving a mid-range card (like the GTX 750Ti) a PCIe3.0x8 slot and limit the airflow of a high-end card and limit its PCIe bandwidth to x8 at the same time.
4) Message boards : Number crunching : PCIe 2.0 and GTX 970 (Message 42766)
Posted 4 days ago by Profile Retvari Zoltan
I just picked up a GTX 970 to add to this project and am contemplating where to place it in the computer. I have three PCIe slots I could place it in on my Asus Sabertooth Z97 board. The top two are PCIe 3.0 x16 and the bottom is PCIe 2.0 x16.
The bottom slot is mechanically x16, but electrically it's only x4, so it's 8 times slower than the top slot.

The top slot has a GTX 980Ti in it and I was thinking about placing the new 970 in the bottom slot to give the cards a little more "breathing room" but I don't know how much of a difference placing it in a PCIe 2.0 slot makes.
It depends on the workunit type. For example the recent GERARD_A2AR workunits will be significantly slower in that slot, the others won't.

Both cards are EVGA with the ACX 2.0 cooling which means they're dumping heat into the case, but I have a rather large case with really good airflow (Phanteks Enthoo Primo) so I don't know how much of a problem it will be with a second card. For reference, my 980Ti nearly always stays under 60C on GPUGrid with the default fan curve.
For these cards a fan placed on the side plate blowing the hot air outwards is recommended.

So it comes down to, is it better to place both cards in PCIe 3.0 slots and have them run warmer, or to skip a slot and have one card in a PCIe 2.0 slot?
I would place it in the lowest slot to give breathing room for the 980Ti, regardless of the performance decrease it could bring. Moreover if you put the 2nd card in the 2nd PCIe3.0 slot, then both will run at only x8, reducing the PCIe bandwith for the GTX980Ti, which should be avoided.
5) Message boards : Number crunching : Gridcoin (Message 42751)
Posted 8 days ago by Profile Retvari Zoltan
Getting myself setup with this now. Am I correct in thinking that since Gridcoin uses my CPID to track me, I do NOT need to install a client on all my crunching machines?
Yes. But all of your hosts have to have the same CPID.
6) Message boards : Number crunching : Nvidia GPU takes 5 days to complete a project? (Message 42747)
Posted 9 days ago by Profile Retvari Zoltan
any projects you recommend that this GPU would work on just fine?
POEM@home (OpenCL), Einstein@home (CUDA), SETI@home (CUDA).
7) Message boards : Number crunching : Nvidia GPU takes 5 days to complete a project? (Message 42744)
Posted 9 days ago by Profile Retvari Zoltan
I have an older alienware laptop and the model is an m14x r1

it has a gt 555m gpu. why is it that when I go to do a project, that it says 5 days to complete?
Because this GPU's computing capability is this low.

I would not mind except for the fact that it passes the deadline. Are there any ways that I can extend the deadline or make it less than 5 days?
No, there isn't.

I just find this really weird since I have an old gtx 460 that can process faster than this..
The "X" marks the top-end cards in NVidia's naming conventions, so you are comparing a mid mid-range mobile GPU to a low top-end desktop GPU. Generally desktop GPUs can be faster because they have higher TDP limits than mobile GPUs. If you compare their specifications (GTX 460 vs GT 555m) it is quite clear, why the GTX 460 is faster:
GTX 460 GT 555m CUDA cores 336 144 Graphics clock 675 MHz up to 753 MHz Processor clock 1350 MHz up to 1506 MHz Texture Fill Rate 37.8 up to 16.2 Billion/sec
8) Message boards : Frequently Asked Questions (FAQ) : Not getting new work! (Message 42739)
Posted 9 days ago by Profile Retvari Zoltan
Less is more sometimes.
9) Message boards : Frequently Asked Questions (FAQ) : Not getting new work! (Message 42730)
Posted 10 days ago by Profile Retvari Zoltan
So 31 Jan 2016 22:56:29 CET | | [work_fetch] target work buffer: 864000.00 + 864000.00 sec
10+10 days is a very long buffer, as GPUGrid has a 5 days deadline, a 1 day +50% bonus and a 2 days +25% bonus.
GPUGrid gives at most 2 WU per GPU for a host, so there's no point in setting longer work buffer than the time it takes for your host to process 2 GPUGrid tasks.
So 31 Jan 2016 22:56:29 CET | | [work_fetch] --- project states ---
So 31 Jan 2016 22:56:29 CET | WUProp@Home | [work_fetch] REC 0.001 prio -0.000001 can't req work: non CPU intensive
So 31 Jan 2016 22:56:29 CET | Milkyway@Home | [work_fetch] REC 6.593 prio -0.001253 can req work
So 31 Jan 2016 22:56:29 CET | SETI@home | [work_fetch] REC 0.000 prio -0.013288 can req work
So 31 Jan 2016 22:56:29 CET | Albert@Home | [work_fetch] REC 461.612 prio -0.037942 can req work
So 31 Jan 2016 22:56:29 CET | Einstein@Home | [work_fetch] REC 5763.092 prio -0.490314 can req work
So 31 Jan 2016 22:56:29 CET | GPUGRID | [work_fetch] REC 6296.189 prio -0.509021 can req work
So 31 Jan 2016 22:56:29 CET | Collatz Conjecture | [work_fetch] REC 3707.924 prio -0.621823 can req work
So 31 Jan 2016 22:56:29 CET | PrimeGrid | [work_fetch] REC 70349.070 prio -5.706109 can req work
REC stands for Recent Estimated Credit. By default the BOINC manager takes a period of 10 days as a base for this estimation.
You can set a shorter period in the cc_config.xml file by using
<rec_half_life_days>X</rec_half_life_days>
I recommend to you to set the REC half life for 1 day, and check if it results in the desired behavior.
If not, then you should try a lager number, or a multiply of 10.
prio stands for "task requesting priority" (not the same as the "resource share" you set in the BOINC manager)
It is clear from the above list, that GPUGrid is not the highest priority according to the BOINC manager's calculation. The reason for that is GPUGrid gives very high credits for it's (long) tasks compared to other (GPU) projects which have to process much more workunits to achieve the same amount of credits (hence their larger priority).
If you want to focus on GPUGrid, it is advisable to have less GPU projects on the same host, or set the other project's resource share to 0. In this case the BOINC manager will request work from these projects when there's no work available from all of the non-0 resource share projects.
You can try to disable GPU tasks in your profile on other project's webpages in a venue created specially for GPUGrid hosts.
This "task requesting priority" calculation is not well documented, but it can mess up when there are many mixed (GPU-CPU) projects in the BOINC manager.
10) Message boards : Frequently Asked Questions (FAQ) : Not getting new work! (Message 42720)
Posted 12 days ago by Profile Retvari Zoltan
the deadline 2016-Jan-12 is over and the problem of getting no new tasks ist still not solved!!!
Everyone else is getting work, so this problem is *not* the project's fault.

Fr 29 Jan 2016 21:02:09 CET | GPUGRID | Not requesting tasks: don't need (CPU: job cache full; NVIDIA GPU: not highest priority project)
This line explicitly says that your BOINC manager does not request work. The project can't send work to a host which does not request it.

"NVIDIA GPU: not highest priority project" is definitely WRONG!!!

I had set the GPUGRID resources to 1000000 and all the other projects just to 1.
So GPUGRID has definitely the highest priority but my computers do not get new tasks independently how many times I request for new ones!
This is a very good observation, so there's no point in clicking on that button without changing some settings.

GET THAT PROBLEM FIXED!!!
Sure, but you've forgotten to say the magic word.

Otherwise GPUGRID will be dispunged!
Perhaps you've learned from Shakespeare to threaten the forum, but tell me how much loss would be a not working host for the project?
Your threatening is exactly that much serious.

To solve your problem:
I think your work buffer cache settings are inappropriate for the number of projects your host are attached to.
I would increase the work buffer, or reduce the number of CPU projects on that host.


Next 10