Posts by samlepirat
|
1)
(Message 2537)
Posted 17 Feb 2014 by 
samlepirat
Post: Precision refers to the number of bytes used to represent numbers. Single precision numbers use 32 bits whereas double precision numbers use 64 bits. More bits allows larger numbers to be represented but operations (addition, subtraction, multiplication, division, etc.) on double precision numbers are slower than operations on single precision numbers. In other words it takes more time to add 2 double precision numbers than it does to add 2 single precision numbers. Ok, understood, and seams logical! ;o) Some GPUs have far more CUDAs with which to perform double precision calculations and they can do more double precision calculations per second than a GPU with fewer CUDAs. Your GTX 650 has relatively few CUDAS so it is relatively slow when a task has many millions of double precision calculations. In fact, you're right about my GTX, it have 384 Cuda Cores! It appears to be a great number, but, after a few searches, now I know that is not so much. For example, the graphic card I wish to own next (in a few months, maybe for next xmas) contains 1152 Cuda cores, in my words : "C'est genre 3 fois plus trop plus mieux!"* In my previous post I was theorizing that perhaps your card's double precision power is so small that it cannot keep up with Asteroids tasks and the demands of the display and that the WDDM portion of Windows thinks it's so slow the driver has crashed. That's just a theory, it might not be correct. Thank you so much, for the very good explanations you gave me, and for the time you spent to write it down here. I don't know if you're right or not, but it makes sense, and all about the Cuda 32/64bits tasks is more clear to me now. ;) * Trying to translate "my words" myself, I also ask google to do the same ..... Google result : It's kind 3 times more better too! (Pragmatic, Predictable, Synthetic, I don't like it.) My way : It's kind 3 times more too much of more better! (Yeah, that's my words, smily, inflated & fun!) P.S.: Minecraft and Boinc at the same time ;) A larger (and readable) copy of this picture is available here! 
|
|
2)
(Message 2534)
Posted 16 Feb 2014 by
samlepirat
Post: For more details on how the cc_config.xml file works click the Official BOINC Wiki link in my signature below, then click the Client configuration link. Note that you can exclude tasks for only NVIDIA but while allowing tasks for ATI and vice-versa. Ok, I just bookmarked the page about the xml config file and I'll see what it can do :) Thanks!! If you have multiple cards you can exclude tasks for one of your cards but allow tasks for the other(s). There are numerous possible combinations. Sorry ^^, I just have a GTX650 card, on a mini-ITX main-board, in a tiny "Elite 120" computer case who cannot receive more than 1 card (not enough space) ;o)  OK, now I understand, you don't want any CUDA tasks from this project. Yes, but, it's not about what I want or not, but what it works or not. I want to help Asteroids@Home as much as I already do for Seti, Poem or Milkyway@Home projects, but not if it causes troubles to my computer. I hope I'll be able to authorize GPU work units again soon. It is possible that your GTX 650's double-precision power is so slow that it slows your screen updates too. Perhaps the Windows Display Driver Manager (WDDM) notices the slowness and produces the black screen to prevent a complete system lock-up? I'm just guessing, I don't know Windows very well. My GTX650's double-precision power ??? I'm confused, I really don't get what this means. My GTX authorizes me to play Minecraft on a screen (1920x1080 @50-60 fps), during a Cuda Work Unit is on going for Einstein@Home, and displayed on the second screen!! All that is done without any slowing down times (or maybe very little times, like a second at 40fps on the game, 2 or 3 times an hour). About my black screens, they were done by Windows, during the 5 to 8 seconds it takes when it (successfully) kills both nvidia driver and cuda55 task, and relaunches the graphic driver. Bravo Windows 8.1, never crashed, never freezed, never ever... My graphic "slow times" were only made during Asteroids cuda55 WU !! |
|
3)
(Message 2532)
Posted 16 Feb 2014 by
samlepirat
Post: Thank you all for your helping answers ;o) I am not sure but part of your problem could be the 332.21 version of drivers, I am using version 327.23 I don't know if the driver is the problem or if it have a problem. But except on Asteroids@home cuda55 tasks, everything is working very fine, and I really enjoy the "Shadow Play" feature from my actual driver revision (who can record on the fly HD videos from live game sequences, very nice). Another idea would be to upgrade to the newest standard version of Boinc which is now 7.2.39. I'm currently using it ;) If you don't want Asteroids@home to send you CUDA tasks go to Asteroids@home preferences [...] and uncheck: Use NVIDIA GPU Done, thank you ^^ Or you can put the following in cc_config: [...] The main difference is here they use floating point and most of the recent GPU's from both vendors are floating point crippled by their manufacturers (to make you buy their expensive offerings). Hum.. I'm not sure to understand... Inserting the lines you gave me will authorize GPU works from Asteroids@home, but not if they're "cuda types" work units ?? Thanks again, Sam :: from "Pays de Loire" (France) ;) |
|
4)
(Message 2524)
Posted 14 Feb 2014 by
samlepirat
Post: Hello !! I'm new, very new, I subscribed to asteroids@home just a few hours ago, and the really 1st WU I get was the "Period Search Application 101.11 (cuda55)". I left my computer for a moment, and when I came back, mouse and all others little graphic animations were eeeeextreeeemmmeeeeely sloooooowww... My 1st reaction : I suspend all calculations, my screens turned to black during a few seconds, and turned back on with the following error message : "The NVIDIA OpenGL driver detected a problem with the display driver and is unable to continue. The application must close. Error code: 3". I rebooted my machine, and, aaarrrgg, again !! All was so slow, like trying to run a DirectX11 game on a 333MHz Intel Celeron from 1997, it worked at 4 sfasf (seconds for a single frame). 2nd reaction : I suspend GPU calculations, same result, I knew the driver was completely involved!! After searches on the web (with useless visits on many nvidia support pages), I found you, having the same kind of problems, with the same Cuda55 WU. So, I did not understand all what you exactly did, but what (and how) can I do to see my computer completing CPU & GPU WUs for all my projects, without skipping manually all the future Asteroid's cuda55 WU I'll receive ?? FIY, my computer : Windows 8.1 Pro(x64) / Boinc 7.2.39(x64) / nVidia 332.21 Driver/ i5-3470S / GA-H77N-Wifi / GV-N650OC-2GI (GTX650) / 2x8Go@1600MHz DDR3. Thank for all, and sorry for my bad english verbs/times accords, mistakes and errors. Sam :: from "Pays de Loire" (France) ;) |
