Welcome to MilkyWay@home

concurrent blocks / nvidia .. question :)


Advanced search

Message boards : Number crunching : concurrent blocks / nvidia .. question :)
Message board moderation

To post messages, you must log in.

AuthorMessage
Jean-Michel Bruenn

Send message
Joined: 23 May 08
Posts: 4
Credit: 1,037,263
RAC: 0
1 million credit badge13 year member badge
Message 45960 - Posted: 4 Feb 2011, 10:47:33 UTC
Last modified: 4 Feb 2011, 11:08:14 UTC

Hello,

i just found:

Amount of concurrent blocks(NVIDIA GPUs Only)
Higher values lead to shorter completion times at the expense of interface lag (1 .. 50000, default is 128)


Someone could explain that a bit more to me? I googled a bit, tho i'm still unsure about it. Also some suggestion would be nice, i got 2 nvidia 9800 GT (with 1024 mb ram and 112 stream processors each)


02:00.0 VGA compatible controller: nVidia Corporation G92 [GeForce 9800 GT] (rev a2)
03:00.0 VGA compatible controller: nVidia Corporation G92 [GeForce 9800 GT] (rev a2)


What would be a good setting? :)

thanks in advance

Edit:

ah. they're running in SLI mode. Just found out, would probably not work at all:

NVIDIA GPU 0: GeForce 9800 GT (driver version unknown, CUDA version 3020, compute capability 1.1, 1024MB, 336 GFLOPS peak)
NVIDIA GPU 1: GeForce 9800 GT (driver version unknown, CUDA version 3020, compute capability 1.1, 1024MB, 336 GFLOPS peak)


it needs 1.3 right? though there's no message like "your gpu lacks..."
ID: 45960 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Matt Arsenault
Volunteer moderator
Project developer
Project tester
Project scientist

Send message
Joined: 8 May 10
Posts: 576
Credit: 15,979,383
RAC: 0
10 million credit badge11 year member badge
Message 46057 - Posted: 8 Feb 2011, 0:52:11 UTC - in response to Message 45960.  

[quote]
Amount of concurrent blocks(NVIDIA GPUs Only)
Higher values lead to shorter completion times at the expense of interface lag (1 .. 50000, default is 128)
This doesn't apply anymore; that setting isn't there. You don't need to think about GPU minutia. I was considering adding a less confusing option for the OpenCL one.


ID: 46057 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Astromancer.
Avatar

Send message
Joined: 21 Nov 09
Posts: 49
Credit: 20,942,758
RAC: 28
20 million credit badge12 year member badge
Message 46089 - Posted: 9 Feb 2011, 13:39:51 UTC

Besides the setting no longer doing anything, you are correct that you need a compute capability 1.3 or higher card to run the project. So you won't get any work for them from MW.
ID: 46089 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Jean-Michel Bruenn

Send message
Joined: 23 May 08
Posts: 4
Credit: 1,037,263
RAC: 0
1 million credit badge13 year member badge
Message 46275 - Posted: 18 Feb 2011, 0:54:46 UTC

thats sad, especially as i got two cards in this box.
ID: 46275 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : Number crunching : concurrent blocks / nvidia .. question :)

©2021 Astroinformatics Group