Message boards :
Number crunching :
nvidia GT630 rev 2
Message board moderation
Author | Message |
---|---|
Bil Send message Joined: 27 Jan 01 Posts: 76 Credit: 1,887,795 RAC: 0 |
Hello! yesterday buyed subj card ( note that - revision 2 - have an 64 bit memory bus, instead of older 128 bit, but have more CUDA cores - 384 vs 96 - better performance in GFLOPs, and smaller energy consumption - 25w versus about 50w). got an ASUS GT630SL1GD3L - with 1 Gb VRAM and passive cooling. for first -it was really small card. in comparission with nvidia 9400GT it shows about 14000 FPS in glxgears (linux) versus about 4300 fps or so on on 9400GT. on that moment looks like a performance on astropulse in seti is about 4 times better than 9400gt. as so till now i be happy for that good card for only 45 eur for new, with 3 years of warranty. i do not test it long without ventilator, and it reaches about 75 degrees celsium in about 23c room. i think, it can be got a bit higher, but i want to go sleep, and turn on 80 mm ventilator, who on 5 volts push ait in card and chipset radiators direction - then temperature drops to 49 - 54 celsium. http://en.wikipedia.org/wiki/List_of_Nvidia_graphics_processing_units#GeForce_600_Series i have also some questions - what is with VRAM on boinc projects and seti - 1 Gb is always more than enought, or better is get 2 Gb version? on that cards ( 630,640,650) is any gain to run more than one task in parallel? if yes, what is about optimal number? any other advances? |
ivan Send message Joined: 5 Mar 01 Posts: 783 Credit: 348,560,338 RAC: 223 |
i have also some questions - what is with VRAM on boinc projects and seti - 1 Gb is always more than enought, or better is get 2 Gb version? Well, I'm currently running two astropulse jobs on both my GTX 560 and my GT 640. nvidia-smi sez: +------------------------------------------------------+ | NVIDIA-SMI 331.38 Driver Version: 331.38 | |-------------------------------+----------------------+----------------------+ | GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC | | Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. | |===============================+======================+======================| | 0 GeForce GTX 560 Off | 0000:01:00.0 N/A | N/A | | 92% 75C N/A N/A / N/A | 344MiB / 1023MiB | N/A Default | +-------------------------------+----------------------+----------------------+ | 1 GeForce GT 640 Off | 0000:06:00.0 N/A | N/A | | 58% 61C N/A N/A / N/A | 132MiB / 1023MiB | N/A Default | +-------------------------------+----------------------+----------------------+ The 560 is my display card, so it's got some overhead for that, but both are using well under half the available memory. Similarly with my Win7 laptop, currently running 1xAP + 1xMB on a GT 555M and using ~350 MB (its display is probably running on the inbuilt Intel GPU at the moment); IIRC MB workunits take up a bit under 256 MB. I doubt you'll see any advantage running more than two jobs simultaneously, but by all means try it and see. |
rob smith Send message Joined: 7 Mar 03 Posts: 22200 Credit: 416,307,556 RAC: 380 |
Memory is NOT the key, it is the overall throughput - while a card may hold four or five tasks in its memory it may not be able to process them in an efficient manner. You need to look at how fast it is processing a task and to a large extent IGNORE memory. With your GTX630 it is probably 1 or two, a lot depending on your system and what else you are doing on it at the time. Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Memory is NOT the key, it is the overall throughput - while a card may hold four or five tasks in its memory it may not be able to process them in an efficient manner. You need to look at how fast it is processing a task and to a large extent IGNORE memory. With your GTX630 it is probably 1 or two, a lot depending on your system and what else you are doing on it at the time. Different projects vary - GPUGrid have put through some tasks which require more than 1 GB per task. But SETI is very modest in its requirements. |
draco Send message Joined: 6 Dec 05 Posts: 119 Credit: 3,327,457 RAC: 0 |
there is also second very good improvement - when i have 9400gt, and set "gpu work always", the desktop is very slow-responsive. it is a not comfortable to work in KDE. now, 630gt with the same settings and task on GPU is no introduce any visible slowness in window manager ( i now use XFCE manager, but i think, in kde that thing is the same.) |
Oz Send message Joined: 6 Jun 99 Posts: 233 Credit: 200,655,462 RAC: 212 |
Search for "seti performance" - this small program will test your card and tell you how many simultaneous tasks are optimal. Also you may wish to go to kwsn lunatics downloads for an optimized installer. Member of the 20 Year Club |
draco Send message Joined: 6 Dec 05 Posts: 119 Credit: 3,327,457 RAC: 0 |
got of result with "sti performance" in google, when type "seti performance" :D and lunatics optimized installer, as i remember is for microsoft products? i have abandon to use MS products about 8 years ago, and be happy from that time - no viruses, no registry problems and half-life windows encounter....and i definitely no want to go back for get an some doubtful improvements from "optimized installer", excuse me :) |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.