View Full Version : Problem in using nvidia card for OpenGl Program
01-10-2006, 12:56 AM
My program is loading a small model(.ms3d)in a GL window. I am also using 2D texture on it. I am using VC++ 6 to wright my code. With on board graphics card i am getting a good frame rate, but when i am installing the Quadro FX 1300 card on my machine along with the proper diver i am getting a less frame rate. I am not able find out the problem. is it anything i have add in my code to utilized the graphics card, or is it something other , please help
01-10-2006, 08:41 PM
From your brief description it sounds like you may not have the drivers installed correctly, or you may be using "in the box" XP drivers which generally do not include an ICD.
What is returned by glGetString(GL_VENDOR) and glGetString(GL_RENDERER) when the Quadro is installed?
01-10-2006, 09:56 PM
Thanks for your reply,
The return of glGetString(GL_VENDOR) is "NVIDIA Corporation" and glGetString(GL_RENDERER) is "Quadro FX 1300/PCI/SSE2"
01-11-2006, 05:12 AM
Well, that suggests the driver is indeed installed correctly and you should be seeing acceleration.
What kind of frame rates are you seeing with the two adaptors? Is it dropping off a lot, or just a little? Are you vsync-limited?
01-11-2006, 11:54 AM
It expected, as a texture access through PCI bus is slowly then through AGP port. Try disable the DMA mode for graphic card if it is possible (it in BIOS or windows device manager), but i is uncertain. Or maybe graphic card is using one interrupt in common whith other device (see it in windows, if so then (it is conflict) set up manually the interrupts in BIOS). Or maybe in BIOS there are settings for card... e.g. to use or don't use the interrupt for card. (maybe a last is a delirium guess)
01-11-2006, 02:23 PM
AGP or PCI in the renderer string identifies the transport available for the GPU, it doesn't necessarily mean texturing across that transport. In the most common cases, texture maps will reside in video memory, local to the GPU.
01-12-2006, 04:07 AM
gold, you right. Then vertex transfert is speed down the renderer. Or grand number of interrupts for card and DMA - Interesting, does PCI card use a interrupt?
Edit: In non fullscreen mode PCI bus is "high voltage" and maybe it is bottleneck.
Powered by vBulletin® Version 4.2.0 Copyright © 2013 vBulletin Solutions, Inc. All rights reserved.