fed250's forum posts

Avatar image for fed250
fed250

25

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#1 fed250
Member since 2009 • 25 Posts
Just as a heads up, make sure that if you do decide to run a dedicated physx card that the vram speed is equal to or faster than the speed that your man card(s) run at. I had debated about putting a dedicated card in for physx because I had a spare GTS 450 lying around I figured I could use for physx until I had the time to build a mame cabinet or something fun like that. Anyway I was running a GTX 690 at the time so it did very well by itself but I looked online to see how much of a boost I could expect by running the GTS 450 as physx but what I found online surprised me and it was showing people getting worse FPS with certain dedicated physx cards if they were older or slower cards. So after a bit of research and trial on my own system with both the GTX 690 and on another machine running a GTX 580 as it's main card I found all the tests to be accurate and my frame rate lowered considerably. What I found out and it makes perfect sense after I saw an online post pointing out that most peoples main cards are using DDR5 VRAM and the older or lower end cards usually have DDR3 VRAM. The speed difference between the DDR5 VRAM and the DDR3 VRAM caused the whole system to slow down. The higher powered card had to sit and wait while the slower card played catch up. The computing performance you saved by offloading the work to a secondary card was greatly offset by the decrease in speed of the other card. So basically if you have a DDR5 VRAM in your main card if you decide to run a dedicated physx then make sure it is at least running DDR5 VRAM. If your main card is running DDR3 VRAM than either DDR3 or DDR5 would work without slowing the system down but in all honesty if you are running a card with DDR3 VRAM as your main card then sell it as well as the extra card you were going to use for physx and buy a newer/better card, you will get much more out of that.