Question about video cards

[Spyker]

I'm buying a new pc and was looking at video cards but when i was comparing different cards I found that lower grade less expensive cards like the NVIDIA GeForce FX 5900 XT 128MB DDR had a significantly higher gpu speed of 475 MHz vs. the NVIDIA GeForce 6800 GT 256MB DDR3, which only had a gpu speed of 350 MHz.

I don't really know a whole lot about video cards so I was wondering what gpu speed is, and why better cards have a lower gpu speed than cheaper cards.

 

[Nubius]

Cause the new card has DDR3 memory which is one advantage alone...new faster memory...beyond that..its a 16 pipe architecture (if you get the Ultra and I think GT? Ultra is for sure might be only 12 pipes for GT) whereas the older ones had 8 so that means more information going through at a time. So you can have a slower core speed but twice as much info going through it see? Plus that 350MHz is only part of the equation. That sounds like it'd be Core speed there's also memory speed. Ya need them both to have good speeds. The memory on the Ultra is 500MHz and so is on the GT. That 350MHz you spoke of is on the GT, the Ultra has 400MHz and the 128mb has 325MHz but it says for memory speed 350MHz - 700MHz effective....not sure if it means it can handle up to 700MHz memory speeds? cause I was looking at eVGA brand 6800s and they provide the same heatsink for all 3...so thats why you can generally overclock there GT's to Ultra standards.

Conclusion: The guy at a review before the 6800 came out was showing a real-time rendered mermaid that we've all become familiar with and getting 50-60FPS. they said if you used the 5950 Ultra 256mb card it'd only do the same thing at 2-3FPS...so as you can see...the cards architecture has quite a bit to do with it. Bigger and better channels and a leap in graphics processing that hasn't been seen in a few years =). Trust me....they own anything before it....and also 6800 owns X800 don't let anyone tell you otherwise

 

[ATIRAGEPRO]

"Bios Shadowrealm..."

OK. I'm not really shure myself, so tell me how this sounds:

The GPU, (Graphics Processing Unit,) Determines the speed and power of the card itself. Much like a CPU, the GPU is in charge of handleing the data, and displaying it to the monitor. The RAM, (Random Access Memory,) Is to determine how much information the GPU is to handle all together, and at one time. Typically a 450mhz GPU with 128mb of RAM would be able to handle about 4.3 Million Polygons per-second (The Tryangular mesh that makes up 3D objects.) While on the outher hand, a 350mhz GPU with 256mb of RAM would dish-out 6.5 million pps.The ram is really the determining factor when it comes to speed. The more RAM you have, the more poly's you're able to desplay at one time, so smoother graphics with less flickering.

That's the only sense i've made out of it so far.

 

[chadley]

It depends on the card.

I have 2 graphics cards at the mo

A Geforce 5900xt 128mb. Core speed 390mhz.
A Radeon 9600 256mb. Core speed 325mhz.

Guess which ones in the draw twiddling it's thumbs?

Its down to rendering speed and the geforce is simply far faster at it. The radeon's gpu cannot render aswell so having more texture memory is a waste of time.

Benchmarks show that most cards with 256mb ram dont perform much better than the same card with only 128mb. Infact some cards run worse with more RAM.

Its the same with PC's in general, once u have enough system RAM that the cpu needs to reach it's max putting more in wont make a difference.

I'd have 3gb in mine if it did!