What is a Video Card's Maximum Potential?

aaronhart

Honorable
Feb 9, 2014
2
0
10,510
What I mean is, where do the specifications start maxing out (and why if possible)? After reading up on the beginner's guide to video cards I think that I've realized that I need to look mainly at the card's external bus (lanes on the highway), clock speed (the speed limit) and lasty the memory (how much can be stored at any given time). With all that said, I've noticed that a 1Gb card will never have more than 256-bit of external bus. Why is this? And also, what is the fastest clock speed of a 1Gb card and why?

Thanks in advance for your input! :)
 
Solution
So, believe it or not, you basically need to forget most of what you've learned. Clock speed, Memory amount and Memory Bus Width don't really mean a whole lot when doing anything but comparing different models of the same card (XFX 7870 vs HIS 7870). For example, the 780Ti has 3 GB of memory, a 384-bit bus, and a clock of 982Mhz. However, it beats (in most everything) an R9 290x, which has 4GB of memory, a 512-bit bus, and a 1000Mhz clock speed. However, the 780 (not the Ti version) has those same specs, but just less cores on the card, and is slower than the R9 290x.

Basically, the only way to be able to tell performance on GPU's is to just look at benchmarks of real games, and see how the cards you are considering compare. The other...
The reason you never see 1GB cards with a larger bit interface is because its not needed. 1GB cant saturate all 256 bits, where as 2GB can. Some claim that 4GB on a 256 lane (GTX 760) is a waste because the lane is too small.

This is why the 7950 with 3GB carried a interface higher than 256, it was 384 or something like that.

The speeds vary, but it dosent really matter how fast something is if it does minimal work each cycle.
 

chrisso

Honorable
Nov 17, 2013
1,333
0
11,660
This is too bigger subject to take in your way. a bus running at a bandwidth of Y might seem bigger than X, But if X is running on pci-e 3 and Y is running at pci-e 2, X will be performing faster.
The best way to judge is just pull up toms hardware gpu hierarchy chart and then compare retail prices. You wont find a bang/$ bargain your way as technical specifications promises can be skewed by bad drivers, poor hardware assembly, etc etc. Better to trust in several benchmarks done expertly and go from there.
 

TriBeard

Honorable
Jan 13, 2014
183
0
10,710
So, believe it or not, you basically need to forget most of what you've learned. Clock speed, Memory amount and Memory Bus Width don't really mean a whole lot when doing anything but comparing different models of the same card (XFX 7870 vs HIS 7870). For example, the 780Ti has 3 GB of memory, a 384-bit bus, and a clock of 982Mhz. However, it beats (in most everything) an R9 290x, which has 4GB of memory, a 512-bit bus, and a 1000Mhz clock speed. However, the 780 (not the Ti version) has those same specs, but just less cores on the card, and is slower than the R9 290x.

Basically, the only way to be able to tell performance on GPU's is to just look at benchmarks of real games, and see how the cards you are considering compare. The other specs are useful to compare similar cards, but that is about it.
 
Solution

chrisso

Honorable
Nov 17, 2013
1,333
0
11,660
And if you want to keep understanding deeper levels of electronics, join the que,. When you get to the top you will be asking the head of science research at a world university what the answers are he will say;'
you think we taught you all this so you would ask US?
Whats the answer?'