What is the role of 128bit and 256bit in graphics card.

Solution
SHORT ANSWER:
Don't worry about it.

LONG ANSWER:
A 128-bit connection means how many BITS of data can be on the GPU to Video Memory connection at any one time; you can think of it simply in how many wires (or traces) connect the memory to the GPU.

Calculating memory bandwidth is complicated by MEMORY COMPRESSION techniques which can compress the data to transfer more over the connection then decompress it. If you want to know the EFFECTIVE memory bandwidth you must look at:

1) How many BITS the connection is
2) frequency
3) memory compression amount

(how many wires, how fast they are moving data, and whether the data has been compressed)

*HUH?
The short answer is that it's not worth looking at those numbers. You should in turn...
SHORT ANSWER:
Don't worry about it.

LONG ANSWER:
A 128-bit connection means how many BITS of data can be on the GPU to Video Memory connection at any one time; you can think of it simply in how many wires (or traces) connect the memory to the GPU.

Calculating memory bandwidth is complicated by MEMORY COMPRESSION techniques which can compress the data to transfer more over the connection then decompress it. If you want to know the EFFECTIVE memory bandwidth you must look at:

1) How many BITS the connection is
2) frequency
3) memory compression amount

(how many wires, how fast they are moving data, and whether the data has been compressed)

*HUH?
The short answer is that it's not worth looking at those numbers. You should in turn look at BENCHMARKS comparing the same GPU of one card to another.

If the GPU is the same, then the 128-bit may use slower memory (i.e. 128-bit for DDR3 vs 256-bit for GDDR5?) but again it's hard to guess how much this is affected. You need to look at benchmarks.

It's possible that you might see a game run at 90% the performance of the 256-bit counterpart of the same GPU, but on the other hand there is probably a good chance the GPU was also clocked at a lower frequency so that it's not even possible to compare.

**In general, the GPU and Memory is setup in a way that there are no significant bottlenecks. No point in clocking the GPU too high for example, if the memory is a big bottleneck because:
a) your reliability goes down (returned cards), and
b) you need a better cooler

Summary:
Again, it's pretty much a non-issue. I'd need to know specifically what you're talking about as much of the time it's not even apples-to-apples. Different GPU, memory compression technique etc.

If concerned then find benchmarks for your EXACT video card but this is pretty much a non-issue for the reasons discussed.

(It would be FOOLISH for NVidia to sell an expensive Graphics Card then gimp it too much by creating a memory bottleneck. So many claims of "but it only has a 256-bit bus" or "wait for HBM2 then the performance will really go up" are by misinformed individuals.)
 
Solution

TRENDING THREADS