which one memory module is best for me watching the future trend GDDDR5 128bit vs 256bit

[sumit3322]

my rig
crossair 450watt psu

CPU:-intel quad core Q8200 2.33 ghz
MOTHERBOARD:- intel DG35EC
PCIe: 2.0 x16
RAM:- 5 GB ddr2
HDD;- segate STATA 7200rpm 1TB
DISPLAY;- HCL 1080p full hd 21 inc

which graphics card i should i buy one with GDDR5 128 bit or GDDR5 256 bit seeing the future trend and also their is any bottleneck if i run GDDR5 256 bit graphics card on a pcie 2.0 slot...
what is the difference in performance in playing games btw 128bit vs 256 bit memory mode cards having same configuration...
please help....

 

[RobCrezz]

Looking at them specs only is no good. Some 128 bit card can be faster than 256bit cards.

Look at card benchmarks in games to see performance.

Any modern card wont be bottlnecked by pci-e 2.0

 

[sumit3322]

Looking at them specs only is no good. Some 128 bit card can be faster than 256bit cards.

Look at card benchmarks in games to see performance.

Any modern card wont be bottlnecked by pci-e 2.0

based on the official bus speed to interact with different components which one is better..
if we take both cars have same config ex gtx 760 with 128bit vs 256bit then which one give better performance for my rig

 

[RobCrezz]

Looking at them specs only is no good. Some 128 bit card can be faster than 256bit cards.

Look at card benchmarks in games to see performance.

Any modern card wont be bottlnecked by pci-e 2.0

based on the official bus speed to interact with different components which one is better..

Its simply not relevant when comparing different cards of different architecture. Other things can influence the bandwidth from the vram to the GPU
1) Bus width,
2) Vram speed
3) Compression applied (modern Maxwell cards are using smaller bus with faster vram and clever compression to give similar bandwidth with a smaller bus, this means more of the GPU can be used to do processing).

 

[Rookie_MIB]

x2 on what RobCrezz says, there's a lot more to performance than just bus width - there's a ton of different things which can affect your overall performance of any GPU.

1) Bus speed vs bus width. While generally higher is better, total bandwidth is bus speed multiplied by bus width. If you have a 128bit bus clocked at twice the speed of a 256bus for a different GPU, they're running at the same effective speed.

2) GPU resources vs bandwidth. If you have a card with a wider bus, but it doesn't have the resources to utilize the bus at 100%, it's not as effective as a GPU with a narrower bus but utilizes it at a higher percentage (assuming of course that it doesn't exceed that maximum). So - for example an nvidia card with 640 cores and a 256bit bus which only utilizes that bus at 50% is just as effective as a different nvidia card with 1280 cores and a 128bit bus but utilizes that bus at 100%.

3) tricks - as mentioned, many of the newer cards utilize on the fly lossless color compression to reduce traffic on the bus. If those compression tricks on average reduce the traffic by 50%, then your 128bit bus is capable of transferring the same amount of data as that 256bit bus (assuming similar clock speeds of course).

So - there's a lot to it, and the best way to see what's going on is to see actual benchmarks to view the real performance.

 

[sumit3322]

x2 on what RobCrezz says, there's a lot more to performance than just bus width - there's a ton of different things which can affect your overall performance of any GPU.

1) Bus speed vs bus width. While generally higher is better, total bandwidth is bus speed multiplied by bus width. If you have a 128bit bus clocked at twice the speed of a 256bus for a different GPU, they're running at the same effective speed.

2) GPU resources vs bandwidth. If you have a card with a wider bus, but it doesn't have the resources to utilize the bus at 100%, it's not as effective as a GPU with a narrower bus but utilizes it at a higher percentage (assuming of course that it doesn't exceed that maximum). So - for example an nvidia card with 640 cores and a 256bit bus which only utilizes that bus at 50% is just as effective as a different nvidia card with 1280 cores and a 128bit bus but utilizes that bus at 100%.

3) tricks - as mentioned, many of the newer cards utilize on the fly lossless color compression to reduce traffic on the bus. If those compression tricks on average reduce the traffic by 50%, then your 128bit bus is capable of transferring the same amount of data as that 256bit bus (assuming similar clock speeds of course).

So - there's a lot to it, and the best way to see what's going on is to see actual benchmarks to view the real performance.

i really like your trick here