Just Upgraded Last Month buuut can Someone Explain why the GTX 780 Uses Slower RAM?

[MagusALL]

Ok so recently I went from a GTX 650 Ti Boost 2gb to a GTX 760 2gb and then to a GTX 770 4gb. Needless to say, because I am just a weird fellow, I am considering jumping to the GTX 780 3gb but I am wondering something before I do. For the GTX 770 the RAM speed is over 7,000mhz yet the GTX 780 is only at ~6,000mhz. This confuses me for two reasons. Why does the 770 have faster, and more VRAM than the faster and more powerful 780 and inversely why does the more expensive 780 have slower VRAM and lower clock speeds to boot? I am sure the number of shaders etc more than makes up for it but I am wondering why the 780 isn't clocked higher and also why the VRAM isn't ~7,000mhz like the less expensive card. Does anyone know the answer? BTW the reason I am upgrading is I will soon be going up in resolution with a new monitor and so I thought I needed the 4gb from the GTX 770 over the 2gb of the 760 but now I am more concerned with the overall power to push a 2560x1600 screen and not so concerned with VRAM especially if I am able to work with 3gb. Thanks for the replies.

 

[babernet_1]

Your best bet is just to get another GTX 770 4GB and be done with it. A 2560X1600 monitor needs at least 3GB. 4GB is great.

Buying a second GTX 770 4GB is cheaper than getting a GTX 780 and the GTX 770 SLI is faster than even a GTX 780TI.

 

[MagusALL]

Is that memory bandwidth? I thought that was for the shader cores themselves, as the VRAM speed directly affects the memory bandwidth, as in if I overclock my memory the bandwidth goes up. Why does the 780 have a slower speed then? If it has higher bandwidth than why is the speed slower? If what you are saying is the bandwidth difference making it faster than why is the speed itself slower and not the same ~7,000mhz? I am just a tad confused why the 780 isn't also ~7,000mhz is all. I guess I can see that the higher bandwidth would let the more powerful card move more info thus lowering the clock speed to adjust for its threshold but I guess I am still wondering why the lower bandwidth card would have faster speed VRAM. How is 4gb of ~7,000 not as fast as 3gb of ~6,000? Am I seeing this incorrectly?

 

[MagusALL]

thanks for the replies. and for the suggestion to SLI but for right now I don't want to go SLI because it might necessitate a PSU upgrade which will add another $100 to the mix and I am getting what I paid for the GTX 770. I am only paying the price difference which is $100 (MSI Gaming Edition's via Amazon). Also could you conceive of a situation where 3gb isn't enough for that monitor? If thats the case why didnt they just put in 4gb? Why make it so difficult for me? 4gb but its a slower card, 3gb but you might be limited at that resolution. Can't I just have what I need??? Anyhow I am sure this makes sense I am just confused a little on why, if the bandwidth is higher, that the speed is slower because if it is faster bandwidth than wouldn't that be beneficial to faster VRAM? I'm no pro so I'm sure im just complicating things for myself. Anyhow, thanks for all the input. It all makes me a smarter gamer/builder.

 

[babernet_1]

The GTX 780 has a 384-bit memory bus. It has a memory bandwidth of 288,384 MB/sec.

The GTX 770 has a 256-bit memory bus. It has a memory bandwidth of 224,384 MB/sec.

The equation is bus speed x memory clock speed (not the effective speed) / 2 (because it is double data rate or GDDR5 VRAM).

For the 780:
384 x 1,502 = 576,768
576,768 / 2 = 288,384

For the 770:
256 x 1,753 = 448,768
448,768 / 2 = 224,384

Actually the formula is [(# of bits)/8][effective clock rate]

bits/8 = bytes the 6,000MHz etc. is the effective clock rate since it uses both edges of the clock and the actual frequency is halve of the effective clock rate so the actual frequencies are 3GHz and 3.5GHz.

Anyways, you get 384/8[6,000) = 288,000MB/s

and 256/8[7,000] = 224,000MB/s

 

[MagusALL]

OK I think I get it. One last question. Why, when you overclock memory, does the bandwidth go up? Is it simply because in that equation, there is a variable and a constant, with the memory bus being the constant and therefore the speed difference changes the overall output? Than why again isn't the 780 higher? Could it not benefit from having the same speed VRAM as the 770? Was that just to make the 770 better than the 760? Just to put it in the middle, performance-wise? Thanks for the help

 

[babernet_1]

thanks for the replies. and for the suggestion to SLI but for right now I don't want to go SLI because it might necessitate a PSU upgrade which will add another $100 to the mix and I am getting what I paid for the GTX 770. I am only paying the price difference which is $100 (MSI Gaming Edition's via Amazon). Also could you conceive of a situation where 3gb isn't enough for that monitor? If thats the case why didnt they just put in 4gb? Why make it so difficult for me? 4gb but its a slower card, 3gb but you might be limited at that resolution. Can't I just have what I need??? Anyhow I am sure this makes sense I am just confused a little on why, if the bandwidth is higher, that the speed is slower because if it is faster bandwidth than wouldn't that be beneficial to faster VRAM? I'm no pro so I'm sure im just complicating things for myself. Anyhow, thanks for all the input. It all makes me a smarter gamer/builder.

If you could wait a couple weeks for the custom cooled R9 290 you would get a better value. The R9 290 costs $400 and with a custom cooler is as fast as a GTX 780. Plus it has the new DSP and Mantle software that will be a real killer. Plus a 512 bit memory interface to its 4GB VRAM.

 

[sora]

Your best bet is just to get another GTX 770 4GB and be done with it. A 2560X1600 monitor needs at least 3GB. 4GB is great.

Buying a second GTX 770 4GB is cheaper than getting a GTX 780 and the GTX 770 SLI is faster than even a GTX 780TI.

This, the graphics card would be perfect with the monitor that you have.