Why the cost difference between gtx 960 & gtx 970?

[Council_of_Trent]

Hello all. In researching my first build, I wanted an affordable but powerful card. Whilst browsing Newegg, I came across the EVGA GTX 960 card that claims 4 gig vram. My friend said that he has a GTX 970 that is also a 4 gig card, but cost him $100 more. So, why the price difference if they are both supposed to be just as powerful? Is the 960 less stable or significantly worse in some way? Any horror stories? Any advice is much appreciated.

[Guest]

Hello all. In researching my first build, I wanted an affordable but powerful card. Whilst browsing Newegg, I came across the EVGA GTX 960 card that claims 4 gig vram. My friend said that he has a GTX 970 that is also a 4 gig card, but cost him $100 more. So, why the price difference if they are both supposed to be just as powerful? Is the 960 less stable or significantly worse in some way? Any horror stories? Any advice is much appreciated.

960 doesn't have NEAR as much /Cuda/ Cores as the 960 does. The 970 is far more powerful. Except the 960 actually has a full 4GB of VRAM where the 970 technically has 3.5 and .5 gigs.

[TheRandomness]

Basically, a 970 is more powerful than 960, hence the price difference. A 960 may not perform better than a 970. Although if you want slightly performance than a 960 at less cost, R9 380 c:

[Castdeath97]

Never use vRAM to judge the performance of ANY card. A 2GB R9 380 can for instance beat a 4GB r9 270.

[zyntaxable]

Thing is, a 960 can't use all 4GB I believe. It's pure marketing.

[Godlygamer23]

960 doesn't have NEAR as much /Cuda/ Cores as the 960 does. The 970 is far more powerful. Except the 960 actually has a full 4GB of VRAM where the 970 technically has 3.5 and .5 gigs.

I'm pretty sure the 970 then still has 4GB of VRAM. The 4GB of VRAM in the 960 may not even matter because it's not powerful enough to use it. There's more to performance than just CUDA cores. You also need to look at the number of ROPs. (Example: 760 vs 660 Ti)

[theereal]

Hello all. In researching my first build, I wanted an affordable but powerful card. Whilst browsing Newegg, I came across the EVGA GTX 960 card that claims 4 gig vram. My friend said that he has a GTX 970 that is also a 4 gig card, but cost him $100 more. So, why the price difference if they are both supposed to be just as powerful? Is the 960 less stable or significantly worse in some way? Any horror stories? Any advice is much appreciated.

The 970 is WAY more powerful than the 960

[ChasingPluto]

960 doesn't have NEAR as much /Cuda/ Cores as the 960 does. The 970 is far more powerful. Except the 960 actually has a full 4GB of VRAM where the 970 technically has 3.5 and .5 gigs.

Literally makes no sense to me. Wasn't the last .5GB partitioned from the 970 because of a limitation of the cut down GPU? So how it is the 960 can manage all 4GB when it has less than the 970. Unless it has to do with the speed of data transfer and computation and overall throughput. I might be able to actually understand that if that was the case.

[Guest]

Literally makes no sense to me. Wasn't the last .5GB partitioned from the 970 because of a limitation of the cut down GPU? So how it is the 960 can manage all 4GB when it has less than the 970. Unless it has to do with the speed of data transfer and computation and overall throughput. I might be able to actually understand that if that was the case.

It doesn't really matter - In an application you always can max out VRAM, That last .5gig portion was always meant to be there, but how the card was designed it wasn't as easy to be accessed, 

You can see there, where the last gate wasn't functioning or was disabled giving it almost a bottleneck feel, where you could access the VRAM if you needed to but it wouldn't be as fast as the other portions because that last gate is closed off and it would need to go to the next gate, right there (see below) which it would get in the way of the last the first module on the chip kid of like merging traffic, which could be causing the microstutters, and as Godly Gamer contributed, there are other factors that I didn't discuss If there is anything else I missed on the topic correct me, As I didn't go too in depth for the reasoning because the user only thought that VRAM mattered and nothing else. Where the most I feel like they would get without getting extremely overwhelmed with info, because I still cant wrap my head around FLOPS and other things 

 

[ChasingPluto]

-snip-

Fair enough. But wasn't the bottleneck caused by the cut-down in SMs? And wouldn't the 960 have less SMs than a 970? So does it manage all 4GB?

[Guest]

Fair enough. But wasn't the bottleneck caused by the cut-down in SMs? And wouldn't the 960 have less SMs than a 970? So does it manage all 4GB?

that I can't answer you.

[ChasingPluto]

that I can't answer you.

Well, maybe one day I'll learn. For now I shall learn to crochet...or was it knit? Either way it's not designing GPUs

[Godlygamer23]

Fair enough. But wasn't the bottleneck caused by the cut-down in SMs? And wouldn't the 960 have less SMs than a 970? So does it manage all 4GB?

Define manage. The card can access all 4GB. If you're asking if the VRAM is allocated evenly between the memory controllers, you'd have to know how many memory controllers are enabled.

[Council_of_Trent]

960 doesn't have NEAR as much /Cuda/ Cores as the 960 does. The 970 is far more powerful. Except the 960 actually has a full 4GB of VRAM where the 970 technically has 3.5 and .5 gigs.

OK, I see. But now, dumb question maybe, but what are cuda cores?