VRAM or Clock Speed?

[VoyexTech]

What is more important? (I will be gaming) I am thinking of either going With AMD next round or NIVIDIA. But I have a $500 USD budget and NIVIDIA is usually more expensive and has better clock speeds but AMD has lower prices and clock speeds but more VRAM on the higher end. 

 

My choice between GTX 1000 or R9 400 will define if I am team red or team green.

[Enderman]

try it yourself

i have seen people get a larger performance increase with vram speed than core clock speed

other times core clock gives a bigger improvement

 

depends on the game and your GPU, so test it yourself to find out

[Oshino Shinobu]

Clock speeds and VRAM aren't a reliable way to compare graphics cards, especially not across platform. Only way to really decide is to look at benchmarks to see how they compare. 

Can't help with a choice between Polaris and Pascal for obvious reasons. 

[roughavoc]

Your choice is between two cards that haven't been released yet, nobody know how they will perform

[kingatlas]

I am team green through buying used(780Ti) and a gaming laptop (dell 7559). For me it depends what will suit me for 1080p. As i will not being changing my monitor for the foreseeable future. 

[Michamus]

At the end of the day, it comes down to what sort of games you play and whether DX12 performance is important to you. As newer games come out, they're going to really shine on the new DX12 API (AOTS, Hitman, etc), which means the R9 390x is really going to shine. However, if you find yourself not really being reliant on DX12, then I would say the GTX 980 is a good bet for you.

 

To answer your question though, clock speed and vram aren't really the whole story. Architecture is pretty important too. For instance:

 

R9 390x

- 8GB vram

- 1100mhz core clock

- 6100mhz memory clock

 

R9 Fury X

- 4GB vram

- 1050mhz core clock

- 500 mhz memory clock

 

So, if you just looked at those points, you would think people were idiots for paying $200 more for a Fury X. But then you look at the memory bit width:

 

R9 390x

- 512bit GDDR5 memory

 

R9 Fury X

 

- 4096bit HBM memory

 

That massive memory bit width is what really makes the R9 Fury X shine above the R9 390x. This is why you'll see a 780 with a cuda core count that's higher and a higher bit-width, but less performance than a 980. It just really comes down to better performing architecture and combining factors between the various GPUs.

 

So, I'd recommend looking at in-game benchmark comparisons to determine which video card is best for you.

 

Edit: I just realized you were talking about the next generation of cards. The thing is, you're not going to see the "R9 400 series" for at least another year and a half, whereas the "GTX 1000" series is due to be announced this month or next month, which would put it on Q3-Q4 release. So, it may come down to DX12 performance comparison between the 1000 series and 300 series. I get the feeling Nvidia is going to blow the R9 300 series DX12 benchmarks out of the water with the 1000 series though.