GTX 970 VRAM Segmentation Handled Properly in RED Engine?

[Asmodean778]

It's just the naming scheme for the processor architecture. VLIW = "very long instruction word", and GCN = "graphics core next". GCN being the current one. We don't know if there are any dependencies on GCN, yet. But in terms of speed alone. I'd be severely surprised if the 6970 had trouble at 1360x768. It's a fairly low resolution.

Myself, on the other hand. Running 2560x1440, I'm doomed lol. Unless something major releases before the game does - which aint likely at this point.

 

[GuyNwah]

---------- Updated at 09:44 PM ----------

What do you mean VLIW or CGN? . I know the game is not out yet but maybe with all setting's on lowest and maybe get decent frame rates with my 1360x768 resolution 24" screen, what do you think? Thank for for the reply.

AMD made a substantial change in their GPU design. "VLIW" (Very Long Instruction Word) is the earlier design. It's used in the 5xxx and 6xxx GPUs (and a few low-end 7xxx's). It was, to put it politely, not easy to program. "GCN" (Graphics Core Next) is the current design. It's a more conventional architecture and easier to program. Mantle and other recent advances they have made take advantage of GCN.

But for a DirectX 11 game, VLIW GPUs like the 6xxx are still fine. I expect that at a small screen resolution, you should get good performance.

 

[Gilrond-i-Virdan]

http://www.pcper.com/reviews/Graphi...Full-Memory-Structure-and-Limitations-GTX-970

 

[M4xw0lf]

http://bursor.com/investigations/nvidia/

Murrica.

 

[GuyNwah]

http://bursor.com/investigations/nvidia/

Murrica.

We call them "ambulance chasers". Shysters Bursor and Fisher will try to drum up enough disgruntled customers to file a class action lawsuit. It will never get to trial. nVidia will (accurately) counter that their specifications are not warranties. They will settle out of court for a pittance, which will mostly go to these thieves.

 

[onionshavelayers]

Muuuuricaaa! FUCK YA!!!

 

[=Axton=]

I hope this thread doesn't get locked for going too far off topic, I too have an MSI 970 and am waiting for response from MSI or NCIX.com where I bought it to see what my options are.

As for the game, I'm surprised the devs haven't replied yet, seeing how much attention this topic has around the web right now, Witcher 3 was one of the reasons I bought a 970 and I'd sure like to know how long the card is going to be relevant.

 

[erxv]

^ im also planning on getting a 970 (or 2 if the wallet gods are forgiving) 980 seems much worse in price/performance...

would like to know if 3,5GB is enough for 1080p on high/ultra

 

[Zabanzo]

As for the game, I'm surprised the devs haven't replied yet, seeing how much attention this topic has around the web right now, Witcher 3 was one of the reasons I bought a 970 and I'd sure like to know how long the card is going to be relevant.

But what can devs do about this? Force game to use less than 3.5gb of VRAM just because Nvidia screwed things up?

 

[GuyNwah]

But what can devs do about this? Force game to use less than 3.5gb of VRAM just because Nvidia screwed things up?

Games use VRAM budgets internally to control how much of VRAM they use. It's not farfetched to have code that says "if this is a GTX 970, your VRAM budget is the lesser of (3.5GB, original budget)."

But the only way you would ever release something like that is if you had to because the game had unacceptable performance attributable to that flaw, as determined by QA (or from customer complaints, after tech support verification).

 

[MkTama]

Anyway seems it's still unclear that Nvidia screwed up in communicating the right specifications: they have told wrong specifications. But the GPU itself was specifically built that way and is therefore perfectly functioning as intended. I definitely don't want to defend Nvidia in any way but this is not a detail. It's a very different thing speaking about a bad GPU or bad communication and I find that there is still much uncertainty about this.

 

[eskiMoe]

Anyway seems it's still unclear that Nvidia screwed up in communicating the right specifications: they have told wrong specifications. But the GPU itself was specifically built that way and is therefore perfectly functioning as intended. I definitely don't want to defend Nvidia in any way but this is not a detail. It's a very different thing speaking about a bad GPU or bad communication and I find that there is still much uncertainty about this.

Hopefully this controversy does make them rethink their GPU design philosophy though.