News-Official AMD R7 & R9 Naming Unveiled

[3DVagabond]

It would have to be either 1.5, 3, 6, 12.

It doesn't HAVE to be. The amt. of memory can be asymmetrical to the bus, but as soon as AMD does it I'm sure it will cause stuttering, etc...

 

[3DVagabond]

Well, to some extent I care. I see no purpose in having +30% power consumption for +15% performance. I overclock, but only as long as efficiency doesn't fall down a pit. Because honestly, the last 10% don't make a difference in playability, they never have. But they certainly make a difference in noise and to a lesser degree in energy costs.

Yes, there is a point beyond which the design just becomes terribly inefficient. Running the chip beyond that point is fine for benchmarking, but shouldn't be released as the reference speed of the card.

Efficiency is a good thing and designers should strive for it. If individual enthusiasts want to overclock/overvolt/supe-up, or venders want to release O/C'd specialty cards that's fine.

 

[wand3r3r]

Well, to some extent I care. I see no purpose in having +30% power consumption for +15% performance. I overclock, but only as long as efficiency doesn't fall down a pit. Because honestly, the last 10% don't make a difference in playability, they never have. But they certainly make a difference in noise and to a lesser degree in energy costs.

I see your point, but fail to see why it's became a marketing point lately.

I don't run anything intensive on my PC 24/7 but when I do I want everything as fast as possible. I'll overclock my CPU as far as I can since it'll be running idle unless needed. Power consumption is my last factor, actually I don't know how much my GPUs have used anyway. Noise and temperatures do matter to me.

 

[Skurge]

So 2304 SPs
48 ROPs
144TMUs
But 384 bit bus and 4gis memory

I'm guessing Dual GPU is cut down maybe 9950 GPUs instead of 9970 GPUs.

 

[Leadbox]

I'm guessing Dual GPU is cut down maybe 9950 GPUs instead of 9970 GPUs.

Good guess, that memory bus though,typo?

 

[Kippa]

And once those six people (figuratively speaking) buy them, then what? High end gaming is enough of a niche market for a consumer product. CUDA is even more of a niche category. CUDA with no support is even smaller numbers yet. Besides how much better performance is Titan than the 780 running CUDA?

Interesting also how people who seem to think CUDA is so important see no value in the OpenCL prowess of GCN.

The problem with the 780 is that it has 3gb of ram and that could be used up quite fast in large complex 3d scenes. That is the real reason why I got the Titan over the 780 specifically for the extra 3gb of ram not the performance. I agree that performance wise 780 is very close to the Titan. If the 780 had 4gb+ or more ram then I would have gone for that instead. Cuda might be niche category but my Adobe Premiere CS5.5 and 3d app Octane (which is also used in Lightwave 3d) uses cuda.

To be honest I AM NOT a brand loyalist. If my apps change from cuda to opencl and AMD is better for opencl then I would quite happily buy AMD. I can see that happening in a year or so.

 

[SolMiester]

And once those six people (figuratively speaking) buy them, then what? High end gaming is enough of a niche market for a consumer product. CUDA is even more of a niche category. CUDA with no support is even smaller numbers yet. Besides how much better performance is Titan than the 780 running CUDA?

Interesting also how people who seem to think CUDA is so important see no value in the OpenCL prowess of GCN.

OpenCL isnt even close in dev to CUDA..hundreds to one in apps....

 

[3DVagabond]

The problem with the 780 is that it has 3gb of ram and that could be used up quite fast in large complex 3d scenes. That is the real reason why I got the Titan over the 780 specifically for the extra 3gb of ram not the performance. I agree that performance wise 780 is very close to the Titan. If the 780 had 4gb+ or more ram then I would have gone for that instead. Cuda might be niche category but my Adobe Premiere CS5.5 and 3d app Octane (which is also used in Lightwave 3d) uses cuda.

To be honest I AM NOT a brand loyalist. If my apps change from cuda to opencl and AMD is better for opencl then I would quite happily buy AMD. I can see that happening in a year or so.

I'm not saying nobody can benefit from the Titan. You're one of those six people I was refering to. You use Titan for pro apps but require no support from nVidia or in the drivers.

OpenCL isnt even close in dev to CUDA..hundreds to one in apps....

What are these hundreds of apps that people are using Titan in that the 780 wouldn't run virtually just as well. The only thing Titan really beats the 780 in is DP. We've already had this discussion about the actual usefulness of Titan in DP considering it doesn't use ECC RAM and you get no support from nVidia using a Geforce card for this type of work.