AMD Teases Next Enthusiast GPU: Still 28nm, 23% Smaller Than GK110
by Ryan Smith on September 18, 2013 11:00 AM EST
Rumors aside, after AMD’s unexpected preview of their 2013 GPU plans back at the start of this year the company has been mum-as-usual on the matter of their future GPUs. At the time AMD announced that they would have a new GPU microarchitecture by the end of this year, and as 2013 slowly winds down the launch of that new microarchitecture draws near. Normally in these situations we would see AMD continue to stay quiet until they’re ready to fully unveil their products, but with the launch of that new microarchitecture already confirmed, AMD has unexpectedly unveiled a couple of details about one of their forthcoming GPUs in an interview with Forbes.
Interviewing the Corporate VP & GM of AMD’s Graphics Business Unit, Matt Skynner, Matt offered up a couple of basic but important details about AMD’s future plans. First and foremost, AMD’s next enthusiast GPU – already strongly implied to be the launch vehicle for their forthcoming microarchitecture – will be a 28nm product. Explaining why, Matt specifically states that “at 28nm for an enthusiast GPU, we can achieve higher clock speeds and higher absolute performance.” With traditional AMD GPU fab TSMC still ramping up for 20nm anyhow this doesn’t come as a great surprise, but it does put to rest any rumors of a 20nm product launch one way or another.
More interesting perhaps is that Matt also gave Forbes a ballpark number on the die size of their new GPU: GK110 is still 30% bigger than the new GPU, or inverted the new GPU is 23% smaller than GK110. While AMD’s small die strategy has been dead for some time, the company has still shied away from large GPUs for various reasons, their largest GPU since the ill-fated R600 (HD 2900 XT) being the 389mm2 Cayman GPU at the heart of the HD 6900 series. 23% smaller than GK110 would put the die size of AMD’s future GPU at around 425mm2, making it slightly larger than Cayman, or roughly the same size as R600. These are ballpark figures of course, so we’ll know more once the GPU formally launches, including of course how well that large die and new microarchitecture translate into performance.
The full interview can be found over at Forbes. Along with teasing AMD’s next enthusiast GPU, Matt also briefly discusses AMD’s software/driver plans, game bundling, the Radeon HD 7990, and more.
Source: Forbes (via Beyond3D)
Facebook
Twitter
RSS
49 Comments
View All Comments
DanNeely - Wednesday, September 18, 2013 - link
Since the new GPU is coming out in Q4 of this year it still being 28nm isn't that surprising.I'm wondering what it says about their plans though. Earlier in the year when the 8xxx oem rebadging was announced it was said we wouldn't be getting a real new GPU until mid next year when TSMC 20nm was expected to become available.
That has me wondering if this launch is an indication that TSMC's 20nm process is having problems and will be delayed and this is is what they've chosen to launch instead of waiting. Alternately, was the size of the performance bump nVidia got with GF110 and the 7xx refresh significantly larger than expected causing them to release this as a stopgap instead of yielding the performance crown for another half year.
Jumangi - Wednesday, September 18, 2013 - link
Oh I think is plainly obvious TSMC is having more of the same problems getting 20nm going just like it took them forever to get 28nm up to respectable levels. As these processes get lower it looks like it will be longer between jumps.Wreckage - Wednesday, September 18, 2013 - link
They may have spent most of their resources developing for the consoles. This may be a hint at their exit from the high end similar to their CPUs. I doubt they will have a GK110 killer on their hands with 28nm.Goty - Wednesday, September 18, 2013 - link
Just like GK110 wasn't a big step over GK104 just because it's on the same manufacturing process.Oh, wait.
Hector2 - Wednesday, September 18, 2013 - link
TSMC has never met their predicted process schedules and new production-worthy processes always come out at least a year behind schedule. It's just a marketing attempt to get people to wait for something that supposedly "just around the corner". It's always much easier to change a number on a Powerpoint slide than to actually change the process --- especially now that we're below 22nmFrenetic Pony - Wednesday, September 18, 2013 - link
Well, TSMC 20nm already doesn't offer much of anything in improvement in the way of power/performance and isn't any cheaper. The only improvement would then be more silicon per mm squared. We already know Apple signed a big contract for 20nm, so my guess is AMD decided they'd rather go 28nm and get it out in time for Christmas than wait 4+ months for little gain.Da W - Wednesday, September 18, 2013 - link
A wild guess here, we can expect GCN 1.1 with may be 5-10% performance improvement. Some sort of architecture optimisation kinda like we saw from fermi to keppler.With a 23% smaller die than Titan, maturing process and probably a 1.1Ghz clock, this would put Hawaii at par or slightly above GTX780. My guess is that AMD went for higher power consumption improvement rather than pure performance.
Anyhow, this will be the best card for an eyfinity setup, without microsuttering issues.
purerice - Wednesday, September 18, 2013 - link
Agreed, though you may have dropped a t in microstuttering. I hopefully won't have to upgrade for a year or two but it's nice to see this. I am not a fan of either AMD or nVidia so I want both to keep pushing the other, even if in incremental improvements.Frenetic Pony - Wednesday, September 18, 2013 - link
It's been 2 years, I've heard they're calling this an actual new architecture (GCN 2.0) rather than a refresh. Which means their plans would have changed a lot since earlier this year, which is what this article seems to imply anyway.Sabresiberian - Thursday, September 19, 2013 - link
Above GK110? I hardly think so, unless there is a lot more improvement in the "maturing process".Don't get me wrong, I'd love for it to be true - we need the competition to stay alive in the GPU space just as it was good for us in the CPU space - but I have serious doubts. AMD's current single-GPU solution is far below the GTX 780, and it is going to take more than a 10% bump to make them competitive at the top end.