Other sources wrong, Pascal not taped out in June, potentially still in testing

[PlayStation 2]

17 minutes ago, VagabondWraith said:

When has AMD been in the lead in marketshare? It's been since the early 2000's when they actually LED nVidia. That's the last time (and only time imo) that AMD will ever lead again. They don't have the finances to keep up, and they sure as hell aren't passing Intel no matter what Zen does. All AMD can do right now is make smart business decisions so they can stay afloat. They're too far behind to ever overtake nVidia's lead.

AMD could at the very least regain a respectable chunk of market share from Nvidia. Intel... well, Intel has too much money for AMD to overtake them.

 

Just now, AHaskin14 said:

I'll agree that they were likely a rushed product, but fortunately the decreased power consumption brought a fair amount of power with it.

Also, I see your from Huntsville, as am I. That's cool!

Yeah. That's really the only thing I like about Maxwell. Other than that, it's pretty fucking horrible. Kepler pussy-whips Maxwell in compute performance, and Kepler doesn't have memory bus issues, a problem that's pretty persistent with the 960.

 

'eyo from the other side of Redstone Arsenal!

[PerfectTemplar]

23 minutes ago, Dan Castellaneta said:

It was pretty much rushed. Unlike Kepler and Fermi, Maxwell only focused on power consumption and completely neglected something that's biting it in the ass in games like Rise of the Tomb Raider. Wanna know how it was rushed? The 980 and 970.

Eh, maybe 970. Nothing wrong with 980 imo. I have one in my other PC. 970 though, bit annoying about the VRAM thing.

But yeah, gains over the 700 series are pretty small.

[Sakkura]

2 hours ago, zMeul said:

Charlie Demerjian predicts the downfall of nVidia and Intel for 10y now, even when he worked at The Inquirer - amazingly, The Inquirer is now a better read after he left

he is either biased or outright malicious
the same Zauba he bashes as being used as proof, was used in the past to track Fiji AMD; are were proved to be quite accurate

ZMeul accusing people of being biased.

[zMeul]

Just now, Sakkura said:

ZMeul accusing people of being biased.

I don't get paid for bashing on AMD, I do it voluntarily and never hidden it

I also bash on nVidia and Intel when they deserve it

[patrickjp93]

48 minutes ago, DocSwag said:

True that they probably won't overtake Nvidia at least in the near future, but they could gain back quite a bit of market share. They were steadily at around 40% until the last year or two, so they could at least get back there.

As for Intel, the chances of AMD getting ahead ever are very slim. The highest they ever got was 48% or so (which they reached twice, once around 2000 and another time around 2005). If Zen is successful, they could probably bring themselves above 30% or maybe even 40%, but leading Intel in market share would be extremely difficult if not impossible.

I disagree about the impossible to overtake Nvidia part, but I do think passing Intel is probably never gonna happen unless a miracle happens (by that I mean like Jim Keller comes back, a bunch of top Intel execs leave, AMD gets at least a few fabs back, and Intel loses all their's.). Even then it would be tough though.

Would you please change the post title? This in no way confirms that Pascal hasn't taped out. In fact the author doesn't even prove it didn't tape out in June. He only points out that everyone's claims about it from one bad source are terrible journalism (which I agree with). This is a clickbait title.

[Curufinwe_wins]

Lol click bait on alternate rumor crafting.

That's ok, time will tell for sure.

[BuckGup]

They are just waiting to destroy AMD and they are getting cocky about it. Stop it! I want more power for less

[TechGod]

16 minutes ago, SamStrecker said:

They are just waiting to destroy AMD and they are getting cocky about it. Stop it! I want more power for less

Destroy AMD? K den, child. 

[Humbug]

5 hours ago, RikvE said:

Oeeeh, maybe some sweet sweet competition then?

how is it competition? NVIDIA needs to get Pascal out soon after Polaris for there to be competition.

anyway I am not too worried, I don't think they will trail by more than 2-3 months.

[Bouzoo]

5 hours ago, ivan134 said:

Well, we all knew that other article was BS (even WCCF didn't report it, that says something), but I'm not knowledgeable enough to know whether this guy is blowing smoke out of his ass when he says they haven't taped out even after reading the article.

Not this again... WCCF not reporting means absolutely nothing. They are not the first to report everything nor do they report everything. Don't take them as a standard for anything. 

[ivan134]

8 minutes ago, Bouzoo said:

Not this again... WCCF not reporting means absolutely nothing. They are not the first to report everything nor do they report everything. Don't take them as a standard for anything. 

You misunderstand. I meant that WCCF is basically a tabloid that usually reports all kinds of rumors, but this one was a far fetched that even they didn't report it. I'm not using WCCF as some kind of gold standard of reporting since a lot of what they do is click bait.

[Humbug]

NVIDIA has been promoting Pascal for years. So I really look forward to seeing what it can deliver.

AMD had a more low key approach with Polaris. They didn't talk about it and suddenly they showed us fully working 14nm GPUs with new architecture and next gen perf/watt.

Either way it looks like the Polaris vs Pascal battle is gonna be a nice jump after the incremental improvements of the last few years on 28nm.

 

 

[Tic-Tac]

Great "news" ! This is the year of GPU excitement !

 

[Sakkura]

18 minutes ago, Humbug said:

NVIDIA has been promoting Pascal for years. So I really look forward to seeing what it can deliver.

Well, they had to change around their schedule, I think mainly because 20nm didn't happen, but maybe HMC was part of the problem too. So the Pascal we get isn't quite the Pascal Nvidia planned years ago (not saying it's better or worse, just that it's different).

[Bouzoo]

1 hour ago, ivan134 said:

You misunderstand. I meant that WCCF is basically a tabloid that usually reports all kinds of rumors, but this one was a far fetched that even they didn't report it. I'm not using WCCF as some kind of gold standard of reporting since a lot of what they do is click bait.

Don't overthink it, they report major news sometimes after all other sites do.  not talking about this one but in general

[Paragon_X]

Meh, we will see in a few months. The one thing that AMD surely is ahead of nVidia is the actual hands on experience with HBM, it would be possible that they could provide more mature drivers on launch.

[That Norwegian Guy]

8 hours ago, Dan Castellaneta said:

So Nvidia is just sitting on their ass. Sweet.

 

IMO they probably had to revise the design (probably with a PCB bolt-on Frankenstein fix) because of lack of Async compute shader support. Nvidia were so shocked when they saw how much async mattered in DX12, and AMD have had it integral to their design since 79xx series.

If they are delaying because of this, Pascal will take more than one or two months to market after Polaris. Probably 4-5 , or 3-ish if they rush 1 SKU to market, say an 880 or 870.

[Tomsen]

10 hours ago, Notional said:

It takes about a year from first tape out to the product hitting the shelf. Certainly pascal is taped out, or pascal being released this year is improbable.

I do however believe it happened recently. There's a reason why amd scrapped tsmc and went all in on glofo and Samsung. I think tsmc has huge problems with 16nm finfet, just like they did 20 nm.

That entirely depends on the process (not process node) they follow. GPUs tend to use a shorter process than CPU (less validation needed). 
That doesn't mean you can't short it down to ~6 months, if you aren't running into any bigger issues with the process node.

I doubt TSMC is having any major issues with their 16nm node. They have plenty of volume of 16nm currently, without any issues. And who is saying AMD scrapped TSMC? I'm expecting AMD to validate their GPU design on TSMC too for APU/SoC and perhaps some dGPU.

 

10 hours ago, patrickjp93 said:

Author, learn to read. The article's author only claims that all the "proof" about Pascal taping out last June is BS. He does nothing to either speculate or prove that Pascal has yet to tape out at all. Clickbait title is clickbait.

Bingo. However, it gives a great deal of speculation on how far in the process Nvidia is in taping out pascal, but nothing to determining.

[TroubleKlef]

So when can I get my hands on a 1070

[BuckGup]

11 hours ago, TechGod said:

Destroy AMD? K den, child. 

I am saying they think they are going to from past cards but they will probably be in for a shocker which will slap them in the face.

[don_svetlio]

So much for Titan coming in April. 1080 Ti coming 2017 is suddenly becoming more plausible

[Notional]

2 hours ago, Tomsen said:

That entirely depends on the process (not process node) they follow. GPUs tend to use a shorter process than CPU (less validation needed). 
That doesn't mean you can't short it down to ~6 months, if you aren't running into any bigger issues with the process node.

I doubt TSMC is having any major issues with their 16nm node. They have plenty of volume of 16nm currently, without any issues. And who is saying AMD scrapped TSMC? I'm expecting AMD to validate their GPU design on TSMC too for APU/SoC and perhaps some dGPU.

GPU's are much simpler than CPU's in the architecture. That much is true, but afaik they also tend to be bigger. It might cause tape out to shelf to decrease from the normal 1 year, but half it? I seriously doubt that. That would put polaris on the shelves in a month or two, since they already have fully functional cards in demo.

We don't really know what TSMC is capable of atm, as only NVidia are getting big chips made. As mentioned before, phone chips worked fine on 20nm as well. The fact they work on low power, small size, low performance says little of the potential of 16nm ff+ on big high performance chips like dGPU's.

AMD has not officially stated they are using 16nm/TSMC on anything. On the contrary, they have officially stated that Polaris 10 and 11 (low and high end) are 14nm FF, and that they will use Samsung foundries on top of Global Foundries, meaning their projected production exceeds that of Global Foundries capacity.

There is 0 change of any AMD apu being TSMC manufactured. Not gonna happen. Maybe a SOC, but I honestly doubt it. Both ZEN and Polaris seems to be optimised for the 14nm FF node by Samsung/GloFo. It would make little sense to change that. The only scenario I can see, is if AMD exceeds production capabilities of both GloFo and Samsung. I seriously doubt that could happen.

[Sakkura]

16 minutes ago, don_svetlio said:

So much for Titan coming in April. 1080 Ti coming 2017 is suddenly becoming more plausible

Right. If AMD launch in their vague "mid-2016" release window, and Nvidia is just a month or two behind with their mainstream GPUs, then Big Pascal slipping to 2017 seems very reasonable. It took 11 months from the GTX 680 to the GTX Titan.

[don_svetlio]

1 minute ago, Sakkura said:

Right. If AMD launch in their vague "mid-2016" release window, and Nvidia is just a month or two behind with their mainstream GPUs, then Big Pascal slipping to 2017 seems very reasonable. It took 11 months from the GTX 680 to the GTX Titan.

The interesting part is Polaris flagship is supposedly launching first. Wonder how that will end up

[Sakkura]

2 minutes ago, don_svetlio said:

The interesting part is Polaris flagship is supposedly launching first. Wonder how that will end up

Maybe the flagship will be roughly Tahiti-sized. On paper, the planar density should about quadruple, so that size still gives you a heck of a lot of transistors to work with.