AMD Confirms 20nm Products Moved to FinFET, Warns on Q2’15 Earnings
by Ryan Smith on July 6, 2015 4:55 PM EST- Posted in
- AMD
- CPUs
- GlobalFoundries
- 20nm
- GPUs
- 16nm
- 14nm
- Financial Results
A little bit ago AMD sent out an announcement updating their official outlook for the second quarter of 2015. Though we don’t typically publish financial projections, the long and short of it is that AMD is looking to brace investors for a worse than expected Q2, after an already difficult Q1. Soft APU sales are being blamed for dragging down both revenue and gross margins, with AMD now expecting Q2 revenue to be down 8% sequentially, or around $950M, while the non-GAAP gross margin will be just 28%.
Much more interesting however is this little nugget of information buried in the announcement towards the end, offering a short update on AMD’s 20nm plans. AMD had previously announced their intentions to bring out some products at 20nm – these were most likely just APUs, with the only one we explicitly know about being the now-canceled Skybridge. In any case, AMD is now confirming that they have moved several of their 20nm designs to a “leading-edge FinFET node,” and as far as we know AMD no longer has any further 20nm projects in the pipeline. AMD’s press release does not state which foundries these products are now at – or indeed if they’re at multiple foundries – so it’s unknown at this time whether the work is at TSMC, GlobalFoundries, or split between the two of them.
The rationale for announcing this shift at this time comes from the financial aspect. AMD will be taking a $33M charge to their GAAP gross margin as part of the work required to move these designs to a new node. Jumping to FinFET nodes should improve the competitiveness of these products, and greatly so in the case of anything that needs to clock high or is otherwise heavily exposed to leakage, but of course this will take additional time and engineering resources in order to transition these products.
We expect AMD to discuss the issue in at least a bit more depth later next week, when they hold their Q2 earnings call on July 16th.
Source: AMD
Facebook
Twitter
RSS
62 Comments
View All Comments
rocketbuddha - Tuesday, July 7, 2015 - link
The biggest problem is that AMD ends up in really "cheap": notebooks with very little expandability and quality items that it does not do any justice to it.They are saddled with 720p screens, poor keyboard, cheap build quality, 5400 RPM mechanical HDDs as well as downright 4 cell battery.
And for a cheaper price you will see an equally cheap/cheaper celeron model [with the same characteristics]. So if somebody goes by price they can go either way.
If somebody wants quality they would avoid the AMD model altogether and go with the expensive i7.
jjj - Monday, July 6, 2015 - link
It was pretty much a given that they'll have no 20nm products ,except maybe for console and they don't actually state that they won't have anything on 20nm, just that some products planned on 20nm will go smaller.The financial part is far more interesting though (if you actually look at it).
Q1 was 1.03B revenue with Computing and Graphics segment revenue 532 million, while Console,embedded and server was 498 million.
So if you assume Q2 revenue some 950 mil and console flat (could be better than flat but lets say server is down and console up so all in all flat), computing ends up at 450 million so some 15% down on quarter.The compute and graphics segement would also be 46% down on year and 55% down vs 2 years ago (994 million). And this is CPU, APU , GPU while Nvidia does 2x just from GPU.
With Zen rumored to arrive in oct-nov next year and APUs with Zen only in 2017 things will keep going downhill for them for quite some time no matter what.
Zen expected to be just 8 cores and perf somewhat unknown is not very encouraging. If Intel can do 4 cores with 8MB L3 is some 60mm2 on 14nm,then AMD Zen better have 12-16 cores at same per core perf and 300$ price tag or much better perf per core than Intel if they only have 8 cores. Costs at 14nm might be high compared to 28nm but 28nm is cheap because it's old and if 14ff has good yields ,there is no reason not to give us big chips. Zen could catch Intel with it's pants down but only if AMD gives us perf and core count. With Intel wasting die on a GPU ,AMD could do much better if Zen is competitive. Look at Broadwell desktop , 160mm+ die with more than 100mm2 for the GPU and eDRAM interface when they could have fitted 12 cores with 24MB L3 and wider memory interface with no GPU in the same die.
At this point Zen is the last design they can afford given their revenue by the time it launches and they better get not only the core right but the products right. They'll really need a huge jump in revenue with Zen to be able afford 10nm designs.
Nagorak - Monday, July 6, 2015 - link
There is no point having 12 or 16 cores. Few things even take advantage of 4 cores at this point, even 8 will already be overkill. AMD needs better single threaded performance, not just throw in more cores. If their single threaded performance is bad then they are screwed either way.Anyway, I think everyone basically agrees that Zen is AMD's last hurrah. Either it's good and turns the company around, or stick a fork in them.
OrphanageExplosion - Tuesday, July 7, 2015 - link
Virtually every modern game uses at least four cores. Even the horrific Arkham Knight utilises all eight threads on my i7 4790K.An eight-core processor 16/17 makes sense.
frozentundra123456 - Tuesday, July 7, 2015 - link
There is a niche there if the product is competitive, and they could take back some of the server market. But the vast majority of consumer, enterprise, and education users certainly dont need 8 cores. And having no igp will narrow the market dramatically. Plus the market is moving to mobile, where no one is going to want an 8 core 95 watt TDP monster. The only hope i see for AMD to become relevant in the mainstream is a Zen apu with HBM.icrf - Tuesday, July 7, 2015 - link
Are gamers still a driving force behind the CPU performance market? I thought good enough was hit awhile ago with playable frame rates always GPU limited.gamerk2 - Tuesday, July 7, 2015 - link
But the ability to use more cores does NOT give a performance increase. You only get an increase in performance if using more cores removes a CPU bottleneck that existed when fewer cores were used. That's why the i5 can i7 typically give the same performance at the same clock, even when a game scales to 8 cores, because even though the i5's cores are more burdened, they aren't bottlenecked, and thus no performance disadvantage exists.This is a very simple concept in SW design, but one that keeps getting ignored. The ability to use more cores only gives more performance when a bottleneck is removed. No bottleneck, no performance impact.
Morawka - Tuesday, July 7, 2015 - link
but can amd last till 2016-2017 (zen release) when they are losing tens of millions per quarter? hell by then, skylake will be mature, nvidia's Pascal will be out, and ddr4 will be common place.they are just too far behind, and didnt spend the money on smart engineers. amd is getting talent from the bottom barrels of society.
gamerk2 - Tuesday, July 7, 2015 - link
AMD's debt repayments dont begin until 2018, which is when their balance sheets are going to get strained.Zen is AMD's last shot; with total debt equaling their total valuation, and large debt repayments on the horizon, AMD simply can't afford a miss. If Zen misses, things could get bad an a hurry.
jjj - Tuesday, July 7, 2015 - link
You can go ahead and buy fewer cores then, at a much much lower price.Just because you don't need more computing power doesn't mean that people that do must pay thousands of dollars for more cores. Intel is abusing the lack of competition and AMD can exploit that. There are lots of people that need as many cores as they can get.The only reason we don't have more cores is because Intel can sell whatever garbage they want, in a competitive market the landscape would be very very different.
Sure Intel's marketing will try to convince you that you are better off paying for a small die where 2/3 of it is something you don't need. Intel screwed everybody with Gulftown in 2010 and since then things only got worse and worse. Everybody that was around back then is still waiting for a reasonable number of cores at a sane price.
As for single threaded , i never asked for less perf , you just failed to get the point that they can offer a lot more than Intel and they really should, they can't afford to miss opportunities.
Intel could offer 4 cores today well bellow 100$ in retail,easily and AMD (if competitive) can force them to.If yields are very good they could even do 50$ for 4 cores no GPU.