AMD CEO talks of long-term turnaround

[mrmt]

Go POWER8? AMD has got to worry about that, too, if they are looking to break into the server room. If Keller thinks he can challenge POWER8 in SAP benchmarking etc., he may have another thing coming to him.

Perf/watt is the single most important metric on the server market, so if POWER8 is still a power hog then it will still lose share to Xeon, despite being an improvement over POWER7.

 

[ShintaiDK]

Its clear IBM know POWER is dead. Else they wouldnt try to prolong it with the OpenPower attempt. But as said, it will be a slow steady decline.

 

[sm625]

You asked whether or not carrizo beats an 8350 knowing full well carrizo is a mobile only processor. Hence I asked whether or not broad well(also a mobile only processor) is less advanced than haswell a desktop processor too.

Wait so you are saying this machine has a mobile processor? Who knew...

 

[The Alias]

Wait so you are saying this machine has a mobile processor? Who knew...

that vs a 5960 ........ you're missing the point. The point is you can not compare a mobile only processor to a desktop processor and use the fact the desktop wins as grounds to say the desktop processor is more advanced.

 

[AtenRa]

Even if we take desktop, SteamRoller (Kaveri) is more advanced than Vishera. So again No, Vishera is not the most advanced AMD CPU in desktop.

 

[Dresdenboy]

What if speed is just one criteria?

It obviously has been sacrified.

CPU design is a matter of goals and compromises to reach them in the best way. You have multiple possible directions to improve your design. In a simple way one could depict this as trading one unit of speed to get one unit of power efficiency and one of area efficiency.

 

[mrmt]

It obviously has been sacrified.

CPU design is a matter of goals and compromises to reach them in the best way. You have multiple possible directions to improve your design. In a simple way one could depict this as trading one unit of speed to get one unit of power efficiency and one of area efficiency.

I agree here, but doesn't change the fact that for some cases Piledriver is much better than Carrizo, just because Carrizo wasn't engineered for these cases.

Basically Carrizo is designed for a much narrower market than Piledriver was, and Carrizo will probably have far less volumes than any other AMD mainstream products in their recent history. Carrizo is unusable for anyone running a server, or anyone running a reasonably fast desktop. So basically Carrizo is only better for its target market, but due to the narrow product scope, it's not an overall better solution than Piledriver because of the wider span of the latter. Carrizo doesn't just offer a different set of design tradeoffs to serve all the market of its predecessors, but an overall narrower scope on top of the design tradeoffs.

 

[AtenRa]

Basically Carrizo is designed for a much narrower market than Piledriver was, and Carrizo will probably have far less volumes than any other AMD mainstream products in their recent history. Carrizo is unusable for anyone running a server, or anyone running a reasonably fast desktop. So basically Carrizo is only better for its target market, but due to the narrow product scope, it's not an overall better solution than Piledriver because of the wider span of the latter. Carrizo doesn't just offer a different set of design tradeoffs to serve all the market of its predecessors, but an overall narrower scope on top of the design tradeoffs.

Excavator is designed for more than mobile.

 

[mrmt]

As I said, bad numbers are incoming.


AMD issues Q2 warning; shares halted (updated)
04:20 PM ET • AMD • SA Editor Eric Jhonsa
AMD now expects Q2 revenue to be down 8% Q/Q, worse than prior guidance for a 3% drop (+/- 3%) and a consensus for revenue to fall 3% to $1.00B. The company primarily blames "weaker than expected consumer PC demand impacting the company's Original Equipment Manufacturer (OEM) APU sales." (Press Release)
Gross margin (non-GAAP) is expected to be at 28%, below prior guidance of 32% and down from Q1's 32%. AMD blames both lower PC APU volumes and a mix shift towards enterprise, embedded, and semi-custom chip sales.
AMD also forecasts a $33M GAAP charge related to migrating several designs previously set to use a 20nm manufacturing process to FinFET processes (presumably at 14nm or 16nm). The Q2-ending cash balance is expected to be $830M, down from Q1's $906M and in-line with expectations.

 

[Fjodor2001]

Nothing unexpected, really. No one is buying PCs now, because everyone is waiting for Win 10. Also, most people held off buying an AMD GPU until the next gen was presented just recently.

 

[DrMrLordX]

Perf/watt is the single most important metric on the server market, so if POWER8 is still a power hog then it will still lose share to Xeon, despite being an improvement over POWER7.

What I really wanted to see was a perf/watt comparison between Haswell-EX and POWER8. What we got instead was:

http://www.anandtech.com/show/9193/the-xeon-e78800-v3-review/16

No POWER8 perf/watt or watt/job numbers. Boo.

 

[DrMrLordX]

Excavator is designed for more than mobile.

One of the things I don't understand about Bristol Ridge is the timing. Why wait until Q3 2016 to launch it? Is Promontory not ready yet? DDR4 is already on the market. Why not just launch AM4 ASAP with Bristol Ridge? It's 28nm Excavator . . . is there something in there that isn't yet ready for production?

 

[Phynaz]

As I said, bad numbers are incoming.


AMD issues Q2 warning; shares halted (updated)
04:20 PM ET • AMD • SA Editor Eric Jhonsa
AMD now expects Q2 revenue to be down 8% Q/Q, worse than prior guidance for a 3% drop (+/- 3%) and a consensus for revenue to fall 3% to $1.00B. The company primarily blames "weaker than expected consumer PC demand impacting the company's Original Equipment Manufacturer (OEM) APU sales." (Press Release)
Gross margin (non-GAAP) is expected to be at 28%, below prior guidance of 32% and down from Q1's 32%. AMD blames both lower PC APU volumes and a mix shift towards enterprise, embedded, and semi-custom chip sales.
AMD also forecasts a $33M GAAP charge related to migrating several designs previously set to use a 20nm manufacturing process to FinFET processes (presumably at 14nm or 16nm). The Q2-ending cash balance is expected to be $830M, down from Q1's $906M and in-line with expectations.

Wait, I thought selling more of those was supposed to good?!? Wasn't this the strategy AMD just presented that was going to get it out of the hole?

 

[Phynaz]

Holy smokes, stock price is down 15%. Lisa's stock options are tanking.

 

[mrmt]

Nothing unexpected, really. No one is buying PCs now, because everyone is waiting for Win 10. Also, most people held off buying an AMD GPU until the next gen was presented just recently.

It was unexpected for AMD management at least. It was them giving the missed forecast.

 

[Phynaz]

Nothing unexpected, really. No one is buying PCs now, because everyone is waiting for Win 10. Also, most people held off buying an AMD GPU until the next gen was presented just recently.

If no one is buying PCs then why didn't Intel warn? This is strictly an AMD issue, not a market issue.

 

[mrmt]

Holy smokes, stock price is down 15%. Lisa's stock options are tanking.

The drop in gross margins has far more deleterious effects than the the revenue drop. I think finally Intel 14nm parts are coming in full force to the market.

But yes, Lisa has a lot of explaining to do. Carrizo is proving to be a train wreck, and it has to last until 2017.

 

[mrmt]

Wait, I thought selling more of those was supposed to good?!? Wasn't this the strategy AMD just presented that was going to get it out of the hole?

It's a nice complementary income stream, but far lower quality when compared to a healthy bleeding edge cpu business.

 

[JDG1980]

Today's financial news shouldn't be a surprise. AMD doesn't really have much worthwhile to sell at the moment, and won't until FinFET is ready. We've known at least since the 300 series graphics release that AMD was going to be in a holding pattern throughout 2015, with only a handful of new products (Carrizo and Fiji), both of them fairly niche, and years-old rebrands for everything else.

2016 will be the make-or-break year for AMD. I know this has been said before, but AMD is near the end of its rope, and the 2016 products (Zen and Arctic Islands) have to be successful, or AMD is finished as an independent company. Lisa Su has said as much in the interview with Tom's Hardware; they're betting the company on Zen being a success.

The rest of 2015 will likely be almost as bad as this quarter. What did anyone expect? Of course non-console APU sales are down. The wonder is that anyone's buying them at all. At least Carrizo has the potential to be moderately competitive in mid-range laptops, though the fact that it's only shown up in crappy, cut-rate devices so far is not encouraging.

 

[Arachnotronic]

Today's financial news shouldn't be a surprise.

It was obviously a surprise to AMD as they missed their already very modest expectations.

 

[JDG1980]

But yes, Lisa has a lot of explaining to do. Carrizo is proving to be a train wreck, and it has to last until 2017.

Did Lisa Su even have anything to do with Carrizo? Given the lead time on designing CPUs, that seems questionable. Sure, she could have cancelled it, but in that case AMD would have nothing at all even remotely competitive for the laptop market.

So far I haven't seen full benchmarks, but based on the leaked results that have shown up here and there, Carrizo doesn't look too bad in terms of hardware performance. Where AMD has really fallen down (as usual) is in the marketing and management side. We're seeing nonsense like HP offering only a 1366x768 screen and no SSD on Carrizo laptops, while allowing Full-HD and SSD upgrades on the corresponding Intel model. If AMD wants to get their products out of the bargain basement, they have to be more aggressive about pushing back against this kind of thing. They should create their own ultrabook design for Carrizo if they have to - even in their precarious financial situation this is a fairly low-cost, short-range R&D project that could help repair their brand image.

 

[mrmt]

Today's financial news shouldn't be a surprise. AMD doesn't really have much worthwhile to sell at the moment, and won't until FinFET is ready.

What should't be a surprise? The sales drop? Then why didn't AMD management forecast it correctly?

 

[Arachnotronic]

Did Lisa Su even have anything to do with Carrizo?

Sure. She's been a higher-up at AMD since 2012.

 

[ShintaiDK]

Holy smokes, stock price is down 15%. Lisa's stock options are tanking.

Wow that is crazy. Revenue is in the sewer. No more M 4 digit revenue.

 

[Arachnotronic]

Wow that is crazy. Revenue is in the sewer.

Revenue plunging, margins in the gutter.

This is why I find it baffling when some posters here claim that the "doomsayers" with respect to AMD are making completely unfounded claims. AMD's business is falling apart and it's going to take a lot to fix this.