The AMD Execution Thread [2007 - 2017]

[Alexko]

SirEric left to join Qualcomm - wish him the best there.

It will be interesting to see where AMD is in a years time - we are all negative here but AMD stock has risen since Rory Read took the helm. Do investors and analyst know something we don't?

It has risen, but not significantly more than the NYSE, really.

Still, I think the analyst community has had a fairly positive reaction to Read, overall. It may be his experience and track record at Lenovo, his energetic style, or his nearly obsessive (displayed) emphasis on execution—which, frankly, hasn't exactly been AMD's strong suit lately—or perhaps his recent decision to take the company to a kind of SoC-inspired mix-and-match strategy.

AMD has had serious missteps in the past few years, and I think it's reasonable to assume that some of them were due to poor management. So perhaps the recent overhaul isn't such a bad thing, and that may be what analysts think too.

Then again, AMD's financial situation has been steadily improving these past few quarters, so the stock rise could merely reflect that.

 

[Silent_Buddha]

It has risen, but not significantly more than the NYSE, really.

Still, I think the analyst community has had a fairly positive reaction to Read, overall. It may be his experience and track record at Lenovo, his energetic style, or his nearly obsessive (displayed) emphasis on execution—which, frankly, hasn't exactly been AMD's strong suit lately—or perhaps his recent decision to take the company to a kind of SoC-inspired mix-and-match strategy.

AMD has had serious missteps in the past few years, and I think it's reasonable to assume that some of them were due to poor management. So perhaps the recent overhaul isn't such a bad thing, and that may be what analysts think too.

Then again, AMD's financial situation has been steadily improving these past few quarters, so the stock rise could merely reflect that.

Their execution on the CPU side is certainly lacking, but currently they are executing fantastically on the GPU side. They have been releasing product on a regular yearly basis ever since Cypress. It was 1 year 3 months after Rv770 and only 7 months after 4890. And thus far it seems like Keplar is going to be similar to Fermi in being late to the party.

So on the CPU side they are certainly doing worse than Intel while on the GPU side they are doing significantly better than Nvidia with regards to execution.

So it's diffuclt to just use a blanket statement as to whether they ar executing well or not well as a company. And if they can ever get enough supply of their Fusion APU's they might actually be able to make some inroads versus Intel in the consumer market.

Regards,
SB

 

[Alexko]

Their execution on the CPU side is certainly lacking, but currently they are executing fantastically on the GPU side. They have been releasing product on a regular yearly basis ever since Cypress. It was 1 year 3 months after Rv770 and only 7 months after 4890. And thus far it seems like Keplar is going to be similar to Fermi in being late to the party.

So on the CPU side they are certainly doing worse than Intel while on the GPU side they are doing significantly better than Nvidia with regards to execution.

So it's diffuclt to just use a blanket statement as to whether they ar executing well or not well as a company. And if they can ever get enough supply of their Fusion APU's they might actually be able to make some inroads versus Intel in the consumer market.

Regards,
SB

The hardware GPU guys are doing very well, yes, but I wouldn't call their execution "fantastic" on the software side.

Take FirePros, for instance: their hardware is great, but the software leaves so much to be desired that their market share in this segment is almost negligible. That may sound like a small detail, but past financial reports given out by NVIDIA show that Quadros bring the majority of their profits.

In fact, AMD's graphics division may appear to be doing well, and in many ways it is, but in practice it usually barely breaks even, sometimes not even that. NVIDIA, on the other hand, and in spite of its err… snafus, has been quite profitable.

 

[Erinyes]

The hardware GPU guys are doing very well, yes, but I wouldn't call their execution "fantastic" on the software side.

Take FirePros, for instance: their hardware is great, but the software leaves so much to be desired that their market share in this segment is almost negligible. That may sound like a small detail, but past financial reports given out by NVIDIA show that Quadros bring the majority of their profits.

In fact, AMD's graphics division may appear to be doing well, and in many ways it is, but in practice it usually barely breaks even, sometimes not even that. NVIDIA, on the other hand, and in spite of its err… snafus, has been quite profitable.

You have to remember that APU revenue is counted along with CPU revenue and not GPU. So while the GPU division is actually contributing a significant portion to the design of the APU, the revenue and profits are not reflected in their results.

 

[3dilettante]

APUs are relatively recent, though. The graphics group didn't set the world on fire prior to that.

 

[Kaotik]

APUs are relatively recent, though. The graphics group didn't set the world on fire prior to that.

But they were the only part of AMD doing positive results almost every quarter

 

[3dilettante]

It's better than losing money, although it did a few times. The positive results weren't that much, not compared to CPUs, nor even within its own slice: graphics revenue.
We know it's possible to make a good money in graphics, it's just not being made at AMD.

If Demers has moved on to Qualcomm, it may be a better fit since Qualcomm doesn't go after the markets AMD's graphics group services without distinction or reward.

 

[Alexko]

It's better than losing money, although it did a few times. The positive results weren't that much, not compared to CPUs, nor even within its own slice: graphics revenue.
We know it's possible to make a good money in graphics, it's just not being made at AMD.

Exactly. I guess it's possible that AMD was shifting some part of CPU-related expenses to the graphics division, just to make the CPU division look better (or less terrible) but still.

 

[Farid]

It's interesting to see that we'll finally see PowerVR products competing again with NV and exATI again in the high end market. The high-end tablet market, that is.

And while centering the AMD business around SOCs and Cloud appliances/solutions might be a sound politic at this point in time, I still think that the executive core should not forget what made AMD and ATI relevant in the market place in the first place.

By the way, the Eric Demers leaving AMD news should have its own thread, I think.

 

[Silent_Buddha]

In fact, AMD's graphics division may appear to be doing well, and in many ways it is, but in practice it usually barely breaks even, sometimes not even that. NVIDIA, on the other hand, and in spite of its err… snafus, has been quite profitable.

Sure but that doesn't have anything to do with their execution in getting product out, at least in the consumer and OEM space. It has more to do with marketing and low pricing affecting not only their ability to move product but the revenue it generates when sold. This is the first generation since the x19xx series that AMD has been on price parity with Nvidia.

But unfortunately they've lost all revenue from the budget OEM sales as that has move entirely to the CPU side. So a large chunk of revenue and potential profits gone from GPU into CPU. So, it'll be interesting to see how things go forward now that they can price similarly to Nvidia.

But either way I think marketing still has quite a ways to go to catch up with Nvidia in the consumer space.

In the professional space GCN should help them considerably once their professional versions of GCN go through enough validation (in about 6 months to a year, I'm guessing). Well at least in the HPC sector.

It's still going to be difficult to make inroads with Nvidia as entrenched as they are there. It'll be very important (IMO) to start getting some compute integration with some of the prosumer and professional production suites (Adobe products would be a good start).

Regards,
SB

 

[rpg.314]

In the professional space GCN should help them considerably once their professional versions of GCN go through enough validation (in about 6 months to a year, I'm guessing). Well at least in the HPC sector.

I doubt it. They are out of HPC and pro markets due to their sw. New hw won't fix it. Especially with all the layoffs....

 

[Alexko]

Sure but that doesn't have anything to do with their execution in getting product out, at least in the consumer and OEM space. It has more to do with marketing and low pricing affecting not only their ability to move product but the revenue it generates when sold. This is the first generation since the x19xx series that AMD has been on price parity with Nvidia.

But unfortunately they've lost all revenue from the budget OEM sales as that has move entirely to the CPU side. So a large chunk of revenue and potential profits gone from GPU into CPU. So, it'll be interesting to see how things go forward now that they can price similarly to Nvidia.

But either way I think marketing still has quite a ways to go to catch up with Nvidia in the consumer space.

In the professional space GCN should help them considerably once their professional versions of GCN go through enough validation (in about 6 months to a year, I'm guessing). Well at least in the HPC sector.

It's still going to be difficult to make inroads with Nvidia as entrenched as they are there. It'll be very important (IMO) to start getting some compute integration with some of the prosumer and professional production suites (Adobe products would be a good start).

Regards,
SB

Are you sure about the bold part?

Anyway, as far as I know, NVIDIA's professional segment profits come mostly from Quadros, not Teslas. On the hardware side, AMD wasn't particularly disadvantaged in this sector, but their drivers suck so bad that few people even consider FirePros. I have yet to see any sign from AMD that this is about to change. No statement of any kind from execs or anything, and, frankly, marginal improvement from generation to generation. I find the fact that they're losing such a large amount of potential profit (again, looking at NVIDIA) due to an apparent lack of willingness to even try is quite baffling to me.

A wild guess would be that pro drivers are a casualty of AMD's efforts to reduce costs everywhere, and that some key people decided the investment in pro drivers somehow wasn't worth it. I usually try to refrain from armchair-CEOing, but this really seems short-sighted.

 

[Kaotik]

Are you sure about the bold part?

APU profits are indeed counted for the CPU department completely, not GPU

 

[Alexko]

APU profits are indeed counted for the CPU department completely, not GPU

Right, sorry, I read Silent_Buddha's statement a bit too fast and misunderstood it.

 

[Silent_Buddha]

I doubt it. They are out of HPC and pro markets due to their sw. New hw won't fix it. Especially with all the layoffs....

In HPC it's all about hardware, support and how robust programming languages/enronments. Nvidia has had a significant impact here not only with compute focused hardware which significantly outperformed their competition from AMD, but also with a relatively robust and stable CUDA environment.

Of course, when I made my statement in the previous post, I'm assuming that AMD will do a better job with OpenCL/Direct Compute/whatever else is used for HPC than they did with OpenGL. We're still quite a few months away from seeing how their efforts in this space do with GCN hardware. But at least their hardware is generally on par or better currently. We'll see if Keplar (whenever it finally comes out) can give Nvidia a significant performance advantage in compute again or not.

Anyway, as far as I know, NVIDIA's professional segment profits come mostly from Quadros, not Teslas. On the hardware side, AMD wasn't particularly disadvantaged in this sector, but their drivers suck so bad that few people even consider FirePros. I have yet to see any sign from AMD that this is about to change. No statement of any kind from execs or anything, and, frankly, marginal improvement from generation to generation. I find the fact that they're losing such a large amount of potential profit (again, looking at NVIDIA) due to an apparent lack of willingness to even try is quite baffling to me.

A wild guess would be that pro drivers are a casualty of AMD's efforts to reduce costs everywhere, and that some key people decided the investment in pro drivers somehow wasn't worth it. I usually try to refrain from armchair-CEOing, but this really seems short-sighted.

In theory even if the driver team has been cut down, and I haven't really heard that it has. Last news I heard was that AMD have invested more money into the driver team and signficantly more money into developer relations and support. But back to the point, in theory, GCN hardware should be easier to extract performance from in a wider variety of cases in the professional market. A market which, while still dominated by graphical loads, is also moving into compute. Not as fast as the HPC segment, but certainly much faster than the consumer segment. And here it looks like some things are finally looking up for AMD. Adobe moving to OpenCL and eventually dropping CUDA is going to help. I know personally that AMD lost out on a lot of money the past few years due to this. It was almost impossible to sell AMD product to anyone investing in an Adobe production environment. OpenGL is still a concern of course (http://kb2.adobe.com/cps/831/cpsid_83117.html ), but it generally isn't a show stopper for most like it was years ago. R700 and newer architectures don't seem to have OpenGL problems at least with Adobe.

Basically there's an opportunity here for them. Whether they can take advantage of it or not is unknown at this point. And even if they do get drivers and support on par with Nvidia, it's still highly unlikely they'll make much inroads unless their Marketing department does a lot better than it has in the past.

And that's where I think they may still have significant problems as I believe it's the Marketing department that suffered the most cuts, but I could very well be wrong on this.

Regards,
SB

 

[Arty]

AMD lost out on MacBook Air win over Llano yields apparently.

AMD struggled with its new fabless model while trying to crank out “fusion” processors that combined a CPU and a GPU in a single part. On paper the idea was promising. A notebook processor dubbed “Llano” got a close look from Apple for an update to the ultralight MacBook Air, scheduled for launch in mid-2011.

But AMD couldn’t get early working samples of Llano to Apple on time, one former employee says. Several former AMD employees disagree on just how close AMD came. “We had it,” one says. But too many of the Llano parts were faulty. AMD lost the deal.

 

[Zaphod]

AMD lost out on MacBook Air win over Llano yields apparently.

The source AMD profile from Forbes Magazine is worth a read in itself: After Years Of Hard Knocks, AMD Honing Its Killer Instinct.

 

[CarstenS]

Is it? I scanned through it an did not see anything really new or interesting except for the anecdote of Read rewarding his staff with a bumper car basketball match.

 

[Zaphod]

Maybe not from a tech perspective, but on the "3D & Semiconductor Industry" board in the "The AMD Execution Gloom Thread", I'd say yes. Not very in-depth, but with a couple of interesting tidbits, and reading how the business rags sees the industry always provides some perspective.

 

[rpg.314]

Nvidia has had a significant impact here not only with compute focused hardware

...eh, what?....

which significantly outperformed their competition from AMD,

um.., no.