[Digitimes] Kaveri delayed

[mrmt]

I think (know for a fact) that AMD looked at alternatives, including a merger with Nvidia. That didn't work out. Back in 2006, who else was making graphics good enough except ATI?

No, I'm not talking about looking for other company to partner, but of cashing out being the only viable alternative to bring the two companies together.

 

[SiliconWars]

No, I'm not talking about looking for other company to partner, but of cashing out being the only viable alternative to bring the two companies together.

The way I see it, both were too small vs their respective competitors. ATI remained competitive with Nvidia but never had anything like the money.

Cashing out made sure Intel didn't do it first, something that AMD always has to be wary of, like in the SeaMicro purchase.

 

[mrmt]

The way I see it, both were too small vs their respective competitors. ATI remained competitive with Nvidia but never had anything like the money.

Cashing out made sure Intel didn't do it first, something that AMD always has to be wary of, like in the SeaMicro purchase.

And how a merger or a well structured joint venture could be hindered by Intel?

 

[SiliconWars]

And how a merger or a well structured joint venture could be hindered by Intel?

Who knows? Why stay apart when tighter integration helps to cut long term costs and improves efficiency? It's not like AMD is the only company that ever bought another? There are tactical and strategic reasons as well.

AMD overspent if you look at it purely from a scenario where only AMD and ATI exists in a bubble. Other factors need to be considered like what did ATI get from a merger etc etc?

 

[mrmt]

Who knows? Why stay apart when tighter integration helps to cut long term costs and improves efficiency? It's not like AMD is the only company that ever bought another? There are tactical and strategic reasons as well.

AMD overspent if you look at it purely from a scenario where only AMD and ATI exists in a bubble. Other factors need to be considered like what did ATI get from a merger etc etc?

The only reason you cash out the shareholders of an acquired company is when you do not want or cannot share power with the acquired management company. It's easier to cash out everyone than to engineer a power sharing structure that works. When you look at the acquisition, AMD management preferred to spend cash they didn't have than to build a power sharing agreement that worked for both sides.

There is nor tactical neither strategic considerations that could justify wrecking AMD balance sheet the way they did to acquire ATI. Only the lazy arrogant management of the time could think it was a good move. Not surprisingly it was the same management team that gave us bulldozer.

 

[AnandThenMan]

Thanks for keeping it classy and attacking the messenger :thumbsup:

"But I have come to expect extreme negativity from you regarding anything AMD" is an observation on how you constantly post negativity involving AMD. Your "rats from a sinking ship" and other scornful comments awhile back is what really got me noticing.

Not a personal attack, good grief.

I stand by my long held assertion that AMD did not over pay for ATI, it was bad timing, bad management and other factors that magnify what went down after the acquisition. Take a look at how much Skype was sold for (more than once), Intel paid $7.68 billion for McAfee. AMD got an entire patent portfolio, world class graphics tech etc. etc. for $5.6 billion this is good value.

 

[mrmt]

I stand by my long held assertion that AMD did not over pay for ATI, it was bad timing, bad management and other factors that magnify what went down after the acquisition. Take a look at how much Skype was sold for (more than once), Intel paid $7.68 billion for McAfee. AMD got an entire patent portfolio, world class graphics tech etc. etc. for $5.6 billion this is good value.

What you buy is just half of the value equation. The other half is what you can do with what you buy.

Give a 50% discount in a ferrari to a student already drowning in debt and it will still be a bad business, because even if he can leverage himself enough to finance his new toy, he won't be able to adequately maintain or insure the car.

AMD is just like this student. It simply didn't have the money to monetize ATI assets.

 

[pablo87]

What you buy is just half of the value equation. The other half is what you can do with what you buy.

Give a 50% discount in a ferrari to a student already drowning in debt and it will still be a bad business, because even if he can leverage himself enough to finance his new toy, he won't be able to adequately maintain or insure the car.

AMD is just like this student. It simply didn't have the money to monetize ATI assets.

Its not a money issue, its bad company dragging down good one. Its the corollary of what Buffett said ("When a management with a reputation for brilliance tackles a business with a reputation for bad economics, it is the reputation of the business that remains intact") - when a good company is acquired by a bad one, the good becomes bad)

Pretend ATI is still separate, what would have happened?

* they would have retained the Smartphone GPU division - WIN
* they probably still get the Sony and MS console design wins (though the processor might look different) - Good
* their notebook discrete share would probably be much higher - WIN
* they'd be onto the next generation of desktop discrete GPU - WIN
* probably already shipping an ARM based SOC - WIN

Most important, they'd have been hiring, not firing, all this time. Don't believe me? Read the Glassdoor reviews - all bad from ATI side, many from long time employees.

If I was Intel, I'd be tempted to buy AMD and break it up - ATI back to how it was (and a big part of Intel's ecosystem), AMD standalone with its mutually self-destructive "arrangement" with Gloflo, and keep the console business (paying a royalty to ATI for its GPU tech) for itself. There would be no argument from Canadian anti-trust, that's for sure.

 

[SiliconWars]

Pretend ATI is still separate, what would have happened?

* they would have retained the Smartphone GPU division - WIN

Or they would have sold it due to a lack of money. ATI wasn't exactly awash with cash at the time of the buyout.

* they probably still get the Sony and MS console design wins (though the processor might look different) - Good

How so? What advantage did ATI have that Nvidia or even Intel didn't have this time around?

* their notebook discrete share would probably be much higher - WIN

AMD didn't fire all the ex ATI guys and replace them with trained monkeys. It's the same guys making the same chips. AMD is losing discrete notebook market share because the first iteration of GCN wasn't optimized for notebook (something they are currently changing).

* they'd be onto the next generation of desktop discrete GPU - WIN

How so, by using magic pixie dust? There is no reason to believe that ATI would be on the next generation. In fact there is more reason to believe that they'd be a year away, waiting on 20nm from TSMC to be ready for prime time, as being first on the new node was their classic "advantage" (that rarely worked).

* probably already shipping an ARM based SOC - WIN

http://phx.corporate-ir.net/phoenix.zhtml?c=116466&p=irol-reportsannual

There is a link to Nvidia's finanical reports. Feel free to go through them and tell us how much of a "WIN" Tegra has been for the company, then tell us how ATI would have been any different. (No need btw, it piqued my interest and I'm sure you'll be amazed to learn that Tegra has lost Nvidia ~$1/2 billion in the past 4 years, with the worst loss of $208 million coming last year (yes even with the high profile design wins). Now consider Intel lost $3 billion in mobile in the past 2 years. ATI would be bankrupt by now chasing that pot of gold.

Most important, they'd have been hiring, not firing, all this time. Don't believe me? Read the Glassdoor reviews - all bad from ATI side, many from long time employees.

Yep...just like Nvidia. Now compare Nvidia's staff numbers from 6 years ago vs today, compare their ballooning opex to revenue over the same time-frame. Such green grass on the other side!

 

[AnandThenMan]

* they probably still get the Sony and MS console design wins (though the processor might look different) - Good

The processor might look different? It would be completely different. And ATI would be bidding up directly against Nvidia, in this case Kepler would have been a serious contender if not a better choice, assuming ATI had been able to keep up with Nvidia at all.

* their notebook discrete share would probably be much higher - WIN
A market that is destined to be largely extinct.

* they'd be onto the next generation of desktop discrete GPU - WIN
AMD has more often than not beaten Nvidia to a new generation of GPU. When AMD purchased ATI, ATI had badly missed their targets in time to market, power draw, and performance. AMD's fabrication experience made a huge difference to the timely roll out of GPUs.

 

[pablo87]

Or they would have sold it due to a lack of money. ATI wasn't exactly awash with cash at the time of the buyout.

How so? What advantage did ATI have that Nvidia or even Intel didn't have this time around?

AMD didn't fire all the ex ATI guys and replace them with trained monkeys. It's the same guys making the same chips. AMD is losing discrete notebook market share because the first iteration of GCN wasn't optimized for notebook (something they are currently changing).

How so, by using magic pixie dust? There is no reason to believe that ATI would be on the next generation. In fact there is more reason to believe that they'd be a year away, waiting on 20nm from TSMC to be ready for prime time, as being first on the new node was their classic "advantage" (that rarely worked).

http://phx.corporate-ir.net/phoenix.zhtml?c=116466&p=irol-reportsannual

There is a link to Nvidia's finanical reports. Feel free to go through them and tell us how much of a "WIN" Tegra has been for the company, then tell us how ATI would have been any different. (No need btw, it piqued my interest and I'm sure you'll be amazed to learn that Tegra has lost Nvidia ~$1/2 billion in the past 4 years, with the worst loss of $208 million coming last year (yes even with the high profile design wins). Now consider Intel lost $3 billion in mobile in the past 2 years. ATI would be bankrupt by now chasing that pot of gold.

Yep...just like Nvidia. Now compare Nvidia's staff numbers from 6 years ago vs today, compare their ballooning opex to revenue over the same time-frame. Such green grass on the other side!

- are you serious? They got $65MM for Imageon.

- they're Gaming Consoles, remember? The GPU is the product.

- Discrete notebooks are overwhelmingly Intel CPU. Since AMD acquired ATI, ATI's discrete GPU share has been going downhill. Coincidence? I think not. You see it as an engineering issue, I see the Forest from the trees.

- ATI did 2 iterations of gpus @ 40nm.

- considering the stellar work ATI did on Brazos and Kabini, I'd say its a good bet they would have done as well or better with ARM.

- as for Tegra, perhaps their execution wasn't the best...wouldn't be the first time they had unfit product.

- and whatever losses NVIDIA is claiming (and exaggerated by the competition) I'd take with a grain of salt - almost every public company finds a scapegoat division or product whenever earnings are missed to wit Microsoft's $900MM Surface write down...kind of like when AMD wrote off $100MM of Llano inventory, only to peddle them to customers the next 2 quarters with a nice add to gross margins based on $0 cost.

Again, I suggest you go and read the Glassdoor reviews from real ATI ex-employees, the people who really know what's going on at Gemba. They are the territory; here, we're only selling maps and as we know, we live in a world where mapmakers ain't what they used to be.

 

[Enigmoid]

AMD is losing discrete notebook market share because the first iteration of GCN wasn't optimized for notebook (something they are currently changing).

GCN works very well with notebooks, the problem isn't a hardware problem, its a management problem.

(Yes the following is to do a lot with GPUs but its mainly to illustrate the poor execution AMD has done on the mobile side).

The problem isn't that AMD can't field notebook parts its that it appears that AMD has no interest in discrete GPU in notebooks. AMD is focused on their APUs much to the detriment their discrete GPU division in notebooks (unfortunatly while APUs may be the future the margin on a single chip is probably less than they would get from CPU + GPU).

Main Problems with AMD in notebook

1. Drivers are extremely bad compared to nvidia. Enduro doesn't work as well as optimus (though greatly improved). For a year after launch the 7970m had driver problems with enduro. AMD denied the issue and attempted to block forum posters. This is a minor problem and should have been fixed pronto.

Drivers on mobile are extremely finicky and to update catalyst drivers you need the OEM driver release (which most OEMs don't do). Nvidia has this problem too but its not nearly as bad.

Loss of driver support for anything sub Evergreen (HD 4000 and below don't get updates).

Again this is a problem in management.

2. Broken Hybrid Crossfire (and regular crossfire for that matter).

3. Management failure to implement 45 watt CPUs for mobile (which would greatly improve performance in larger notebooks and a lot of the lower end notebooks are 15.6" and can take the heat easily). No brainer when competing with 45 watt intel CPUs.

4. Lack of mobile models. HD 7750/7770 are perfect for mobile but there are very few GCN chips in the 7xxx series that are actually common (only the 7730m and 7970m are popular). This isn't that they don't have the chips its that they simply aren't making the chips available to mobile. A 7790 based mobile chip would be really nice for mobile yet is nowhere to be found. AMD has instead released the 384 shader 87xxm series which generally falls in with nvidia's 730m (behind the curve). 88xx chips aren't found and there is a huge jump between the 640 shader 8870m and the 1280 shader 8970m (nvidia has this nicely filled in). They simply are not releasing the hardware.

5. Rebranding. AMD on the mobile side has been an absolute mess. 7730m performs better than the 8730m; the 6750m/6770m was a winning chip in mobile being significantly stronger than nvidia's mainstream chip the 540m at a similar price but what does AMD do for the 7xxx series rebrand? Drop clocks and GDDR5 support.

3D 11 mark gpu score (notebookcheck)

6770m - 1328
6750m - 1158
7670m - 1079

Thats right the Turks mobile refresh dropped the clock rate and GDDR5 support. The predecessor of the exact same chip performed significantly better than its successor (though the 7670m is found in much cheaper notebooks).

6xxx series was quite competitive vs nvidia but vs the 7xxx series nvidia performs much better (and is available) and nvidia's 7xxm series just walks away. (730m, nvidia's low end model just walk away from AMD's mid range, 1777 points vs 1534 for the 8750m).

Again AMD's management needs to get the chips on the table. There is literally no point to releasing a 384 core mainstream chip that's WEAKER than pretty much your competitor's weakest chip that isn't the barest entry level. POOR BUSINESS DECISIONS. They instead should have worked on a 7790 based chip. (And yes they spent the R&D to design this chip which pretty much is only found in mobile (also a few OEM desktop chips) probably only so they could try to hybrid crossfire with trinity/richland with their broken hybrid crossfire when they should have been working simply on the drivers for hybrid crossfire with Kaveri (and crossfire in general).

Other problems

1. The release of Bulldozer (obvious failure and should have cut their losses and just moved an impoved phenom 2 to 32 nm-- die shrunk phenom II would probably be more efficient than piledriver given some newer powersaving features).

2. Need to work with OEMs to produce convincing products. Kabini thin and lights for example.

So you can easily see that AMD is a complete mess mainly because of management. Hardware (GPU, Kabini/Temash are nice improvements) is very good but support is generally poor.

 

[pablo87]

The processor might look different? It would be completely different. And ATI would be bidding up directly against Nvidia, in this case Kepler would have been a serious contender if not a better choice, assuming ATI had been able to keep up with Nvidia at all.

* their notebook discrete share would probably be much higher - WIN
A market that is destined to be largely extinct.

* they'd be onto the next generation of desktop discrete GPU - WIN
AMD has more often than not beaten Nvidia to a new generation of GPU. When AMD purchased ATI, ATI had badly missed their targets in time to market, power draw, and performance. AMD's fabrication experience made a huge difference to the timely roll out of GPUs.

Point #1 makes little sense. These are GAMING Consoles, not PCs. They're not dealing with AMD because of their processor competence. AMD beat NVIDIA on the basis of their GPU tech and competence, period.

#2, I'm not convinced by your argument. Its so difficult to buy these, and they're so overpriced. Every business is a growth business.

#3, it did? that's news to me. Considering all the issues AMD has had on the CPU front...

 

[wlee15]

- are you serious? They got $65MM for Imageon.

- they're Gaming Consoles, remember? The GPU is the product.

- Discrete notebooks are overwhelmingly Intel CPU. Since AMD acquired ATI, ATI's discrete GPU share has been going downhill. Coincidence? I think not. You see it as an engineering issue, I see the Forest from the trees.

- ATI did 2 iterations of gpus @ 40nm.

- considering the stellar work ATI did on Brazos and Kabini, I'd say its a good bet they would have done as well or better with ARM.

- as for Tegra, perhaps their execution wasn't the best...wouldn't be the first time they had unfit product.

- and whatever losses NVIDIA is claiming (and exaggerated by the competition) I'd take with a grain of salt - almost every public company finds a scapegoat division or product whenever earnings are missed to wit Microsoft's $900MM Surface write down...kind of like when AMD wrote off $100MM of Llano inventory, only to peddle them to customers the next 2 quarters with a nice add to gross margins based on $0 cost.

Again, I suggest you go and read the Glassdoor reviews from real ATI ex-employees, the people who really know what's going on at Gemba. They are the territory; here, we're only selling maps and as we know, we live in a world where mapmakers ain't what they used to be.

Imageon wasn't a product that last for much longer because ATI didn't have a in-house baseband solution. Even the former king of the PDA/early smartphone era application processor TI had to leave the marketplace.

 

[ShintaiDK]

GCN works very well with notebooks, the problem isn't a hardware problem, its a management problem.

(Yes the following is to do a lot with GPUs but its mainly to illustrate the poor execution AMD has done on the mobile side).

The problem isn't that AMD can't field notebook parts its that it appears that AMD has no interest in discrete GPU in notebooks. AMD is focused on their APUs much to the detriment their discrete GPU division in notebooks (unfortunatly while APUs may be the future the margin on a single chip is probably less than they would get from CPU + GPU).

That what happens when you compete with yourself. APU sales canibalize dGPU sales. And unlike the dGPU, the APUs are essentially sold with free IGP as a value added part. Due to the competition from Intel. And again the management issue of AMD prefering to focus on free IGPs than dGPUs. Hence why nVidia completely dominates the very lucrative segment.

 

[ShintaiDK]

http://vr-zone.com/articles/new-confirmed-details-on-amds-2014-apu-lineup-kaveri-delayed/47455.html

More confirmation on that Kaveri is delayed into 2014:

According to this revised schedule we obtained, engineering samples of Kaveri won’t be ready until August, production candidate samples will go out in October and initial production will begin in December. The target launch into channel is now mid-February 2014.

If production first starts in december, then volume will be extremely low at february.

And excavator is a 2015 product. Most likely new socket as well.

 

[rainy]

ShintaiDK, I'm just curious: you're bashing AMD directly or indirectly just for fun (as it was said by IDC)?

 

[Sweepr]

http://vr-zone.com/articles/new-confirmed-details-on-amds-2014-apu-lineup-kaveri-delayed/47455.html

More confirmation on that Kaveri is delayed into 2014:


If production first starts in december, then volume will be extremely low at february.

And excavator is a 2015 product. Most likely new socket as well.

Good find. Indeed its now delayed to 2014, and they dont mention if desktop or mobile Kaveri is coming in February. Both could be available in February or one of them could be available even later.

 

[cytg111]

ShintaiDK, I'm just curious: you're bashing AMD directly or indirectly just for fun (as it was said by IDC)?

Being around for some time my best bet is that he gets his paycheck, if not directly then indirectly, not from AMD but another competing source! Given the intel, depth and drive it really doesnt make sense otherwise.l

 

[frozentundra123456]

ShintaiDK, I'm just curious: you're bashing AMD directly or indirectly just for fun (as it was said by IDC)?

He looks like a paragon of objectivity compared to some of the Intel bashers.

 

[cytg111]

He looks like a paragon of objectivity compared to some of the Intel bashers.

Personally I dont get it, it is as dark to me as religion is.

 

[bgt]

Being around for some time my best bet is that he gets his paycheck, if not directly then indirectly, not from AMD but another competing source! Given the intel, depth and drive it really doesnt make sense otherwise.l

Haha, what a nice guess!!! He is a bit dark sometimes but he is honest:thumbsup:

 

[bgt]

He looks like a paragon of objectivity compared to some of the Intel bashers.

Is some1 bashing Intel...............did not notice it here really()

 

[rainy]

He looks like a paragon of objectivity compared to some of the Intel bashers.

When I've recently registered here at Anandtech forum I was hoping to avoid seeing stupid/idiotic fans' (more precisely fanboys) wars that I see so often on Polish forums (I'm from Poland).

Looks like I was wrong, maybe not completely but at least partially: many people here behave exactly this way.

 

[Erenhardt]

GCN works very well with notebooks, the problem isn't a hardware problem, its a management problem.

Sorry for such a SNIP
GCN work well with notebooks, but that doesn't negate the point that GCN is not optimized for notebooks. No one needs so much compute performance on a mobile device. Bitcoin mobile mining is not a real thing, is it? Kepler is striped from this YET useless when it comes to gaming fat. That give a little edge over GCN in power consumption and thermals