[silent_guy]

NVidia has reported at some point that its consumer GPU BU was operating at a slight loss, but that doesn't mean it'd be more profitable to leave that market. The professional business uses that roughly the same silicon and since silicon developments accounts for the vast majority of NRE, the positive gross margins of the consumer silicon fuels development of the professional business.

This has been discussed before: there are various ways in accounting to allocate R&D to BUs. If all silicon development costs were put on the account of the professional BU, the consumer BU would suddenly be crazy profitable at the expense of professional. The bottom line wouldn't change: at decently profitable company with great synergy between 2 product lines.

One of ATI/AMD's major failures is that they have never been able to achieve this, relying only on consumer GPUs. And since high-NRE technology is ultimately a winner-take-all industry, it's really just a matter of time.

[3dilettante]

Originally Posted by AlexV

Ridiculously so, and this may be part of the core of their issues. The deal was dolt-ish (far too cash heavy), and it was clearly a result of either panic (why?) or interesting pressures being exerted. I'm sure that people will point out that the GPG is the only group that's sort of performing within AMD, but that's really not saying much when factoring in just how overvalued ATI at the time of acquisition (the trillion or so goodwill write-offs hint to it strongly).

The character of the company's leadership has been less than stellar.
Ruiz in the years after the purchase showed a serious lack of ethics and a brush with an insider trading scandal when it came to fab-light. His selection as CEO came after Hector oversaw Motorola giving up the ghost on its semiconductor division, which is an odd precedent to bring to a semiconductor company.
Hector, his friends, and those underwriting the purchase probably made out all right from it.

Ruiz would've been a good chap to manage fabs, but just that (having a preeminently Ops man decide upon acquisitions is unwise, given that he'll want to get everything in-house and he'll want the expensive best), and Dirk would've been better off left to manage small / medium-sized teams of engineers as opposed to figuring out where to place the company's chops (he was truly awful at that, if we look at the current situation which is strongly influenced by the calls he made).

I'm genuinely curious about where Dirk ran into the Peter Principle. I suppose the determination would depend on just how much credit he deserved for K7 and K8, especially after losing or possibly driving out a good number of the people involved in it.
I would say it is likely that Barcelona is near that inflection point.

It's unclear where Read fits in, although I honestly think that him doing some cobweb cleaning in the upper tiers of the hierarchy and the super-underperforming areas (see the Marketing dept. cuts) was a good call. It remains to be seen if he's anything more than a one trick pony because if you slice up everybody you end up alone.

It seems like he's cutting AMD down so it can be small enough to linger as a company at the margins.
I'm not even sure that is possible given that there are better-run or better-backed marginal players.

edit:
Or as some speculative posts I've seen elsewhere, Read's trying to slash off anything that a takeover wouldn't need. Sure the x86 stuff would be a likely loss, but it already is.

[3dcgi]

Also, Nvidia pushed 3dlabs out of the business because 3dlabs didn't have the consumer business to finance workstation.

[AlexV]

Originally Posted by 3dilettante

...Hector...Dirk...

It is also interesting to note, and rather illustrative of one of the wrongs of modern society, that these fellows got very good severance packages. For utterly failing at the job they were supposed to do. Also, it took many years to clean other people that were intensely underachieving (see their stream of CMOs).

[Raqia]

Rory Read seems serious about addressing the fiscal situation of the company and has a plan: steer it to profits by aggressively focusing on servers (SeaMicro acquisition, Jim Keller hiring). Any gains there would trickle down to the consumer space. It would be a real shame if he let the ATI division flounder by missing its consumer release dates.

The up coming cuts could have been forseen from all the executive departures and the relatively worse PC environment. I think any talk about an acquisition of AMD would be complicated by the existing clause that its X86 license would be annulled if it were purchased by a third party. There would have to be a refocusing toward other segments or complex negotiations with Intel for a reworked agreement. A potential buyer could try to leverage past anti-trust allegations, but with iPhone and Droid bolstering ARM so much, this might not have the bite it used to.

Best of luck to AMD, it deserves better for what it has done than to just fade into History.

[AlexV]

Originally Posted by Raqia

Rory Read seems serious about addressing the fiscal situation of the company and has a plan: steer it to profits by aggressively focusing on servers (SeaMicro acquisition, Jim Keller hiring). Any gains there would trickle down to the consumer space. It would be a real shame if he let the ATI division flounder by missing its consumer release dates.

I would hope that AMD's CEO isn't autistical, because aggresively focusing on servers based on a fringe purchase and hiring a single man that seems to have gotten forumites panties up in a bunch (damn you Theo Valich and your "best engineer of all time" crap - even if he were, which is hardly the case or even endowed with relevance, life is not the movies and a single ninja does not defeat armies).

AMD has neither the brand nor the product stack to aggressively focus on servers, they're far less than noise in that space. They might as well focus on convincing consumers that they should team up and buy AMD products just because supporting the underdog is nice and helps the market in some nondescript way. The latter has a snowball's chance in hell to succeed, the former not at all.

There may be method to the madness though, and 3dilletante's remark about Read wanting to slim AMD into a fringe player appears quite valid. IMHO, at this point 3 scenarios seem relatively likely (all hinging on the cut reports being accurate to some degree):

1. Read really really really believes the whole handheld SOC hoopla is the path to greatness, and is working his butt off to emulate everything SOC people do - one of the things being outsourcing their SW stack development to somewhere in Asia, for great justice (a commonly held mis-belief is that leads to great cost savings)! That would render quite a few jobs redundant, especially in Canada where IIRC lots of the ATI software work is being done;
2. Read is cleaning up house for a purchase by some party that already has the units that are being made redundant, and which doesn't want to be in the same stupid position AMD was in after the ATI purchase, when they had lots of overlap that took a long time to clean;
3. Read is cleaning up house for a purchase by some party that, whilst not necessarily being active in tech or having overlapping units, feels that AMD's current cost structure is dumb, and that some modern techniques like aggressive outsourcing in (what is perceived as) non-key areas can lead to great cost efficiency - see first scenario for where this leads to.

I think that at this point it's fairly easy to accept that Project WIN is not about taking any particular market by storm, and some token hirings of publicly visible people is quite irrelevant.

[3dcgi]

A couple points. AMD already does a lot of software development in Asia though there's still a lot in North America as well. I'm in agreement with your mis-belief assessment.

Second point is there wasn't much personnel overlap between AMD and ATI. The overlap was in tools and methodologies which you can't clean up ahead of time unless there's an industry standard you can start using.

[ninelven]

Originally Posted by AlexV

Read is cleaning up house for a purchase by some party that already has the units that are being made redundant, and which doesn't want to be in the same stupid position AMD was in after the ATI purchase, when they had lots of overlap that took a long time to clean;

I still think MS buying them makes sense. Of course, MS has a way of doing things that don't make any sense, so I am probably way off base.

[caveman-jim]

Originally Posted by AlexV

AMD has neither the brand nor the product stack to aggressively focus on servers,

Do they need to aggressively focus on servers?

I think the strategy for AMD is co-existence, being part of the big picture, not trying to win a pissing match. They can't 'win' on Intel's terms, so they're redefining what game they're playing. This is can be known as 'giving up' or 'losing' but also 'changing the game' or 'playing by your own rules'. Filter by your own level of pessimism.

Almost as an aside, it means x86 CPU single thread performance and big discrete gaming GPU's are collateral damage. AMD are transforming into a different company, HSA platform is their outline and consumer devices / notebooks and cloud are the markets they're going after, while getting out of big desktop, workstation and small server.

Charlie has some more oped on semiaccurate this morning.

[3dilettante]

Even in the great big mobile cloud future everyone hopes to put their destiny and data on, there was going to need to be a need for servers to underpin the whole structure/house of cards.
As piss-poor as the value and licensing argument AMD's server offerings have been, they are still a much better fit for the pool of IP and experience AMD is resting on.
The virtualization, interconnect, I/O and memory support work best there, and prior to the very high density solutions Sea Micro convinced AMD to shell out money for, it was comparatively forgiving for the power penalty of using the overprovisioned x86 cores and AMD's inferior process and physical design. The servers still have a latency floor in many loads--for which some power needed to be expended; Intel helps advance software there--which provides coat-tails, and x86 is now established for large memory systems--which ARM-based systems cannot claim.
For comparison, AMD's latest for mobile is the warmed over and very warm AMD "tablet" chip that is based on an architecture that wound up targeting netbooks way past their sell-by date.

The big picture is that AMD's looking to back itself into markets populated by companies not-Intel, that are still better-run than AMD. Some are possibly less able to strongarm OEMs, but they aren't really more forgiving, especially not the ones like Samsung and Apple, which are examples of a likely extremely vertically integrated future for which AMD and its other competitors (Intel?) have no place.
None have an interest or need for AMD's continued existence to placate regulators.

AMD's "playing by its own rules" after a brutal record of failing its partners also confirms that it cannot be trusted to play by the rules it was using just prior, and any number of companies needing a supplier or a constant partner have been hurt significantly by AMD's execution, and then let down when it manages to execute on uncompelling products.

The latest SA editorial seems to point out that AMD's latest efforts appear to be re-brewing tea from its already used bags.

[AlexV]

Moreover, AMD is in no position to change any game, unless it wants to primarily play with itself. That only works as the only game in town until you grow up...

[DieH@rd]

S|A is predicting total self-destruction of AMD.
http://semiaccurate.com/2012/10/15/a...emiconductors/

[jimbo75]

It doesn't take a genius to figure that out tbh.

One would hope that if there is any truth in the SA article, it would be that the near term projects were abandoned, leaving the longer term ones in place. The industry has long been talking about skipping over a generation at 20nm in order to try to catch up with intel at 14nm, so if those are the projects that were cancelled then it might not be all bad.

Sadly, the sheer amount of high quality talent that has left AMD over the past 6 months makes me believe that they knew this was a stupid plan that was just going to kill the company.

[AlexV]

Originally Posted by jimbo75

IThe industry has long been talking

Yes, that's generally what the industry is doing, talking. They've been catching up with Intel for years now...only that the gap is constantly widening. Perhaps that instead of chasing unicorns, the industry should play to its strengths. Also, AMD is in no position to say: OK guys, we'll take a break and catch up with you later. Either they generate revenue or they're crunched, they have no fat to rely on and no creditor confidence to begin with.

[jimbo75]

Originally Posted by AlexV

Yes, that's generally what the industry is doing, talking. They've been catching up with Intel for years now...only that the gap is constantly widening. Perhaps that instead of chasing unicorns, the industry should play to its strengths. Also, AMD is in no position to say: OK guys, we'll take a break and catch up with you later. Either they generate revenue or they're crunched, they have no fat to rely on and no creditor confidence to begin with.

Well when you look at what they have upcoming on the cpu side it's not exactly hard to figure out why slowing it down or dropping some of it makes sense.

I mean, does anyone care about Piledriver? They might as well have dropped that completely instead of wasting money on what is destined to be a failure from birth. I doubt Steamroller will close the gap enough as well. Kabini should have been ready yes...Trinity is decent but it's not like it's a massive upgrade on Llano or anything.

Accepting you are hopelessly outgunned and cutting your losses isn't necessarily a bad thing - they really do need a new cpu core at 14nm to get back into this (not saying they'll have one, just saying it's what they need) - and until then they'll always be a distant second to intel. AMD's main problem is they can't seem to take advantage of where they are winning - the Kabini and graphics delays are just mind-boggling.

[Davros]

Has anyone posted this ?
http://www.bloomberg.com/news/2012-1...-340-jobs.html

[jimbo75]

On the cuts, it just shows what is wrong with the company anyway. They cut 10% of the workforce last year (1000 people or so) and they were back above the old number within months. AMD currently has more staff (11700?) than any time I can remember in recent history. Far, far too many staff for a company with these cash-flow issues.

So while 20-30% looks bad, it's not on top of last years 10%.

edit - I just checked and they had actually ballooned to 12000 staff in Q3 2011, which is just above the current figure of 11700.

[3dcgi]

Originally Posted by jimbo75

AMD's main problem is they can't seem to take advantage of where they are winning - the Kabini and graphics delays are just mind-boggling.

What graphics delays are you referring too? The entire SI stack launched this year.

[I.S.T.]

Originally Posted by Davros

Has anyone posted this ?
http://www.bloomberg.com/news/2012-1...-340-jobs.html

It's been talked about for the last few pages lol.

[AlexV]

Originally Posted by 3dcgi

What graphics delays are you referring too? The entire SI stack launched this year.

I believe that is in relation with the other SI, which can be perceived as delayed if it doesn't launch this year, based on what ones expectations are. TBF, Tahiti isn't in dire need of refreshing.

[jimbo75]

Yeah basically, although it's a hard one to figure out. With Fermi, Nvidia turned it around in 7 months from 480>580 while AMD was 13 months with 5870>6970. Now the talk is they will both be out in March so that's a similar kind of turnaround again. I dunno if that is a delay for AMD because of that, but it sure feels like a delay when Nvidia can basically catch up so many lost months in one refresh. You have to really fear for AMD's graphics division if they lose the TTM and die size advantages they've had.

Why Kabini is taking so long is harder to figure out. I assume this is the same team that got Bobcat out at A0 silicon so why it's taking 30+ months for the replacement to happen is just plain weird. It can't be the process as that's proven and 1 year old. The chip itself doesn't appear to be a massively different undertaking even with the added southbridge. For AMD to purposefully delay it is criminal neglect, I just can't believe any company would act so ineptly.

[Dave Baumann]

480>580 was a silicon spin. 5870>6790 was an architectural change (with a shift of process from its original designed process).

[UniversalTruth]

Posted, no?! I'm taking popcorn and bacon, it's gonna be interesting season.

AMD to reduce previous-generation APU prices

http://www.digitimes.com/news/a20121015PD217.html

After announcing its latest FM2-based A series APUs (Trinity) on October 1, AMD is set to reduce prices of its previous-generation APUs with the FM1-based A4-3300 to be reduced from US$46 to US$30, according to sources from motherboard players. AMD Taiwan declined to comment about its price strategy.

In addition, AMD will also reduce its A4-3400 from US$48 to US$35. As for its newly launched FM2-based A4-5300, AMD plans to reduce the CPU's price from US$53 currently to US$30 a quarter later to strength its competitiveness against the Intel Pentium series, the sources noted.


Now, please, the videocards too.

[3dcgi]

Originally Posted by AlexV

I believe that is in relation with the other SI, which can be perceived as delayed if it doesn't launch this year, based on what ones expectations are. TBF, Tahiti isn't in dire need of refreshing.

I assume by the other SI you mean Sea Islands (CI). It's easier if people use a shorthand other than SI.

[Silent_Buddha]

Originally Posted by jimbo75

Yeah basically, although it's a hard one to figure out. With Fermi, Nvidia turned it around in 7 months from 480>580 while AMD was 13 months with 5870>6970.

And how long between 580 and 680? Versus 6970 and 7970? Or between 280 and 480 and between 4870 and 5870?

If you're going to cherry pick some numbers, at least show the other side as well.

Regards,
SB