The AMD Execution Thread [2007 - 2017]

[Geo]

So my question is: how did they do it? Or is it just financial smoke and mirrors?

For those who lean towards the cynical, it's impossible to say with 100% certainty so long as a sizeable "one time charge" like that is on the report. That's why "GAAP" exists, after all, to minimize leeway for hanky panky. (Which, I should add, I am not accusing anyone at AMD of engaging in such tactics).

But assuming we've now seen the last of the one time charges for awhile, 1Q should confirm these numbers.

 

[jimmyjames123]

It's actually damn good as the non-GAAP operating loss was only $9M... I'm waiting on the detailed PDF for some info including cash position etc. before I news this though (unless somebody else on staff beats me to it due to timezones!)

Arun,

It looks like the non-GAAP operating loss is only $9 million mainly due to the fact that there is $1.669 billion in ATI acquisition-related and severance charges, of which $1.608 billion is due to ATI impairment of goodwill and acquired tangible assets.

Do you know what they mean by "ATI impairment of goodwill"???

 

[nutball]

So, Barcelona *is* actually shipping in very decent volume.

Shipping to whom? Where have they gone? Who is buying them?

Maybe I'm talking to the wrong three-letter companies, but to me they're saying "we're not shipping AMD quads until *much* later". Maybe they're shipping to four letter companies?

 

[dizietsma]

So in effect, neglecting yet another one off ATI charge, they lost $170m which is not much better than my $200m loss forecast above.

Now that it's the subdued Q1 period I think they will stay at $200m loss for this period and then go out to a $300m loss in Q2 as Intel 45nm takes hold and Intel starts cutting the price. nvidia have new products out to further apply pressure.

How much do all the charges related to the ATI buy add up to now by the way? Is it $5b yet?

 

[MTd2]

Ok folks, now back to the reality (bad news):

However, he said that the company is targeting a H2 ramp of the 45 nm generation, which indicates that AMD will miss the originally promised mid-2008 launch of the processors. This statement also contradicts Meyer’s statement from the Q3 2007 earnings call that AMD would begin ramping 45 nm CPUs in H1 2008.

http://www.tgdaily.com/content/view/35672/135/

 

[AlNom]

Shipping to whom? Where have they gone? Who is buying them?

Maybe I'm talking to the wrong three-letter companies, but to me they're saying "we're not shipping AMD quads until *much* later". Maybe they're shipping to four letter companies?

Look no further than two-letters? HP has a number of Phenom systems being offered.

 

[ZioniX]

Do you know what they mean by "ATI impairment of goodwill"???

It means that AMD is acknowledging that it has paid way too much for ATI and is now writing off its goodwill to reflect that.

Here's what AMD said about it, some time ago:

“The acquisition took place at the moment, when ATI was not really leading in terms of technology… It is not like we acquired ATI and we lost market share. It was just a consequence of [ATI’s execution]: Nvidia had better graphics than ATI back then and that is why AMD lost market share. We also should consider [hardware] cycles of OEMs: if you are missing their cycles, you are out for a while. You have to be [ready] with the right [product] part at the right moment to get a cycle. When you are in, you are going to stay in for a long time. So, the main reason behind the share loss is missing OEM cycles. But we are regaining them now,” recently said Vincenzo Pistillo, director of consumer business development in EMEA region for AMD.

“This conclusion was reached based on the results of an updated long-term financial outlook for the businesses of the former ATI Technologies as part of AMD’s strategic planning cycle conducted annually during the company’s fourth quarter and based on the preliminary findings of the company’s annual goodwill impairment testing that commenced in the beginning of October 2007,” a statement concerning the write-down reads.

From X-bit labs

 

[aca]

This may break a few hearts...

http://www.semiconductor.net/article/CA6512566.html?industryid=47298&nid=3572

ZRAM is dead, long live ZRAM II.
The charge carrying capacity for the technology AMD wanted to use for its caches was apparently insufficient and scaled poorly.

Second-gen ZRAM is apparently a reprise of the bipolar transistor, or will be if it comes to fruition.
The description leaves me kind of skeptical, as it seems timing and a number of other factors need to be juggled. It sounds like a fragile solution.

It's also using a bipolar transistor, which has functionality that requires current flow at all times, unlike CMOS that only has flow during transitions.
On the other hand, at small geometries and high frequencies, everything leaks anyway. ZRAM would have to be refreshed, so power consumption might be a wash.

This does put off AMD's using ZRAM for a while, if ever.

The idea of having charge stored on the floating body is still a solid idea, I reckon. Using the lateral bipolar instead of mos makes sense as the injection mechanism from bip is much better defined than MOS characteristics especially on smaller scales, imho.

 

[vertex_shader]

http://biz.yahoo.com/bw/080117/20080117006120.html?.v=1

Graphics

Graphics segment revenue was $259 million, a three percent sequential increase. The growth was due to demand for AMD’s new ATI Radeon HD™ 3800 series and continued adoption of the ATI Radeon HD 2000 series of graphics processors.

Not much.

 

[MTd2]

AMD is getting really, really serious in building that 3.2 billion fab in Albany, NY

http://www.bizjournals.com/albany/stories/2008/01/14/daily47.html

 

[ShaidarHaran]

I must admit I'm glad to have been wrong. Good job AMD. Now keep this ship headed in the right direction and you might just last long enough to be competitive again

 

[Mat3]

The second-generation ZRAM product has only been recently disclosed, and no working samples have been demonstrated.
No characteristics were given and no timelines are involved.

ZRAM gen2 was announced over a year ago...

http://www.us.design-reuse.com/news/14802/z-ram-gen2-ultra-dense-memory-technology-isi-significantly-improves-speed-power.html

 

[Sound_Card]

What are the possibilities of IBM buying AMD? Unless I'm missing something, it makes the most sence if anyone is to buy them out.

-They already work very close and depend on each other for new processes
-x86 licence would not be nullified considering IBM already has one
-hold their foot in the console market and expand
-Combine expertise on:

• cell
• fusion
• K10

-they have just under 14 billion in cash, no debt

I think IBM is just waitng for the right time. I don't think they want to buy a 3 billion dollar company and pay additional 5 billion to cover their debt. AMD would need to restucture a bit, perhaps selling dresden.

 

[3dilettante]

ZRAM gen2 was announced over a year ago...

http://www.us.design-reuse.com/news/14802/z-ram-gen2-ultra-dense-memory-technology-isi-significantly-improves-speed-power.html

I guess that's a good thing.

It just leaves the lack of details for over a year and no disclosed working samples.
With the time frame given, AMD would have been working on gen1 technology in 2006.

It seems iffy that AMD could shift so quickly from Gen1, which the article said it had licensed by that point, to Gen2 and fit it within a few years when chip design takes even longer.

 

[AlphaWolf]

I think IBM is just waitng for the right time. I don't think they want to buy a 3 billion dollar company and pay additional 5 billion to cover their debt. AMD would need to restucture a bit, perhaps selling dresden.

If a company like IBM showed interest in buying AMD, they'd never be able to do so for so little. You can't just go by stock values when you start talking acquisition.

 

[_xxx_]

I don't think they'll sell Dresden with the subventions they get there.

 

[Arun]

3dilettante, I still don't understand why you claim Z-RAM Gen1 is effectively unusable. This problem has been known for many years and I would in fact be rather surprised if AMD didn't know about it when they got their original technology license.

It is a trade-off, and arguably a problematic one that makes the technology less attractive for certain purposes. Same for Gen2, which might also not be ideal for everything as you pointed out given how it works, but I'm skeptical that's really a big problem. Even if it was, it's not intended to be capable of replacing anything and everything. It's called Z-RAM, not Magic RAM...

 

[3dilettante]

3dilettante, I still don't understand why you claim Z-RAM Gen1 is effectively unusable. This problem has been known for many years and I would in fact be rather surprised if AMD didn't know about it when they got their original technology license.

It's more of a "where's the beef" thing for me.

AMD wasn't planning on Z-RAM gen1 for the 65nm node, but the planned 45nm products don't seem to show any outsized cache capacities that would scream Z-RAM to me.
If it is there, the benefits seem pretty modest.

Past 45nm, it seems the buzz is more on gen2, and the time frame seems to be relaxed enough to allow AMD to skip the first gen.

It is a trade-off, and arguably a problematic one that makes the technology less attractive for certain purposes. Same for Gen2, which might also not be ideal for everything as you pointed out given how it works, but I'm skeptical that's really a big problem. Even if it was, it's not intended to be capable of replacing anything and everything. It's called Z-RAM, not Magic RAM...

I understand that, and I've believed for some time that it could be compelling in a last-level large cache.
The problem I see is that for all its density advantages, little has been said on what it takes to actually manufacture a serious product on it, particularly a high-performance CPU on AMD's process. It doesn't help that AMD hasn't exactly impressed on either the CPU or process front even before we add Z-RAM to the mix.

Maybe a GPU could benefit from it. It would clock lower, and it would be on the same processes the sample Z-RAMs have been fabbed on.

 

[MTd2]

Heh, today AMD got a 10.7% stock raise due to a take over by IBM rumor.

 

[Sound_Card]

Heh, today AMD got a 10.7% stock raise due to a take over by IBM rumor.

A IBM take over makes sence though. Got to give it some credit.