Is AMD Really a Sinking Ship?

[Nedjo]

AMD Phenom won't hit 2.6GHz until 2008

Written by Lars-Göran Nilsson
Thursday, 08 November 2007 09:42

65nm production issues

According to Digitimes, AMD is having some problems hitting high clock speeds of the upcoming Phenom processors and although this isn't big news in itself, it seems like we won't see a 2.6GHz part until next year.

There will aparently only be two launch parts as well, the 2.2GHz 9500 and the 2.3GHz 9600, with the 2.4GHz 9700 arriving some time in December.

AMD need to step up and sort our these production issues as soon as possible, if the leaked benchmark figures we've seen is anything to go by, Intel doesn't exactly have a lot to worry about.

Hopefully AMD is just having some teething problem with its 65nm production process, but we really hoped to see more competitive parts from AMD at launch.

So far DigiTimes has proved to be one of the most un-reliable news sites (we all remember "native Yorkfield. So why to believe that this "news" is accurate?

[Epicenter]

You can't compare K6 period and actual period because new Fab, new technology costs explode. I don't hope amd go bankrupt but I'm not sure you could refer to the past to say they will not go. If they have not enought money they can't invest enought for the future

People seem to assume no one is buying AMD hardware because they're not on top. Such is not the case. Your average consumer thinks with his wallet first and benchmark numbers second, if he's even savvy enough to look at those rather than following whatever he read in a magazine or saw in an ad. Having a less competitive project doesn't mean you go out of business, otherwise every industry in the world would have only one player.

That said I think AMD's acquisition of ATi was probably not the best timed, and ATi's poor position competing with nVidia is a part of that. I don't see AMD's business plan being in the enthusiast sector, I see it being in servers and 'least common denominator' PCs.

Think about it-- AMD stands to create a line of ICs that essentially are a desktop version of what's so popular in mobile devices, a System-on-a-Chip. A multi-core CPU, a modest GPU, and a northbridge all in one; a southbridge could easily be integrated too. You're looking, then, at AMD producing a chip that composes the vast majority of the components in your average PC. Sold at low cost to OEMs/System builders, they could really stand to clean up. That sort of innovation may very well lead to tiny ITX-style machines (well, look at DTX!) at extremely low prices, low power consumption (look at Geode) .... it's not hard to see where they're going with this. That's potential market share not to be scoffed at, even if it doesn't win benchmark comparisons against Intel.

Guess who sells the most GPUs? .. It's not nVidia or ATi/AMD .. it's Intel and their onboard video. So imagine the kind of market share AMD could have if they are selling boards with an onboard AMD-made CPU, chipset, GPU, northbridge, and a handful or two of small parts (capacitors, resistors, etc.) and all the OEM needs to supply is a case, PSU, HDD and optical drive? I think you see what I'm getting at here.

[nemrod]

People seem to assume no one is buying AMD hardware because they're not on top.

I don't assume nobody buy AMD hardware. I just see than since a lot of quater they lose several hundred of millions dollars each quater.
what I see is that to keep sale, they have to sale at a much too low price. As consumer we enjoy of low price, but this will perhaps what could finally kill AMD.

but you're right on that point:

Guess who sells the most GPUs? .. It's not nVidia or ATi/AMD .. it's Intel and their onboard video. So imagine the kind of market share AMD could have if they are selling boards with an onboard AMD-made CPU, chipset, GPU, northbridge, and a handful or two of small parts (capacitors, resistors, etc.) and all the OEM needs to supply is a case, PSU, HDD and optical drive? I think you see what I'm getting at here.

if they could sale with enought margins.

[nemrod]

So far DigiTimes has proved to be one of the most un-reliable news sites (we all remember "native Yorkfield. So why to believe that this "news" is accurate?

September 2006:

Intel plans to launch its second-generation quad-core processors - the Yorkfield series–as replacement parts for the upcoming Core 2 Quad (codenamed Kentsfield) CPUs in the third quarter of 2007, motherboard makers familiar with Intel's latest roadmap revealed. Intel is on track for a roll-out of its Kentsfield series on November 16, according to the makers, adding that the first-generation quad-core lineup may be considered a transitional product segment due to insufficient design capability.

Since Kentsfield appears to have two separate L2 caches, shared separately by each pair of processors, bus bandwidth consequently increases, the makers said. In contrast, Yorkfield will have one L2 cache shared directly by each pair of chips, enabling more efficient quad-core operation, with less FSB bandwidth, the makers noted.

Intel's Yorkfield will be paired with the next-generation Bearlake chipset family, which will support a 1333MHz FSB and a PCI Express 2.0 interface, according to the makers. The first Yorkfield-based systems, which will utilize the Bearlake X chipset and DDR3-1333 memory, will target the high-end gaming market, said the makers.

Intel's Yorkfield will be manufactured on 45-nanometer (45nm) process technology, the makers indicated.

It looks not so bad. The only really wrong information is
"Yorkfield will have one L2 cache shared directly by each pair of chips, enabling more efficient quad-core operation, with less FSB bandwidth, the makers noted."

Do you want to compare with informations we had in september 2006 about the barcelona? Even the name was wrong.. K8L...

You think unbelievable that there will not be 2.6GHz phenom before Q1 2008. I 'd like to have your story for this:
http://www.spec.org/cpu2006/results/...903-01948.html

[Nedjo]

You think unbelievable that there will not be 2.6GHz phenom before Q1 2008. I 'd like to have your story for this:
http://www.spec.org/cpu2006/results/...903-01948.html

I guess that only IBM knows why did SPEC done that.

[Hornet331]

I guess that only IBM knows why did SPEC done that.

i guess your wrong.

SPEC has determined that this result was not in compliance with the
SPEC CPU2006 run and reporting rules. Specifically, the submitter
reported that the result would not meet the 3 month availability
requirement in the SPEC CPU2006 run rules due to a change in the
availability date of the system.

ibm just cant deliver the system cause there are not enough cpus.

[nemrod]

I guess that only IBM knows why did SPEC done that.

Apparently that was IBM who ask, not spec. But spec rule, ask that vendors be able to ship their system within 90 days of issuing a fresh test score.
Why can't IBM provide those barcelona servers to customer in the 3 months after the 10 september, I don't know exactly but I assume we should not laugh too high when we have information like no 2.6 GHz phenom before Q1 2008.

[Ace123]

Good find, and good point.

Theoretically, if AMD produces a product thats far cheaper to produce, then they could win a price war if they could get their chips prices a little more competetivley than Intels.

Now... this doesnt take into account that AMD owes alot of money.

If GM can make comparable cars to fords, and produce them for cheaper, it would make since that GM would be able to price them out of the market.

Gm has been successful with this for some time, regardless of the fact that everyone has been spouting the "GM is a sinking ship" song for years, decades even.