Saturday, April 07, 2007

AMD's klling game

As I estimated here, AMD is capable of producing 37 million dual core CPUs per quarter. This might explain why AMD is selling dual core CPUs at $65.

The drastic price drop of Athlon X2 CPUs indicates that there will be soon a dramatic performance hike on AMD CPUs. Intel's BK may be sooner than I think.

Behold.

16 Comments:

Blogger enumae said...

But it doesn't explain why they were only at 30% dualcore at the and of Q4 2006.

7:44 PM, April 07, 2007  
Blogger lex said...

Pretender math.. has some serious flawed assumptions

1) Wafer starts / month are not at stated max

2) Die size isn't square and thus not close to 492 die per wafer

3) Yields for AMD are not close at all to 90%.. Dude where are you getting your info.. LOL

End result is far fewer die and far higher cost and no profit for AMD.

AMD going BK in 2008..

Sharikou PHD what is your prognosis on INTEL BK and what happenened in 2007 and why won't INTEL go BK in 2008...

10:18 PM, April 07, 2007  
Blogger Randy Allen said...

We all know how Sharikou's previous "predictions" have gone. Intel was supposed to suffer losses in Q2'06 and onwards. AMD was also supposed to exit 2006 with 40% market share.

This is the post in question:

ntel may suffer losses in 2Q06, 3Q06, 4Q06, 1Q07, 2Q07, 3Q07

For Intel, the common scene for the next 12 months will be impariment of goodwill (abandonment of acquired technologies and businesses), writing off inventory, writing off assets (old FABs, etc), one time charge for layoffs, price crash, unit share loss, etc, etc. Intel's stock holder equity will vaporize faster than that of Enron. GAAP loss is a certainty, but I also see operating loss from 3Q06 onward. I previously projected mass layoffs in 2007, but since Intel failed to execute a survival strategy, I projected a faster collapse in Feb 2006.

2Q06: 40% price drop, 20% volume drop, impairment on goodwill, one time charges for massive 16,000 layoff. GAAP loss in the bag. Expect Intel to warn soon. Currently, Wall Street expects Intel to earn $0.15 per share. Expect ($0.05) loss at minimum.

3Q06: Conroe ramps to 10%. 20% additional price drop of P4, mobile CPU price crash, 10% additional volume drop. Operating loss expected. Plus more impairment on goodwill, plus one time charge on inventory write off.

4Q06: Conroe ramps to 20%. Merom ramps to 10%. Further reduction of P4 and Core Duo price. Revenue flat from 3Q06 in a seasonally up quarter. Operating loss. AMD exit 2006 with 40% market share (run rate).

1Q07: Conroe ramps to 35% of desktop units. Revenue down more than seasonal as AMD's 65nm parts flood the market.

2Q07: Intel server market share drops to 40% as AMD ramps Rev H quadcore. Bulldozer hits hard.

3Q07: AMD FAB36 ramps to 20,000wspm at 65nm. FAB38 ramps. Chartered FAB7 ramps. AMD grabs >50% of market share.

4Q07: Go figure.

4:50 AM, April 08, 2007  
Blogger sharikouisallwaysright said...

While i do forsee that AMD will strike intel back there is no BK from either corp to expect.

Intel may lay off workers or close some facilities.

AMD will survive 2007 and laugh about intels shrinking process...but nothing more will happen.

Ok, doomsday could change everything ;)

6:00 AM, April 08, 2007  
Blogger lex said...

INTEL isn't shutting anything..

THey are adding more and more and more capacity.

I expect the China fab will go like gain busters.. Damm 52,000 wafers per month will crush ATI/AMD. That will be a lot of graphics and chipsets.

Now you got ANOTHER 300mm factory coming..

Eat your hard out PhD...

If you aren't making money and how many quarters has AMD ever made money? You are on the road to bankruptcy...

http://www.theinquirer.net/default.aspx?article=38782

7:37 AM, April 08, 2007  
Blogger Dahaka said...

Intel can't victory over barcelona with these old fsb architechture.
Intel needs integrated memory controller ,higher bus speed and advanced core.AMD will bring end of Intel.Now wait for barcelona and imagine what barcelona can over Intel.

1:19 PM, April 08, 2007  
Blogger yomamafor2 said...

the reason why AMD is dropping prices on their dual core processors is because AMD simply cannot charge more for an inferior processor. AMD can only stay competitive with extremely low prices, or they'll lose more market share.

look at AMD's financial status now. this price war has basically drained their money supply. If nothing is done, the upcoming Sept. price war from Intel will seriously cripple AMD. Let's just hope that Agena and Kuma will be the cavalry to save AMD before AMD crashes.

the title of this thread should be, "AMD plays bloody defensive against Intel's killing game."

3:08 PM, April 08, 2007  
Blogger yomamafor2 said...

as for the FSB vs. HyperTransport, FSB may in some instances bottleneck the CPU. However, that's exactly the reason why Intel generally has larger L2 cache than AMD: to hide the bottleneck.

http://www.hardwareanalysis.com/action/
printarticle/1833/

The fact that there’s less traffic on the FSB helps keep the overhead low, hence Intel hasn’t opted to bump the FSB speed up any further, it’ll remain at 1066Mhz for now. Intel will do this on the processors based on the same architecture that are dubbed Woodcrest which are meant for server use, but these have different requirements all together and might well run at near maximum bus capacity most of the time. Although the new Core 2 architecture is a data hungry one due to the short pipeline and its ability to process four instructions per clock cycle Intel’s shared L2-cache alleviates much, if not all, of the bottleneck created there.

One of the disadvantage of using FSB is the lack of scalability. Intel simply cannot compete with AMD on 4-way system or above. However, normal home users (us), or anything below 2-way servers, FSB clearly does not present a significant problem.

http://www.anandtech.com/
printarticle.aspx?i=2772

The Dual Independent Bus (DIB) will not make much difference for Woodcrest and Dempsey as only some HPC applications are really limited by the FSB bandwidth. Three years of benchmarking tell us that most server and workstation application are not bottlenecked by the modern FSB speeds. The Opteron platform does not scale so much better thanks to NUMA in dual and quad core configurations. No, in most applications, the low latency integrated memory controller makes the difference, not FSB/NUMA bandwidth. Of course, with Clovertown, or two Woodcrests on one chip, a shared FSB might become a bottleneck, and in that case a DIB is a good idea.

not only that, while AMD's touts its direct connect HyperTransport, its bandwidth is only 6.4Gb/s, compared to DIB 8.5Gb/s @ FSB 1066, or 10.5Gb/s @ FSB 1333.

you can take a look at the benchmarks at the end of the site. While Quad Opteron takes the lead, there is really no competition for dual core Woodcrest @ 3.0. Thus, it all comes down to processor architecture.

now we'll just have to wait for Barcelona to see if AMD can take back the lead.

3:30 PM, April 08, 2007  
Blogger Azmount Aryl said...

lex said...
I expect the China fab will go like gain busters.. Damm 52,000 wafers per month will crush ATI/AMD. That will be a lot of graphics and chipsets.


Thats 90nm fab and it will be online by 2010, at that time even chipset for intel systems will be made on 65nm.

Here what TSMC chairman Morris Chang said:
"For the impact on TSMC, the direct impact won't be big," TSMC chairman Morris Chang told a media gathering. It would take at least two to three years (for the plant) to start producing, he said.LINK

That plant would compete with what TSMC produces, however, intel's plant will be making 90nm in 2010 while TSMC has better technology today (and i quote):
Taiwan Semiconductor Manufacturing Co. Ltd. (TSMC) has been taking a front seat in driving process technology to 90-, 65- and 45-nanometre.LINK


lex said...
Eat your hard out PhD...


Are you in 8th grade Lex? You know, intel-fans aren't all morons, there are plenty of well educated people among them. Many of them wont in fact be happy to identify themselves with 8th graded poppies making harsh, idiotic comments.

9:32 PM, April 08, 2007  
Blogger Azmount Aryl said...

Braking news:

Here i found one more of those "Crazy Hector" deals on the Egg:
Buy Athlon 6000+ and get a free ASUS MoBo

9:44 PM, April 08, 2007  
Blogger Giant said...

Wow. AMD must really be desperate to sell some CPUs now!

Meanwhile, Intel has raised the bar again: http://techreport.com/reviews/2007q2/core2-qx6800/index.x?pg=1

Quad Core 2.93Ghz while consuming less power than AMD 6000+. Now that is impressive.

10:29 PM, April 08, 2007  
Blogger Dahaka said...

"not only that, while AMD's touts its direct connect HyperTransport, its bandwidth is only 6.4Gb/s, compared to DIB 8.5Gb/s @ FSB 1066, or 10.5Gb/s @ FSB 1333."

You are wrong.

Hypertransport is not 6.4Gb/s.
It is 8.0Gb/s.It is 2000 mhz and 16 bit wide.1000 up 1000 down speed totally 2000 mhz bidirectional bus.

FSB is a huge bottleneck for intels processors.Because they haven't got memory controller.If they integrate a memory controller in the processor FSB won't be bottleneck for intel.

1:26 AM, April 09, 2007  
Blogger Randy Allen said...

This comment has been removed by the author.

4:18 AM, April 09, 2007  
Blogger Evil said...

So why did AMD lower their numbers for this Q?

AMD is in some serious trouble, not Intel. Intel's product line keeps getting stronger and stronger and pushing the performance envelope, much like AMD did serveral years ago.

AMD is once again losing market share in both the desktop and server market to Intel.

8:27 AM, April 09, 2007  
Blogger Ho Ho said...

Anyone remember how Sharikou was saying some time ago how Intel kills its Netburst sales by not having enough C2Ds on the market? IIRC he said that people see, that C2D is vastly superior to Netburst and don't want the older one any more. Since there is not enough C2Ds they take AMD CPUs. Isn't it pretty much the same for AMD this year?

12:27 AM, April 10, 2007  
Blogger Giant said...

Because, Sharikou said, that AMD will very rapidly transition to K10. Hector Ruiz stated that K10 won't make up a large portion of the CPUs they sell until next year though. (He said this in the "K8L frag starts in April" post)

Speaking of which, it's April now, and it doesn't seem like K10 is ready yet.

1:40 AM, April 10, 2007  

Post a Comment

Links to this post:

Create a Link

<< Home