Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Advanced Micro Devices - Moderated (AMD) -- Ignore unavailable to you. Want to Upgrade?

To: Elmer Phud who wrote (118360)	4/5/2004 1:06:38 PM
From: Dan3	Respond to of 275872

Re: don't forget to include the 100s of millions of chipset components If you're counting on 100s of millions of chipsets for last quarter, you're in for a disappointment. And given the tiny size of most of the other parts you listed, and the fact they aren't produced using a 90nm process, what you posted makes no sense - it's as though you claimed that GM may not have produced many cars, but the factories were busy because they shipped a lot of OnStar systems to other companies. The facilities the make cars aren't the facilities that make OnStar components, and the OnStar components are much smaller, anyway.

To: Elmer Phud who wrote (118360)	4/5/2004 4:42:49 PM
From: pgerassi	Read Replies (1) \| Respond to of 275872

Elmerp: Re: "don't forget to include the 100s of millions of chipset components, XScale, uControllers, ethernet controllers, network processors, cell phone, wireless and other comms and of course Flash" None of these are using 90nm. So what would a 90nm 300mm fab be doing producing these? Especially since Intel has many other fabs that do. CPU division has the bulk of the revenue and more than 100% of the profits. $5 billion divided by 38 million CPUs is $132 each. That allows for $2 billion for everything else. I think that Intel needs more like $6 billion from their CPUs to break even and that gets us to about $158 especially when those 300mm fabs begin to depreciate. If SEC forces the expensing of options, add $25 to these ASPs. AMD had $50 million of profit at $566 million so that gets us to about $500 million of revenue generated by 7.2 million CPUs or about $70 ASP. Options expensing will add about $10 in ASP, if only given to the CPU division to make up and $5, if split between flash and CPUs. As to SPEC, where are the benchmarks using normal production compilers with normal options like those that are used to make production software. SPEC doesn't allow for BLAS and other specific optimized libraries to be used unless they are folded into the compiler which is contradictory to how software is developed in these areas. Also SPEC does not allow for any algorithm to be used that performs the task or check that the results are accurate (the latter allows for reduced precision to speed things up or even to produce garbage which can be quickly accomplished). According to SPEC, 3.4GHz P4 XE is supposedly faster at floating point, yet SciMark has it well behind Athlon 64. And when x87 is used (both for its better precision and for its widespread use), the P4 is atrocious. On real database loads, Xeon performs about 88% as fast as Opteron, anandtech.com , and Xeon MP is even worse, 83% (4 way). And these systems did not use PC3200 for the Opteron 848. Other benchmarks like linuxhardware.org show that using 64 bit mode, Opteron can go over 50% faster than 32 bit Xeon. In the mySQL benchmarks, aceshardware.com , Athlon 64 FX53 (Opteron 150) goes 50% faster than 3.4GHz P4 XE and probably will be 80% faster than 3.0GHz 2MB Xeon Nacona. Many have noticed that good SPEC scores do not seem to translate to good application performance in the real world. Now you can see that it is true. Pete