Replies to post #9498 on Intel Corp (INTC)

[Mysef]

SemiconEng,
re:fwiw, It is Certainly more cost efficient, to make 90nm devices on 300mm wafers, then 130nm devices on 200mm wafers.

Guess it depends how you define efficient. If your goal is to gain marketshare from AMD with Celerons in $399 DELLs, knock yourself out.
Mysef


http://www.siliconinvestor.com/stocktalk/msg.gsp?msgid=19844633

HP recommending Opteron over Xeon?

From the HP DL585 Q&A:
http://h18004.www1.hp.com/products/servers/proliantdl585/que...

Q1. What types of applications see the most performance benefit from Opteron technology?
A1. Applications that require high bandwidth and low latency access to system memory benefit from Opteron technology, especially true for high performance cluster computing but also applicable to a large number of commercial applications as well.

2-way and 4-way platforms based upon Opteron technology provide more than the system memory bandwidth at lower latencies than current Xeon and Xeon MP based platforms. Also, Opteron technology allows memory and I/O resources to scale with additional processing power providing increased levels of system resources as computing needs increase. Combining resource scaling with Opteron's Hyper Transport technology allows for those system resources to be shared effectively across all processing units benefiting many multi-threaded applications and provides for high levels of scalability from UP to DP to 4-way processing.

Additionally, customers who need to run existing 32-bit applications along with 64-bit applications can benefit from the simultaneous 32-bit and 64-bit computing capabilities of Opteron technology. In those environments, 32-bit application performance is not compromised. Also, applications that need large memory addressing such as those with large datasets (financial and scientific modeling applications) or must manage a large number of concurrent users or application threads such as large databases and data warehouse applications for solutions in customer relationship management (CRM), supply chain management (SCM), enterprise resource planning (ERP), and digital rights management (DRM) systems can benefit from Opteron's low latency memory architectures.

Seems like HP is touting the Opteron as a superior server architecture vs. Xeon to me.

Gary.

[Dan3]

Re: Therefore, the introduction of 2.40GHz processor with Prescott core may be an indicator of 90nm massive ramp.

2.4ghz P4's were EOL'd some months ago. That they are rising from the dead is probably due to binsplit issues, not a sudden desire on Intel's part to get less money for its chips.
http://www.theinquirer.net/?article=12040

[Elmer Phud]

Semi -

In case we suppose that there are yield issues with the Pentium 4 CPUs at 90nm technology process, then, the appearance of 2.40GHz model is quite logical, as Intel now can sell chips that do not fit into the original frequency envelope.

In other words, 90nm Celerons. Despite his misunderstanding of what yield and binsplit means he is quite insightful.

fwiw, It is Certainly more cost efficient, to make 90nm devices on 300mm wafers, then 130nm devices on 200mm wafers.

It sure is. That's obvious to you but you know the business.

Here's a little teaser for those here. Simple calculations show that if Intel has good yields (you be the judge) they get upwards of 400 good Prescott die per wafer. Eventually those 300mm fabs will be capable of between 5 & 10K wafer starts per week. Do the math. Some people claim Intel's yields are in a class by themselves. The output could be even higher. As for me, I can't comment... Clearly Intel has more capacity than they can fill with processors alone, assuming decent yields, and a 4th 300mm fab, F12, under conversion. What is the plan here? I might also add that there is another area of constraint but we won't get into that right now...