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: graphicsguru who wrote (241458)	9/29/2007 1:31:40 PM
From: chipguy	Read Replies (2) \| Respond to of 275872

Exactly what will be Itanium's market once MP x86 CSI parts come into the world? Itanium's ability to work in large SMP configurations will then be shared with x86. I think you are quite mistaken in thinking that after CSI is implemented in both Xeon and IPF processors that they will be usable interchangeable and otherwise identical except for the instruction set, i.e. pop one in a socket or the other. IPF and Xeon processors will be still be designed with distinct trade-offs between RAS and scalability vs cost and ease of implementation. Even CSI itself will have feature subsets and form factors that would differ for different sizes and classes of machines. Even with an identical CSI pinout an IPF processor designed to scale to 512 socket systems will dedicate much more mm2 of silicon to internal functionality for routing, buffering, RAS, and ability to track virtual channels compared to Xeon processors designed for low cost and optimized for four socket systems. Someone asked the chief Intel architect at the June IPF roadmap presentation/press conference whether there will be common Xeon and IPF motherboards based on CSI. He replied that is possible to design motherboards that could accept both Xeon and IPF processors but OEMs were unlikely to do so. If the board implemented all performance, RAS, and power delivery capabilities to use IPF processors to their full advantage then it would be too expensive and overkill to use with Xeons compared to a mainstream Xeon board. If it was cost oriented and targeted Xeon CSI mainstream capability then plugging IPF chips into it would deliver much lower performance and functionality than a full featured IPF motherboard. IMO this is similar to the dynamics in play when attempts were made to create common Alpha and K7 mother boards in the late 1990s. It was a failure despite the two processor families nominally sharing the same interface. DEC's high performance Tsunami chipset was overkill and too expensive for K7 while AMD and third party K7 chipsets left way too much Alpha performance on the table. The purpose of CSI isn't to unify IPF and Xeon systems and motherboards. It is to share silicon IP and chipset development costs between IPF and Xeon within Intel and OEMs. IPF processors will still ship mainly in systems optimized for performance and scalability while Xeon processors will still ship mainly in systems optimized for low cost and power.