SI
SI
discoversearch

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)
AMD 252.09-0.3%3:59 PM EST

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: Joe NYC who wrote (231471)4/25/2007 2:04:31 PM
From: nuclearRead Replies (1) of 275872
 
Joe,

Volume of DC is certainly much higher than QC. A perfect QC yield is 1/4 that of DC, so pretty disastrous for cost.

For non-fully-functional QC to be disabled, it need to be designed that way. The non-functional core must be isolated.
I have not look at the architecture in any detail, so I don't know for certain.

With a large die, it is something that engineering would consider when they architect it. Redundancies have been attempted in the past to improve yields and often fail to work as expected or the redundant logic affected speed. The DRAM tried it. (I have read papers on it. Not sure whether it was commercialized) ATI tried it successfuly with R300 and Nvidia adopted this technique with NV40. With this, die sizes have gone up R600 and G80.

ATI had horrible yields with R420. There was a severe shortage of X800XT for a few months with only X800 Pro available. Far more X800 Pro(12 pipes) were sold than X800XT (16 pipes). Nvidia also sell more 8800GT (disabled) than 8800GTS. In the early X800 Pro, it is truly a non fully functinal die but in later X850 (a respin of R420 called R423), a fair number have successfully flash their video BIOS to enable all quads making it behave like a X850XT.

In GPU, Video BIOS contain info on which quad is disabled. For CPU, it will have to be on package itself. If done correctly, combined with disabled cache, cost savings can be significant.

Nuclear
Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext