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: Plissken who wrote (199714)	6/1/2006 10:19:54 AM
From: Joe NYC	Read Replies (2) \| Respond to of 275872

Armin, Actually that's a point where the coprocessor approach would be more sensible. The die area required (and yield issues implied) by integrating graphics onto the CPU are non-trivial. However providing a dedicated HT link might be quite compelling. Graphics card like memory bandwidth. CPU (with DDR2) has a lot of it. HT only lets a fraction of it through (4 GB/s), unless AMD was willing to widen the HT link to 32 bits, and go a bit higher with clock speed. BTW, as I posted before, when AMD goes to HT 3.0, the width of the links should increase to 32 bits (and possibly go from 3 to 4). The links are splitable, for higher socket systems, and keep them together for low socket count. 2 way server would be best served with a single 32 bit link. Even though HT 3.0 is 10.4 GB/s, it is still under 12.8, and memory technologies will advance in the future. There is no reason HT should be the bottleneck. I don't know what the performance hit is even currently, when the memory controller of one processor is able to feed 6.4 GB/s of data (some 5 GB/s effective) but HT link is able to transmit only 4 GB/s (minus some overhead). The performance hit may be small, but it is something... Joe