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 : All About Sun Microsystems -- Ignore unavailable to you. Want to Upgrade?

To: Charles Tutt who wrote (27543)	2/10/2000 10:39:00 AM
From: rudedog	Read Replies (2) \| Respond to of 64865

Charles - Does that mean they'll run the Itanium in emulation mode? No, they could do that with the 32 bit code base with NO changes but performance would be pretty poor, I suspect. This is native IA64 code. It's just that most of the OS operations don't care much about 64 bit operations - outside of memory management. If 64 bit pointers are handled correctly then those routines should work just fine in either 32 bit or 64 bit mode. The big benefit would be a flat memory model for big memory - but even there, the current Win2K allows many programs to "see" a large address space even in the 32 bit world, with the OS masking the translation to the 36 bit extended addressing space. MS has actually done a pretty good job of providing APIs which will naturally take advantage of large addressing. The improvements in floating point operations should likewise be fairly transparent - most programmers don't go much below the surface with FP code and so the calls will just run faster. The biggest problem I see with merced performance is the lack of hardware support if the compilers do not find and exploit parallelism - that was an assumption of early merced design which did not prove out, and the silicon won't really address those shortcomings until McKinley, so there will need to be a lot of hand-tuning of merced implementations. Given the short life of the product, it seems doubtful to me that more than a handful of ISVs (the ones being heavily supported by Intel) will take the time to do that for merced. That implies that merced sales will be narrow in focus, probably mostly in areas where there is already a shift to Intel technology. I would expect the database vendors and perhaps the ERP guys to do something, but I expect broader optimization for IA64 to be in the McKinley timeframe, which I think is probably mid-2002 or so.