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 : Rambus (RMBS) - Eagle or Penguin
RMBS 95.26+3.1%Nov 14 9:30 AM EST

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: blake_paterson who wrote (43126)5/29/2000 10:00:00 AM
From: SBHX  Read Replies (2) of 93625
 
blake_paterson,

Willy with SSE is 10x faster than anything out there.

Just to be clear, SSE (Streaming SIMD extensions) are supported by all PIII (even non willy PIIIs). They are mostly 4-way floating point instructions that help 3D geometry, transform and lighting --- features that are now standard in the new graphics processors. There's also a single instruction that also helps motion estimation by calculating sum of abs diff. The data streaming prefetch and fence instructions make up the rest of SSE. Useful, but hardly critical to run well.

The departure from MMX to SSE (or MMX-II as it used to be called) is that MMX is intel's public open architecture supported even by AMD, and universally well supported by most s/w vendors. MMX-II is a closed architecture and disliked universally by s/w vendors. Intel generally has to spend more money for people to write MMX than x86 native, and now they have to spend even more to get people to write MMX-II code.

The people who watch the PC industry for fun also have noted that intel's latest architectural innovations in areas of CPU bus, IO and memory interconnects have not been open and they have erected barriers that make them essential controlled and proprietary. Each new innovation requires even more and more stringent barriers as their legal department hone and polish the terms of their NDAs to the point where other vendors working with them are at their complete mercy. Did you ever try to get a Slot1 or socket 370 license recently? Well, that is going to get MUCH MUCH worse moving forward.

This really is the part that most people don't like. For those who don't have a financial interest long or short, this thing is really about open vs closed proprietary architectures. Sometimes closed architectures will win and the customers end up paying for it, but that doesn't mean we have to like it.

Also, the graphics guys charge whatever thay can get for each board. If you really want a 64MB DDR graphics card, then you might as well really really really really pay for it >;->. The wonders of non-upgradeable graphics boards.
Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext