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 : CYRIX / NSM -- Ignore unavailable to you. Want to Upgrade?


To: Scumbria who wrote (26063)4/20/1998 11:18:00 AM
From: Steve Porter  Read Replies (1) | Respond to of 33344
 
Scumbria,

What are you somking man..

Listen MP is great is I have two separate applications each needing 100% of my CPU. But spliting 2 threads accross CPU's isn't worth it. Case in point:

Novell netware loads it's tasks to a CPU.. all of the threads for that tasks live on that CPU.

NT splits threads accross CPUs.

Guess which one scales better (as much as I hate to say it). Netware!.. Why.. because of cache coherency.. it doesn't matter how fast the MP bus is, it is still 400% slower than L1 cache..

As for multiple pipes, they are PERFECT for multithreading. The problem is that compilers still don't take full advantage of them, thus we have things like branch predicition etc.

Wake up man. I love the idea of MP, but you are talking like it will be the saviour of the world. I could buy MP 486's 5 years ago.. did anyone buy them???????.. There's a REASON.. COST, the big $$$$$$$$$$$$$, MP is VERY expensive.. why well:

Glue logic,
Power consumption (yeah MP will work in a notebook.. pfftt.)
HEAT! (any idea how hot to PII bricks are side by side or how about 4 PPro's)..
not to mention a increase in risk of failure (normal riskxnumber of cpus)..

There is a reason why MP hasn't caught on you know.

Steve

Cyrix/NSM Rule!