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: jim kelley who wrote (24982)	12/17/1999 2:06:00 PM
From: Thomas Mercer-Hursh	Read Replies (1) \| Respond to of 64865

I was not trying to design a SCSI driver. No, but you were tending to imply that there was little difference in disk subsystems as a part of your "benchmarks tell the whole story" pitch.

To: jim kelley who wrote (24982)	12/17/1999 3:30:00 PM
From: rudedog	Respond to of 64865

Jim - RE: That is best left to members of the SCSI union who will do this for anyone for a buck. Well, as a guy who has designed silicon for controller systems, the controllers themselves, and done optimized drivers for Sun, HP, DEC Alpha, SGI, and NT, I may be a card-carrying member of the SCSI union. But although I was among the best doing that back when I worked for a living, I could not make a silk purse out of a sow's ear. The 3-tier I/O bus architecture on the HP T500, for example, did not allow for creation of a direct-connect driver - the hardware forced the driver to take at least 4 interrupts. The driver work is done either by the OS people, or working closely with them, and is very specifically designed for the particular strengths and weaknesses of the OS and hardware environment. So the architecture, and the skill of the OS team, can have huge impact on the capability of the system to do "real-world" jobs. And one can not just go out and hire a "drivers jock" and get that level of performance - there are only a handful of people in the industry who can do more than a mediocre job, and many of them work for Sun. Finally, regarding benchmarks, the big "real world" applications are tuned for the hardware architectures that are selling best and which show the best performance. It is a big job. So when IBM brings out a new "hot box", the big ISVs will maybe do some token porting work but the real optimization will be driven by customer demand, which happens months or sometimes years down the road. And there is no reason to assume that a new architecture will outperform a system with slower CPUs, or even one which has lower performance on standard benchmarks, even with optimization. Large companies with an in-house application developed for a specific architecture are even less likely to shift without a major justification. That's the reason that getting the "architectural dominance" in new application environments is so important - the results can put you in clover for years.