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 : All About Sun Microsystems

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: QwikSand who wrote (28526)3/3/2000 7:59:00 PM
From: rudedog  Read Replies (1) of 64865
 
QS - re: I guess it's just a matter of what one means by 'fast'
unless this is something a lot different than it looks, it means "fast" is about 25 years. RSX-11M for the PDP-11 had the ability to support page-selective multi-instance overlay compiles... if the MSFT programmers think it is important to save storage when gigabytes are available, they would have had a REALLY hard time when the whole OS, tools suite and data had to fit on a 2.5MB disk pack. The RSX compilers supported features which allowed developers to segment the overlay structure to optimize both base code and overlay size, and was smart enough to know when a required overlay was already in memory on a context switch, even when it was from another user's session, and whether the overlay was re-entrant. That stuff got a lot more sophisticated with VMS for the VAX, where one of the tricks to improve multi-user performance was to have as little code in the base segment as possible which greatly reduced swap times and made it more likely that an overlay segment would already be resident.

Of course RSX or VMS programmers could also explicitly call global shared libraries, and the system designer could lock pieces of those in memory - this was most common in the real-time instrumentation support, since the code to run a data acquisition front end pretty much had to be ready to run when the interrupt hit, there was no intelligence in the hardware front end in those days beyond simple silo storage.

There must be something more to this MSFT stuff than we're seeing - on the surface it just looks like globally available file aliasing - something done at a much more sophisticated level with globally available virtual storage page aliasing, which has had several implementations in the last few years and will probably be a key building block of SANs in the future.

Rick Rashid, who now heads up MSFT research, is certainly aware of all of this stuff - he wrote the MACH kernel, after all, and was one of the smarter guys I knew when he was at Carnegie-Mellon. He can recite the architectural heritage of every Unix flavor (and most other modern OS variants) the way an evangilist can quote scripture, and also knows the technical components of each of those lines, and when they occured. I doubt that they run press releases past him, though.
Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext