home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.benchmarks:1738 comp.arch:10923
- Path: sparky!uunet!bcstec!silverm
- From: silverm@bcstec.ca.boeing.com (Jeff Silverman)
- Newsgroups: comp.benchmarks,comp.arch
- Subject: Re: DEC ALPHA Performance Claims
- Message-ID: <4288@bcstec.ca.boeing.com>
- Date: 19 Nov 92 05:51:40 GMT
- References: <BxH7s7.5Cv@inews.Intel.COM> <4248@bcstec.ca.boeing.com> <1992Nov16.174912.22905@ryn.mro4.dec.com>
- Followup-To: comp.benchmarks
- Organization: Boeing Computer Services, Seattle
- Lines: 38
-
- >In article <4248@bcstec.ca.boeing.com>, silverm@bcstec.ca.boeing.com (Jeff Silverman) writes...
- >>
- >>1) Newer implementations of the alpha may need or take advantage of new ways
- >>of ordering instructions. This in turn means that different alpha
- >>
- >Optimal instruction scheduling for newer processors may indeed be different.
- >Old binaries will run correctly, but probably somewhat slower. New binaries
-
- I have received several mail messages from several people at DEC who have
- told me that I am right and wrong. They tell me that DEC is making the
- code generators available to everybody (obviously, a solution that the
- silicon people dreamed up), so that this won't be a major issue. DEC is
- expecting that as more experience is gained with the chip, newer and better
- code generators will come out.
-
- I have also received several message from several people at customer sites.
- A lot of them are number crunchers, and they work in FORTRAN. Why FORTRAN?
- Because it's fast (FORTRAN tends to be faster than C because you don't have
- to build argument lists on the stack - whether this is important or not,
- I'm not sure - I like to program in "C"). So they aren't so concerned with
- the speed of the chip, they're concerned with the speed of the computer,
- which is the compiled code, memory, disk, the whole ball of wax.
-
- I also got one message which was fascinating - it was pointed out to me that
- the activities associated with having more demand for RAM than RAM are a
- major performance drag. Paging is bad news if you are into speed, swapping
- is even worse. So how about a computer system where there is enough physical
- RAM so that everything that must be RAM resident is, and it doesn't swap or
- page. This operating system would be something like RT-11, but it would be
- called RT-64 or similar. I remember that on the DECsystem-10 (KA-10 CPU)
- you could get a performance goose by turning off the relocation logic, but
- that made it tough to run TOPS-10.
-
- I am really excited about the Alpha, but we're not going to get one anytime
- soon.
-
- Jeff Silverman, Boeing
-
-