home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.amiga.misc,comp.sys.amiga.hardware
- Path: news-1.csn.net!ub!dsinc!scala!news
- From: dave.haynie@scala.com (Dave Haynie)
- Subject: Re: PowerPC !
- Sender: news@scala.scala.com (Usenet administrator)
- Message-ID: <1996Mar5.185907.12755@scala.scala.com>
- Date: Tue, 5 Mar 1996 18:59:07 GMT
- Reply-To: dave.haynie@scala.com (Dave Haynie)
- References: <Pine.SUN.3.91.960220162223.624A-100000@scsscsc1.reading.ac.uk> <4gdgsa$bbm@gina.zfn.uni-bremen.de> <4gpot8$l1h@murphy2.servtech.com> <4h57l1$rc5@hobbes.compusult.nf.ca> <4h6hgp$fqo@rs18.hrz.th-darmstadt.de>
- Nntp-Posting-Host: gator
- Organization: Scala Computer Television, US Research Center
-
- In <4h6hgp$fqo@rs18.hrz.th-darmstadt.de>, kraemer@rzri6f.gsi.de (Michael Kraemer) writes:
- >In article <4h57l1$rc5@hobbes.compusult.nf.ca>, mworden@public.compusult.nf.ca (Mark Worden) writes:
- >> nyle landas (nyle@cyber1.servtech.com) wrote:
- >>
- >> : I have read the PowerAmiga is slated for 1st quarter 1997. PowerPC cards
- >> : for exsisting Amigas should be out this summer from Phase 5.
-
- >> What kind of speed will it have compared to an actual PowerPC?
-
- The PowerPC boards from Phase 5 are actual PowerPC boards. The
- question to ask is not "how fast is it compared to a real PowerPC"
- (since it is one), but "how fast is it compared to a real 68K", since
- you'll be running 68K code much of the time.
-
- >> I had an EMPLANT Deluxe in my A3000/030 awhile ago, and it ran
- >> dreadfully slow.
-
- Real Macs are dreadfully slow on 68030 systems, so you can't expect
- the MacOS running on an Amiga to be much different. The A3000 does
- typically run the MacOS a tad faster than the nearest Mac equivalent,
- the Macintosh IIci. That's primarily because the hard disk is much,
- much faster on the A3000. The IIci was the first Mac II to have a
- decent memory system, so it's comparable in other ways to the A3000.
-
- >Well, my experience with A3000 vs PowerPC e.g. with a curve fitting program
- >(floating point math + array addressing + local gfx)
- >and looks like follows:
- >
- >A3000 (68030/882 @ 25 MHz): 29.2 secs
- >Linux-PC (80486 @ 33 MHz): 7.2 secs
- >RS6000/AIX (PPC601 @ 80 MHz): 0.8 secs
- > (PPC604 @120 MHz): 0.4 secs
-
- The 68030/68882, to put it as kindly as possible, is lame when it
- comes to floating point. Typical benchmarks put it at 0.3-0.5 MFLOPS,
- versus 3.5-5.0 MFLOPS on a 68040 at the same speed. If the benchmark
- is heavily floating point intensive, an '040-25 should easly beat a
- '486-33 (not that anyone uses '486-33s anymore).
-
- >Although the PPC seems to blow away the 68k stuff I've decided to
- >upgrade my A3000 with an '060 board for the simple reason that I
- >don't expect the PPC upgrade boards to work smoothly with all progs
- >in emulation mode.
-
- There's no reason to believe the PPC emulation won't provide a
- perfectly compatible 68K programming model. What it can't do is match
- the expected instruction timings of any 68K processor. So things that
- are very dependent on your system's speed, typically interactive
- multimedia and game software, may have problems. Other tools should
- run fine, though there's a fair chance floating point won't be all
- that great, if they chose to implement the 68K floating point
- instructions at all in the PPC model (you might do just as well with a
- PPC native IEEE library).
-
- Dave Haynie | ex-Commodore Engineering | for DiskSalv 3 &
- Sr. Systems Engineer | Hardwired Media Company | "The Deathbed Vigil"
- Scala Inc., US R&D | Ki No Kawa Aikido | info@iam.com
-
- "Feeling ... Pretty ... Psyched" -R.E.M.
-
-