home *** CD-ROM | disk | FTP | other *** search
- Path: news.netspace.net.au!usenet
- From: astroboy@netspace.net.au (Paul Dossett)
- Newsgroups: comp.sys.amiga.applications
- Subject: Re: Is MUI processor intensive?
- Date: 15 Feb 1996 08:28:05 GMT
- Organization: NetSpace Online Systems
- Message-ID: <4862.6619T1138T2039@netspace.net.au>
- References: <4f80sp$m2h@due.unit.no> <1468.6611T653T370@datashopper.dk>
- <1394.6613T995T2132@netspace.net.au> <2445.6614T1150T1609@imaginet.fr>
- NNTP-Posting-Host: dialup-a1-29.mel.netspace.net.au
- X-Newsreader: THOR 2.22 (Amiga;TCP/IP) *UNREGISTERED*
-
- UNREGISTERED VERSION (yeproy@imaginet.fr) said:
- >Paul Dossett wrote :
-
- >> So what? The machine will still be faster. On the whole, the AGA chipset
- >> doesn't particularly speed things up. AIBB rates my ECS system as running
- >> graphics tasks faster than a 1200, thanks to a faster processor.
-
- >Which graphics tests? WritePixel and EllipseTest, because they are CPU
- >intensive, but a Chip RAM only A1200 (14 MHz '020) leaves in the dust your
- >33 MHz '040 A2000 when it comes to LineTest!!! Your machine performs only
- >12% better than the stock A2000 with such a test anyway...
-
- Yes, that's the ECS set giving up the ghost there! :(
- Damn its eyes! Somebody sell me a Piccolo! PLEASE!
-
- >Ever tried to compare your '040 A2000 with a '040 A1200, just to be fair?
- >I did, with a 25 MHz Falcon 040 (thanks G÷ran): the 1200 performs 40% better
- >than your A2000 with WritePixel, with 25% less CPU power.
-
- I did mention that I had a faster processor than a 1200, so don't try and use
- that against me.. ;)
-
- You do realise what you're saying here, don't you? If not, wait a couple more
- lines.
-
- >The standard 25 MHz '040 A4000 is even better, with 51% instead of 40%.
-
- Something's wrong there, all things being equal, the 1200 board should have
- been faster due to the A4000's memory bus..?
-
- >AGA is not a great performer, for sure. But it is still 50% faster than
- >ECS.
-
- Here's the kicker, my friend!
-
- ECS: AGA:
-
- 16 bit bus 32 bit bus AGA=2x speed
- 7.09 MHz clockrate 14.2 MHz clockrate AGA=2x again, making a
- bandwidth of *4* times ECS!
-
- Do you see how shocking those benchmarks are now? You're comparing a chipset
- that has FOUR TIMES the bandwidth of my 1986 A2000, yet you're not even
- DOUBLING the speed in real-world benchmarks!
-
- Now do you understand my statement that AGA doesn't speed things up
- incredibly?
-
- Ladies and gentleman, I thank you.
-
- (bows)
-
- >> Most programs (word processors, raytracers, samplers) need CPU power.
- >> Graphics throughput is so completely unimportant to the argument that you
- >> can discount it. The A600 with the A620 *will* be a lot faster than an
- >> A1200 for
-
- >Ever tried ShapeShifter? Don't if you don't have a graphic board in your
- >A2000, unless you accept a monochrome display. Graphics throughput is not
- >something to neglect with modern applications (and I didn't even mentionned
- >MUI, Paul... ;-)).
-
- Hmm.. I run ShapeShifter in 640x400, 16 colours. It's useable. Sure, it'll
- be faster with a GFX board, but until I can find one, SS is capable of playing
- Oxyd or whatever at the moment.
-
- As you seem obsessed with the lack of speed of my computer, Yann, could you
- answer me something (not flaming, just a legitimate question)? Do you have
- the PD program MacBench 2.0 for the Mac, and could you tell me the speed your
- 28Mhz Picasso gets on the graphics test (in 256 colours, preferably)?
-
- I'm *very* interested in how a 2000 with a 16bit graphics bus compares to a
- 32bit Quadra. Right now in monochrome my system gets something like:
-
- Processor : 105%
- FPU: 98% (not sure why this is lower, but oh well)
- Graphics: 118% (!! This is on a chipset with half the bandwidth! I'd love
- to know what Christian has done to get this mark so good!).
-
- I'm really hoping my system can run Doom, etc. nicely, but it seems illogical
- that it can with such a slow Zorro-II bus. Can you give me any information?
- Anyone?
-
- >> These are the facts, try using the systems in comparison and running
- >> benchmarks if you don't believe them.
-
- >Being the author of CompilAIBB, I have studied *many* benchmark results,
- >and I do not share your conclusions about AGA speed improvement, nor
- >about usefulness of quick graphics.
-
- Then explain the shoddy results you're turned up.
-
- >Now, to be honest, it is highly probable that the 28 MHz A620 performs
- >as much as 2 times faster than a 14 MHz A1200, except for graphic tests.
- >So the overall feeling of speed should be in favor of this A620.
-
- That is what I said...
-
- >Anyway, it could help clarify things if a user of a A600+A620 could send me
- >an AIBB module from his machine.
-
- Does such a creature exist?
-
- >Of course one could argue that benchmarks represent vaguely (at best...)
- >the feeling of speed that people have in everyday life, but I guess it is
- >another debate...
-
- Yep, and I'll keep on bitching for as long as you like! This is fun.. ;)
-
- >Regards,
-
- >Yann
-
- >PS: Paul, I'm sorry to contradict you twice in two messages, I can assure
- > you there's nothing personnal in it... ;-)
-
- The feeling is mutual, mate! I have no problems with differing opinions, in
- fact I thrive on them... :)
-
- --
- Paul Dossett | Yamaha, Epiphone, Pearl, Paiste | Amiga 2000/040/21/365/3.1 __
- -------------| Bubble Bobble Deluxe Dvlp. Team | Amiga CD32/020/2/CD/3.1__///
- boo! : astroboy@netspace.net.au.. | '76 Toyota Corolla 1.2 \XX/
-
-