From: ortmann@informatik.tu-muenchen.de (Mathias Ortmann) Newsgroups: comp.sys.amiga.advocacy Subject: Re: Phase IV's A/Box Date: Mon, 28 Jul 1997 20:43:09 GMT Organization: Leibniz-Rechenzentrum der Bayerischen Akademie der Wissenschaften Message-ID: <5rj07j$k5r@arcadia.informatik.uni-muenchen.de> References: <92524-870108464@mindlink.bc.ca> NNTP-Posting-Host: dial116.ppp.lrz-muenchen.de Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Newsreader: Forte Agent 1.5/32.451 Lines: 125 Path: 195.95.96.2!news.unisource.be!News.Amsterdam.UnisourceCS!newsgate.unisource.nl!bullseye.news.demon.net!demon!dispatch.news.demon.net!demon!news.maxwell.syr.edu!newsfeed.nacamar.de!wuff.mayn.de!wuff.franken.de!news-nue1.dfn.de!news-mue1.dfn.de!rzg.mpg.de!lrz-muenchen.de!news.informatik.uni-muenchen.de!usenet Terry_Palfrey@mindlink.bc.ca (Terry Palfrey) wrote: >>>By *us* he means the AMIGA advocates. > >>Right. The not-overly-convincing zealots. > >Hmm, having used an AMIGA since it came out, that was the shop's >in 85/86 because of my finances, I can only say that not only is >it flexible and fluid but in every example I've seen it is capable >as well. I'm not arguing the Amiga's superiority in the '80s and the early '90s. > Toaster'd systems, people with the 1900$ 040's with 32 megs >of ram for rendering and imaging, programming systems and of course >setups like mine for home use that did and still do despite the fact >that the rest of the world uses 100MHz to 200MHz machines that just >don't seem as good. They don't _seem_ as good because you refuse to build a balanced system consisting of quality components, something that you have to do with an Amiga in order to make it usable, too. My PC is ancient in terms of pure CPU power (100 MHz Pentium with a motherboard that was released in mid-94!), but beats many cheap 200 MHz systems in terms of ergonomy, usability and stability. Why? Because of the display (1280x1024x24, 80 Hz), memory (96 MB), disks (7200 RPM SCSI, _no_ IDE), CDROM (3 MB/s peak), 4mm DAT streamer, high-quality sound card, high-quality keyboard, high-quality mouse (3 buttons :-), quality OS (NT/Linux, _no_ Win95/DOS). It's the peripherals that account for the perceived power of a given system, not so much the actual CPU speed. Of course, you have to approach any alien system open-mindedly, or you'll always find a reason to hate it. >>>Not the pretenders, correctors, hangers-on, quoters of magazines, >>>other OS deviants and fringers. > >>LOL! > >As well you might, but in this newsgroup we have people who actually >built LUCAS boards, who know what TRANSACTOR was, ARP'd machines, >HACKED them, ENHANCED them, USE ARexx properly, EXPANDED machines at >great cost and were simply THERE during the whole time. Assuming that you are right, why don't they answer dumb anti-Amiga propaganda with educated, rational and substantiated followups? >>>So let us in on it, how fast is UAE vs the real AMIGAS? > >>In terms of hardware emulation: A PentiumII-300 can just keep up >>with an A500 in extreme video modes (50 Hz dual-playfield scrolling, >>HAM). > >But in terms of simple displays like DOpus, Pagestream and a browser >you would get a nice clean hi-res display with the options of screens >like a normal AMIGA? On a P200 (a $99 chip), UAE feels like a slow A3000 (even the artifacts when scrolling in interlaced mode are emulated authentically). I hope that by the end of the year, a virtual graphics card will increase application performance and allow for arbitrary resolutions and color depths beyond 4 bit. >>In terms of pure M68K emulation: Surprisingly fast. > >Which class of M68K are we talking about? So many claims have >been foisted on this forum. This depends on too many parameters to be answered with a single number. However, here is a very rough, conservative sample calculation that is in accordance with reality: a = average number of x86 instructions per 68K instruction: 40 b = average number of x86 instructions per clock cycle: 1.2 c = average clock cycles per 68000 instruction: 3 Assuming a 100 MHz host CPU, the virtual 68000 runs at a speed of 100 MHz/a*b*c = 9 MHz UAE will spend a lot of time in the custom chip emulation if the frame rate is below 50 Hz. Note that the frame rate cannot get any higher than that, beyond which the M68K emulation benefits from all surplus host CPU power in a non-linear manner. Therefore, you cannot linearly scale speed estimates. >>Combined: Not very usable on anything slower than a Pentium-200. >>Very impressive on a PPro-200. I'll post the results of my tests >>on a PentiumII-300 (-337.5?) in a few weeks. > >Pentium 200 being the lower end of use-ability, okay a good starting >benchmark. On such a beast AdPro and ARexx combinations could be >expected to have what kind of performance? Assuming little screen activity, somewhere between a 10 and 20 MHz 68030, i.e. usable, but not really snappy (file system access will be comparatively fast, by the way). >>>Does it make economical sense to run it rather than the real thing? > >>If you have a powerful host machine anyway, yes, because it's free. >>If you are considering buying a high-end PC just to have a replacement >>for your Amiga, forget it. > >Just examining the possibility for a dual pent-pro 200MHz system to >get some functional utilities working in the abysmal NT environment. >There's 12gigs of fast AV drive and 128 megs of dimm ram available plus >all the video goodies and scsi ultrawide on the system drives. UAE won't benefit much from the second CPU, but a PPro-200 is exactly the type of CPU UAE starts making sense on. I wish I had a version of gcc that produces code which takes maximum advantage of the PPro/PentiumII. >The only thing that is not an option is instability or flakiness. Assuming that all Amiga software performs legal operations, UAE is perfectly stable. >Rock solid AMIGA apps even in a co-operative environment would be >a boon at this point in time. Try UAE on your PPro-200 and see if it fits your needs. -- Mathias Ortmann ortmann@informatik.tu-muenchen.de UAE 0.6.9 Win32/DirectX @ http://www.informatik.tu-muenchen.de/~ortmann/uae/