Surely the speed factor was only of benefit when we were competing with 286 PC's with no 3D hardware.
The speed of the Amiga enhanced the quality of demo's (some games) etc. when comparing it to early PC's

When it came to 3D stuff the ST, at the time, produced faster running software - even faster than the PC.

My point is - The speed benchmark only served it's purpose when we needed to Texturemap polygons perform math intensive functions and achieve 25fps on Robocop3 (For example). These days it's the graphics cards that do the grunt - lets face it, take the 3D acceleration from the graphics cards and what do you have, not much.
The PC is what Amiga could have been. The Mac is what the Amiga could have been. It is true - these computers use custom chipsets. Our only hurdle is that our Amiga has not been developed for for a number of years. That is changing.
We now have AROS, which is usable. Give me sound and a TCP stack and I'll get rid of XP, in an instant.

UAE in AROS is fine, it needs a little work, but it does what the title suggests. Amiga Emulator. Not "Amiga emulator with a kick ass speed boost" On the right system a 68020 with FPU knocks the pants off a 060 many times over. You need to know very little more.

[Sorry about the rant]

Quote:

Well I have no problems with my WinUAE setup, only SPEEEDDD!

I agree with that. The problems I have running UAE aren't usually speed, but rather it fails here and there. For example, in the Linux version, if you run it on "maxiumum" speed, or have the JIT on, all sorts of funny things will happen. The only way to get it to work properly is "cycle exact" emulating an Amiga 500, at which point its only useful for old games and demos. Still, that's what it's for anyway, as far as I'm concerned. We have AROS for speed.

If you assume UAE under AROS runs as decently as UAE anywhere else and you haven't tried it, well I would just say that UAE is much faster than any REAL Amiga I have ever owned (provided that PC it's running on is at least 1 GigHz which is pretty common on modern PCs.

I stopped using the Amiga for day to day work with the release of windows 95. Looking back at it, I really missed some things like the cool demos and paint and animation software. I kinda skipped over Amiga high res graphics cards because for a while they were very non-standard and the software wasn't mature.

I like AROS and I am not really worried about uae for it being fast enough because under WinUAE it's a better Amiga than any other amiga I had was.

I understand that it's important for some Amiga folks who are considering moving to AROS to retain their software, but considering the number of other PC OS's out there I don't think that anyone will be having issues finding good software to run, even if you share the hard drive with another OS till more AROS software gets out..

I think the focus of the A1 and AROS and even the Peg machines are focusing way too much on being compatible and less on new function and bringing things to the OS that makes it truly usable and a better solution.

I love UAE but I think people are too focused there instead of why they are making a change in the first place..

http://wuarchive.wustl.edu/~aminet/aminet.cgi?string=benchmark

is another place to go nosing around. I'm particularly fond of AIBB, because it has a lot of benchmarks, and it's fairly easy to tease out the custom chip set stuff from everything else.

I'll run my poor little, slow hush on them with the new (and improved!) Linux UAE with JIT and post results. Of course, keep in mind, artifical benchmarks are notoriously misleading. But it can give you an idea of what kind of Amiga-ish behavior is difficult for the emulator (on your machine) and which is easy.

Thanks!!

For pure chip emulation, I hear that a 5:1 ratio is common for well-written, JIT based emulators, equalizing for all variables and not doing any direct execution (everthing gets jitted with no "pass-through" of native code on native targets). A look at the benchmarks for QEmu, an x86 emulator with JIT, running on an Intel target, may be informative:

http://fabrice.bellard.free.fr/qemu/benchmarks.html

Naturally, the case is significantly worse for UAE since it's emulating a very differnet CPU. The target CPU has more registers than the source (which is a drawback for emulation), and it's the opposite endianness (another hit). In addition, one is trying to emulate the Amiga's custom chips. This isn't nearly as bad as one may think -- I've heard estimates of about 5% of resource consumption for emulating the custom chips on modern Intel hardware -- however -- the real damage is when you take timing into account. Without timing, emulating the Amiga's custom chips on any modern machine would be trivial, and have almost no effect on total system speed. But for accurate emulation, things must be in the right order. The emulator spends some time doing nothing, even when it's at 100%.

Here is an article which takes a rather optimistic look at it, and it's linked to another article which gives a more in-depth look at the custom chip emulation.

http://www.amigaforever.com/kb/3-119.html

Things look better when you use the Picasso drivers. This reduces a lot of the burden on emulation concerning custom chips and their related timing issues. The speed bump is palatable. Even better is something like the Amithlon approach, where custom chips are simply removed from the picture, native drivers are written for video, and the emulator focusses on the relatively easy task of just emulating the CPU as fast as it can.

I was curious as to the benchmarks for UAE. I don't mean "..it runs 100 times faster than a 500.." I am more curious how much does one lose running a program in UAE vs running it native? If there were any native x86 Amgia programs. How does JIT effect this?