The Demoscene might not be about limitations anymore
category: general [glöplog]
Quote:
As for cheaper alternatives, there's also Win UAE, which probably isn't much less accurate than a Pi Storm anyway?
The PiStorm doesn't emulate the chipset, so not sure what you mean by "accuracy". Obviously the instruction timings won't be exactly like on a real CPU, but the various generations of "real" 68k chips also differ greatly.
But if you want the cheapest, nothing beats just running UAE on a 15 year old PC laptop indeed. (Just remove the swollen batteries first, heh.) But that's also not "real hardware", where this discussion started.
Quote:
Stock A1200 potentially provides a similar level of uniformity and stability.
One of the main takeaways for me from the Gerp No CPU Challenge compo was, that most Amiga coders refuse to do AGA stuff, even if you take away their CPU. :) Having done my sarcastic remark, may it be different this time around!
Quote:
While 060 has succeeded in providing an almost uniform demo platform for many years, there are noticeable variations in performance, even for the same clock frequency and accelerator brand.
While I 100% understand the allure of an uniform demo platform, the '060 was really "close enough". Since it always will be the matter of where you draw some lines. Just ask the NTSC C64 lovers over the pond... :) Also, let me be Captain Obvious for a second and point it out, that the least uniform platform ever, the PC, also became the largest ever by far. Anyway. I get it, of course. By not having to worry about hardware differences, one can push the limits further and harder, in another direction.
Quote:
The PiStorm doesn't emulate the chipset, so not sure what you mean by "accuracy".
You still have to verify that it runs correctly on real hardware.
Like Blueberry said:
Exactly - This is what I love, too. Even with everyone watching on youtube, to me it matters that other coders are writing code on the same spec platform. And it's also why I prefer making games for consoles over PC and mobile platforms - if something runs 60fps for me, it runs 60fps for everyone.
Grip said:
This level of variation seems close, but it isn't good enough if you try to push limits, set records, or do cycle exact magic. You write your demo on a machine that runs at 2.05x speed, and your 50fps effects may slow down and ruin your timing on a 1.99x expansion machine - that just sucks. So if those numbers are correct, I'm a fan of sticking to the unexpanded vanilla a1200 spec.
For the 030/040/060 demos, variation is the norm. That's perfectly fine, those demos are great, but they are a different tier of platform, much more similar to the PC scene, and not what I love.
Quote:
A huge part of the allure of the "standard" A500/512+512/8372A config is that if it runs on your machine, you can be pretty sure it runs in exactly the same way on your neighbor's machine as well.
Exactly - This is what I love, too. Even with everyone watching on youtube, to me it matters that other coders are writing code on the same spec platform. And it's also why I prefer making games for consoles over PC and mobile platforms - if something runs 60fps for me, it runs 60fps for everyone.
Grip said:
Quote:
Trapdoor fastmem expansions report slightly different speedups in E.G. SysInfo. I've seen values between 1.99 and 2.32 times the speed of an unexpanded 1200 with various cards.
This level of variation seems close, but it isn't good enough if you try to push limits, set records, or do cycle exact magic. You write your demo on a machine that runs at 2.05x speed, and your 50fps effects may slow down and ruin your timing on a 1.99x expansion machine - that just sucks. So if those numbers are correct, I'm a fan of sticking to the unexpanded vanilla a1200 spec.
For the 030/040/060 demos, variation is the norm. That's perfectly fine, those demos are great, but they are a different tier of platform, much more similar to the PC scene, and not what I love.
Maybe I lost a bit track of the discussion when the PI-Storms got in (which run much faster than 060s ever did).
When discussing "stock" 1200, fast-mem should be out of question because it never existed "out of the box" - at least not in the same box as the 1200 ;)
When it comes to a hard drive, it did exist out of the box and makes a big difference:
The machine comes with 2MB RAM and during the run-time of a demo you want to fill that up a few times. That can easily fill 3 or more disks and that simply means you are creating a demo which runs at a quite different pace.
For me (looking from the 060-perspective) that's basically like doing an A500-demo with a bit more beef and colors. That's an interesting extra challenge - but in a way, also a different discipline.
When discussing "stock" 1200, fast-mem should be out of question because it never existed "out of the box" - at least not in the same box as the 1200 ;)
When it comes to a hard drive, it did exist out of the box and makes a big difference:
The machine comes with 2MB RAM and during the run-time of a demo you want to fill that up a few times. That can easily fill 3 or more disks and that simply means you are creating a demo which runs at a quite different pace.
For me (looking from the 060-perspective) that's basically like doing an A500-demo with a bit more beef and colors. That's an interesting extra challenge - but in a way, also a different discipline.
