PC games that made the Amiga look like Shit

[rKickrkds]

Originally Posted by Bruce Abbott

That doesn't sound right. The 286 should be faster.

Clock for clock the 68000 is faster than a 286 in mips, my own benchmark and other results on the web match.

Well we know that the game was written for the PC first then and ported to the Amiga. I bet they didn't make much effort to optimize it. Still the Amiga version is very playable on an 1200 with FastRAM (which is similar to a 12 MHz 286) and doesn't look much different.

I tried this on Winuae, default A1200@14mhz it still runs slower. I upped processor speed to 25mhz and it is as close to the same speed it runs on my 286 12mhz.

Another case of a game tuned for one platform and badly ported to another. The Amstrad CPC+ version is also quite remarkable, except for being a bit slow in the character animation (which is to be expected from a 3MHz Z80 with no hardware acceleration).

?? It wasn't released on the Amstrad CPC. PC version only came out a few months after the Arcade in 1991 and would not have had a lot of development time to get it out that fast. It was ported by Elite Systems as well, not known for arcade perfect ports. Amiga version came out a whole year later in december 1992 and was not a pc port. The point about this game on the PC is you had an arcade perfect clone of an $2499 Arcade game the same year of release in 1991 on a 286. That PC version blows away the best Amiga arcade conversion of any action arcade game I have played on the system. It's like the x68000 arcade conversions I have played. Just makes you think how massively under utilized the 286 was from 1984 to 1991 with the awful conversions it usually had.

[Megalomaniac]

Are you thinking of Prehistorik rather than Caveman Ninja on the Amstrad Plus? That is seriously impressive, the sequel (which the Amiga and ST never got) even more so.

From the brief play I've given it, Amiga Caveman Ninja is an embarrassing juddery mess that US Gold might have thought twice about releasing. Can't comment on the PC version, but I'd assumed that a 286 wasn't capable of 256-colour 2D games with anything like Amiga-quality graphics at a decent speed, so I wonder how the programmers did it?

[pandy71]

Originally Posted by Bruce Abbott

That doesn't sound right. The 286 should be faster.

but it was not faster - try to convert jpeg to 16 color bitmap with generally available djpeg (part of standard jpeglib) this can be quite representative - 386SX@16MHz was slightly faster but 286@12MHz slower.

[Bruce Abbott]

Originally Posted by pandy71

but it was not faster - try to convert jpeg to 16 color bitmap with generally available djpeg (part of standard jpeglib) this can be quite representative - 386SX@16MHz was slightly faster but 286@12MHz slower.

SPEC CPU92 MIPS V1.1

AT&T PC6300+ 80286 6MHz 0.81 MIPS
Amiga 1000 68000 7.16MHz 0.54 MIPS

That means a 12MHz 286 PC with no wait states should do 3 times more MIPS than a 7.16MHz 68000 Amiga.

How that relates to real-world programs depends on the instructions being executed. I'm not sure how decoding jpegs relates to a 2D platform game, but I'm betting the instruction mix is quite different.

[AestheticDebris]

Originally Posted by Megalomaniac

Are you thinking of Prehistorik rather than Caveman Ninja on the Amstrad Plus? That is seriously impressive, the sequel (which the Amiga and ST never got) even more so.

Prehistorik was a standard CPC game, only Prehistorik 2 got a "Plus enhanced" version. It's impressive, but somewhat held back by the fact the base code has to also run on the standard CPC models so the Plus features are limited to largely decorative effects over the top (it doesn't take advantage of smooth hardware scrolling etc).

[gimbal]

Originally Posted by Megalomaniac

but I'd assumed that a 286 wasn't capable of 256-colour 2D games with anything like Amiga-quality graphics at a decent speed, so I wonder how the programmers did it?

How did they do it? They probably read some books on hardware registers and wrote the code. Assumptions don't tend to be correct, oh mighty Amiga warrior.

[Aardvark]

Youtube videos of Caveman Ninja are a bit misleading. I did a quick test on various PCem configs and steady 60fps gameplay requires a 486DX2 66MHz which wasn't even available at the time, and with a 386DX 33MHz it starts to get quite choppy.

[d4rk3lf]

Originally Posted by Aardvark

Youtube videos of Caveman Ninja are a bit misleading. I did a quick test on various PCem configs and steady 60fps gameplay requires a 486DX2 66MHz which wasn't even available at the time, and with a 386DX 33MHz it starts to get quite choppy.

Thank you for mentioning that.
People always forgets how PC struggles with constant fps.
This is why games to me looked so much more polished on my friend A500, then on my other friend 386DX/40.
And these days.. even Amiga fans discuss if A1200 can beat 286.

[amilo3438]

Originally Posted by Bruce Abbott

When the Amiga was released in 1985 its graphics were light years ahead of the PC. However that soon changed. In 1987 IBM released VGA, with 256 colors and hardware scrolling. Super VGA arrived soon afterwards, with 256 colors in 640x480. Still no sprites or copper effects, but the CPUs were getting faster to make up for it. In 1986 Compaq released the Deskpro 386, with a 16MHz 80386DX and fast 32 bit RAM. By 1991 386 PCs were all the rage, and 486 machines were starting to appear.

The question is how much do you get for your money?

How much would a PC cost that would have similar capabilities to the Amiga500 at the time?

It means that for the PC of that time, it was necessary to invest in a better cpu, a better gfx card and a sound card!
And if that gfx has a better resolution and more colors, and the sound of more channels and 16-bit, of course everything will look better.

So I don't see the point of this. (unless you are comparing a PC of similar price)

[gimbal]

"At the time" is very vague. Just like the Amiga 500 survived well into the A1200 era, the 286 survived well into the 386 era.

[pandy71]

Originally Posted by Bruce Abbott

SPEC CPU92 MIPS V1.1

AT&T PC6300+ 80286 6MHz 0.81 MIPS
Amiga 1000 68000 7.16MHz 0.54 MIPS

That means a 12MHz 286 PC with no wait states should do 3 times more MIPS than a 7.16MHz 68000 Amiga.

How that relates to real-world programs depends on the instructions being executed. I'm not sure how decoding jpegs relates to a 2D platform game, but I'm betting the instruction mix is quite different.

You are experienced and you are fully aware that synthetic test usually rarely shows real life performance - using not heavily optimized plain C code is usually better to show real speed (so not only CPU but complete system). And we are also aware that Intel real life performance was usually worse than Motorola. 386 changed a lot in this picture but still up to 68040 or even perhaps 68060 Motorola CPU was faster than comparable Intel at the same clock.

[Bruce Abbott]

Originally Posted by pandy71

You are experienced and you are fully aware that synthetic test usually rarely shows real life performance... And we are also aware that Intel real life performance was usually worse than Motorola.

I was responding to this:-

Originally Posted by rKickrkds

I know MIPS is not the best benchmark but the Motorola 68000@8mhz benchmarks 1.4MIPS and a 286@12mhz is only 1.28MIPS.

SPEC is the definitive MIPS standard, and they found that the 80286 did more MIPS per MHz than the 68000. This is to be expected because the 286 uses fewer clocks per memory cycle. OTOH the memory has to be faster to keep up. That's why the 6MHz PC/AT had 1 wait state inserted, and why many 286 benchmarks are slower than expected from the CPU alone.

When someone says they are surprised by how well a late model 12MHz 286 plays certain games I am not surprised. Compared to a 7.1MHz 68000 It's doing more MIPS, and if it's dealing with byte or word data the 68000's 32 bit registers are no advantage. There aren't a lot of games that show this off though, because most 286's had slower memory and CGA or EGA, not VGA.

[Bruce Abbott]

Originally Posted by gimbal

"At the time" is very vague. Just like the Amiga 500 survived well into the A1200 era, the 286 survived well into the 386 era.

Yes, and in later years the Harris 80C286 was clocked at up to 25MHz. With a fast motherboard and RAM, good graphics card and overclocked ISA bus the performance difference between that and a 6MHz PC/AT is staggering.

But that's the way it was with the PC vs Amiga. Everybody compared the latest model PC with all bells and whistles to a stock Amiga 500 - not an A2000 with 25MHz 030 or an A3000 with its 32 bit ChipRAM, and certainly not any 3rd party addons. Sadly this was justified because only the stock A500 had a large enough userbase to attract developers.

There were millions of older lower spec PCs out there too, that users tried to play games on badly. But that didn't matter because most would soon upgrade to higher spec machines - unlike Amiga fans who stuck with their old machines to the bitter end (not that this was necessarily a bad policy - it was certainly a lot cheaper than upgrading your PC every few years).

[redblade]

Originally Posted by Bruce Abbott

But that's the way it was with the PC vs Amiga. Everybody compared the latest model PC with all bells and whistles to a stock Amiga 500 - not an A2000 with 25MHz 030 or an A3000 with its 32 bit ChipRAM, and certainly not any 3rd party addons. Sadly this was justified because only the stock A500 had a large enough userbase to attract developers.

There were millions of older lower spec PCs out there too, that users tried to play games on badly. But that didn't matter because most would soon upgrade to higher spec machines - unlike Amiga fans who stuck with their old machines to the bitter end (not that this was necessarily a bad policy - it was certainly a lot cheaper than upgrading your PC every few years).

Maybe if more piggy back Accelerator boards with FAST RAM were available people would have upgraded to a 14MHZ 68k/68010 with 512K fast ??

[Megalomaniac]

Originally Posted by Bruce Abbott

But that's the way it was with the PC vs Amiga. Everybody compared the latest model PC with all bells and whistles to a stock Amiga 500 - not an A2000 with 25MHz 030 or an A3000 with its 32 bit ChipRAM, and certainly not any 3rd party addons. Sadly this was justified because only the stock A500 had a large enough userbase to attract developers.

'Everybody' includes most of the Amiga scene too, sadly. When you read reviews of those games that were designed to need more than a 68000, reviewers for games-based mags rarely mention that, and almost never factored it into the score, even if the box recommended a faster Amiga - although hard disks seemed to be factored into the scoring. Similarly, A1200 games were almost always rated based on their 020 2Mb Chip no fastRAM performance.

With hindsight, that was frustratingly short-sighted, although it remained expensive to accelerate an A500 for a bit too long, the GVP A520 in mid-1992 was a bit of a gamechanger but was still £750 (with a hard drive an 50Mhz processor). The whole thing feels quite chicken-and-egg though, maybe if the Amiga games mags had said "this is slow on a 68000, but on a 68020 it's fast enough to be playable, and on an 030 it really flies - and if it sells well and you fill in the reference cards to say you have an 030, who knows what else they'll make for us?".

The bigger difference would have been an affordable model with a faster processor out of the box, at least a year before the A1200, to prevent the drift of those players towards PCs. Maybe instead of the A1500 we could have had an A500-form-factor 030 and more memory ECS system for say £700, maybe with 2 floppy drives?

I know its contentious, but I honestly think the A600 sent the wrong message to especially fans of flight sims, with the missing keypad and lack of accelearatability (if that's a word) in its heyday. The six months were it was that or a clearance A500+ must have been offputting to fans of those games (who were big buyers of legal commercial games, flight sims always charted well even if they looked uncommercial - A320 Airbus was number 2 in Amiga Format's charts the month before the A600 launched)

[sokolovic]

Originally Posted by redblade

Maybe if more piggy back Accelerator boards with FAST RAM were available people would have upgraded to a 14MHZ 68k/68010 with 512K fast ??

Upgrading wasn't pushed at all by Commodore, the Amiga press and, most importantly, by the vast majority of the developpers in the Amiga market.

[oscar_ates]

Originally Posted by gimbal

Come on dude. This hurts. The rules of this thread were re-quoted on this page.

Oops sorry! Jumped too quick on title to answer. On topic, pc games really took off after 1993 when Doom was released. After that Amiga went nose down to ground. Before 1993, Amiga was quite relevant still against PC.

[gimbal]

Originally Posted by Aardvark

Youtube videos of Caveman Ninja are a bit misleading. I did a quick test on various PCem configs and steady 60fps gameplay requires a 486DX2 66MHz which wasn't even available at the time, and with a 386DX 33MHz it starts to get quite choppy.

You did not need a bloody DX2 66 to run Caveman Ninja smoothly, on such a machine you could run Duke 3D for peet's sake. Using an emulator to conjure up numbers :/ If a bloody 2D platformer ran choppy on anything above a 386 SX, you had a bottom of the barrel video card with slow as molasses video memory.

[Aardvark]

Originally Posted by gimbal

You did not need a bloody DX2 66 to run Caveman Ninja smoothly, on such a machine you could run Duke 3D for peet's sake. Using an emulator to conjure up numbers :/ If a bloody 2D platformer ran choppy on anything above a 386 SX, you had a bottom of the barrel video card with slow as molasses video memory.

PCem is accurate for these kind of tests, and I used fast video card in my configs.

Never said it's unplayable with 386, but to reach steady 60fps like shown in youtube videos requires at least DX2 66mhz which wasn't available in 1991.

[TCD]

I think the 'problem' is that in 1991 'smooth' didn't mean 60 FPS. One video about Doom's performance sets the bar between 20 to 25 FPS. Of course that's subjective, but I think the goalpost for 'smooth' is more like 30 FPS which might help put things into perspective.