Voodoo 3 3000 -vs- Voodoo 4 4500, strange

[redrumloa]

Strange results. P96 speed shows bizzare ultra terrible benchmark results for the Voodoo 4 4500 and good results for the Voodoo 3 3000. However in Quake II the Voodoo 4 4500 wins in all resolutions.

I realise P96speed is mostly if not totally aimed at 2D, but the results for the V4 is just absurd. I'd think results that bad would be seen in the 'seat of your pants' feeling. The results are like 50X slower in some areas.

Hmm.. After running the last tests I feel the need to dispell Tjoaz. Hmm.. new post or news item? I think news..

[Karlos]

Since the graphics subsystem of amiga os is not fully unified (we have different, unrelated systems for 2D and 3D), I'm not really that surprised that things like this are possible.

The only 2D aspect I can think of that would affect 3D games would be blitting speed (for windowed displays) and screenbuffer switch times (for fullscreen displays).I very much doubt that there would be any significant difference between the two cards for these jobs. Blitting and framebuffer switching tend to be more than fast enough on any half decent graphics card ;-)

[redrumloa]

@Karlos

I should have run these benchmarks long ago. Tjoaz loves to jump in every Prometheus thread spreading old benchmark comparisons from Prometheus' Alpha drivers days on a system with A3640. Those numbers are so far off it isn't funny. Comparing my numbers to numbers submitted to Amigaspeed.de, it is fair to say the prometheus/Voodoo is the FASTEST gfx card solution on the market. My benchmarks beat Mediator in every single category by a margin of as much as 300%! If I compare it to his old numbers, which would be unfair but along the same level of him, it'd be beating it by a margin of 1000%.

[Karlos]

It's RTG Jim, but not as we know it :-D

[adolescent]

@red

I don't think it's fair to say that.  The methods used on Amigaspeed.de are dated, and un-scientific.  The only way to have an accurate comparison is to do the tests on identical systems (ie. same software, patches, memory, etc.   The only variable should be the PCI solution).  I'd guess there would be nominal differences.  (Actually, I'd guess that the G-Rex would be a little quicker).

[redrumloa]

@adolescent

Well I need to get these numbers out there. There really is not alot of comprehensive Prometheus benchmarks on the net. There's a certain someone who loves pointing to highly biased, old and innacurate information. I'll but out my numbers, and I welcome new benchmarks from the other side.

[Ilwrath]

The methods used on Amigaspeed.de are dated, and un-scientific. The only way to have an accurate comparison is to do the tests on identical systems (ie. same software, patches, memory, etc. The only variable should be the PCI solution)

Yeah... And even then, why so much emphasis on 3d performance?   I mean, I'm interested in a Prometheus, and have never even bothered to install Warp3D as it's basically useless except if you want to play ports of 5-7 year old games at framerates that only look when compared to 5-7 year old x86's.

I'd really like to see some fairly good apples-to-apples comparisons of 2D performance vs. both Mediator and a Z3 graphics card for reference.  The same for network card performance would be cool, too.

I think that would finally put the FUD to rest.

[redrumloa]

And even then, why so much emphasis on 3d performance?

P96Speed is a 2D benchmarking program.

useless except if you want to play ports of 5-7 year old games at framerates that only look when compared to 5-7 year old x86's.

I dunno, Quake II here makes a pretty playable game. I really don't think a 7 year old PC could touch it.

and a Z3 graphics card for reference

You have one handy? Run P96speed and post the results for comparison.

[Ilwrath]

P96Speed is a 2D benchmarking program.

Yes, but none of the P96 tests that I saw on Amigaspeed.de.vu even mentioned the Prometheus?

I dunno, Quake II here makes a pretty playable game. I really don't think a 7 year old PC could touch it.

You forget that the Pentium II 300mhz was released over 7 years ago.  (May 7, 1997)  Toss that Voodoo card in with a P2/300 and an Intel 440LX chipset (available August 1997 -- so ALMOST 7 years), and I bet it'll match or probably even beat the Amiga.  :-(  

And who would use that rig, anymore?  I mean, charitable organizations won't even take them as donations!  In comparison to anything resembling a modern machine, the Amiga has virtually no 3D power, with or without a Voodoo card.  Why try to press something that the Amiga just isn't suited for?

You have one handy? Run P96speed and post the results for comparison.

Which version are you using?  I have a module created for my baseline A4000 reference.  (A4000 Cyberstorm Mk2 CV64/3D)  

If it's the right version, I'll e-mail it to you.  :-)  (If not, let me build a stock workbench, and build one -- I wouldn't want to use my normal workbench, as it's a little corrupted/flaky after a nasty disk corruption.)

I'd also happily donate a baseline A2065 Network baseline.  Just supply me with the test.  :-)  

I'm all for finding actual performace facts.

[redrumloa]

Yes, but none of the P96 tests that I saw on Amigaspeed.de.vu even mentioned the Prometheus?

I guess no one submitted one :-?

Toss that Voodoo card in with a P2/300 and an Intel 440LX chipset (available August 1997 -- so ALMOST 7 years), and I bet it'll match or probably even beat the Amiga.

Possible, go dig up some benchmarks and let's find out:-D

Amiga has virtually no 3D power, with or without a Voodoo card.

Uuuh Amiga is lacking titles, but don't tell my system it doesn't have 3d. It would get upset;-) Playing Quake II with a solid 33FPS is no slouch and does not handicap someone playing against a PC.

Which version are you using?

1.2

If it's the right version, I'll e-mail it to you.

Post it here!

I'd also happily donate a baseline A2065 Network baseline. Just supply me with the test.

I don't know a test off hand :-?

[Ilwrath]

on pathetic framerates:

Possible, go dig up some benchmarks and let's find out

The RIVA 128 AGP on a 300mhz P2 with 440LX chipset pulls 49 fps in Quake II GL. Source: Tom's Hardware -- December 1997 (And if Tom's had a review board back then, it's a good chance he bought it from the local store.  His website was hardly known.)  So if we go by your 33fps, a nearly 7 year old computer bests the Amiga by 16 FPS, or 48%.

Why would this be a selling point?  It's like trying to sell a new ultra-cool fuel-efficient hybrid car by advertising that it can "accelerate almost as fast as a 15 year old Yugo" instead of playing up the environmentally friendly aspects.  

Contrary to what modern PC benchmarking entails, I think there are performance numbers more useful than 3D benchmarks.  Especially in cases like this, where the 3D power is so woefully behind modern systems.  

Playing Quake II with a solid 33FPS is no slouch and does not handicap someone playing against a PC.

Well, it kind of does.  The PC user can crank some resolution and pick off the Amiga user before the Amiga user can even pick out the PC player from the blotchy wall texture.

And, outside of Amiga users, does anyone play Quake II anymore?  The FPS genre doesn't really have "nostalgia."  I mean, no one wants to go back to the "good ole days" of shooting at poorly formed, low polygon objects with blotchy textures and little to no atmospheric effects.  I don't think anyone really chooses to play Quake II over Quake III, do they?  Especially as the multi-player game goes, Quake III isn't so much as sequel as it is a drop-in replacement.

on P96 benchmarks:
I'm feeling a bit of insomnia.  I'll run the set tonight.  (I can't find my P96 module right now, and I want to post this before my session times out...)

[adz]

I dunno, Quake II here makes a pretty playable game. I really don't think a 7 year old PC could touch it.

I wouldn't go as far as saying that, I've got a P200 thats at least 7 years old that can play Unreal/Quake II in high res and high detail with little or no framing. Mind you the only reason I can do that is because the machine is running dual voodoo II's.

[Insanity]

I've got a p2/300 on a 440BX(I think) using a Voodoo 3 3000 AGP (I think it has 128 MB of ram, dunno about in what configuration though) . It has w2k installed.

Do you want me to install q2 and run some tests?

If so, then I want settings and all the specifics for the tests that are to be run.

[zipper]

> Strange results. P96 speed shows bizzare ultra terrible benchmark results for the Voodoo 4 4500 and good results for the Voodoo 3 3000. However in Quake II the Voodoo 4 4500 wins in all resolutions.

Easy explanation: Voodoo4/5 "support" was added to P96 about May -03 but is highly unoptimized and has not been updated ever since. I don't know if the hardware acceleration works at all - the difference between V3000 and V4500 is staggering. In 3D they are about as fast - no wonder as the clock is the same and the better hw in 4500 is not utilized.

[redrumloa]

Especially in cases like this, where the 3D power is so woefully behind modern systems.

 I emplore you to fly to Florida and play Quake II on my system and say it's not playable, or you'd be at a serious handicap against a PC. That's not the point though.

I really think you are missing the point here. Most people using classic Amigas aren't trying to beat modern PCs speed in 3D games. Why shouldn't a classic Amiga user get the most out of his box though? If it was all about benchmarks between PCs and Amigas, there wouldn't be a single classic user left in the world. Why do people restore classic/antique cars? Why do people customize classic/antique cars? They aren't as fast as modern cars, get much worse gas mileage, handles far worse and often it costs alot more to restore an old classic then it is to buy a new production car.

3D is done because it can be done, and it does give the possibilty to play some modernish games. With SDL and hopefully other ports, the possibilities grow even larger. I am looking forward to playing some original Duke 3D on this system, and I hope Hyperion realses some more ports in the future when OS4 is out. I will keep tweaking my system to ring out every ounce of performance possible:-)