Test System:
Quicksilver 2002
7448 @ 2.0Ghz
Radeon 8500LE (275Mhz, 64MB)

Using parameters by Piru for benchmarking the driver:
setenv TGLSYNC 0
flushlib tinygl.library

And under Quake 3 (All details maxed out):
r_primitives 2
timedemo 1
demo four

Results:
640x480 - 130.0 FPS
800x600 - 108.6 FPS
1024x768 - 83.5 FPS
1280x1024 - 64.2 FPS

> Quicksilver 2002 [...] 2.0Ghz
> Radeon 8500LE [...]
> 800x600 - 108.6 FPS

Mac mini 1.5 GHz
Radeon 9200
800x600 - 140 FPS

What gives?

Same Quake?

Never a 140 FPS. On my Mini 1.5 i get like:

640x480 - 76,6 FPS
800x600 - 59,4 FPS
1024x768 - 44,8 FPS

Andreas_Wolf,
Quote:

Hmm,. got me... Are you using the executable from MorphOS Files? I think that is the one I am using, but I will check later.

It could also be possible that this is a result of the poor Quicksilver RAM benchmarks? MEMTEST gives Quicksilvers very poor RAM speed.

> Are you using the executable from MorphOS Files?

The 140 FPS at 800x600 resolution is what was reported in 2010 by visitors of a show where the new R200 3D drivers were presented in action. Just click on the text of my previous posting and you should know more.

> It could also be possible that this is a result of the poor
> Quicksilver RAM benchmarks?

The 60 FPS result for 800x600 on Mac mini reported by koszer (as well as the 70 FPS reported by Cim on Mac mini in the other thread) doesn't make this appear likely.

Quote:

Andreas, I know for a fact that you in at least one other thread have seen the response that for Quake 3 to use the optimal rendering path, you need to type /r_primitives 2 in the Quake 3 console. If you don't do that, Quake 3 will pick a rendering path that is suboptimal. You know this already, it has been said to you several times. Yet you either choose to ignore it, which makes you a troll, or you're incapable of comprehending said fact, which makes you retarded. Which is it?

your nickname choice is kinda retarded.

Close to locking the thread. Watch it.

I've just looked into my q3cfg and here's what I've found:

Quote:

What do you know? Looks like I don't really need to set it though the console every time I do the benchmark (did you really think I haven't done that in the first place? Moreover - suprise, suprise - redrumloa did just the same, still no 140 FPS though).

Edit: I've managed to get 78,5 FPS by tweaking some in-game prefs (haven't touched the .cfg file). But still, 78,5 is not 140...

Edit 2: even more tweaking with .cfg file gave me 102 FPS (no gun, no shadows, no gibs, no gun/ammo 3d models...). All right, maybe 140 is doable, but at a (great) cost.

[ Edited by koszer 11.06.2012 - 16:20 ]

> you in at least one other thread have seen the response that for
> Quake 3 to use the optimal rendering path, you need to type
> /r_primitives 2 in the Quake 3 console.

That's correct.

> If you don't do that, Quake 3 will pick a rendering path that is suboptimal.

Yes, that's why redrumloa did exactly that I guess. Quote from posting #1:

"And under Quake 3 (All details maxed out): r_primitives 2"

As you can see, the comparison is with this option enabled in both cases.

> You know this already

True.

> it has been said to you several times.

No, it was only once:

https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=8500&forum=11&start=7

> Yet you either choose to ignore it, which makes you a troll, or
> you're incapable of comprehending said fact, which makes you
> retarded. Which is it?

None of that, as I neither ignored it nor non-comprehended it, which everybody can easily see who has at least basic reading comprehension skills. So maybe you want to ask your question to yourself?

> redrumloa did just the same, still no 140 FPS though

And that's with both faster CPU and faster GPU.

> even more tweaking with .cfg file gave me 102 FPS (no gun, no
> shadows, no gibs, no gun/ammo 3d models...). All right, maybe 140 is
> doable, but at a (great) cost.

Incidentally, said presentation of Quake 3 running with alleged 140 FPS has been recorded on video:

https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=7464&forum=11&start=15
https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=7464&forum=11&start=21

The quality is rather substandard, but maybe you can recognize from one of the recordings which of the tweakings that you just did may or may not have been applied there.

I did a lot of benchmarking of Quake III in last two days and here are my results:
All tests were done on Morphos 3.0 with TGLSYNC 0 and r_primitives 2, all settings maxed out (32 bit textures and rendering, max textures and geometry, trilinear filtering):

Mac Mini 1.83Ghz Radeon 9200 64MB:
512*384 104,8 FPS
640*480 75,4 FPS
800*600 60,0 FPS
1024*768 42,7 FPS
1280*1024 30,5 FPS

Mac Mini 1.5GHz Radeon 9200 32MB
512*384 93,0 FPS
640*480 68,9 FPS
800*600 52,6 FPS
1024*768 28,8 FPS
1280*1024 4,4 FPS
On this Mac I was expecting better performance in 1280*1024 when it gets out of memory (because of new driver and swap mechanism) but actually it was worse than with MorphOS 2.7. Also 1024*768 was worse.

And finaly results from Mac OS X port of IOQuake III 1.36 on Mac Mini 1.83GHz Radeon 9200 64MB:
512*384 123,0 FPS
640*480 116,8 FPS
800*600 90,3 FPS
1024*768 60,8 FPS
1280*1024 40,6 FPS

I also noticed that I can't use 320*240 screenmode in Morphos 3.0 (I get black screen) while it works in 2.7.

Actually I could achieve over 140FPS but in 640*480 or lower on lowest possible settings. I would really like to know how someone managed that on Mac Mini in 800*600...

Andreas_Wolf,
Quote:

So let met get this straight... You're complaining that some random third party using a different computer with a different graphics card using different Quake 3 settings on the MorphOS 3.0 release gets different results than a third party report of a demonstration of a development version of MorphOS 2 years ago? And you're feeling so strongly about the issue that you seem to think it's a good idea to spam every single thread on MorphZone with your nonsensical complaints?

> You're complaining

No, I'm merely curious.

> that some random third party

The result is what counts, not who is reporting the result.

> using a different computer

No, identical computer in koszer's case, and better computers in redrumloa's and Cim's cases.

> with a different graphics card

No, identical graphics card in Cim's and koszer's cases, and better graphics card in redrumloa's case.

> using different Quake 3 settings

Funny "counter argument", as even with the lowest possible settings the claimed 140 FPS cannot be reached apparently. Trying to mimic the settings applied during the 2010 presentation can thus only lead to even worse results, don't you think?

> on the MorphOS 3.0 release

Yes, as that's the one said to come with the improved R200 3D driver which was shown back in 2010.

> gets different results

Much lower results actually.

> than a third party report

The result is what counts, not who is reporting the result.

> of a demonstration of a development version of MorphOS 2 years ago?

One would think that during those 2 years of further development the driver has become even better, or kept the same at least, not become worse, right?

> you're feeling so strongly about the issue

As said, I'm merely curious as to where the performance presented 2 years ago has gone. Do you know it? I'm looking forward to your valuable contribution to this thread's topic. Maybe with your help we can solve the mystery.

> that you seem to think it's a good idea to spam every single thread on MorphZone with
> your nonsensical complaints?

Again, no complaints from my side at all, just being curious. And looking at the outcome of this thread so far, my curiosity has been anything but nonsensical, quite contrary to any of your "contributions" to this thread so far. Besides, this "every single thread" you're fantasizing about is actually exactly 2 (verbal two) threads where I was showing my curiosity about the matter at hand, the one being this very "MorphOS 3.0 Quake III benchmarks" thread and the other being a thread where in posting #2 the poster decided to talk about the Quake 3 performance he's getting with the improved R200 3D driver.

> So let met get this straight...

As shown above, getting things straight is obviously not your finest skill ;-)

From what I've spotted on the video - the gun, 3d ammo/weapon models and shadows are on. I will make some more benchmarks to compare with my previous ones (this time with Q1 and 2, so we could see what kind of speed boost in 3d we've got with 3.0), but it seems that the "improved 3d drivers" we could see on this video are still ahead of us.

I'm guessing not everything made it into the 3.0 release. Oh well, maybe it will make the 3.1 release. I do see an improvement, especially in Payback which I still play all the time. There is no longing for the WOS executable any longer as the 68K executable plays great with all bells and whistles in high res.

> From what I've spotted on the video - the gun, 3d ammo/weapon models and
> shadows are on. [...] it seems that the "improved 3d drivers" we could see on
> this video are still ahead of us.

Thanks for your investigations. Now I'm curious (still not complaining, hehe) as to what might have been the reasons for holding back the driver that was shown on the presentation and if we might get it with a MorphOS 3.x update.

redrumloa,
Quote:

When i start Payback, it just displays a nearly white screen and nothing happens after.
What did do to run Payback?

eliot,

Do you have an original CD or the demo? If you have an original, have you applied the updates? Use payback setup to select select a resolution supported by your monitor. I am currently using hardware rendering with a screenmode of 1024x768x24b. I am using the standard (68k) executable and not the WOS executable.

[ Edited by redrumloa 11.06.2012 - 14:34 ]

And now, something completely different: some more benchmarks:

Quake I:
640x480, 32 bit, windowed:
2.7: 117,5 FPS
3.0: 143,5 FPS

1024x768, 32 bit, windowed:
2.7: 52,1 FPS
3.0: 67,8 FPS

1024x768, 32 bit, fullscreen:
2.7: 60,1 FPS
3.0: 60,1 FPS

GLQuake running in fullscreen mode seems to ignore the "TGLSYNC 0" so the same FPS count in the last run is no suprise. In overall, current 3D drivers give us something like 20-30% more FPS than the older ones. Not bad, if I may say so, but I'd still prefer that 100% more

Andreas, I know you like to blah blah every criticism of your crap to death, so I'm simply gonna conclude that you choose option 1, that you're trolling. I'm guessing that it is more likely than that you're legally retarded.

You get pointed out that you're crying over the FPS values from two different computers, with two different graphics cards, with different Quake 3 settings, with versions of operating systems that differ by about 2 years, don't reach exactly the same FPS value, but are different by about 8%. I think you're a lost cause, but could you please start a blog where you can keep crying over it non-stop, please? Anything to keep your spamming off MorphZone at least.

As far as I'm concerned:

1. The computers were the same (Mac Mini 1.5 GHz)
2. The graphic cards were the same (Radeon 9200 64 MB)
3. The Quake settings were pretty much the same (no FPS-tweaking with .cfg file)
4. The difference is more like 100%, not 8% (140 FPS versus 70)

In case you're referring to the post #2 - it seems to me that Andreas was curious why a faster computer (faster CPU and much better gfx card) performs slower than expected. Soon (post #4) it turned out that the Mini doesn't live up to expectations too.

> Andreas, I know you like to blah blah every criticism of your crap to death

The only one uttering blah blah in this thread has been you and no one else. Your blah blah has been accompanied by false statements, intentional false statements (i.e. lies), misconceptions and ad hominem attacks.

> so I'm simply gonna conclude that you choose option 1, that you're trolling.

And so you're concluding wrong, as has been usual for you in this thread.

> I'm guessing that it is more likely than that you're legally retarded.

You still fail to see that my curiosity is neither trolling nor retardation. Pity.

> You get pointed out that you're crying

And you got pointed out that I'm neither crying nor complaining.

> over the FPS values from two different computers

No, identical computer in koszer's case, and better computers in redrumloa's and Cim's cases.

> with two different graphics cards

No, identical graphics card in Cim's and koszer's cases, and better graphics card in redrumloa's case.

> with different Quake 3 settings

Funny "counter argument", as even with the lowest possible settings the claimed 140 FPS cannot be reached apparently. Trying to mimic the settings applied during the 2010 presentation can thus only lead to even worse results, don't you think?

> with versions of operating systems that differ by about 2 years

One would think that during those 2 years of further development the driver has become even better, or kept the same at least, not become worse, right?

> don't reach exactly the same FPS value, but are different by about 8%.

2010 result: 140 FPS
2012 result (as reported by koszer): 78.5 FPS

According to the math I know, 78.5 is 44% less than 140, and 140 is 78% more than 78.5. How did you arrive at your obscure "8%" figure?

> could you please start a blog where you can keep crying over it non-stop, please?

I'm not interesting in starting a blog, and even less so in starting to cry over this matter.

> Anything to keep your spamming off MorphZone at least.

Let me suggest you then to simply ignore my postings and keep to threads you're actually able and willing to contribute something substantial to. This thread is apparently none of those.

Can someone just ban this Wolf_Andreas guy so my thread doesn't get locked?