Forums | MacLife

More later maybe, but a couple things before I'm off:

To test platform/CPU power, bench at low-medium res, not high. High-rez tests put the strain on the graphics subsystem. 640x480 sometimes overstresses half-baked drivers.

Altivec helps mainly FP ops, and sound is basically integer. It can help with physics. AI is likewise mostly INT.

Oh, TG posted while I wrote...

Of course I looked at it. In some depth. Are you aware that hardware T&L is basically exclusive with SIMDsets, and QIII uses hardware transform, but not lighting accel? And Devine worked a good year on his optimizing. I'm wondering what an extra year could've gotten out of HaloPC (PS2.0b/3.0 paths, etc) with only Sean Green working on it. Throws comparisons into something of a cocked hat, especially at 16*12.

I find your excess of faith disturbing.

dj phat 2000 wrote:

...
I realize there is only so much magic that Altivec can do but, it seems like it goes wasted.  And only that one game ever really used it.  And it helped a decent amount.

Once more, the NASCAR sim used Altivec. Framerates and sys reqs were similar.

SIMDsets like Alti could've helped more when T&L was done in the CPU. Now that that's usually offloaded to the grfx card, it's useful mainly for physics, perhaps a couple of other things. It was good, not magic. The 'Velocity Engine' moniker was RDF that overly raised expectations. Alti is powerful, but somewhat hard to use well. That may be why it isn't used more. It took Devine a long time to get those gains, perhaps because it was mainly good for lighting and physics, not transform ops.

Later.

Just from the looks of things though.  Even if we had faster CPU's (clock speed wise), it would still be slower to game on a Mac.  With something like a 4GHz G5 and 2GHz FSB's, with PCI-E graphics.  It would still be slower.  While it would play games great.  It would not hold up to that of the PC side with similar specs.  I realize now that Altivec is not really gonna help here but, at the times when it could have made a difference, it didn't get used.  And if only one guy was optimizing the code to run on it like Devine did.  What would happen if a team of guys were on it?  I guess my thinking is not correct but, it makes sense to me that it could have been MUCH more useful at that time.  Now a days it seems that when people complained that the FSB on the G4's were way too slow to feed a graphics card and what not.  As we now have WICKED fast FSB's up to 1.25Ghz (Dual).  The PC is still gets faster Frames.  We have the same video cards as well.  Al-bet a bit slower then the PC side clocks them at.  However, it is not like a 5 FPS difference.  More like 15 to 30 or more.   

Another thing that is definitely bugging me is that the performance for ATi and Nvidia cards on the Mac are reversed as to what the PC side gets.  The PC side see's higher performance from Nvidia cards then ATi cards, while ATi cards kick Nvidia's butt on the Mac side.  I find that strange really.  I know we don't use Direct X, however does that mean that Open GL is that much better on ATi cards then Nvidia cards?  If so then why does DIII run way better on Nvidia on the PC then ATi does?  I thought that was an Open GL based game.  Will that translate to the Mac side too?  (Once again looking for benchmarks on the game   )

I don't know.  just rambling........

Bat wrote:

Your bank's windows are still intact!

Bat is referring to something I said elsewhere.  "When Apple releases a 3Ghz G5 with PCI-e I will spend 3,000 dollars so fast the windows in my bank will implode"

The Cat's gone nuts!

With dual-core chips on the horizon from both Intel and AMD, there likely will be more optimization for SMP in the future.

The problem is that the "megahertz myth" isn't entirely a myth.  It does matter when it come to games.  The G4 just couldn't keep up.  The improved architecture of the G5 helps, but the only consumer machine with a G5 is the iMac - and that's hamstrung with a low-end GPU.  The cheapest G5 tower (which is essentially an iMac G5 in a larger case) is $1499 - and that's still with the same low-end card.  The only passable game rigs are the DP G5s - and they're too far out of reach for most consumers.  Until Apple can get some G5 muscle and a good GPU in a consumer machine for a reasonable price, the Mac game market isn't going to take off.  Even then, I can buy and Xbox and Doom 3 combined for less than the cost of just a video card upgrade on the Mac.  If you're using the machine for other things that require a fast GPU, great - but just for gaming?  Hard to justify the expense.

I do wonder of using a dedicated sound card like the Revolution 7.1 helps at all with performance.  If so, might be time to try and get Creative Labs to put out a Mac version of their products again.

Doom 3 looks cool, but it's going to cost a lot of cash to get a Mac to run it.  On the other hand, an Athlon64 with a mid-range video card (like the 6600GT) can run Doom 3 just fine and at half the cost.

I do appreciate the work that Mac game developers go through and I'm sure they do everything they can.  Apple just needs to get capable hardware in the hands of consumers.

Whoohoooo.  Well, I feel better again.  Thanks Glenda!!! 

GlendaAdams

Today, 03:33 PM

"There will be benchmarks galore on various web sites over the coming weeks...

As far as Altivec, how much does it help- I don't have a hard number, since we haven't built a non-altivec version in several months. Its faster than without Altivec, but we're not getting double the speed or anything like that. It helps a few percentage points, since its mostly for sound and physics.

Most of the application level optimizations for Doom3 are also like this- they gain a few percentage points here and there. The engine is well engineered and optimized to start with, so it is feeding data to OpenGL about as fast as the drivers/cards can handle it. The PowerPC architecture does have some inherent overhead issues (float to int conversion, data alignment problems) that can't be optimized away, so most of the app level things we did were spot fixes.

Glenda"

"One other quick answer about network play - Doom3 Mac has the same net play as the PC, and plays between the PC & the Mac."   

Still would like to know if there is a way to see how much like really does Altivec help a game these days.  Just wondering if it will be something that will almost get phased out due to it not really being able to do anything other then maybe sound processing.  Games will push GPU's more and more I am sure, so..  Altivec looks, at least to me as being almost a non relevant instruction set.  Well, for gaming that is.  Unless the Physics of a game can be helped by it.  But, then wouldn't having a 64 Bit CPU (math) be able to help as well? Just wondering. 

I will be buying the game though.

Also, she referred to inherent overhead issues of the PPC architecture.   I didn't know there were any, lol. 
Could one of you kind lads explain what she means.  Like, how bad is that problem she described.  I mean, the G5 has 2 float and 2 interger units right?  I thought that was a REALLY good THING.  But, I am sure I am off on what it really means as to what those units do. 

Thanks...

Last edited by dj phat 2000 (2005-03-02 10:53 pm)

benches: http://www.macworld.com/news/2005/03/02/doom3/index.php

Not great, but decent. 20fps for us non G5ers.

Running the same demo at 640 x 480 with 3D rendering turned off -- which can be done by setting the r_skiprender value to 1 in the Doom 3 console, then running the timedemo -- returns an FPS average of better than 106 frames per second. Setting that command gets all the 3D stuff out of the way, telling us how fast the core game engine itself is running. And that number compares pretty well from Macs to PCs. That implies that the performance difference we're seeing has something to do with the way the 3D graphics are rendered, as opposed to any specific shortcoming in the game engine or overall slowdown that would account for a difference.


So where does this performance drop come from? Some of it, at least, can be explained by the way the Mac works. Mac game developers agree that it's much harder to get the Mac to pay attention exclusively to a game, even if it's the only application running, than it is on a PC running Windows XP. That has some payoffs -- you can play this game in a window, for example, seeing only a minor performance drop -- because "OpenGL's tendrils run much deeper" in the Mac OS X kernel than they do in Windows, according to one source with whom we spoke.


Will further driver optimization help? It's entirely possible, although it's hard to say how much. We expect such improvements will be iterative and gradual. But if Quake 3 Arena's history is any indicator, we'll see continued improvements in both driver speed and graphics chip speed until both Macs and PCs are on a level playing field. At least until the next great game engine comes along.

Last edited by mo' ron (2005-03-03 12:15 am)

And those benchmarks are on a computer costing over three grand.  Imagine how "well" consumer Macs and lesser PowerMacs will run it...

In another forum Peter Cohen has said not to be discouraged, the demo he ran put the greatest amount of strain on the hardware as the game could offer.  He reports that during actual game play it rarely dropped below 40fps.  The entire game is capped at 60fps

No. Only in benchmark mode will it run faster than 60.

I am a bit concerned about this. It may mostly run above 40, but the dips are liable to be when you least want them, e.g. during firefights. I hear there's a lot of time spent walking- sneaking- around.

Still, these are early benchmarks, and performance during gameplay on lesser machines remains to be seen. Driver tweaks will improve matters over time.

I better go look for Peter's remarks.

Oh, and the game is said to look quite good even in low-res, like 640x480.