OK, I'll add mine into the mix. 19.4fps/21.6fps for the two runs, as suggested on barefeats. I ran command 'timedemo demo 1 1', all details turned off at 640x480 - no, not at the defaults, I went in and turned all the snazzy stuff off. My machine is a Powerbook rev.

• Doom 3 Timedemo
• Doom 3 Download Pc

1.33GHz G4/256MB/FX5200/10.3.8 Much, much better than expected. Considering I'm below both the minimum CPU and RAM totals, using the FX5200.

1. Sadly didn't realize this until after buying the game and trying to figure out why timedemo wasn't working, and was also trying to get RBDOOM-3-BFG to run on the NVIDIA Jetson TK1. Michaellarabel commented May 24, 2014.
2. Doom 3 BFG Edition & More 0 If you’re like me, than you were a happy man on October 16th, 2012 as Id Software re-released an enhanced and updated version of the shooter classic from 2004 Doom 3.

We've posted the definitive guide to benchmarking with DOOM 3 which includes information. Now we are ready to create the actual timedemo. Load up Doom 3 and fire.

Ran it in a window so I could see my MenuMeters performance; the CPU was not maxed out, but I was disk swapping. I played through the first portion of the game, it was a little jerky but playable - again, down to the lack of memory.

So, in conclusion. This game will certainly run on your older machines, just deal with low details. Can't wait to get another 1G stick in this thing. OK, I'll add mine into the mix. 19.4fps/21.6fps for the two runs, as suggested on barefeats.

I ran command 'timedemo demo 1 1', all details turned off at 640x480 - no, not at the defaults, I went in and turned all the snazzy stuff off. My machine is a Powerbook rev. 1.33GHz G4/256MB/FX5200/10.3.8 Much, much better than expected. Considering I'm below both the minimum CPU and RAM totals, using the FX5200. Ran it in a window so I could see my MenuMeters performance; the CPU was not maxed out, but I was disk swapping. I played through the first portion of the game, it was a little jerky but playable - again, down to the lack of memory.

So, in conclusion. This game will certainly run on your older machines, just deal with low details. Can't wait to get another 1G stick in this thing. Thanks for showing more of the PC numbers so that people could see. This game is normally very GPU-limited.but clearly the dual 2.0 with RAID 0 still isn't getting past the CPU limitation.that's why the FPS is still pretty low on low resolutions.I don't think the OpenGL implimentation is the biggest hurdle anymore.it looks like the G5 is the slow-down.ugh. As you can see from the PC numbers, at 640x480, the game should be FLYING with a fast processor and good graphics card.there is a difference between that and 1600x1200 on the Mac, but not nearly as much as there should be.Which tells me that at low resolutions, the G5's graphics card.at least when using a 6800.is waiting around for things to do. Thanks for showing more of the PC numbers so that people could see.

This game is normally very GPU-limited.but clearly the dual 2.0 with RAID 0 still isn't getting past the CPU limitation.that's why the FPS is still pretty low on low resolutions.I don't think the OpenGL implimentation is the biggest hurdle anymore.it looks like the G5 is the slow-down.ugh. As you can see from the PC numbers, at 640x480, the game should be FLYING with a fast processor and good graphics card.there is a difference between that and 1600x1200 on the Mac, but not nearly as much as there should be.Which tells me that at low resolutions, the G5's graphics card.at least when using a 6800.is waiting around for things to do. Click to expand.I do agree with you that at lower resolutions such as 640x480 the game rather utlizes the main cpu than the gpu but somehow I have a feeling that the game wasnt G5 optimized at all or if it was then they need to hire better programmers. Anyway, one example of a game that quickly entered my mind when I saw theses benchmarks was like is this a Quake 3 deja vu again? Quake 3 back in its day was the most sophisticated graphics engine of its time.

Anyway, the first version of that game for a mac the frame rates were definately lagging behind the PC counterparts. But through better programming after a few patches that same game on a Mac suddenly gave PC counterparts a real run for its money where the top high end Macs with top high end GPUs were actually giving faster frame rates than high end PCs. So I will make this short, graphics drivers are one thing but optimizations are another and somehow I know that Doom developers didnt put too much heart in developing for the Mac version simply because it costs money to really do so and knowing that Apple has a small market share to begin with they just wanted to get rich the easiest possible way. I do agree with you that at lower resolutions such as 640x480 the game rather utlizes the main cpu than the gpu but somehow I have a feeling that the game wasnt G5 optimized at all or if it was then they need to hire better programmers. Anyway, one example of a game that quickly entered my mind when I saw theses benchmarks was like is this a Quake 3 deja vu again? Quake 3 back in its day was the most sophisticated graphics engine of its time. Anyway, the first version of that game for a mac the frame rates were definately lagging behind the PC counterparts.

But through better programming after a few patches that same game on a Mac suddenly gave PC counterparts a real run for its money where the top high end Macs with top high end GPUs were actually giving faster frame rates than high end PCs. So I will make this short, graphics drivers are one thing but optimizations are another and somehow I know that Doom developers didnt put too much heart in developing for the Mac version simply because it costs money to really do so and knowing that Apple has a small market share to begin with they just wanted to get rich the easiest possible way. The problem with the quake 3 example is that by the time Macs were as fast as PCs in Quake 3, PCs had moved on and added all of the shader-based graphics stuff that you see today, and the engines (even the Q3 engine) that were in use relied on things that Macs still don't do very well. An example of this that PC gamers will understand: 3D Mark 2001 has been a very useful benchmarking tool for years, but as time goes on, it becomes less and less relevant.

When you are averaging 200 fps in most of the tests, then the results become less important, because you don't care if you are getting 200 or 240 fps.you can't see the difference. Let's say there are 2 machines that you can buy. One of them gets 24,000 3dMark01 points and 2,000 3dMark05 points. The other one gets only 20,000 3dMark01 points, but gets 3,500 3dMark05 points. If you play Quake 3, then the first computer will be of more interest to you. If you play doom 3 or HL2, then you'll be wanting the 2nd machine, because it does better when processing more complex and modern graphical elements like shaders, etc. I made those numbers up, so there isn't a real-world example of a graphics card that does Dx7-level stuff slower and Dx9 stuff faster.

In general, they are faster on both fronts. But the time is coming soon when, because of complex calculations and specialized programming, a computer will be able to do good on modern tests, but still fall behind an 'older' one with less recent technology.

So what if you get 400 fps in Quake 3. That doesn't improve your Doom 3 experience.indeed, focusing on squeezing out some more Q3 frames is a total waste of time. The technologies that Doom 3 uses test parts of the computer/graphics card that didn't even exist when Q3 was released.

Today's game graphics are beginning to focus more on feature sets and less on raw speed. There will always be people who want to play everything at 100+ fps.and they can go buy SLI motherboards and put 2 6800 Ultras in there.woo hoo. I'd rather buy a car with that money, though, because a DX9-level graphics card that can do 200 fps in Doom 3 benchmarks won't mean anything when Doom 5 comes out and runs better on 150 dollar then-modern graphics cards that can only manage 100 fps in doom 3.

I find it very interesting what really made the big difference between the two Dual 1.8 GHz G5s running @ 640 x 480, Medium quality. 37.8 on these low settings on a similar GPU seems like a huge difference to me. Is the Radeon 9600 the bottleneck here? Or the RAID0 vs. I think it is the 9600 vs.

Increasing the rez to 800 x 600 with the same medium settings, made no difference on the 'faster' Mac, but did show quite dramatic framerate drops on the Mac with the Radeon. Anyone have a DP 1.8 GHz G5 with Radeon 9800 (OEM or SE or XT) who would care to run the same tests? People hated Quake when it was first introduced. The graphics were so far beyond Doom that the systems at the time couldn't deal with the game efficiently enough. It was only after 2 years when the PC and graphic cards were fast enough to handle Quake that the game started to take off. I'm sure the same goes for Doom 3.

Doom 3 Timedemo

The graphics engine of this game is completely amazing. Very realistic, especially if they could in the future support s-curve and bezier splines inside of those chunky polygons. In 2 years as well, Doom 3 and their subsequent mods will be very popular. It probably won't be for another decade before John Carmack and the folks at Id come up with a new, more amazing graphics engine to drive the next generation 3D shooter.

Greeting to all unreal fans reading this. As 227g seems to support so-called 'real time dynamic shadows' I've got a question about them: what should these 'shadows' look like? It is easy to notice that the performance drops hugely in case dynamic shadows are enabled, even with low resolution set in preferences dialog.

But I can barely notice any difference in the rendered picture with shadows on versus shadows disabled. Yeah, I can see the shadow casted from the game character downwards to the floor. I can see the shadows casted from the vegetation decor like 'palm-like-trees' at the NyLeve's Falls. But all of these don't look too cool to be worth the performance drop they cause. Light direction seems to be always upwards down, and looks like there are no shadows being cast from the dynamic lights sources like Dispension Pistol bolt and alike. I mean, while these shadows are really seem to be 'realtime' they are in no ways competitive to the shadows in Doom 3, DarkPlaces, Tenebrae or Nexuiz. Am I missing something (possibly hitting some bug with shadows not being rendered as they should on my PC) or should the shadows in 227g look just like that?

Shadows are a nice add on. Of course they are not light source dependent, but it adds a realistic flair.

I like them very much, especially after adding the smoothing they look very realistic, even more than in many other games where they often look unnaturally sharp. The performance drop is acceptable if on a newer machine and not playing some mod with a very huge number of monsters. No doubt, they should be in the GPU as well as the Lighting (which is the main reason many things can't be done in GPU, counts for the shadows as well) and many other things, but this would mean a complete rewrite from scratch of the renderer, and I am not allowed to share the code with someone more skilled in these things, like dots, as mentioned. If you like them use them, if not (or if they drain to much fps) leave them off. Thanks on making this clear, guys. My question wasn't intended to be a 'complain' in any way, I just was suspecting that I'm possibly hitting some bug as shadows were not looking as good as they should be when rendered using Carmak's reverse shadow volumes algorithm but instead were looking more like simple planar projection shadows rendered into texture go get 'smoothing effect' later on at alphablending stage.

Sure, it's bad for all the community that the engine hadn't been open-sourced by Epic. I would bet that if it would go GPL it would be enhanced with all the fancy features like realistic GLSL-shaders based water in less than a week. Nevertheless there's no point in barking at wind about Epic's position, at least we've got 'something' allowing Smirf to release new patches and that's clearly better than nothing. As for shadows looking too sharp in games like Doom 3 - that's unfortunate downside of Carmack's shadow volumes approach.

There's a way to get a smoother picture still using shadow volumes that involves a couple of extra render passes (stencil buffer into backbuffer with alpha-channel, backbuffer into texture, and then blur filter implemented as a shader or even totally in software) but the performance drop it would cause is extremely huge. There are some other shadow-casting algorithms out there allowing to get more realistic picture with 'soft' shadows, but they are either too slow to be used in realtime game or would produce not so realistic picture after all. It seems to me that we won't get a realistically looking shadows in games until the day when the hardware would allow to do realtime reverse ray-tracing rendering. Getting back to 227g, for me the shadows in game, despite looking not so bad, don't add a lot to the image to be worth enabling.

Doom 3 Download Pc

Even with medium resolution shadows I get only around 40 FPS on average with drops down to 20 at some locations (4xSSAA + 2xMSAA/16xAniso). My hardware isn't a decent one but it's still not so bad nVIDIA GeForce GTS250 with 1Gb VRAM, AMD Phenom II X4 955 Processor running at 3.2GHz and 6Gb RAM. Disabling shadows gives me huge performance boost allowing to play game at 1680x1050 resolution with 4x supersampling antialiasing + 2x MSAA + 16x Anisotropic filter + hardware texture sharpening and still get insane FPS numbers around 190 FPS on average with 80 minimum and 400 maximum at 'flyby timedemo' having FPS limiter turned off. For me it's more important to get staircase-on-edges-free picture with sharp distant textures and have FPS to be more than 60.

Enabling shadows in their current form would force me to stick back to MSAA variants that produce worse picture in orger to have comfort FPS. Still it is all a matter of a personal preferences and depends on a hardware a player has.

In any case it is always better to have a choice to enable something or not than not having something at all. I want once again to thank all the members of devteam for making an amazing progress with 227g release! You've done an excellent job and I'm excited looking forward for any future improvements to the classic Unreal you'd be able to add in. Thanks a lot! Wow, talking about engrish on-topic: the only compaint i have about the realtime shadows is that they are a very low quality on corpses. Doesn't matter on which setting you are running, the quality decreases as soon as a pawn finished its dead animation.

I get this is usefull to get higher fps but i would be nice if we could change it. @LeXa: the reason shadows take away this much performance to probably because of your cpu (like said before, realtime shadows uses the cpu). Altough i don't really get why. I ones had an AMD processor single core that ran on 1.2GHz and real time shadows ran on low quality around 40 to 60 fps on the default maps. Also i don't see alot of quality improvement on anything higher than low. Only when taking a really close look i could see some difference which i never do in Unreal anyway.

Thing is, as said, all calculations are done in CPU, especially the smoothing eats up a lot of power. While I have to agree that if using low res (which is still not bad due to the smoothing) the performance is not bad.

Either way, all rendering in Unreal is still in CPU unlike modern games, it should be done in GPU which again would allow it 'easily' to implement realistic light dependent shadows- at least compared to the mess it would result in, doing that in current rendering method. Unfortunately this is a hard task, especially without having the sources, since I can't give them to anyone. So everyone who wants to give it a try needs to work out a real huge project with very limited information.