halo needing a powerful PC

LisaJoy said:
My Son insisted we get this game, so I folded, I always do..
we just have it in default settings, which looks great to me..
did our first timedemo

Date / Time: 10/5/2003 4:33:46 AM (75218ms)
2000MHz, 1024MB, 128M ATI Radeon 9800 PRO (DeviceID=0x4e48) Driver=6.14.10.6378 Shader=1.4
C:\Program Files\Microsoft Games\Halo\halo.exe -timedemo (Version=1.0.0.564)
Frames=4700
Total Time=97.15s
Average frame rate=48.38fps
Below 5fps= 11% (time) 0% (frames) (11.393s spent in 11 frames)
Below 10fps= 12% (time) 0% (frames)
Below 15fps= 12% (time) 0% (frames)
Below 20fps= 12% (time) 0% (frames)
Below 25fps= 13% (time) 0% (frames)
Below 30fps= 13% (time) 0% (frames)
Below 40fps= 21% (time) 6% (frames)
Below 50fps= 43% (time) 26% (frames)
Below 60fps= 72% (time) 59% (frames)
Memory used Max=161MB, Min=131MB, Ave=150MB


certainly not bad

Well Lisajoy, your computer is certainly handling better than mine, but you might see a performance boost by adding in the line -use 20 both during the time demo and during the game.(heck, radeon 9600/9700/9800 and maybe 9500 are pixel shader 2.0 capable, might as well use it)

BTW, the halo timedemo is mostly cutscenes(there is one part that is playable in the game but they just leave you watching, which is actually like the fastest part of the timedemo). Also, it does have a loading thing that the benchmark isn't smart enough to ignore, but on my computer it loads so fast that it happens in under a second.

Also, it may say 0% of the frames, but that's just cause it's such a small number of the frames, timewise it is still taking up 11% at below 5 fps, with another 2% lost somewhere between that and 30 fps.
 
cthellis42 said:
don't quite know what the "time" measurement is looking at, but since everything "Below 30fps" on down has 0% on frames, I assume that means they were never touched.

Errr... I don't think it gets any more explicit then this...
Chart said:
Frames=4700
Total Time=97.15s
Below 5fps= 11% (time) 0% (frames) (11.393s spent in 11 frames)
 
I don't quite know what the "time" measurement is looking at, but since everything "Below 30fps" on down has 0% on frames, I assume that means they were never touched.

Well they attribute an actual value of 11 frames to the lowest. If it's truncating at the whole number; 11 of 4700 frames would not register as a % of the frames rendered, but would still make a pretty big impression in game play if it took 11seconds to render them.

Otherwise tell me what this:

Below 5fps= 11% (time) 0% (frames) (11.393s spent in 11 frames)

Is supposed to mean.
 
could it be, if the benchmark actually counts the loading screen, that's where those numbers came from. if some frames took ~1 second to draw it would be pretty obvious to the naked eye.
 
I don't quite know what the "time" measurement is looking at, but since everything "Below 30fps" on down has 0% on frames, I assume that means they were never touched.

Though MAN as that chart confusing to follow without the accompanying documentation. Nothing adds up!

Well, let me show you how to read it.

Frames=4700
Total Time=97.15s
Average frame rate=48.38fps
Below 5fps= 11% (time) 0% (frames) (11.393s spent in 11 frames)
Below 10fps= 12% (time) 0% (frames)
Below 15fps= 12% (time) 0% (frames)
Below 20fps= 12% (time) 0% (frames)
Below 25fps= 13% (time) 0% (frames)
Below 30fps= 13% (time) 0% (frames)
Below 40fps= 21% (time) 6% (frames)
Below 50fps= 43% (time) 26% (frames)
Below 60fps= 72% (time) 59% (frames)

Notice Frames and Total time. In this time demo there are 4700 frames. Total time is the total time that the PC takes to finish going through that 4700 frames. Average frame rate is Frames/Total time.

Now on to that table. The percentage on time is actually cumulative.

below 5 fps = 11% (time)
below 10 fps = 12% (time)

This mean there are only 1% of that Total time (97.15s) that the frame rate is around 5-10 fps.

Below 10fps= 12% (time)
Below 15fps= 12% (time)
Below 20fps= 12% (time)

This mean there is almost no time that the frame rate is around 10-20fps.

Below 20fps= 12% (time)
Below 25fps= 13% (time)
Below 30fps= 13% (time)

This mean there are 1% of that Total time, the frame rate is between 20-30fps.

So and so. On this PC you only get 13% of the time that frame rates dips below 30fps.

Frame percentage are just that, many showed 0% because its truncated. Obvioulsy when fps drops the number of frame render during that instances also reduced because each frame takes longer to render.

If the timedemo is representative of the game, this PC can pretty much run Halo at higher frame rate than Xbox.
 
Does it really suprise you that it performs so bad? Gearbox have never done any good ports from PC to console, what makes you think they could go from console to PC and not screw it up?

I think it could also be the fact that they had to recode alot of the shaders to work on the average PC card since the xbox code was to customised.
 
Hehe. Things also make a lot more sense when one hasn't been up so damn long, too. ;) I kept trying to do the "obvious" thing and add percentages together. :oops: :p

Seems rather like there are a few places that just CRUSH the card down to 1fps (or maybe just stall and freeze in general with a a few frames in the low surrounding it which average out) which mainly makes me wonder what the heck is going on in that timedemo. Could it be a bug, or did they put in an area to take just about any card to its knees on purpose? I guess one would have to watch it and pay close attention to make the best guess... (Do they have a line graph?)

On the whole, though, I'm mainly skeptical over how well it reflects actual gameplay. Maybe we'll have some people FRAPSing level runthroughs a few times to get some approximates.
 
well, i still dont get it, but not like i care much, i aint gonna spend 2 grand for a PC anytime soon, and certainly not to play Halo... so there..
 
Well I need a core upgrade, and those you can do pretty well without going through the roof. Pick up a 9800 non-pro (current price/performance champ, imho. And seemingly can hit Pro-levels of performance if you're willing to risk flashing the BIOS), get a solid mobo and either a Barton 2500 XP or 2.4 ghz P4 (which one seems to be able to kick up a lot if one wants), and whatever RAM you want. You can upgrade whatever parts when you want in stages, and the total cost from a place like NewEgg is under $450 even for trustable brands. (512MB of DDR-400 RAM is all I looked at to start)

IMHO, there are plenty of ways to make sure you have a solid system with plenty of headroom that even by itself would do you well and last a long time without spending anywhere NEAR the amounts people think.
 
Back
Top