<strong>Considering your eyes can't distinguish individual frames at speeds above 80 fps, going over 100 fps is pretty much irrelevant anyway. So if it's 105 FPS or 195 FPS it really doesn't matter... it'll look the same to your eyes.
Actually what you need in order to adequately fool the eye is about 24fps and 72hz -- basically what you get from a theatre projector. Each frame is flashed three times to reduce your perception of flicker, the 24 frames are sufficient to create the illusion of motion. Same thing on a computer. 24-30 "constant" frames per second through a 72Hz refresh rate would be all the temporal resolution you need. I think, though, that the problem comes in heavy action or changing video system demands where the machine would then slow down in a hurry and you would notice.
When your game is running at 200fps and you hit a very demanding sequence that drops it to 100, you shouldn't notice anything since both rates are well beyond the capabilities of your eyes, and the display.
I guess that's why people need high frame rates, fps must drop dramatically as visual complexity is added ??? Aren't new cards supposed to somehow adress this in that while they may not push significantly more frames, they can make each of the given frames much more complex?
And what makes you think 1024x768x32 isn't equal to 1024x768x32? Apple used high quality settings, that's about as sure as it can be.
Plus, last time they had Q3 benchmarks in there, they weren't with s_mixahead and chunksize adapted. (two tweaks without effect on quality of the game, but compensating for certain programming shortcomings.)
So, in what respect are these benchmarks not comparable?
G-News
<hr></blockquote>
1024x768x32 isn't equal to 1024x768x32 if the one demo is set in a narrow hall with few or no opponents and the other is in an outdoor setting with 50+ opponents and a deep horizon.
The benchmarked demo must be exactly the same or else it is pointless to compare them.
1024x768x32 isn't equal to 1024x768x32 if the one demo is set in a narrow hall with few or no opponents and the other is in an outdoor setting with 50+ opponents and a deep horizon.
The benchmarked demo must be exactly the same or else it is pointless to compare them.</strong><hr></blockquote>
People usually use the standard four.dm_6x-demo when benchmarking Q3, though there are exeptions. I don't know in this case.
This is a media making machine. I ripped MP3s last night at the Jaguar Premiere in Palo Alto. the dual 867 ripped a range of 9x to about 18x. The dual GB did 12x to 22x. I expect the dual 1.25 will do 15x/28x FROM the SuperDrive. Once I add the Yamaha CRW-F1 which reads @44X, It may be even faster. No prototype/first build 1.25 GHz Macs are in the store to test yet.
Comments
<strong>Considering your eyes can't distinguish individual frames at speeds above 80 fps, going over 100 fps is pretty much irrelevant anyway. So if it's 105 FPS or 195 FPS it really doesn't matter... it'll look the same to your eyes.
[ 08-19-2002: Message edited by: Xaqtly ]</strong><hr></blockquote>
eyes are a LOT slower than that. its more like 20 or 30 fps where you can't see the diff anymore.
<a href="http://arstechnica.infopop.net/OpenTopic/page?a=tpc&s=50009562&f=8300945231&m=2000938035"; target="_blank">MacAchaia</a>
[QB]Here is a newer Barefeets benchmark for the new machines, Actuall a reasonable comparison.
In this eweek article, Apple actually refutes some of BF's claims. Not too typical of Apple's shutmouth policy. Time for a smackdown.
<a href="http://www.eweek.com/article2/0,3959,478975,00.asp"; target="_blank">eweek article</a>
Oh, it's on, baby. It's on.
Alex.
When your game is running at 200fps and you hit a very demanding sequence that drops it to 100, you shouldn't notice anything since both rates are well beyond the capabilities of your eyes, and the display.
I guess that's why people need high frame rates, fps must drop dramatically as visual complexity is added ??? Aren't new cards supposed to somehow adress this in that while they may not push significantly more frames, they can make each of the given frames much more complex?
And what makes you think 1024x768x32 isn't equal to 1024x768x32? Apple used high quality settings, that's about as sure as it can be.
Plus, last time they had Q3 benchmarks in there, they weren't with s_mixahead and chunksize adapted. (two tweaks without effect on quality of the game, but compensating for certain programming shortcomings.)
So, in what respect are these benchmarks not comparable?
G-News
<hr></blockquote>
1024x768x32 isn't equal to 1024x768x32 if the one demo is set in a narrow hall with few or no opponents and the other is in an outdoor setting with 50+ opponents and a deep horizon.
The benchmarked demo must be exactly the same or else it is pointless to compare them.
<strong>
1024x768x32 isn't equal to 1024x768x32 if the one demo is set in a narrow hall with few or no opponents and the other is in an outdoor setting with 50+ opponents and a deep horizon.
The benchmarked demo must be exactly the same or else it is pointless to compare them.</strong><hr></blockquote>
People usually use the standard four.dm_6x-demo when benchmarking Q3, though there are exeptions. I don't know in this case.
People usually use the standard four.dm_6x-demo when benchmarking Q3, though there are exeptions. I don't know in this case.
<hr></blockquote>
The x86 tests G-News refers to uses 'Time Demo 4' and Apple uses what?
We don?t know because they don?t tell, therefore the comparisons are useless.
Plus, you could argue that they'd run it on Mac OS X, that means they'll have to use a fairly up to date version of it already, 1.30 or 1.31b5.
In these versions there's only demo four left, all other tests have to be added manually, which Apple certainly wouldn't do (or manage to do).
So it's extremely likely that these benchmarks are 1:1 comparable, even if you ultra-nay-sayers don't like it.
I know I do like it.
G-News
<strong>
The x86 tests G-News refers to uses 'Time Demo 4' and Apple uses what?
We don?t know because they don?t tell, therefore the comparisons are useless.</strong><hr></blockquote>
'Time Demo 4' = four.dm6x (x= 6 or 7)
At least the test results sounds like they've used timedemo four...