Munx wrote on 2021-02-02, 19:37:

Run some Unreal and Q3 benchmarks and compare them to a Pentium 3!

Apple kept touting how PowerPC is superior, but actual direct benchmark comparisons are a rarity.

That's a fun idea and something I will do when I have the 6200 in the Mac, I have a PC Geforce 6200 around somewhere as well.

I'm still waiting for the new battery and ide-cf adaptor to come as well.

Munx wrote on 2021-02-02, 19:37:

Run some Unreal and Q3 benchmarks and compare them to a Pentium 3!

Apple kept touting how PowerPC is superior, but actual direct benchmark comparisons are a rarity.

I've actually done a few PIII vs G4 benchmark sessions. A single PIII-S overclocked to 1628MHz is faster than a dual G4 overclocked to 1500MHz in two of the three games I tried. You can find the full specs of both systems in my signature. The PIII-S was running XP Pro SP3, and the dual G4 was running OS X Tiger (10.4.11)

Quake III Arena timedemo0001 at 1024x768 with all settings cranked:
PIII-S: 208.1 FPS
Dual G4: 244.7 FPS
However, it's worth noting that the OS X version of Quake III can take advantage of SMP. Activity Monitor was reporting roughly equal usage across both CPUs.

Doom 3 timedemo1 at 1024x768 with the Ultra preset:
PIII-S: 50.8 FPS
Dual G4: 27.9 FPS

Unreal Tournament 2003 botmatch at 1024x768, max detail settings:
PIII-S: 56.4 FPS
Dual G4: 36.0 FPS

So yeah, unless the game engine has the smarts to put both CPUs to work, the G4 really sucks at gaming. However, one area in which the G4 does really well is AltiVec-optimized H.264 decoding. Using CorePlayer, which is basically hand-tuned for AltiVec, the dual G4 can handle a 8 mb/s 1080p H.264 clip in software. The single PIII-S running CoreAVC can handle 5 mb/s 720p H.264 just fine, but really struggles with 1080p. Of course, a dual PIII-S @ 1628 could probably handle the 1080p test clip just as well as the dual G4. 😀

Standard Def Steve wrote on 2021-02-03, 04:37:

I've actually done a few PIII vs G4 benchmark sessions. A single PIII-S overclocked to 1628MHz is faster than a dual G4 overcloc […]
Show full quote

Munx wrote on 2021-02-02, 19:37:

Run some Unreal and Q3 benchmarks and compare them to a Pentium 3!

Apple kept touting how PowerPC is superior, but actual direct benchmark comparisons are a rarity.

I've actually done a few PIII vs G4 benchmark sessions. A single PIII-S overclocked to 1628MHz is faster than a dual G4 overclocked to 1500MHz in two of the three games I tried. You can find the full specs of both systems in my signature. The PIII-S was running XP Pro SP3, and the dual G4 was running OS X Tiger (10.4.11)

Quake III Arena timedemo0001 at 1024x768 with all settings cranked:
PIII-S: 208.1 FPS
Dual G4: 244.7 FPS
However, it's worth noting that the OS X version of Quake III can take advantage of SMP. Activity Monitor was reporting roughly equal usage across both CPUs.

Doom 3 timedemo1 at 1024x768 with the Ultra preset:
PIII-S: 50.8 FPS
Dual G4: 27.9 FPS

Unreal Tournament 2003 botmatch at 1024x768, max detail settings:
PIII-S: 56.4 FPS
Dual G4: 36.0 FPS

So yeah, unless the game engine has the smarts to put both CPUs to work, the G4 really sucks at gaming. However, one area in which the G4 does really well is AltiVec-optimized H.264 decoding. Using CorePlayer, which is basically hand-tuned for AltiVec, the dual G4 can handle a 8 mb/s 1080p H.264 clip in software. The single PIII-S running CoreAVC can handle 5 mb/s 720p H.264 just fine, but really struggles with 1080p. Of course, a dual PIII-S @ 1628 could probably handle the 1080p test clip just as well as the dual G4. 😀

Huh. That's quite a performance difference. I wonder if part of that is OSX overhead?

That's a definite possibility, as WinXP really does "feel" lighter and more responsive than OS X. Could also be sloppy/lazy game ports, or unoptimized graphics drivers. I guess it's also possible that the Motorola chipset has weaker memory and/or AGP performance than x86 chipsets, but unfortunately it's hard to measure that kinda low-level stuff on a Mac.