Mac Benchmarks....

[agent302]

Ok, my scores are as follows:

Math: 27034
Memory: 23142
Graphics: 31416
Disc: 2253

Which leads me to be very confused. Why? Because my system (a Powerbook 667 DVI) is faster than the included set of an iMac 350 for Math, Memory, and Disc scores. But it's slower on graphics, which makes absolutely no sense (Mobility Radeon 7500 slower than Rage128?).

Also, mc68k, a neat benchmark from Apple (that you can get if you install these debug tools) called Skidmarks (it's in the bashpack.dmg file installed in /Developer/Applications) that measures processor performance in Integer, Floating Point, and Vector calculations. Now, if a G3 and G4 are equal in non-vector calculations, then my Powerbook should have scored 67% better than my iMac 400. Instead, on integer calculations, the G4 was 115% faster, which leads me to think that, even though this is probably a finally tuned application, the G4 is faster than the G3 at the same clockspeed.

 

[Catfish_Man]

Re: G3 is faster clock for clock..

Originally posted by Funkatation
on unoptimized code. Remember, a G4 is not a G3 with altivec anymore. Ever since the new G4's came out they are on a 7 stage pipeline. The G3's are still on a 4 stage pipeline. So for unoptimized code, it would make sense that a G3 is faster.

A G3 might outperform a G4+ on well optimized scalar integer code that isn't limited by memory speed. At everything else (FP, especially double precision FP, memory, and vector[Altivec]) the G4 will win. I'm still kinda surprised that the iBook is doing so well, this seems to be a pretty shoddy benchmark (perhaps if we figured out how to run it in single user mode it would be more accurate).

I would suggest that we try it with disk basher, cache basher, graphics basher, and skidmarks GT. These are from apple and can be downloaded from their website. It may require a free developer registration, I'm not sure. If someone gets them, post a link to it (or if you can, mirror it). I'll post my computer's stats with heartbeat as soon as I test it. I want to see how much of a difference having omniweb in the background makes.

 

[Catfish_Man]

I've attached my G3 233's scores in heartbeat. Pretty abysmal aren't they? Also, I see someone posted a link to the bashpack while I was typing my previous post, I'll run those tests soon.

 

[mc68k]

I think we should create another benchmarking thread with some "real" tools from Apple, so we can get a better sense of how good/crappy our machines really are.

 

[Backtothemac]

Skidmarks GT

Well, this is for the iMac X.1.4

 

[Backtothemac]

Here are the iBook's. iBook 600 with 384 of ram.

X.2 WWDC Build

 

[Zenith]

Here's my X.2 results in Heart Beat X.

With iTunes running and playing a MP3:

Math: 16819
Memory: 22733
Graph: 24801
Disk: 0 (a bug)

Without iTunes:

Math: 17894
Memory: 19840
Graph: 20106
Disk: 0 (the bug again)

Yesterday I got lower scores than I do today, and then I ran the benchmarks without IE running. It seems that I can get better scores with a few apps running. Strange about the iTunes-scores. I don't trust this program...

 

[resm]

Mac Benchmarks

here my iMac 800 :

Test Math : 20250
Test Memory : 19098
Test Graphics : 30085
Test Disc : 1580

 

[chibianh]

here's my skidmarks gt test..

 

[mmmdreg]

I did the best of three while Mozilla and foldingcontrol were open but hidden..Scores as follows:
15014 Maths
25280 Memory
46374 Graphics
7618 Disc...
thats on an iMac 600 (G3) with 640mb RAM.

 

[Backtothemac]

Well, that makes it official. This program is not trustworthy. Lets see the first clue was that the G3's were smoking the G4's, and now a G3 with less memory smoking a G3 with a clean install of X.1.4

Oh, well. Back to the original question.... anyway to benchmark besides counting bounces.