Apple Benchmarking Tool... and iMac Doc

[arn]

An Xlr8yourmac reader email describes (http://www.xlr8yourmac.com/OSX/os_x_benchmark_tools.html) some OS X Benchmarking apps from Apple:

What I thought might interest your readers is the bashpack.dmg that is installed in Developer/Applications. This disk image contains a suite of four benchmarking applications; CacheBasher for testing cache/memory bandwidth, DiskBasher for testing drive performance, GraphicsBasher for testing Quickdraw performance and Skidmarks GT for basic processor performance.

The application is available from Apple (ftp://ftp.apple.com/developer/Tool_Chest/Testing_-_Debugging/Performance_tools/CHUD_1.1.1.pkg.tar.gz) - use at your own risk.

Of note, Apple had reposted a final version of their Taking Apart Your iMac service document... but it has since been pulled off their site. Fortunately, a quick Google search (http://www.google.com/search?hl=en&q=imac_flatpanel.pdf) reveals other sources (albeit Draft versions) of the same... for those who are intent on opening your iMac.

[menoinjun]

This stuff is really confusing, but I'm going to try it tonight. I think I'm just going to do the Skidmark test because all those numbers from the other tests dont mean anything to me.

-Pete

[mac15]

damn right is confusing as hell

[MacAztec]

I dont get "Use at your own risk." Is this a bad app? I am testing it on a G4 500, 256 MB RAM,and an 18GB Ultra SCSI 160 HD.

[Rower_CPU]

Here are my Skidmarks results:
Integer: 1031
Floating Point: 1033
Vector: 1017

TiBook 400

[Xapplimatic]

My Skidmark results (iBook G3/466):

Skidmarks GT - Processor Performance Benchmark
(c) 2002 Apple Computer, Inc.
Note: Power Mac G4/400 (PCI Graphics) = 1000 for all tests.
...
Integer: 1167
Floating Point: 991
Vector: N/A

[Skorobeus]

My SkidMark Numbers (DPG4 1ghz):

Skidmarks GT - Processor Performance Benchmark
(c) 2002 Apple Computer, Inc.
Note: Power Mac G4/400 (PCI Graphics) = 1000 for all tests.
...
Integer: 3025
Floating Point: 2118
Vector: 2619

Skidmarks completed succesfully.

[macintoshse]

There's a link to the pdf at the bottom of this page:
http://www.macnyt.dk/default.tpl?news=20020112030528
(the link says "Service Source: Flat Panel iMac")

[afonso]

Gathering system information...
Starting performance tests...

Integer: 3029
Floating Point: 2117
Vector: 2606

Skidmarks completed succesfully.

Take care,
Afonso

[abe]

Skidmarks GT - Processor Performance Benchmark
(c) 2002 Apple Computer, Inc.
Note: Power Mac G4/400 (PCI Graphics) = 1000 for all tests.

Gathering system information...
Starting performance tests...

Integer: 2422
Floating Point: 1694
Vector: 2095

Skidmarks completed succesfully.

[TechLarry]

Every link I've found so far has been for the DRAFT version of the document.

Where is the FINAL version that was reported on?

Thanks :)

TL

[Backtothemac]

Here are mine for my iBook 600 Combo w/384MB.

Skidmarks GT - Processor Performance Benchmark
(c) 2002 Apple Computer, Inc.
Note: Power Mac G4/400 (PCI Graphics) = 1000 for all tests.

Gathering system information...
Starting performance tests...

Integer: 1485
Floating Point: 1274
Vector: N/A

Skidmarks completed succesfully.

[Falleron]

My Ghz DP PowerMac gets:

Integer : 3028
Floating Point : 2118
Vector : 2619

[menoinjun]

Ok, this is VERY disheartening!!

Gathering system information...
Starting performance tests...

Integer: 1380
Floating Point: 1377
Vector: 1356

Skidmarks completed succesfully.

DUAL PROCESSOR 533 w/768mb ram! Shouldn't my results be a little higher than an iBook 600?

-Pete

Do I have my settings wrong? Is it not seeing my second processor?

[Catfish_Man]

...I get

Int: 599
Float: 480 (roughly, I'm typing from memory)
Vector: N/A


btw, the iBook 600 scoring higher on int and float than a G4 533 is to be expected. The speed gain of the 533 comes from the other processor and the vector math (the G4 got 1300 on vector, the iBook got N/A because it doesn't have Altivec). I *think* that the DP533 scoring lower than an iBook 600 shows that it only tests one processor, but I could be wrong. Try setting the number of threads up to two.

Also, the G4's more advanced FP unit gave it a higher float score, while its lower clock speed gave it a worse integer score.


For cachebasher it shows this (I think)
1st column: size of the data being transferred
2nd column: something
3rd column: number of bus clock cycles it took
4th column: number of clock cycles it took
5th column: bandwidth (available bandwidth? used bandwidth?)


The ratio of the third column to the fourth shows your bus multiplier. On mine the 4th column was always 3.5 times more than the third (66*3.5 is roughly 233). On a DP1GHz it would be 7.5 (if I remember correctly). On a 2GHz P4 (if you could run the benchmark on a P4) it would be 5. It really shows how slow busses eat up processor clock cycles.

[Backtothemac]

Actually, PT, the scores are very accurate. My iBook is 77 MHZ faster than your Dual 533. That would explain the INT. score being higher. You have a better floating point unit, and you have the vector Unit :( wish I did. This little iBook is reall quick and I love it, but with multi proc aware apps, and with OSX supporting Multi proc, you have got to love your system.

[Backtothemac]

Hey, redid the test, and got 1519 on the INT score. Redid it again, and look what I got!

http://homepage.mac.com/chuck_stacey/PhotoAlbum6.html (http://)

This is either the fastest iBook ever, or it is somehow been sent through time, and came back with a G7 :p

Don't know how it happened, but I thought my brothers would like a good laugh.

[Rower_CPU]

Originally posted by Backtothemac
Hey, redid the test, and got 1519 on the INT score. Redid it again, and look what I got!

http://homepage.mac.com/chuck_stacey/PhotoAlbum6.html (http://)

This is either the fastest iBook ever, or it is somehow been sent through time, and came back with a G7 :p

Don't know how it happened, but I thought my brothers would like a good laugh.

Gotta love those bugs that give you crazy scores like that!

Here's my scores on my G3 450 @ work:
Integer: 1127
Floating Point: 952
Vector: N/A

[G4scott]

iBook 600 results:

Skidmarks GT - Processor Performance Benchmark
(c) 2002 Apple Computer, Inc.
Note: Power Mac G4/400 (PCI Graphics) = 1000 for all tests.

Gathering system information...
Starting performance tests...

Integer: 52689
Floating Point: 45127
Vector: N/A

Skidmarks completed succesfully.


Uh, i guess my iBook is pretty darn fast. This was the first time I ran it. The second time results were:

Integer: 1511
Floating Point: 1274
Vector: N/A

[slingn]

Everybody seems really excited to use Skidmarks, but before it 's results are misinterpreted I would like to point out a couple of things:

1) it is a processor benchmark pure and simple - no system level stuff at all. As such it tends to scale linearly with processor frequency.

2) it does not take advantage of dual processors at this time

Nathan Slingerland
Architecture and Performance Group
Apple Computer, Inc.

[menoinjun]

<whew> That calms my nerves a little bit. My 533 still isn't nearly as fast as the Dual 1gig would be, but it must wipe the floor with a single G4 400. (no offense to G4 400 owners)

-Pete

[mr.me]

My results on a 667mhz tower.

Integer: 1999
FP: 1402
Vector: 1741

[Xapplimatic]

Here are my Skidmarks results:

Integer: 6502
Floating Point: 7455
Vector: 9000
3-D Matrix: 1984

New PowerMac G5 Sphere.

[Rower_CPU]

Originally posted by Xapplimatic
Here are my Skidmarks results:

Integer: 6502
Floating Point: 7455
Vector: 9000
3-D Matrix: 1984

New PowerMac G5 Sphere.

If only it were true...;)