Power Mac G5 Quad Benchmarks

MacRumors

macrumors bot
Original poster
Apr 12, 2001
7,439
8,501


One of our Wales UK-based members has reported receiving his Power Mac G5 Quad earlier in the week, and has run a preliminary set of Xbench benchmarks on it. While full details are in the forum thread, the overall score for the machine with 3 GB RAM running Mac OS X 10.4.2 weighs in at 151.86.
 

Mudbug

Administrator emeritus
Jun 28, 2002
3,809
1
North Central Colorado
Xbench Results:
Results 151.86
System Info
Xbench Version 1.2
System Version 10.4.2 (8E90)
Physical RAM 3072 MB
Model PowerMac11,2
Processor PowerPC G5x4 @ 2.50 GHz
L1 Cache 64K (instruction), 32K (data)
L2 Cache 1024K @ 2.50 GHz
Bus Frequency 1 GHz
Drive Type WDC WD2500JS-41MVB1
CPU Test 125.82
GCD Loop 125.52 6.62 Mops/sec
Floating Point Basic 125.96 2.99 Gflop/sec
vecLib FFT 129.52 4.27 Gflop/sec
Floating Point Library 122.46 21.32 Mops/sec
Thread Test 246.38
Computation 252.23 5.11 Mops/sec, 4 threads
Lock Contention 240.79 10.36 Mlocks/sec, 4 threads
Memory Test 129.76
System 123.99
Allocate 116.39 427.43 Kalloc/sec
Fill 172.55 8389.84 MB/sec
Copy 101.94 2105.60 MB/sec
Stream 136.10
Copy 134.55 2779.01 MB/sec [G5]
Scale 134.42 2777.05 MB/sec [G5]
Add 137.73 2933.98 MB/sec [G5]
Triad 137.79 2947.68 MB/sec [G5]
Quartz Graphics Test 150.30
Line 124.14 8.27 Klines/sec [50% alpha]
Rectangle 141.05 42.11 Krects/sec [50% alpha]
Circle 140.17 11.43 Kcircles/sec [50% alpha]
Bezier 133.71 3.37 Kbeziers/sec [50% alpha]
Text 284.92 17.82 Kchars/sec
OpenGL Graphics Test 144.50
Spinning Squares 144.50 183.30 frames/sec
User Interface Test 229.01
Elements 229.01 1.05 Krefresh/sec
Disk Test 71.27
Sequential 101.86
Uncached Write 116.11 71.29 MB/sec [4K blocks]
Uncached Write 106.60 60.32 MB/sec [256K blocks]
Uncached Read 76.82 22.48 MB/sec [4K blocks]
Uncached Read 121.06 60.84 MB/sec [256K blocks]
Random 54.81
Uncached Write 20.79 2.20 MB/sec [4K blocks]
Uncached Write 135.37 43.34 MB/sec [256K blocks]
Uncached Read 97.78 0.69 MB/sec [4K blocks]
Uncached Read 137.66 25.54 MB/sec [256K blocks]
 

Lacero

macrumors 604
Jan 20, 2005
6,639
2
Not sure how fast 151.86 on Xbench is. Any real world tests using everyday apps such as Shake?

Sounds great if the quads can run 70-80% faster than the dual-cores.
 

Kobushi

macrumors 6502a
Jun 7, 2005
540
0
Right behind you.
Lacero said:
Need more comparison data.

Not sure how fast 151.86 on Xbench is. Any real world tests using everyday apps such as Shake?
I concur.

I was just thinking "oooo pretty numbers" then realized I have no idea what they mean. Any hardware savvy folks care to explain?
 

Mr Maui

macrumors 65816
Jul 19, 2002
1,152
0
AoWolf said:
Yeah I would like to see it next to a duel 2.7
Simply put ... it's faster!! :D

Actual comparisons in real world activities will come out as more people receive their boxes.
 

liketom

macrumors 601
Apr 8, 2004
4,167
23
Lincoln,UK
i was expecting more to be truefull?

but real world tests would be cool

itunes lossless speeds like on ZDnet UK lol

he wants to do the photoshop test !!!
 

Chaszmyr

macrumors 601
Aug 9, 2002
4,265
76
Xbench is a lowsy benchmark program, imo. It weighs in everything, so if your dual 2.7 has the same hard drive, faster video card, and more RAM, you might well score almost as high even if your processor is a lot slower... And on top of that, I don't think the numbers are all that accurate or consistent.

With that said, here are the results from my dual 2.7 with 10.4.2, 3gb RAM, and a GeForce 6800 Ultra:

Results 109.87
System Info
Xbench Version 1.2
System Version 10.4.2 (8C46)
Physical RAM 3072 MB
Model PowerMac7,3
Processor PowerPC G5x2 @ 2.70 GHz
L1 Cache 64K (instruction), 32K (data)
L2 Cache 512K @ 2.70 GHz
Bus Frequency 1 GHz
Video Card GeForce 6800 Ultra
Drive Type Hitachi HDS724040KLSA80
CPU Test 136.73
GCD Loop 135.74 7.16 Mops/sec
Floating Point Basic 135.24 3.21 Gflop/sec
vecLib FFT 140.73 4.64 Gflop/sec
Floating Point Library 135.35 23.57 Mops/sec
Thread Test 136.73
Computation 139.18 2.82 Mops/sec, 4 threads
Lock Contention 134.36 5.78 Mlocks/sec, 4 threads
Memory Test 102.83
System 104.67
Allocate 142.29 522.53 Kalloc/sec
Fill 139.42 6778.70 MB/sec
Copy 69.16 1428.42 MB/sec
Stream 101.05
Copy 101.85 2103.59 MB/sec [G5]
Scale 101.74 2101.95 MB/sec [G5]
Add 100.63 2143.73 MB/sec [G5]
Triad 100.00 2139.22 MB/sec [G5]
Quartz Graphics Test 135.14
Line 132.53 8.82 Klines/sec [50% alpha]
Rectangle 131.38 39.22 Krects/sec [50% alpha]
Circle 132.36 10.79 Kcircles/sec [50% alpha]
Bezier 132.24 3.34 Kbeziers/sec [50% alpha]
Text 148.70 9.30 Kchars/sec
OpenGL Graphics Test 136.03
Spinning Squares 136.03 172.56 frames/sec
User Interface Test 133.26
Elements 133.26 611.58 refresh/sec
Disk Test 58.46
Sequential 94.22
Uncached Write 101.07 62.06 MB/sec [4K blocks]
Uncached Write 90.15 51.00 MB/sec [256K blocks]
Uncached Read 87.06 25.48 MB/sec [4K blocks]
Uncached Read 100.18 50.35 MB/sec [256K blocks]
Random 42.38
Uncached Write 15.43 1.63 MB/sec [4K blocks]
Uncached Write 89.89 28.78 MB/sec [256K blocks]
Uncached Read 93.40 0.66 MB/sec [4K blocks]
Uncached Read 129.35 24.00 MB/sec [256K blocks]
You'll notice I scored higher on CPU, which obviously means that tests a single CPU. Looks like one of the biggest hits i took was on HD, which isn't surprising because hard drives slow down once you get a bunch of stuff on them. I scored a lot lower on memory, which is how it should be. The only real surprise is that I scored lower on graphics tests... The 6800 Ultra drivers really bite. The thing you really need to take note of, is I scored WAY lower on the thread test.
 

risc

macrumors 68030
Jul 23, 2004
2,756
0
Melbourne, Australia
Everyone knows XBench is a load of crap, I put a faster video card and faster HDD in my Power Mac and the score went down. ;) If you run it 10 times you'll get 10 varied answered, and I absolutely hate the way the people that made XBench constantly change the base line so you update and you go from 220 points down to 120.

edit: I wouldn't be suprised either if XBench doesn't work correctly with the 4 cores of the Quad yet either.
 

d.perel

macrumors regular
Feb 3, 2005
204
0
macbaseball said:
That's what I thought too, but I just tested my Dual 2.3 and I got 115.
Does anyone know what some comparative mac or pc setups get?
 

iGary

Guest
May 26, 2004
19,583
0
Randy's House
They just changed the scoring in the software - my dual 2.7 used to get in the 260's and just now got 124 something.

Seems like the quad would blow the dual 2.7 out of the water. A Cinebench would be better, me thinks.

Whatever. :)
 

iGary

Guest
May 26, 2004
19,583
0
Randy's House
For example:

Gary Reich's iMac G5 156.42 iMac (Flat Panel) 2004-11-02 14:55:43.0
Gary Reich's iMac G5 152.28 iMac (Flat Panel) 2004-12-27 13:45:22.0
Gary Reich's iMac G5 155.70 iMac (CRT) 2005-03-04 10:32:45.0
Gary Reich’s Computer 69.78 PowerMac G4 (Sawtooth) 2003-11-20 11:25:21.0
Gary Reich’s Computer 135.86 PowerMac G4 (MDD) 2003-11-20 15:04:31.0
Gary Reich’s Computer 134.24 PowerMac G4 (MDD) 2003-11-20 15:08:53.0
Gary Reich’s Computer 135.84 PowerMac G4 (MDD) 2003-11-20 15:59:39.0
Gary Reich’s Computer 135.84 PowerMac G4 (MDD) 2003-11-20 16:03:13.0
Gary Reich’s Computer 137.33 PowerMac G4 (MDD) 2003-11-20 16:38:36.0
Gary Reich’s Computer 130.30 PowerMac G4 (MDD) 2003-11-24 20:11:24.0
Gary Reich’s Computer 69.13 PowerMac G4 (Sawtooth) 2003-12-04 12:44:59.0
Gary Reich’s Computer 141.95 PowerMac G4 (MDD) 2004-03-16 18:32:06.0
Gary Reich’s Computer 141.16 PowerMac G4 (MDD) 2004-09-15 17:06:19.0
Gary Reich’s Computer 139.72 PowerMac G4 (MDD) 2004-10-04 17:37:37.0
Gary Reich’s Computer 140.87 PowerMac G4 (MDD) 2004-10-07 14:17:39.0
Gary Reich’s Computer 101.30 iBook (White) 2004-10-29 12:20:46.0
Gary Reich’s Computer 101.82 iBook (White) 2004-10-29 12:29:32.0
Gary Reich’s Computer 101.94 iBook (White) 2004-11-02 12:25:53.0
Gary Reich’s Computer 135.05 PowerMac G4 (MDD) 2004-12-15 19:47:10.0
Gary Reich’s iMac G5 156.98 iMac (Flat Panel) 2005-05-09 12:41:24.0
Gary Reich’s Power Mac G4 137.04 PowerMac G4 (MDD) 2005-05-02 20:18:02.0
Gary Reich’s Power Mac G5 267.48 PowerMac G5 (Orig) 2005-09-01 17:50:31.0
Gary Reich’s Power Mac G5 268.28 PowerMac G5 (Orig) 2005-08-13 12:52:26.0
Gary Reich’s Power Mac G5 269.37 PowerMac G5 (Orig) 2005-07-14 18:22:15.0
Gary Reich’s Power Mac G5 271.43 PowerMac G5 (Orig) 2005-07-16 03:49:40.0
Gary Reich’s Power Mac G5 260.94 PowerMac G5 (Orig) 2005-07-29 16:09:26.0
Gary Reich’s Power Mac G5 124.87 PowerMac G5 (June 2004) 2005-11-10 12:02:37.0
 

Bear

macrumors G3
Jul 23, 2002
8,089
4
Sol III - Terra
What's worse is..

You need to make sure the same version of XBench was run on the systems you are comparing, otherwise the numbers are can be meaningless.
 

jhu

macrumors 6502a
Apr 4, 2004
854
0
why don't we run benchmarks with linpack? that way we can also compare results with supercomputers.
 

zakatov

macrumors 6502a
Mar 8, 2005
501
0
South Florida
I believe the latest version's baseline is: dual 2Ghz = 100 Xbench points.

EDIT: "Re-calibrated 100 point baseline to a 2.0 GHz G5 running Tiger"

Now the real question is whether Xbench is hitting all the CPUs hard enough, or whether the bottleneck is elsewhere.
 

risc

macrumors 68030
Jul 23, 2004
2,756
0
Melbourne, Australia
darthkuru said:
i thought the bus speed on the quad 2.5s was 1.25? it says 1.0 GHz in the results.
XBench probably has no idea how to correctly benchmark a quad, that is why this thread is pretty pointless. What I'd like to see from the OP is pictures of the machine, and get his comments on how it "feels".