DIY Benchmark

[MagicBoy]

OK, here's one for you. Ran the original benchmark twice on my 2009 Mac mini 2GHz (RAM and storage upgraded).
Leopard - 3m 13s
Yosemite - 48s

Checked the log and Leopard is running 11 encode passes vs 3 on Yosemite. Wonder why?

I'm going to try the same handbrake test on the Mac mini just out of curiosity.

EDIT : 55m32s for the Mac mini. So 4.5 times slower which is in line with the GeekBench 2 (64) scores.

 

[lowendlinux]

Here’s the dual 5675 Hackintosh on 10.9.5 so I’m looking a 2:25 but I also don’t see 11 passes.

Second run is much faster but I did it to see how much it would spike the processors and it never went above 10.83 percent.

 

[flyrod]

...so I’m looking a 2:25 but I also don’t see 11 passes.

Did you check the "Multipass" box? I'm seeing slightly different behavior with this program on 10.4 vs 10.5 but I'm booting 10.5 from a firewire disk. Maybe that is causing a problem...

 

[flyrod]

15:15 On my 12" powerbook running 10.4.11. It's much faster with "Multipass" unchecked.

 

[flyrod]

I think there are some discrepancies in the results posted so far related to the settings. From the default window, I count 5 changes needed to match settings.jpg

Quality to 100%
Multipass checkbox
Sound to 128kbps
Width to 640
Height to 360

If we're not using the same settings, the results are not really comparable. For example, if you don't change Quality or check the Multipass, it takes fewer passes and completes much faster (20 or 30s on my powerbook).

I think the weirdness I was seeing on 10.5 was because I was booting it on an external drive, and of course it wanted to immediately index the universe (never mind what I wanted), then it wanted to run the flurry screen saver on top of that....

 

[Dronecatcher]

If we're not using the same settings, the results are not really comparable.

Yeah, it's vital to use the settings outlined in the instructions for the results to have meaning.

I've just bought a 1.67 15" DLSD Powerbook, its result is 10:07

 

[Dronecatcher]

G5 Quad Leopard 7Gb 00:01:59 (3559)

 

[flyrod]

G5 Quad Leopard: 00:03:12

I wonder why the disparity? Some of the other results seem repeatable.

What if you run the G5 in "reduced" mode?

 

[Dronecatcher]

What if you run the G5 in "reduced" mode?

Reduced mode will certainly have an impact - not sure on the figures but I think it clocks the CPU down to around 60/70%.
I alway set to Highest for benchmarking and leave on Automatic for rest of the time...unless the fans are noisy in which case I leave on Reduced.

 

[flyrod]

Right, I was wondering if running in reduced mode would duplicate the slower result fuzzface got with the same computer.

 

[jbarley]

Just tried this benchmark test on my G5 quad, I've posted the log results hoping someone will interpret them for me.

 

[flyrod]

Looks like 1:58 (27:59 - 26:01), which is what Dronecatcher got on his quad. Nice to see a repeatable test; thanks for trying it!

 

[jbarley]

Thanks, I thought that might be the result but wanted to make sure I had calculated it correctly.

 

[MacCubed]

My Quad got around 3 mins on Max Processor

 

[MacCubed]

Although that may be the RAM, 3GB and some of the PC2-5300 is getting downclocked to PC2-3200 speeds... And if the graphics card is also involved, that may be it too. Its a 6600

 

[QSDP-User]

Thought I'd throw my two fastest PPCs' results into this thread.
And some machine specs, if that matters.

Sawtooth/Tiger 10.4.11
Max RAM (2 GB)
Single 1 GHz upgrade (7450)
Flashed Radeon 9800 128MB
Flashed Wiebe SATA3112 PCI card -- SATA drive running benchmark

File: test.mp4
Saturday, 18:57:34: Export Stream started
Saturday, 18:58:12: Encoding
Saturday, 18:59:13: Encoding
Saturday, 19:01:43: Encoding
Saturday, 19:04:32: Encoding
Saturday, 19:07:20: Encoding
Saturday, 19:10:09: Encoding
Saturday, 19:12:58: Encoding
Saturday, 19:15:47: Encoding
Saturday, 19:18:35: Encoding
Saturday, 19:21:24: Encoding
Saturday, 19:24:14: Encoding
Data rate 4.70 Mbps (4.57 video, 0.13 audio)
Saturday, 19:24:37: Movie completed

19:24:37 Completed
18:57:34 Started

The math
18:84:37
18:57:34
--------
00:27:03

=====================================

Quicksilver 2002 Dual 1 GHz (7450) / Leopard 10.5.8
Max RAM 1.5 GB
GeForce4 MX 64MB
IDE drive/default built-in IDE

File: test.mp4
Saturday, 18:27:42: Export Stream started
Saturday, 18:28:01: Encoding
Saturday, 18:28:23: Encoding
Saturday, 18:29:06: Encoding
Saturday, 18:29:53: Encoding
Saturday, 18:30:41: Encoding
Saturday, 18:31:28: Encoding
Saturday, 18:32:15: Encoding
Saturday, 18:33:02: Encoding
Saturday, 18:33:50: Encoding
Saturday, 18:34:37: Encoding
Saturday, 18:35:24: Encoding
Data rate 4.79 Mbps (4.67 video, 0.13 audio)
Saturday, 18:35:30: Movie completed

18:35:30 Completed
18:27:42 Started

The math
18:34:90
18:27:42
--------
00:07:48

I hope my damn math's correct
Posted from Quicksilver running Leopard WebKit.

 

[Dronecatcher]

Just tried this benchmark test on my G5 quad, I've posted the log results hoping someone will interpret them for me.

View attachment 601308

What version Quicktime are you running? Now that this test takes twice as long for me, a possible culprit might be Quicktime - I might have been at 7.5.5 when I tested originally.

 

[jbarley]

What version Quicktime are you running? Now that this test takes twice as long for me, a possible culprit might be Quicktime - I might have been at 7.5.5 when I tested originally.

Quicktime version 7.7

 

[Dronecatcher]

Quicktime version 7.7

Thanks...same version as I have now - back to the drawing board on reasons for the slow down!

 

[jbarley]

Something odd has happened and I have no idea as to what.
But last fall when I first ran your DIY benchmark I got a result of 1.58 min.
Now the best I can get is 4:09 min.
Same G5 quad with all the same settings.

 

[Dronecatcher]

Something odd has happened and I have no idea as to what.
But last fall when I first ran your DIY benchmark I got a result of 1.58 min.
Now the best I can get is 4:09 min.
Same G5 quad with all the same settings.

Now that is strange - exactly the same issue as me - mine went from 1.59 to 3.27. Apart from the possibility I might have had Quicktime 7.5.5 to begin with (though doubtful) the only other codec related app I've installed is Handbrake.

 

[jbarley]

What version Quicktime are you running? Now that this test takes twice as long for me, a possible culprit might be Quicktime - I might have been at 7.5.5 when I tested originally.

It seems you were spot on about Quicktime.
I just downgraded back to QT 7.5 using Pacifist and the resulting improvement was instantaneous.
59 seconds, I can live with that.

 

[Dronecatcher]

It seems you were spot on about Quicktime.
I just downgraded back to QT 7.5 using Pacifist and the resulting improvement was instantaneous.
59 seconds, I can live with that.

Great stuff! I didn't know about Pacifist - I wouldn't have attempted to downgrade without it. Back to 7.5.5 and now clocking in at 01:58
I'm guessing the h264 codec in 7.7 is more fully featured and being from 2011 not really sympathetic to PPC.

 

[jbarley]

Something else I just noticed, and I have no idea whatsoever as to why,
My "TenFourBox" apps in the dock now start with just 1 sometimes 2 bounces, whereas before with QT 7.7 they bounced
3 or 4 times to start.
I've checked this out several times, I can boot from different partitions, one with QT 7.5 and the other with QT 7.7.

 

[Dronecatcher]

Something else I just noticed, and I have no idea whatsoever as to why,
My "TenFourBox" apps in the dock now start with just 1 sometimes 2 bounces, whereas before with QT 7.7 they bounced
3 or 4 times to start.
I've checked this out several times, I can boot from different partitions, one with QT 7.5 and the other with QT 7.7.

Unfortunately I've had to put 7.7 back on - rolling back to 7.5.5 had broken Safari and iTunes. When I do a fresh install will be sticking to the older version though.
I wonder how many other apps and UI features are held back a little by 7.7?