First G5 Dual 2GHz owner photos/initial tests: See this page posted Sept. 4th, 2003.
First Impressions from WWDC 2003:
A reader at WWDC posted his comments on the new Dual 2GHz G5 system.
"I finally got a chance to look at one today, and made sure to drill the engineer guarding it for questions.
The case appears slightly bigger than the old quicksilver cases were. The front is quite plain...just the three ports,
button, and cd tray- no more. The side is unlocked via a back latch which can be secured so there's no way to break in.
Once the side is unlocked, you can completely remove the side panel, revealing the guts of the machine. A translucent
plastic piece covers the lower three quadrants, forcing their venting to work properly. The system will not boot if this is
Once that final plastic cover is removed, you gain access to every part of the computer except the power supply which
is hidden in the lower compartment. Rotating storage (optical drive (singular) and hard drives (up to two) are stored in
the uppermost compartment. The next compartment down holds the 8X AGP slot and 3 PCI-X slots...2 100MHz, 1
The third compartment houses the massive heat sinks on top of the G5's (further mentioned later) and the 8 memory
slots. The memory in this machine MUST be upgraded in pairs, working from the outside inwards. This is not a
suggestion, it's a requirement.
(dual channel DDR/PC3200)
The heat sinks on the processors are enormous- a good 4 inches tall. One thing he said is that it will be extremely
difficult to upgrade the chips in this machine. Each computer is uniquely paired with its processors to have perfect
venting. Changing out processors would throw off the chip that handles the G5 cooling, resulting in (at best) the
computer running for a while then shutting off, or (at worst) a non-booting machine.
Though I didn't get a chance to run XBench on the G5, I was very impressed. The boot screen came up just long
enough for me to know that it came up- it was VERY fast. The case design isn't my favorite, but the quietness of the
fans is well worth it. Keep in mind the machine design is not finalized yet, and Apple will still make some small
modifications before shipping.
Another reader at WWDC sent an email with Xbench 1.0 benchmark results, but I wanted to wait and post them after I could include results from a previous Dual G4 system (preferably a dual 1.42GHz model) for comparison. Granted it's just a benchmark... (Note - he later said it was run with several other apps in the background, so the results are not useful for comparison. Also some Xbench tests may use instructions that have poor performance on the G5 mentioned in the Tech note/PowerPC G5 Performance Primer.)
G5 Superdrive Model: Although Apple often uses more than one OEM brand/source for their drives - this reader with a new G5 system said the superdrive was the OEM Sony DW-U10A. (I think the same drive has been used in some iMac G4s or other mac models in the past - it has native burn support in 10.2.6 and perhaps even an earlier OS X version).
"I don't know if anyone has let you know, but I just received my 1.8GHz G5. It has the Sony DVD RW DW-U10A for the SuperDrive. This was a pleasant surprise.
From what I've heard in the past, the Apple shipped OEM drives did not support DVD+R/+RW burning like the standard DW-U10A does. I asked Greg if he could test for DVD+R/+RW support. (For past reports on that drive - search the drive database here selecting Sony as the drive brand and "DVD+R/RW + DVD-R/RW" as the drive type. Omit mac model to not limit search results to a specific model.)
G5 System Photo Galleries:
High-Res G5 Photo Gallery: (from the Sept. 3rd www.xlr8yourmac.com news page)
Thought I'd share some "artsy" high resolution G5 photos with you and your readers:
Very well done... this guy should be on Apple's payroll.
Apple G5 Photos and Movie: Apple's Powermac G5 pages link to an Apple G5 movie and a few photos in the Apple photo gallery (more detailed photos are linked below).
(July 2003 links, before G5s shipped) Macitynet has some pix of the new G5 systems taken at WWDC at their G5 photo gallery and later posted page with more detailed G5 system photos. 99mac.com also has some nice photos of the new G5 systems.
Apple G5 Tech Notes/Docs:
Apple White Papers on G5 System and G5 CPU: (added Aug. 26th, 2003) Apple has
Kbase docs with links to PDF files of G5 related white papers (English and Japanese PDF doc links there):
G5 Developer Doc Finialized: The Apple G5 Developer Doc (previously preliminary) has been finalized and released this morning. The PDF revision date is August 20th.
The final version of the PowerMac G5 Developer Note is now available
The developer note for the newly introduce PowerMac G5 is now available in
HTML and .pdf format.
(Actual direct links to HTML version is
here and here is the PDF doc.)
This developer note provides technical, hardware architecture information on
the PowerMac G5 announced last month.
It includes information about the major components on the logic board: the
microprocessor, the other main ICs, the buses that connect them to each other
and to the I/O interfaces."
G5 Preliminary Developer Doc: (from the July 3rd, 2003 main site news page)
A reader posted a link to the preliminary developer doc (PDF file) for the new PowerMac G5. (Includes notes on system design, Power supply (450W and 600W), CPU module, interfaces, PCI slots, chipset, ram and more.)
Tech Note: G5 Performance Primer: Apple has a Tech note/PowerPC G5 Performance Primer that outlines the design differences in the G5 vs G4 CPUs. Sections include:
- Quick Comparison: G4 & G5
- Summary of Differences between the G4 and G5
- Performance Do's for the G5
- Performance Don'ts for the G5
- How to make your code run best on the G5
Apple G5 Customer Installable Parts Guides: Apple has posted a kbase doc page on Power Mac G5 CIP (Customer Installable Parts) Instructions with links to PDF guides to installing/replacing boards, drives, cables, etc.
Apple G5 Kbase Docs:
G5 Performance Tests/Articles:
G5 vs G4 vs G3 Codewarrior/GCC Benchmarks (from the Sept. 4th, 2003 news page)
(from a reader email)
Per your request for G5 benchmarks (requested real-world tests actually-Mike) here are some comparisons between G3, G4,
& G5 for one specific compute intensive task. I would not have predicted some of the results. For example, better speed in classic and differences between CodeWarrior (which I thought used gcc ??) and gcc. The lowly 1.6GHz G5 came out looking pretty good though.
What I havn't included here is that even the G4 trounces a 2.4 GHz P4 Xeon and some 900 MHz (many $$) Sun Blades :)
Best regards, Brent
- Task is a numerically intensive (double precision floating point) field
- Single task running except where noted. Disk I/O is excluded from timing. Small footprint (about 2.5 Megs).
- iBook running 10.2.3, B&W and G4 running 10.2.6, G5 running 10.2.7.
Case 1: CodeWarrior v8 compiler, G3 target, O3 optimisation, classic
Clock Time Speed
Processor Clock Ratio secs Ratio
G3 B&W 450 MHz 0.32 164 0.34
G3 iBook 800 MHz 0.56 92 0.56
G4 DP 1.42 GHz 1.00 56 1.00
G5 1.6 GHz 1.13 37 1.51
Case 2: CodeWarrior v8 compiler, G4 target, O3 optimisation, MachO
Clock Time Speed
Processor Clock Ratio secs Ratio
G3 B&W 450 MHz 0.32 170 0.35
G3 iBook 800 MHz 0.56 98 0.60
G4 DP 1.42 GHz 1.00 59 1.00
G4 DP 1.42 GHz 1.00 62 0.95 (2 tasks running)
G5 1.6 GHz 1.13 42 1.40
Case 3: gcc 3.1 compiler, G4 target, O3 optimisation, MachO (terminal) application.
Clock Time Speed
Processor Clock Ratio secs Ratio
G3 B&W 450 MHz 0.32 198 0.39
G3 iBook 800 MHz 0.56 94 0.82
G4 DP 1.42 GHz 1.00 77 1.00
- No speed hit when running in classic (actually 5% faster on CWv8_G3O3 !) as opposed to OSX native (MachO).
- CodeWarrior produces faster code than gcc.
- Target specific compiler optimisation (eg: running G4-targeted as opposed to G3-targeted code on a G4) did not make much difference.
- For both CodeWarrior and gcc compilers, no performance gains (and sometimes some degradation) from higher than O3 optimisation.
- Dual G4 takes only about a 5% speed hit per task when running 2 identical tasks simultaneously.
- G3 faster (between 0% and 46%) than G4 when normalise by clock.
- G5 faster (between 24% and 34%) than G4 when normalise by clock.
CAVEAT - These results and conclusions valid for this particular benchmark only !
Reader Tests on AIFF-MP3 conversion with G5 vs G4s (from the Sept. 3rd, 2003 www.xlr8yourmac.com news page)
I wanted to let you know about a test I did at the Apple Store in
Glendale, CA. I called and they said they had a G5 1.6 on display, so
I went down there with an audio CD to test the speed of an iTunes
AIFF->mp3 conversion against other machines at the store. I ripped an
audio CD track to AIFF format on each machine's hard disk to eliminate
the bottleneck caused by optical drive speed. Here are the details and
Audio CD: Herbie Hancock "Headhunters"
Track: "Chameleon" - funk/jazz tune, 15 minutes and 44 seconds long,
ripped to hard disk as an uncompressed 16 bit, 44.1 kHz stereo AIFF
file using iTunes.
Task: Using iTunes, convert the AIFF file to mp3 ("Higher Quality"
setting, 192 kbps). I didn't have a stopwatch on me, so what I'm reporting here is
the highest speed multiple iTunes reported during conversion.
Machines tested (all had 256 MB RAM and iTunes 4.0.1 (117), and on all
I quit all other applications, turned off every sound effect in iTunes,
and unchecked "play songs while importing" and "create file names with
- G5 1.6: - 23.2x
- G4 Dual 1 GHz (MDD): - 20.6x
- G4 1.25 (single processor): - 18.4x
- iMac 1 GHz: - 13.1x
- iMac 800 MHz:- 10.4x
I'm surprised at how well a single 1.25 G4 holds up compared to a
single 1.6 G5. If one speculates mathematically what a 1.6 G4 would
do, it's 23.5x, which is the same as the 1.6 G5. I wonder how much of
a speed increase we'll see when iTunes is optimized for the G5. Hope
the above is useful to other readers of your site.
If any G5 owners have other real-world test results from production systems (not pre-release Dual G5, etc. - systems they actually own. i.e. iMovie Tutorial export, QT DV->MPEG conversion, FCP or FCExpress tests, etc.) let me know. (Include system details in reports - thanks.)
Dual G5 2GHz Tests with PShop7, FCP4 and Soundtrack: (from the Aug. 15th, 2003 news page - note this was before Adobe released the G5 PS7 Plugin update) There's a page at soundtracklounge.com with Photoshop7/Final Cut Pro 4/Soundtrack tests using a "pre-production" Dual 2GHz G5. The comparison G4 machine was a single 933MHz CPU. (It would have been nice to see how it compared to a dual G4 1.25GHz or dual G4 1.42GHz model, but the results are still interesting.)
After Effects (short) Comparison Test with G5 Dual 2GHz (from the June 26th main news page)
I'm getting this 2nd/3rd hand, so I can't verify this, but just a FYI since in the past Mac vs PC After Effects tests (G4 duals vs PCs) showed the PCs much faster as I remember (reportedly Adobe's software didn't seem to take advantage of dual CPUs nearly as well as Apple's Final Cut Pro - but that app is not cross-platform). I personally despise the platform war debates (it brings out the worst in everyone it seems) but here's an email I received today. (Note also it's not a full Night Flight test - only the first 10 frames.)
Hi Mike, I belong to the After Effect-List, which exchanges information, tips, and techniques for both the PC and Mac versions of Adobe After Effects.
One of the listees was at WWDC, and ran a specific AE project that we all use to evaluate processor speed and performance.
Here is a copy and paste from his post this morning (Thursday):
"Hey everyone, I've been at Apple's developer conference and had a chance to install and try out After Effects on a new G5.
I ran the Night Flight file that has come to be the standard for AE
benchmarking. Since I didn't want to sit there and watch it render for
hours, I ran just the first 10 interlaced frames from the project's
pre-set render queue...
Here are my results for this test on the three computers I have available to me:
1 x 1.0 GHz G4 PowerBook 17" - ~30 minutes (3 min/frame)
2 x 2.66 GHz Pentium Xeon from Boxx - 11 min, 39 sec (1.2 min/frame)
(actually 1.165 min/frame)
2 x 2.0 GHz PowerMac G5 - 6 min, 1 sec (0.6 min/frame)
I ran the Xeon test on a couple different identical machines to make sure mine wasn't just running slowly, but got identical results.
Of course my Mac bias is well-documented, but I'm sure many people here can vouch for me as an honest person. If the results had gone the other way, I'd just keep my mouth shut and let someone else break the bad news.
Other observations about this test that may ultimately work in the Mac's favor:
1) The machine was not running 64-bit Panther, but only a tweaked version of 32-bit Jaguar. Likewise, AE is obviously not yet compiled to take advantage of the G5 chip in any way. Both or these situations will automatically be rectified in the future.
2) Night Flight is very CPU-intensive, but not very disk I/O intensive.
I think the 1 GHz system bus and other details on the G5 will provide
greater gains for typical projects that rely more heavily on I/O."
All pretty interesting!!!!
Another mail from an AE List subscriber said the performance test above was
from Brendan Bolles, a respected AE plug-in developer and included a direct link to the post at
It will be interesting to see the results of more complete tests when the G5 systems ship.
G5 vs G4 Java Benchmark Performance: (from the June 30th, 2003 main site news page)
Hello, I write you to try to clear some things up a bit.
In many forums there are arguments about how the G5 compares to the G4. The Xbench results may have caused the G5 look a bit poor against the G4. (this may be due to compiler or issues with some instructions on the G5 mentioned in the Apple Tech note/G5 Performance Primer-Mike)
Since I was at the live streaming event from Berlin to SF, I was able
to test the new G5 Dual 2Ghz a little bit.
The benchmark I was able to run is a web based Java Linpack benchmark.
While it is actually a benchmark of how good the implementation of Java is on the specific OS it is a good way to compare the old G4 to the G5.
In this list are the results of different CPU's performing this bench.
If you scroll down at around 90Mflop/s you will find the top of the
line G4 Dual 1.42Ghz.
My machine a G4 933Mhz scores 66Mflop/s under OS X 10.2.6 or under 9.2.2 77.4Mflop/s. If you now scroll to the very top of the chart you will find the Dual 2Ghz G5 scoring an amazing 236Mflop/s.
This benchmark is very dependant on the memory bandwidth, this is why nearly all P4's score very high.
Despite the fact that the Java implementation lacks a bit under OS X it
clearly shows the power of the G5. This benchmark is in no way optimized for the G5 and also lacks SMP
features. But the real strength of the G5 would be demonstrated in a Linpack double precision 1K x 1K because of its 64Bit capabilities.
Regards, Christian D.
One reason for some lower-than-expected G5 results in Xbench (and some other apps/benchmarks perhaps) may be due to the instruction issues noted in Apple's Tech note: G5 Performance Primer noted here in last week's news page. (compares G5 to G4 architecture, notes on deeper pipelines in the G5, problems with some instructions, pipeline 'bubbles' and dos/don'ts for G5 programmers.)
The Fast and Furious G5 Case Mod:
New G5 case mods already in the works:
Thought you might appreciate the lighter side. Made it myself. ;)
G5 intercooler and Nitrous optional... Just needs a few more decals and lowering the case until it drags the table when moved then it'll be ready for "Super2NR TV" or "High Rev Tuners"...