Power Mac G5 Quad Benchmarks

[melgross]

That high-speed memory costs $$$$. I imagine that the unit cost would rise to unacceptable levels if they added 1MB more cache...especially for the Quad.

It does cost more. But the cost to Apple of the new dual core chips WITH double the L2 cache is LESS that the cost of two single core chips with 512KB L2 cache each. The cost of both dual core chips, therefore, is less than the cost of the two chips that preceeded them.

That last sentence is not what I meant. What I did mean is that the Two dual core chips costs Apple less that if they had purchased four chips to replace the two that preceeded the Quad. Sorry.

 

[jhu]

it floors me that IBM crippled the G5. They truly could have made something great. oh well... my new DC 2.3 get's the job done just fine. i basically use my mac for music production (using Logic), surfing and stuff like that. i know that logic would benefit from the quad since it's a multi-threaded application, but my projects don't require that much horsepower....or at least that's how i reason to myself for not spending the extra $$$$ on the quad...i used it for Logic.

and encroach on the performance of their power4/5 chips? that's probably not a good idea either.

 

[melgross]

and encroach on the performance of their power4/5 chips? that's probably not a good idea either.

The POWER 4 is already gone, no problem there. The 5 is also gone, replaced with the 5+.

Before too long, that will be replaced with the 6, which is much more powerful.

The 970 is based on the Power 4.

I don't think that IBM has to worry about encroachment.

 

[xoct]

FCP on Quad vs. 2.7 Dual

We just got our Quad today to accompany our 2.7 Dual. Both have the same 4.5GB RAM and the same standard graphics card. In the latest update of FCP 5, we took a 1080i HD shot and applied a Gaussian Blur. The render time on the Quad was 50% faster than the 2.7 Dual. I can't remember how long the clip was, but it took about three minutes for the Quad, and about six minutes for the 2.7 Dual. We were all very pleased. The next big test I'd like to do is to take a QT movie and compress it as an H.264 because that's so darn slow to do on the 2.7 Dual.

 

[melgross]

looks good

We just got our Quad today to accompany our 2.7 Dual. Both have the same 4.5GB RAM and the same standard graphics card. In the latest update of FCP 5, we took a 1080i HD shot and applied a Gaussian Blur. The render time on the Quad was 50% faster than the 2.7 Dual. I can't remember how long the clip was, but it took about three minutes for the Quad, and about six minutes for the 2.7 Dual. We were all very pleased. The next big test I'd like to do is to take a QT movie and compress it as an H.264 because that's so darn slow to do on the 2.7 Dual.

Can you try an iTunes rip? Those that I've seen show it to have the same speed on a Quad as on a dual cpu. I'm wondering, if that's true, if iTunes is properly threaded, or if it only has two threads to work with.

Thanks, in advance.

 

[groovebuster]

It actually is single-threaded!

iTunes relies completely on the QuickTime engine and QT can handle only one processor core at a time... So no gain by a multi-processor machine at all. 1 core or 100 cores... no difference.

I wonder when Apple will sort that out finally...

BTW... an application that truly is designed for multi-threading scales automatically with the number of available processors.

groovebuster

Edit: Do the test yourself... switch off one of the processors and rip the same song in iTunes again. Should take almost exactly the same time...

 

[anthonylambert]

I recieved my QUAD yesterday... I noticed that the bus speed reported by XBench says only 1gz not 1.25gz as per the specs. The earlier benchmarks reported here say the same.... hey what's going on???

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

I'm running the standard apple supplied memory...

Tony

 

[anthonylambert]

Any recomendations...

Anyone got any recomendations for free/shareware apps that will really graphically show off my lovely new Quad machine?

I saw a program which used to model the sea or waves anyway in realtime in an apple show a year or two back any idea what it was? might of been a screen saver.

Tony

 

[BakedBeans]

I recieved my QUAD yesterday... I noticed that the bus speed reported by XBench says only 1gz not 1.25gz as per the specs. The earlier benchmarks reported here say the same.... hey what's going on???

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

I'm running the standard apple supplied memory...

Tony

The answer?

xbench sucks!

 

[RaVeNChAoS]

Using Photoshop CS or CS2 (or PS 7)
==================================================
1.) Download the test image from http://www.quicklance.com/test.jpg
2.) Save it to the computer and then open it up in Photoshop
3.) From there please apply a ‘radial blur’ with the settings at:
Amount = 100
Blur Method = Spin
Quality = Best
Using a stop watch / ps timer see how long it takes to apply this filter
I just want to see what these new cpu’s can really do.

Results:
iMac G5 1.8GHz, 1GB - 2:00
Athlon XP3200+, 1GB - 2:15
Athlon64 4000+, 1GB - 1:25
Dual 2.5 Running 10.4.2 with 2.5 GB RAM 40 seconds
PowerMac Dual 2.7 Dell 2405 FPW, 2.5 gigs of ram, Radeon 9650 42 seconds
Dual Core 2.0 GHz G5 with 2.5GB ram Photoshop CS2 47.4 seconds

Quad 2.5GHz G5 2.5GB RAM 10.4.3 22 seconds !!!!! It's an amazing huh??!!!

Just got my new G5 Quad, couldn't afford ram yet lol so here are the results.
Photoshop test as stated above:
Quad 2.5GHz G5 | 512mb RAM | Nvidia 6600 | OS 10.4.3 19.7 seconds (I don't know why yours took longer with more ram.
However so far the cinebench results kind of suck imo. I have kept a database of results for cinebench on the following systems I have tested before.

CINEBENCH 2003 v1
****************************************************

Tester : RaVeN

Processor : quad g5
MHz : 2.5
Number of CPUs : 4
Operating System : 10.4.3

Graphics Card : 6600 nvidia
Resolution : <fill this out>
Color Depth : <fill this out>

****************************************************

Rendering (Single CPU): 354 CB-CPU
Rendering (Multiple CPU): 903 CB-CPU

Multiprocessor Speedup: 2.55

Shading (CINEMA 4D) : 351 CB-GFX
Shading (OpenGL Software Lighting) : 1045 CB-GFX
Shading (OpenGL Hardware Lighting) : 1850 CB-GFX

OpenGL Speedup: 5.28

****************************************************

CINEBENCH 2003 v1
****************************************************

Tester : RaVeN

Processor : G5
MHz : 2.5 DP
Number of CPUs : 2
Operating System : 10.3.8

Graphics Card : 9600
Resolution : <fill this out>
Color Depth : <fill this out>

****************************************************

Rendering (Single CPU): 356 CB-CPU
Rendering (Multiple CPU): 635 CB-CPU

Multiprocessor Speedup: 1.78

Shading (CINEMA 4D) : 336 CB-GFX
Shading (OpenGL Software Lighting) : 937 CB-GFX
Shading (OpenGL Hardware Lighting) : 1641 CB-GFX

OpenGL Speedup: 4.89

****************************************************

CINEBENCH 2003 v1
****************************************************

Tester : RaVeN

Processor : Dual Opteron 264
MHz : 2ghz
Number of CPUs : 2
Operating System : xp pro sp1

Graphics Card : ati 9600
Resolution : <fill this out>
Color Depth : <fill this out>

****************************************************

Rendering (Single CPU): 272 CB-CPU
Rendering (Multiple CPU): 507 CB-CPU

Multiprocessor Speedup: 1.86

Shading (CINEMA 4D) : 304 CB-GFX
Shading (OpenGL Software Lighting) : 1399 CB-GFX
Shading (OpenGL Hardware Lighting) : 2754 CB-GFX

OpenGL Speedup: 9.06

****************************************************

CINEBENCH 2003 v1
****************************************************

Tester : RaVeN

Processor : p4
MHz : 3.2
Number of CPUs : 2
Operating System : xppro

Graphics Card : geforce5200
Resolution : <fill this out>
Color Depth : <fill this out>

****************************************************

Rendering (Single CPU): 280 CB-CPU
Rendering (Multiple CPU): 334 CB-CPU

Multiprocessor Speedup: 1.19

Shading (CINEMA 4D) : 359 CB-GFX
Shading (OpenGL Software Lighting) : 1183 CB-GFX
Shading (OpenGL Hardware Lighting) : 1134 CB-GFX

OpenGL Speedup: 3.29

****************************************************

CINEBENCH 2003 v1
****************************************************

Tester : RaVeN

Processor : G4 MDD 2gb
MHz : 1.25 DP
Number of CPUs : 2
Operating System : 10.4.2

Graphics Card : 9800 pro
Resolution : <fill this out>
Color Depth : <fill this out>

****************************************************

Rendering (Single CPU): 120 CB-CPU
Rendering (Multiple CPU): 218 CB-CPU

Multiprocessor Speedup: 1.82

Shading (CINEMA 4D) : 150 CB-GFX
Shading (OpenGL Software Lighting) : 381 CB-GFX
Shading (OpenGL Hardware Lighting) : 712 CB-GFX

OpenGL Speedup: 4.73

****************************************************

CINEBENCH 2003 v1
****************************************************

Tester : RaVeN

Processor : G4
MHz : 1 ghz dp
Number of CPUs : 2
Operating System : 10.3.8

Graphics Card : Geforce 4 MX
Resolution : <fill this out>
Color Depth : <fill this out>

****************************************************

Rendering (Single CPU): 96 CB-CPU
Rendering (Multiple CPU): 172 CB-CPU

Multiprocessor Speedup: 1.80

Shading (CINEMA 4D) : 117 CB-GFX
Shading (OpenGL Software Lighting) : 334 CB-GFX
Shading (OpenGL Hardware Lighting) : 172 CB-GFX

OpenGL Speedup: 2.85

****************************************************

CINEBENCH 2003 v1
****************************************************

Tester : RaVeN

Processor : mac mini 256mb ram
MHz : 1.25
Number of CPUs : 1
Operating System : 10.38

Graphics Card : ati 32mb
Resolution : <fill this out>
Color Depth : <fill this out>

****************************************************

Rendering (Single CPU): 113 CB-CPU
Rendering (Multiple CPU): --- CB-CPU


Shading (CINEMA 4D) : 141 CB-GFX
Shading (OpenGL Software Lighting) : 345 CB-GFX
Shading (OpenGL Hardware Lighting) : 439 CB-GFX

OpenGL Speedup: 3.11

****************************************************

CINEBENCH 2003 v1
****************************************************

Tester : RaVeN

Processor : imac G4 15" 256mb ram
MHz : 800
Number of CPUs : 1
Operating System : 10.3.9

Graphics Card : nvidia 32mb
Resolution : <fill this out>
Color Depth : <fill this out>

****************************************************

Rendering (Single CPU): 70 CB-CPU
Rendering (Multiple CPU): --- CB-CPU


Shading (CINEMA 4D) : 87 CB-GFX
Shading (OpenGL Software Lighting) : 215 CB-GFX
Shading (OpenGL Hardware Lighting) : 103 CB-GFX

OpenGL Speedup: 2.47

****************************************************

 

[Multimedia]

100% Faster Not 50%

We just got our Quad today to accompany our 2.7 Dual. Both have the same 4.5GB RAM and the same standard graphics card. In the latest update of FCP 5, we took a 1080i HD shot and applied a Gaussian Blur. The render time on the Quad was 50% faster than the 2.7 Dual. I can't remember how long the clip was, but it took about three minutes for the Quad, and about six minutes for the 2.7 Dual. We were all very pleased. The next big test I'd like to do is to take a QT movie and compress it as an H.264 because that's so darn slow to do on the 2.7 Dual.

Twice as fast is 100% faster not 50% faster. Started using EyeTV's new iPod encoding scheme and it takes FOREVER to do with a dual 2 or dual 2.5. I am sure the Quad will still be too slow. We need so much more power to efficiently compress video it is not funny.

 

[maestro55]

I have never personally cared about benchmark tests, anyhow. People want to get the fastest set-ups, but do everyday users really realize the difference? I mean unless you use the latest versions of very bulky software, you won't realize a difference. Seeing the difference in points then, means nothing.

 

[Harry322]

Just got the email that my Quad (with the 7800) is shipping! i'll get it today. Anyone else have a quad with the 7800 yet? I wonder how different the benchmarks are from the 6600?

 

[ottonian]

Jan. 19th for Quad w/7800 delivery...

Wow...

When did you order?

I'd love to get mine before the end of the year BUT...

I was one of the souls who bought a Dual 1.42 5 months before the G5's were released...ouch [keep head down, move forward...]

Could this be a blessing in disguise if they make an additional upgrade to the Quad at the Macworld expo and I can cancel and reorder?

 

[Harry322]

I ordered about a month ago, which is odd. I thought I'd get mine a few weeks after Christmas, but it should be here any minute!

 

[rickvanr]

I ordered about a month ago, which is odd. I thought I'd get mine a few weeks after Christmas, but it should be here any minute!

Congrats! I know how excited I was waiting for my dual 2.7 to come. It'd be even grander if had happened close to December.

 

[jbaugh]

Poor Quad G5 Cinebench 2003 results!!?

Here are my Cinebench 2003 results with my quad G5. They are considerably worse than any other benchmarks I've seen posted for the quad! Especially worrisome are the low scores for both single and multiple CPU rendering. What does this mean. Do I have defective CPUs? Any suggestions?
John

CINEBENCH 2003 v1
****************************************************

Tester : Jb

Processor : Quad G5
MHz : 2.5 GHz
Number of CPUs : 4
Operating System : OS 10.4.3

Graphics Card : GeForce 6600
Resolution : <fill this out>
Color Depth : <fill this out>

****************************************************

Rendering (Single CPU): 273 CB-CPU
Rendering (Multiple CPU): 835 CB-CPU

Multiprocessor Speedup: 3.06

Shading (CINEMA 4D) : 345 CB-GFX
Shading (OpenGL Software Lighting) : 918 CB-GFX
Shading (OpenGL Hardware Lighting) : 1827 CB-GFX

OpenGL Speedup: 5.30

****************************************************

 

[groovebuster]

Congrats! I know how excited I was waiting for my dual 2.7 to come. It'd be even grander if had happened close to December.

I am also really excited...


I ordered my Quad sunday evening at a little dealer here in Germany who offered the machine with an extra 1GB of RAM for free and it will ship today with UPS.

This is premature X-mas for me!!!!

groovebuster

 

[sauria]

Odd

Try reinstalling the app

Here are my Cinebench 2003 results with my quad G5. They are considerably worse than any other benchmarks I've seen posted for the quad! Especially worrisome are the low scores for both single and multiple CPU rendering. What does this mean. Do I have defective CPUs? Any suggestions?
John

CINEBENCH 2003 v1
****************************************************

Tester : Jb

Processor : Quad G5
MHz : 2.5 GHz
Number of CPUs : 4
Operating System : OS 10.4.3

Graphics Card : GeForce 6600
Resolution : <fill this out>
Color Depth : <fill this out>

****************************************************

Rendering (Single CPU): 273 CB-CPU
Rendering (Multiple CPU): 835 CB-CPU

Multiprocessor Speedup: 3.06

Shading (CINEMA 4D) : 345 CB-GFX
Shading (OpenGL Software Lighting) : 918 CB-GFX
Shading (OpenGL Hardware Lighting) : 1827 CB-GFX

OpenGL Speedup: 5.30

****************************************************

 

[jbaugh]

Try reinstalling the app

I'm very embarrassed to admit it, but I goofed. The Cinebench results I obtained were run on the non-G5 optimized version of Cinebench 2003. I didn't even know there was a G5 optimized version until it was pointed out to me by Mike at xlr8yourmac.com.

Now my Cinebench 2003 results are every bit as good as others have posted for their quad G5s.
John

 

[sauria]

Great

I'm very embarrassed to admit it, but I goofed. The Cinebench results I obtained were run on the non-G5 optimized version of Cinebench 2003. I didn't even know there was a G5 optimized version until it was pointed out to me by Mike at xlr8yourmac.com.

Now my Cinebench 2003 results are every bit as good as others have posted for their quad G5s.
John

Good news! You love it now?

 

[jbaugh]

Good news! You love it now?

To say I love it is an understatement.

For some bizarre reason the computer actually was a bit lethargic on many common tasks. Thus it was easy for me to believe the suboptimal Cinebench scores. It took 3- 4 seconds to open Mail, while opening Mail on the identically configured quad G5 at the Apple Store is virtually instantaneous. Similarly, the quad G5 at the Apple Store could open Microsoft Word in less than 2 seconds while my computer took 5 seconds or longer to do this.

I kept getting alert messages from the Belkin UPS software (Bulldog Plus) about exceeding the maximum load. I changed to a higher rated UPS device and deleted the Belkin software. Immediately, my computer is a speed demon. I didn't do anything else differently and I can only assume that the lower capacity UPS device and/or the Belkin software was causing a bottleneck somewhere. Now Mail opens essentially instantaneously. MSWord bounces once. By the time the icon in the Dock has completed its single bounce and returned to baseline, MSWord is open and ready to go.

So, yes, I'm one happy camper. The speed up in working with Final Cut Pro is phenomenal. I can have a 45 minute FCP video ready (from export via Compressor) for DVDSP now in 45 minutes when it took me 8- 12 hrs. to do this on my Titanium PB 800 MHz.
John

 

[sauria]

Great

Perhaps the finder is not optimized for 4 CPUS?

To say I love it is an understatement.

For some bizarre reason the computer actually was a bit lethargic on many common tasks. Thus it was easy for me to believe the suboptimal Cinebench scores. It took 3- 4 seconds to open Mail, while opening Mail on the identically configured quad G5 at the Apple Store is virtually instantaneous. Similarly, the quad G5 at the Apple Store could open Microsoft Word in less than 2 seconds while my computer took 5 seconds or longer to do this.

I kept getting alert messages from the Belkin UPS software (Bulldog Plus) about exceeding the maximum load. I changed to a higher rated UPS device and deleted the Belkin software. Immediately, my computer is a speed demon. I didn't do anything else differently and I can only assume that the lower capacity UPS device and/or the Belkin software was causing a bottleneck somewhere. Now Mail opens essentially instantaneously. MSWord bounces once. By the time the icon in the Dock has completed its single bounce and returned to baseline, MSWord is open and ready to go.

So, yes, I'm one happy camper. The speed up in working with Final Cut Pro is phenomenal. I can have a 45 minute FCP video ready (from export via Compressor) for DVDSP now in 45 minutes when it took me 8- 12 hrs. to do this on my Titanium PB 800 MHz.
John

 

[longofest]

To say I love it is an understatement.

For some bizarre reason the computer actually was a bit lethargic on many common tasks. Thus it was easy for me to believe the suboptimal Cinebench scores. It took 3- 4 seconds to open Mail, while opening Mail on the identically configured quad G5 at the Apple Store is virtually instantaneous. Similarly, the quad G5 at the Apple Store could open Microsoft Word in less than 2 seconds while my computer took 5 seconds or longer to do this.

I kept getting alert messages from the Belkin UPS software (Bulldog Plus) about exceeding the maximum load. I changed to a higher rated UPS device and deleted the Belkin software. Immediately, my computer is a speed demon. I didn't do anything else differently and I can only assume that the lower capacity UPS device and/or the Belkin software was causing a bottleneck somewhere. Now Mail opens essentially instantaneously. MSWord bounces once. By the time the icon in the Dock has completed its single bounce and returned to baseline, MSWord is open and ready to go.

So, yes, I'm one happy camper. The speed up in working with Final Cut Pro is phenomenal. I can have a 45 minute FCP video ready (from export via Compressor) for DVDSP now in 45 minutes when it took me 8- 12 hrs. to do this on my Titanium PB 800 MHz.
John

Hmm... It really could have been either that slowed down your quad. Since OS 10 has built-in software for UPS management, I'd definitely suggest just going with that to reduce unwanted lag. It could have been the software.

However, having an under-rated UPS could have also been the culprit. If it wasn't conditioning the power correctly, its possible your CPU's weren't getting the propper voltage, which would have slowed them down. It's a stretch, but possible.

What size UPS did you have to end up going with? My quad just shipped. I use an APC Back-UPS 800 for my Dual-1Ghz Quicksilver, but if the rumors were true about this thing having over 1000 watt power supply, I think I might need an even bigger one.

 

[longofest]

Perhaps the finder is not optimized for 4 CPUS?

Apple SAYS the finder is SMP-aware, but after using the finder since OS 10.0.x, I can say without a doubt that Apple is smoking dope. I get WAY too many spinning beach-balls to call the finder SMP-aware.