Welcome to the MacNN Forums.

Eug Wanker · 02:30 AM

The Cinebench UB is now out.

I don't have an Intel Mac, but for reference, an iMac G5 2.0 gets:

269: "Highest" Energy Saver setting
265: "Automatic" Energy Saver setting

The iMac Core Duo 2.0 should get over 500.

EDIT:

Intel Core Duo iMac 2.0: 566 (or 308 for a single core)

ug.mac · Feb 18, 2006, 03:07 AM

That score close to Pentium 840 EE in 64bit WinXP (606 in extremetech.com's test), faster then Dual 2Ghz Powermac G5. Very impressive!!

Eug Wanker · 01:24 PM

Yeah, they're in the same ballpark but the Core Duo may have something like a 4-5% advantage on a per-clock basis over the G5.

BTW, my iBook G4 1.33 gets 138 (on both "Highest" and "Automatic").

In other words, A 1.33 GHz G4 has roughly the speed of a Yonah single-core 1.0 GHz. ie. On a clock-for-clock basis, the G5 and Intel Core are both roughly 30%+ faster as compared to the G4, and both the G5 and Intel Core are much higher clocked.

So, despite having "only" 3X the clock speed of the iBook, the Core Duo 2.0 MacBook Pro has 4 times the performance of the current iBook. It also has 3 times the performance of the fastest PowerBook. Impressive.

iMac Core Duo 2.0: 566
Power Mac G5 2.0: 543
iMac G5 970FX 2.0: 269
PowerBook G4 1.67: ~172 (estimated)
iBook G4 1.33 GHz: 138

Horsepoo!!! · Feb 18, 2006, 01:38 PM

Steve wasn't senile afterall. 3-4 times performance increase over the last gen PPC Powerbooks is about right for some apps.

Eug Wanker · 02:25 PM

Originally Posted by Eug Wanker

iMac Core Duo 2.0: 566
Power Mac G5 2.0: 543
iMac G5 970FX 2.0: 269
PowerBook G4 1.67: ~172 (estimated)
iBook G4 1.33 GHz: 138

To put it another way:

iBook G4 1.33 GHz: 161 s
PowerBook G4 1.67: ~130 s (estimated)
iMac G5 970FX 2.0: 84 s
Power Mac G5 2.0: ~42 s (estimated)
iMac Core Duo 2.0: ~40 s (estimated)

So, has anyone else here tried Cinebench 9.5 yet? The benchmark app is here. When you run the Single-processor or Multi-processor benches, it gives both a time and a score. Note that it seems the numbers are a bit different than the ones that came from Cinebench 2003, so you can't really directly compare those numbers from previous reviews, etc.

EDIT:

Holy crap!

Barefeats ran Cinebench, and got 605 on the iMac Core Duo:

I would estimate then that Cinebench completes in 37 seconds on the Core Duo 2.0.

P.S. The iMac 2.1 gets 294 in his tests, which is 9% faster than my iMac 2.0, despite having only a 5% clock-speed advantage. His Power Mac G5 2.0 scores aren't so impressive though.

ug.mac · Feb 18, 2006, 03:16 PM

Well, I don't know if it's OS or the processor or the Cinebench, since a 3.2Ghz HT 800Mhz FSB with 2X1MB L2 cache dual core PENTIUM 840 EE processor running Cinebench under WindowsXP(32bit/64bit ver) only hits 620 (highest score in benchmarks)! 840 EE have 160% advantage over Core Duo on clock speed, but only gain 2.4% in the test.

Eug Wanker · 06:20 PM

Originally Posted by ug.mac

Well, I don't know if it's OS or the processor or the Cinebench, since a 3.2Ghz HT 800Mhz FSB with 2X1MB L2 cache dual core PENTIUM 840 EE processor running Cinebench under WindowsXP(32bit/64bit ver) only hits 620 (highest score in benchmarks)! 840 EE have 160% advantage over Core Duo on clock speed, but only gain 2.4% in the test.

Where did you see the benchmarks?

Anyways, I'm not terribly surprised, The Pentium D 840 non-EE wasn't uber stellar with Cinebench 2003 either, while the Athlon64 X2 chips also got over 600 in Cinebench 2003 and are also much lower clocked (although faster clocked than Yonah).

EDIT:

BTW, an iBook G4 1.07 gets a score of 111 (191 seconds). So it seems:

The G4 gets about 104 points per GHz.
The G5 gets about 135-140 points per GHz. (Less for the Quad.)
Yonah gets about 151 points per GHz.

harrisjamieh · Feb 19, 2006, 08:14 AM

Just tried this on my 17" Core Duo with 1.83 Ghz.

CPU benchmark

1 CPU: 283
2 CPU: 520

Multiprocessor Speedup: 1.84 x

Graphics Benchmark

C4D shading: 327
OpenGL SW-L: 893
OpenGL HW-L: 1566

OpenGL speedup: 4.79 x

Hornet · Feb 19, 2006, 01:11 PM

Eug your threads are always good, time to contribute. Heres my PB 15 1.67ghz 1.5gb ram 64mb 9700 on cinebench 9.5:

Rendering (Single CPU): 164 CB-CPU
Rendering (Multiple CPU): --- CB-CPU

Shading (CINEMA 4D) : 176 CB-GFX
Shading (OpenGL Software Lighting) : 489 CB-GFX
Shading (OpenGL Hardware Lighting) : 888 CB-GFX

OpenGL Speedup: 5.05

So if a dual core 2.0 Yonah gets 605, single core 308 (maybe BF gets higher?), the Yonah is doing 57% better clock for clock single core against a G4. Total cpu comparison, the 2.0ghz Yonah ends up 3.68x the speed of a single core 1.67 G4, impressive!

Cadaver · Feb 20, 2006, 01:03 AM

DP 2.5GHz G5, 3GB RAM, X800XT:

Single CPU: 374
Dual CPU: 670
MP Speedup: 1.76x

C4D shading: 374
OpenGL SW: 1049
OpenGL HW: 1956
OpenGL Speedup: 5.23x

Not that much faster than a 2.0GHz CD-based Mac. Wonder how the 2.16GHz units will do...

Eug Wanker · Feb 20, 2006, 12:23 PM

Sonnet 1.7 GHz G4 7447A with PC100 memory: 158 (144 seconds)
P4 1.6 GHz 32-bit Win2000 with PC133 memory: 145 (157 seconds)

IIRC, the G4 score was 150 for Cinebench 2003

Anyways, it seems for this app that the G4 is clock-for-clock is only slightly faster than a P4 of the sub 2 GHz era (ie. several years ago).

CatOne · Feb 20, 2006, 01:14 PM

CINEBENCH 9.5
************************************************** **

Tester :

Processor : Quad 2.5 G5
MHz :
Number of CPUs : 4
Operating System : 10.4.5

Graphics Card : Nvidia 7800 Ultra
Resolution : 30" Cinema
Color Depth : <fill this out>

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

Rendering (Single CPU): 386 CB-CPU
Rendering (Multiple CPU): 1157 CB-CPU

Multiprocessor Speedup: 3.00

Shading (CINEMA 4D) : 386 CB-GFX
Shading (OpenGL Software Lighting) : 1297 CB-GFX
Shading (OpenGL Hardware Lighting) : 2858 CB-GFX

OpenGL Speedup: 7.40

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

Eug Wanker · Feb 20, 2006, 01:26 PM

CatOne. How long did the multi-CPU test take? 19 seconds? (It's at the bottom left of the screen if you run it separately.) Just wondering.

Eug Wanker · Feb 20, 2006, 06:55 PM

Pentium 4 2.8 HT:

Single CPU (HT off): 238 (93 seconds)
Multi CPU (HT on): 280 (79 seconds)

Originally Posted by ug.mac

Well, I don't know if it's OS or the processor or the Cinebench, since a 3.2Ghz HT 800Mhz FSB with 2X1MB L2 cache dual core PENTIUM 840 EE processor running Cinebench under WindowsXP(32bit/64bit ver) only hits 620 (highest score in benchmarks)! 840 EE have 160% advantage over Core Duo on clock speed, but only gain 2.4% in the test.

From the above results...

P4 2.8 HT ~ G5 2.1 ~ Core Solo 1.83

So your dual-core 3.2 GHz HT being comparable to Core Duo 2.0 sounds about right.

parsec_kadets · Feb 21, 2006, 04:31 PM

Originally Posted by harrisjamieh

Just tried this on my 17" Core Duo with 1.83 Ghz.

CPU benchmark

1 CPU: 283
2 CPU: 520

Multiprocessor Speedup: 1.84 x

Graphics Benchmark

C4D shading: 327
OpenGL SW-L: 893
OpenGL HW-L: 1566

OpenGL speedup: 4.79 x

Originally Posted by Cadaver

DP 2.5GHz G5, 3GB RAM, X800XT:

Single CPU: 374
Dual CPU: 670
MP Speedup: 1.76x

C4D shading: 374
OpenGL SW: 1049
OpenGL HW: 1956
OpenGL Speedup: 5.23x

Not that much faster than a 2.0GHz CD-based Mac. Wonder how the 2.16GHz units will do...

I actually find it more interesting that the Core Duo appears to have a better speedup than the G5. Makes you wonder where the additional overhead on the G5 system is coming from.

Eug Wanker · Feb 21, 2006, 05:05 PM

Originally Posted by harrisjamieh

Just tried this on my 17" Core Duo with 1.83 Ghz.

CPU benchmark

1 CPU: 283
2 CPU: 520

And the MacBook Pro owners are reporting 563 for the Core Duo 2.0 GHz MBP. In fact, that's almost exactly the same as the score for the iMac Core Duo 2.0 too.

Furthermore, that's almost exactly what one would expect extrapolating from that 520 on your Core Duo 1.83. Thus, I don't understand how Bare Feats got their 605 score for the iMac Core Duo 2.0.

BTW, the dual 2.0 G5 Power Mac gets 552.

Scores so far:

Quad G5 2.5 GHz: 1187 (383)
Quad G5 2.5 GHz: 1179 (385)
Quad G5 2.5 GHz: 1157 (386)
Dual G5 2.5 GHz: 670 (374)
Pentium D 3.2 840 EE: 620
iMac Core Duo 2.0: 605 (Bare Feats)
iMac Core Duo 2.0: 566 (308)
iMac Core Duo 2.0: 563 (308)
MacBook Pro 2.0: 563 (305)
Power Mac G5 2.0: 552 (304)
Power Mac G5 2.0: 543
iMac Core Duo 1.83: 520 (283)
Pentium 4 2.8 HT: 280 (238)
iMac G5 970FX 2.0: 269
PowerBook G4 1.67: 164
Sonnet G4 1.7 7447A: 158
Pentium 4 1.6 GHz: 145
iBook G4 1.33 GHz: 138
iBook G4 1.07 GHz: 111

Eug Wanker · 11:32 AM

Quad G5 2.5 GHz: 1187 (383)
Quad G5 2.5 GHz: 1179 (385)
Quad G5 2.5 GHz: 1157 (386)
Dual G5 2.5 GHz: 670 (374)
Pentium D 3.2 840 EE: 620
MacBook Pro 2.16: 608 (329)
iMac Core Duo 2.0: 605 (Bare Feats)
iMac Core Duo 2.0: 566 (308)
iMac Core Duo 2.0: 563 (308)
MacBook Pro 2.0: 563 (305)
Power Mac G5 2.0: 552 (304)
Power Mac G5 2.0: 543
iMac Core Duo 1.83: 520 (283)
Pentium 4 2.8 HT: 280 (238)
iMac G5 970FX 2.0: 269
PowerBook G4 1.67: 164
Sonnet G4 1.7 7447A: 158
Pentium 4 1.6 GHz: 145
iBook G4 1.33 GHz: 138
iBook G4 1.07 GHz: 111

Added the 608 score for the MBP Core Duo 2.16. It seems like nobody can reproduce that 605 score that Bare Feats got for the Core Duo 2.0.

Still, it's an astounding improvement in performance after just one model update of the Mac laptops. I really hope the 13" gets a widescreen Core Duo 1.83 or 1.67. I'm still not ruling out single-core for the 13" though.

jamil5454 · Feb 22, 2006, 01:22 PM

The render took 1:28

CINEBENCH 9.5
************************************************** **

Tester : jamil5454

Processor : Athlon64 2800+, 1GB PC3200
MHz : 1.8
Number of CPUs : 1
Operating System : Windows XP SP2

Graphics Card : Radeon 9600 Pro (128MB)
Resolution : 1280x1024
Color Depth : 32bit

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

Rendering (Single CPU): 259 CB-CPU
Rendering (Multiple CPU): --- CB-CPU

Shading (CINEMA 4D) : 279 CB-GFX
Shading (OpenGL Software Lighting) : 1172 CB-GFX
Shading (OpenGL Hardware Lighting) : 2708 CB-GFX

OpenGL Speedup: 9.70

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

If Apple really wanted to improve their marketshare, they'd spend more time on getting better video card performace (especially on the Intel Macs). My OpenGL hardware score almost matches CatOne's 7800 GT, which has twice the memory, 5x the pixel pipelines, double the system bandwidth, and roughly 6x the memory speed! With Apple now on Intel Macs, there's no excuse why we shouldn't have much better graphics performance. Come on ATI and Nvidia.

ehchan · Mar 1, 2006, 07:50 PM

These results from my Sony Core Duo 1.83GHz laptop

Pretty laptop cool in that it has both an Nvidia Geforce 7400 Go GPU and the Intel GMA950 integrated video... below are results from both video cards.

CINEBENCH 9.5
************************************************** **

Tester : Ed

Processor : Sony VAIO SZ120P
MHz : 1833
Number of CPUs : 2
Operating System : Windows XP Pro SP2

Graphics Card : Nvidia Geforce 7400 Go
Resolution : 1280x800
Color Depth : 32-bit

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

Rendering (Single CPU): 271 CB-CPU
Rendering (Multiple CPU): 498 CB-CPU

Multiprocessor Speedup: 1.84

Shading (CINEMA 4D) : 319 CB-GFX
Shading (OpenGL Software Lighting) : 1261 CB-GFX
Shading (OpenGL Hardware Lighting) : 2942 CB-GFX

OpenGL Speedup: 9.22

************************************************** **
CINEBENCH 9.5
************************************************** **

Tester : Ed

Processor : Sony VAIO SZ120P
MHz : 1833
Number of CPUs : 2
Operating System : Windows XP Pro SP2

Graphics Card : GMA950
Resolution : 1280x800
Color Depth : 32-bit

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

Rendering (Single CPU): 270 CB-CPU
Rendering (Multiple CPU): 496 CB-CPU

Multiprocessor Speedup: 1.83

Shading (CINEMA 4D) : 319 CB-GFX
Shading (OpenGL Software Lighting) : 959 CB-GFX
Shading (OpenGL Hardware Lighting) : 1156 CB-GFX

OpenGL Speedup: 3.62

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

york28 · Mar 2, 2006, 04:48 AM

Does Cinebench do anything that would benefit from having a 64 bit chip? Otherwise, it doesn't really have too much to offer over the new Intel chips, judging from the benchmarks.

Eug Wanker · 11:28 AM

Originally Posted by ehchan

These results from my Sony Core Duo 1.83GHz laptop

Pretty laptop cool in that it has both an Nvidia Geforce 7400 Go GPU and the Intel GMA950 integrated video... below are results from both video cards.

Interesting test.

The GeForce 7400 Go is approximately 2.5X as fast as Intel GMA 950. This should be of interest to the Mac mini owners, which has been just benched. A Core Solo 1.5 Mac mini gets 213, which is about 30% faster than a G4 1.67 PowerBook, and 40% faster than a G4 1.5 Mac mini.

Also, it's interesting to see that the iMac Core Duo 1.83 is slightly faster than the Sony laptop with the same CPU. Probably some of this has to do with memory bandwidth, but it suggests that the Mac version is well-optimized. BTW, my iMac G5 2.0 is the same speed as an Athlon64 2800+ 1.8 GHz, as expected.

Quad G5 2.5 GHz: 1187 (383)
Quad G5 2.5 GHz: 1179 (385)
Quad G5 2.5 GHz: 1157 (386)
Dual G5 2.5 GHz: 670 (374)
Pentium D 3.2 840 EE: 620
MacBook Pro 2.16: 608 (329)
iMac Core Duo 2.0: 605 (Bare Feats)
iMac Core Duo 2.0: 566 (308)
iMac Core Duo 2.0: 563 (308)
MacBook Pro 2.0: 563 (305)
Power Mac G5 2.0: 552 (304)
Power Mac G5 2.0: 543
iMac Core Duo 1.83: 520 (283)
iMac Core Duo 1.83: 518 (283) (xlr8yourmac)
Sony Core Duo 1.83: 498 (271)
Pentium 4 2.8 HT: 280 (238)
iMac G5 970FX 2.0: 269
Athlon64 2800+ 1.8: 269
Mac mini Core Solo 1.5: 213
PowerBook G4 1.67: 164
Sonnet G4 1.7 7447A: 158
Mac mini G4 1.5: 152
Pentium 4 1.6 GHz: 145
iBook G4 1.33 GHz: 138
iBook G4 1.07 GHz: 111

Clock for clock, it seems that Intel Core is slightly faster than the G5, but they're close. However, Intel Core is lower power, and the G5 is faster clocked. And the G4? Well, it's just slow. Clock for clock, Intel Core is about 40% faster than the G4, and Intel Core is faster clocked.

Originally Posted by york28

Does Cinebench do anything that would benefit from having a 64 bit chip? Otherwise, it doesn't really have too much to offer over the new Intel chips, judging from the benchmarks.

Not sure what you're getting at, but 64-bit Cinema 4D on Windows is much faster than the 32-bit version. There is no 64-bit version for Mac OS X, but that's because of the 64-bit limitations in 10.4. There will most likely be a 64-bit version for Mac OS X after it hits 10.5, since 10.5 will likely get full 64-bit support.

---

P.S. The OpenGL hardware lighting scores for the Mac minis are interesting:

Mac mini G4 1.5 GHz:

Rendering (1 CPU): 152

C4D: 159
OpenGL SW-L: 414
OpenGL HW-L: 506
OpenGL Speedup: 3.18

Mac mini Core Solo 1.5 GHz:

Rendering (1 CPU): 213

C4D: 259
OpenGL SW-L: 885
OpenGL HW-L: 441
OpenGL Speedup: 3.41

The GMA 950 truly is unimpressive. In fact, for the Intel Mac mini, the OpenGL software lighting score is twice as high as the hardware score. This is unusual, and suggests the possibility that there is poor optimization for GMA 950 graphics (at least on OS X), but nonetheless the score as it stands now is quite poor.

Voch · Mar 2, 2006, 11:44 AM

Wow, excellent Cinebench roundup, Eug (or is that Mr. Wanker?).

rnb2 · Mar 2, 2006, 03:32 PM

Just posted by ururk in the "Core Solo Oddities" thread:

------------------------------------------------------------------------------
CINEBENCH 9.5
************************************************** **

Tester : John (ururk)

Processor : Mac mini 1.66ghz Core duo
MHz : 1.66
Number of CPUs : 2
Operating System : 10.4.5

Graphics Card :
Resolution : 1280 x 1024 @ 60 Hz
Color Depth : Millions

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

Rendering (Single CPU): 256 CB-CPU
Rendering (Multiple CPU): 471 CB-CPU

Multiprocessor Speedup: 1.84

Shading (CINEMA 4D) : 301 CB-GFX
Shading (OpenGL Software Lighting) : 1079 CB-GFX
Shading (OpenGL Hardware Lighting) : 547 CB-GFX

OpenGL Speedup: 3.59

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

Eug Wanker · 04:02 PM

Quad G5 2.5 GHz: 1187 (383)
Quad G5 2.5 GHz: 1179 (385)
Quad G5 2.5 GHz: 1157 (386)
Dual G5 2.5 GHz: 670 (374)
Pentium D 3.2 840 EE: 620
MacBook Pro 2.16: 608 (329)
iMac Core Duo 2.0: 605 (Bare Feats)
iMac Core Duo 2.0: 566 (308)
iMac Core Duo 2.0: 563 (308)
MacBook Pro 2.0: 563 (305)
Power Mac G5 2.0: 552 (304)
Power Mac G5 2.0: 543
iMac Core Duo 1.83: 520 (283)
iMac Core Duo 1.83: 518 (283) (xlr8yourmac)
Sony Core Duo 1.83: 498 (271)
Mac mini Core Duo 1.66: 471 (256)
Pentium 4 2.8 HT: 280 (238)
iMac G5 970FX 2.0: 269
Athlon64 2800+ 1.8: 269
Mac mini Core Solo 1.5: 213
PowerBook G4 1.67: 164
Sonnet G4 1.7 7447A: 158
Mac mini G4 1.5: 152
Pentium 4 1.6 GHz: 145
iBook G4 1.33 GHz: 138
iBook G4 1.07 GHz: 111

Also, Mac mini Core Duo 1.66 - Shading (OpenGL Hardware Lighting) : 547 CB-GFX <-- Ouch

ehchan · Mar 2, 2006, 08:34 PM

Originally Posted by Eug Wanker

C4D: 259
OpenGL SW-L: 885
OpenGL HW-L: 441
OpenGL Speedup: 3.41

The GMA 950 truly is unimpressive. In fact, for the Intel Mac mini, the OpenGL software lighting score is twice as high as the hardware score. This is unusual, and suggests the possibility that there is poor optimization for GMA 950 graphics (at least on OS X), but nonetheless the score as it stands now is quite poor.

I think Mac OS's driver for the 950 is really unoptimized... my Sony's GMA950 score was more than double the HW test. So, hopefully, they will optimize those drivers, then it will be faster than the 9200

One other note-- notice how using the on-board video doesn't really affect the CPU scores. In other testing, it doesn't affect memory bandwidth either. Probably due to the dual-channel memory controller in the 945G chipset. I verified this by removing one DIMM from my laptop, and running more memory bandwidth tests. With 1 DIMM installed, the GMA950 took about 15% of my bandwidth, and it varies with resolution, so the higher res, the more it steals. With 2 DIMMs, no losses at all. The lesson to be learned? Make sure you install RAM in pairs in your Core Duo computers!

Eug Wanker · 09:08 PM

One other note-- notice how using the on-board video doesn't really affect the CPU scores.

Err... The GPU is completely irrelevant for the CPU scores.

I think Mac OS's driver for the 950 is really unoptimized... my Sony's GMA950 score was more than double the HW test. So, hopefully, they will optimize those drivers, then it will be faster than the 9200

True, but then again, using GMA 950 on Windows nets you only 1.2X the speed compared to using just the CPU:

Shading (OpenGL Software Lighting) : 959 CB-GFX
Shading (OpenGL Hardware Lighting) : 1156 CB-GFX

Contrast that to the results with your 7400 Go:

Shading (OpenGL Software Lighting) : 1261 CB-GFX
Shading (OpenGL Hardware Lighting) : 2942 CB-GFX

That's 2.3X the speed vs software OpenGL.

I will also point out that a single-core P4 3.0 with Radeon X300 gets 1842 with the lowly Radeon X300, a 60% advantage over your Sony Core Duo's GMA 950, despite the fact that P4 machine only gets 252 in the CPU bench, which is slower than one core of your Sony Core Duo.

jamil5454 · 09:10 PM

Originally Posted by Eug Wanker

True, but then again, using GMA 950 on Windows nets you only 1.2X the speed compared to using just the CPU:

Shading (OpenGL Software Lighting) : 959 CB-GFX
Shading (OpenGL Hardware Lighting) : 1156 CB-GFX

Hardware(accelerated) transform and lighting, introduce in the original GeForce 256 in 2000, has been the biggest thing to hit 3D video cards. Nearly every single video card manufactured from Nvidia and ATI since 2000 has it. The GMA950 lacks this.

Not arguing with you, just pointing out that if the GMA9650 had hardware T&L then it wouldn't suck so much. Hence the little to no (or on the Mac side a decline in performace due to the additional overhead of running through unoptimized drivers) improvement over the Software score. Both tests basically rely on the CPU for the grunt of the work.

It's a lazy graphics chip.

From Intel's site:

Microsoft* DirectX* 9 Vertex Shader 3.0 and Transform and Lighting supported in software through highly optimized Processor Specific Geometry Pipeline (PSGP)

rnb2 · Mar 2, 2006, 09:02 PM

Basically, the Core Duo mini is over 3x the G4 1.5 mini in raw CPU score, but only 8% better in hardware OpenGL shading. I'm wondering how much better that score would be if they'd just thrown the 9200 in the Intel mini? I know the 9200 was nothing to write home about, but it's hard to imagine that it wouldn't be better than the GMA950.

Ah, what might have been........

Eug Wanker · Mar 2, 2006, 09:12 PM

Originally Posted by rnb2

Basically, the Core Duo mini is over 3x the G4 1.5 mini in raw CPU score, but only 8% better in hardware OpenGL shading. I'm wondering how much better that score would be if they'd just thrown the 9200 in the Intel mini? I know the 9200 was nothing to write home about, but it's hard to imagine that it wouldn't be better than the GMA950.

Ah, what might have been........

Indeed, but in truth, I actually prefer the GMA 950 over the Radeon 9200 for other reasons. The Radeon 9200 does not support the hardware requirements for full CoreImage functionality. I want that Dashboard ripple effect dammit!

However, they both suck.

Originally Posted by jamil5454

Hardware(accelerated) transform and lighting, introduce in the original GeForce 256 in 2000, has been the biggest thing to hit 3D video cards. Nearly every single video card manufactured from Nvidia and ATI since 2000 has it. The GMA950 lacks this.

Not arguing with you, just pointing out that if the GMA9650 had hardware T&L then it wouldn't suck so much. Hence the little to no (or on the Mac side a decline in performace due to the additional overhead of running through unoptimized drivers) improvement over the Software score. Both tests basically rely on the CPU for the grunt of the work.

It's a lazy graphics chip.

Ah, thanks for the explanation. I would guess the nVidia and ATI integrated chipsets also include hardware T&L then, as they both also outperform GMA 950 in most tests.

ehchan · Mar 3, 2006, 01:40 AM

Originally Posted by Eug Wanker

Err... The GPU is completely irrelevant for the CPU scores.

It's usually irrelevent, but the older generation integrated graphics had quite an effect on CPU and memory scores in intel machines... Moreso for memory, since the systems were starved with single channel DDR333 RAM before the 945 chipset. My point is that the Intel 945GM chipset has a MUCH better overall design than the older 855GM intel architecture. Simply due to the much higher memory bandwidth (dual channel DDR667, the integrated graphics don't hamper the cpu/memory performance of the machine like the older intel chipsets did.

fhoubi · Mar 3, 2006, 07:39 PM

Nice to compare...

I am hesitating between a mini + 20" or an Imac 17" BTW ...
Current (don't laugh):

CINEBENCH 9.5
************************************************** **

Tester : fhoubi

Processor : iBook G3
MHz : 700
Number of CPUs : 1
Operating System : MacOS X 10.4.3

Graphics Card :
Resolution : 1024 * 768
Color Depth : millions

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

Rendering (Single CPU): 71 CB-CPU
Rendering (Multiple CPU): --- CB-CPU

Shading (CINEMA 4D) : 78 CB-GFX
Shading (OpenGL Software Lighting) : 187 CB-GFX
Shading (OpenGL Hardware Lighting) : 127 CB-GFX

OpenGL Speedup: 2.40

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

toed · Mar 4, 2006, 11:21 AM

Originally Posted by Eug Wanker

Quad G5 2.5 GHz: 1187 (383)
Quad G5 2.5 GHz: 1179 (385)
Quad G5 2.5 GHz: 1157 (386)
Dual G5 2.5 GHz: 670 (374)
Pentium D 3.2 840 EE: 620
MacBook Pro 2.16: 608 (329)
iMac Core Duo 2.0: 605 (Bare Feats)
iMac Core Duo 2.0: 566 (308)
iMac Core Duo 2.0: 563 (308)
MacBook Pro 2.0: 563 (305)
Power Mac G5 2.0: 552 (304)
Power Mac G5 2.0: 543
iMac Core Duo 1.83: 520 (283)
iMac Core Duo 1.83: 518 (283) (xlr8yourmac)
Sony Core Duo 1.83: 498 (271)
Mac mini Core Duo 1.66: 471 (256)
Pentium 4 2.8 HT: 280 (238)
iMac G5 970FX 2.0: 269
Athlon64 2800+ 1.8: 269
Mac mini Core Solo 1.5: 213
PowerBook G4 1.67: 164
Sonnet G4 1.7 7447A: 158
Mac mini G4 1.5: 152
Pentium 4 1.6 GHz: 145
iBook G4 1.33 GHz: 138
iBook G4 1.07 GHz: 111

Also, Mac mini Core Duo 1.66 - Shading (OpenGL Hardware Lighting) : 547 CB-GFX <-- Ouch

Being new to Cinebench, the higher the numbers the better right? I want to run this on my 20" intel imac, but alas, it is in the shop... the display won't power on.

Eug Wanker · Mar 4, 2006, 06:06 PM

Originally Posted by fhoubi

Current (don't laugh):

Processor : iBook G3
MHz : 700

Rendering (Single CPU): 71 CB-CPU

Heh. I got something like 49 on the older version of this bench with my G4 450 Cube.

Originally Posted by toed

Being new to Cinebench, the higher the numbers the better right?

Yes. The number in brackets is the single-core speed on multiprocessor systems.

I want to run this on my 20" intel imac, but alas, it is in the shop... the display won't power on.

Hmmm... So is mine. Mine powers on, but the power shuts off randomly. One of the local repair guys seems less than impressed with the repair record of the G5 iMacs. Mind you, I talked to him a few months after the exploding capacitors fiasco came to light.

toed · Mar 8, 2006, 10:38 PM

Got my 20" imac back and here are my Cinebench results:

CINEBENCH 9.5
************************************************** **

Tester : TJ

Processor : Imac Core Duo 20"
MHz : 2.0
Number of CPUs : 2
Operating System : Mac OS 10.4.5

Graphics Card : Radeon X1600
Resolution : <fill this out>
Color Depth : <fill this out>

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

Rendering (Single CPU): 308 CB-CPU
Rendering (Multiple CPU): 568 CB-CPU

Multiprocessor Speedup: 1.84

Shading (CINEMA 4D) : 359 CB-GFX
Shading (OpenGL Software Lighting) : 977 CB-GFX
Shading (OpenGL Hardware Lighting) : 1694 CB-GFX

OpenGL Speedup: 4.72

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

Eug Wanker · Mar 9, 2006, 12:51 PM

Quad G5 2.5 GHz: 1187 (383)
Quad G5 2.5 GHz: 1179 (385)
Quad G5 2.5 GHz: 1157 (386)
Dual G5 2.5 GHz: 670 (374)
Pentium D 3.2 840 EE: 620
MacBook Pro 2.16: 608 (329)
MacBook Pro 2.16: 607 (324) (Hardmac)
iMac Core Duo 2.0: 605 (Bare Feats)
iMac Core Duo 2.0: 568 (308)
iMac Core Duo 2.0: 566 (308)
iMac Core Duo 2.0: 563 (308)
MacBook Pro 2.0: 563 (305)
Power Mac G5 2.0: 552 (304)
Power Mac G5 2.0: 543
iMac Core Duo 1.83: 520 (283)
iMac Core Duo 1.83: 518 (283) (xlr8yourmac)
Sony Core Duo 1.83: 498 (271)
Mac mini Core Duo 1.66: 471 (256)
Pentium 4 2.8 HT: 280 (238)
iMac G5 970FX 2.0: 269
iMac G5 970FX 1.8: 244 (Ars)
Athlon64 2800+ 1.8: 269
Mac mini Core Solo 1.5: 217 (Ars)
Mac mini Core Solo 1.5: 213
PowerBook G4 1.67: 171 (Ars)
PowerBook G4 1.67: 164
Sonnet G4 1.7 7447A: 158
Mac mini G4 1.5: 152
Pentium 4 1.6 GHz: 145
iBook G4 1.33 GHz: 138
Mac mini G4 1.25 GHz: 129 (Ars)
iBook G4 1.07 GHz: 111
iBook G3 700 MHz: 71

mduell · Mar 9, 2006, 08:22 PM

Eug: 1459 on an "octo" (2 CPU x 4 cores/CPU) Clovertown box (clockrate unknown, probably 2.66Ghz)

Eug Wanker · Mar 9, 2006, 08:34 PM

Originally Posted by mduell

Eug: 1459 on an "octo" (2 CPU x 4 cores/CPU) Clovertown box (clockrate unknown, probably 2.66Ghz)

Not that I know how well an octo is supposed to perform, but given that the Quad G5 gets over 1100, I'm a little disappointed with the 1458 score for the octo.

We also don't know if it's 32-bit or 64-bit Cinebench. All the benches in this thread so far have been 32-bit.

fhoubi · Mar 20, 2006, 08:39 AM

This (german only) is interesting, Mini Core Solo running OS X AND Windows:

Rendering (Single CPU):
Mac OS X: 220 CB-CPU
Windows: 225 CB-CPU

Shading (CINEMA 4D):
Mac OS X: 268 CB-GFX
Windows: 263 CB-GFX

Shading (OpenGL Software Lighting):
Mac OS X: 912 CB-GFX

Windows: 796 CB-GFX

Shading (OpenGL Hardware Lighting):
Mac OS X: 458 CB-GFX

Windows: 967 CB-GFX[/]

moonmonkey · Mar 20, 2006, 07:49 PM

Originally Posted by fhoubi

This (german only) is interesting, Mini Core Solo running OS X AND Windows:

Rendering (Single CPU):
Mac OS X: 220 CB-CPU
Windows: 225 CB-CPU

Shading (CINEMA 4D):
Mac OS X: 268 CB-GFX
Windows: 263 CB-GFX

Shading (OpenGL Software Lighting):
Mac OS X: 912 CB-GFX

Windows: 796 CB-GFX

Shading (OpenGL Hardware Lighting):
Mac OS X: 458 CB-GFX

Windows: 967 CB-GFX[/]

Gulp, the wrong windows Graphics drivers are faster than the correct Mac ones.
It shows there is a lot of room for improvement.

oni · Mar 20, 2006, 08:13 PM

Heres mine

CINEBENCH 9.5
************************************************** **

Tester : Oni

Processor : Powermac G5
MHz : 1800x2
Number of CPUs : 2
Operating System : 10.4.5

Graphics Card : Radeon9600XT
Resolution : 1280x1024
Color Depth : Millions

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

Rendering (Single CPU): 277 CB-CPU
Rendering (Multiple CPU): 490 CB-CPU

Multiprocessor Speedup: 1.77

Shading (CINEMA 4D) : 275 CB-GFX
Shading (OpenGL Software Lighting) : 797 CB-GFX
Shading (OpenGL Hardware Lighting) : 1556 CB-GFX

OpenGL Speedup: 5.67

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

Catfish_Man · Mar 20, 2006, 08:14 PM

Originally Posted by moonmonkey

Gulp, the wrong windows Graphics drivers are faster than the correct Mac ones.
It shows there is a lot of room for improvement.

Actually iirc they got the drivers working right for XP on minis. Still, there's definitely a lot of room for improvement >.<

moonmonkey · Mar 20, 2006, 10:05 PM

Originally Posted by Catfish_Man

Actually iirc they got the drivers working right for XP on minis. Still, there's definitely a lot of room for improvement >.<

I thought they got them working by using the driver for another card, could be wrong though.

Tenacious Dyl · Apr 6, 2006, 12:49 AM

Thought I would add mine, as there was only 1 other DP 2.5 G5 Result, I fell a little short for some reason. (All other apps closed, ran it 4 times to about same results)
(Dual 2.5 G5, 3.5 GBS RAM, 6800 GT)
************************************************** **

Rendering (Single CPU): 362 CB-CPU
Rendering (Multiple CPU): 639 CB-CPU

Multiprocessor Speedup: 1.76

Shading (CINEMA 4D) : 346 CB-GFX
Shading (OpenGL Software Lighting) : 1019 CB-GFX
Shading (OpenGL Hardware Lighting) : 1853 CB-GFX

OpenGL Speedup: 5.36

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

harrisjamieh · 06:59 AM

I know this thread is getting kind of old, and I have already submitted my results, but I dcided to re-run the software on my iMac, and I got almost double the result on the OpenGL speedup!

CINEBENCH 9.5
************************************************** **

Tester :

Processor : iMac Core Duo
MHz : 1.83 Ghz
Number of CPUs : 2
Operating System : Mac OS X 10.4.6

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

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

Rendering (Single CPU): 283 CB-CPU
Rendering (Multiple CPU): 534 CB-CPU

Multiprocessor Speedup: 1.89

Shading (CINEMA 4D) : 330 CB-GFX
Shading (OpenGL Software Lighting) : 1324 CB-GFX
Shading (OpenGL Hardware Lighting) : 2688 CB-GFX

OpenGL Speedup: 8.13

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

I also have XP installed via bootcamp, and I am just about to try the marking on that

Edit: Scores from XP

CINEBENCH 9.5
************************************************** **

Tester :

Processor : iMac Core Duo
MHz : 1.83 ghz
Number of CPUs : 2
Operating System : Windows XP SP2

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

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

Rendering (Single CPU): 269 CB-CPU
Rendering (Multiple CPU): 492 CB-CPU

Multiprocessor Speedup: 1.83

Shading (CINEMA 4D) : 320 CB-GFX
Shading (OpenGL Software Lighting) : 1296 CB-GFX
Shading (OpenGL Hardware Lighting) : 2595 CB-GFX

OpenGL Speedup: 8.10

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

The CPU scores under XP seem slightly slower, but apart from that, they seem pretty good.

rjayf98 · Apr 7, 2006, 04:40 PM

Originally Posted by harrisjamieh

I know this thread is getting kind of old, and I have already submitted my results, but I dcided to re-run the software on my iMac, and I got almost double the result on the OpenGL speedup!

Any idea how you got the OpenGL scores so much higher?

Catfish_Man · Apr 7, 2006, 06:38 PM

Originally Posted by zany-ninja

Any idea how you got the OpenGL scores so much higher?

10.4.6, perhaps?

power142 · Apr 7, 2006, 09:55 PM

Results from my PC - Athlon64 X2 4200+, Geforce 6800GT, 2GB RAM:

Rendering (Single CPU): 327 CB-CPU
Rendering (Multiple CPU): 613 CB-CPU

Multiprocessor Speedup: 1.87

Shading (CINEMA 4D) : 351 CB-GFX
Shading (OpenGL Software Lighting) : 1452 CB-GFX
Shading (OpenGL Hardware Lighting) : 3680 CB-GFX

OpenGL Speedup: 10.49

Can't even get it to run on my G5 - have version 9.5 of Cinebench and I'm running 10.4.6 on the G5 - "Splash Screen not found"

harrisjamieh · Apr 8, 2006, 02:04 AM

Originally Posted by Catfish_Man

10.4.6, perhaps?

Possibly, could it also be te firmware upgrade for the iMac that I installed 2 days ago?

Tenacious Dyl · Apr 9, 2006, 02:47 AM

I thought for comparison's sake, I would try out my old 12" Alubook, 1 GHz, 768 RAM, 10.4.6
Here are its results,

CINEBENCH 9.5
************************************************** **

Tester : Tenacious Dyl

Processor : G4
MHz : 1000
Number of CPUs : 1
Operating System : 10.4.6

Graphics Card : GeForce FX Go5200 32MB
Resolution : 1024 x 768
Color Depth : 32-bit Color (Millions)

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

Rendering (Single CPU): 104 CB-CPU
Rendering (Multiple CPU): --- CB-CPU

Shading (CINEMA 4D) : 112 CB-GFX
Shading (OpenGL Software Lighting) : 316 CB-GFX
Shading (OpenGL Hardware Lighting) : 594 CB-GFX

OpenGL Speedup: 5.31

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

Horsepoo!!! · Apr 11, 2006, 06:36 PM

Originally Posted by Catfish_Man

10.4.6, perhaps?

People are quick to blame Apple but apparently Maxon was responsible for this problem. The consequence however is that older computers will see lower OpenGL scores since Cinebench has removed deprecate code.