[slipjack]

Cinebench 2000 V1.0 Performance
************************************************** **

Tester : slipjack

Processor : G4 1000Mhz
Number of CPUs : 2
Physical Memory : 512 MB CL3 133 Mhz
Operating System : OS 9
Graphic Card : ATI Radeon 7500
Resolution : 1280x1024
Color Depth : millions

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

Shading (CINEMA 4D) : 8.59 CB
Shading (OpenGL) : 10.64 CB
Raytracing (Single CPU): 12.35 CB
Raytracing (Multiple CPU): 21.77 CB

OpenGL Shading is 1.24 times faster than CINEMA 4D Shading!
2 CPUs are 1.76 times faster than 1 CPU !

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

Whatever the benches though.. this puppy flys.

[juanvaldes]

Glad to see your new baby has arrived
Can anyone who knows what this means translate it into english?

[MacGorilla]

It means it is fast...1.76 times faster than 1 cpu

[TNproud2b]

and almost as fast as a single 1.4GHz Athlon Thunderbird.

[milhous]

Cinebench 2000 V1.0 Performance
************************************************** **

Tester : milhous

Processor : P/// 933
Number of CPUs : 1
Physical Memory : 256 X 2 PC133 CL=2
Operating System : WinXP Pro (Build 2600.xpclient.010817-1148)


Graphic Card : ATI All-in-Wonder Radeon
Resolution : 1280 x 1024
Color Depth : millions

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

Shading (CINEMA 4D) : 7.38 CB
Shading (OpenGL) : 11.83 CB
Raytracing (Single CPU): 10.22 CB
Raytracing (Multiple CPU): --- CB

OpenGL Shading is 1.60 times faster than CINEMA 4D Shading!

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

I don't know if lower values are better, but I would've really thought that the dualie would've smoked this Homebrew box in Cinebench, except in raytracing.

[ 02-13-2002: Message edited by: milhous ]

[TNproud2b]

Cinebench 2000 V1.0 Performance
************************************************** **

Tester : Athlon 1.33 @ 1.43

Processor : K7 Thunderbird
Number of CPUs : 1
Physical Memory : 128MB
Operating System : Win98SE

Graphic Card : V3/3000 AGP
Resolution : 800x600
Color Depth : 32bit

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

Shading (CINEMA 4D) : 13.52 CB
Shading (OpenGL) : 12.92 CB
Raytracing (Single CPU): 19.26 CB
Raytracing (Multiple CPU): --- CB

OpenGL Shading is 1.05 times slower than CINEMA 4D Shading!

[FormerLurker]

Originally posted by TNproud2b:

Resolution : <STRONG>800x600
</STRONG>

Notice that all the other benchmarks posted before these were using 1280x1024 ??

[ 02-14-2002: Message edited by: FormerLurker ]

[TNproud2b]

it doesn't make much difference when set at a higher resolution. My LCD is limited to 800x600, anyways.

Here's a dual Pentium3 900:

Cinebench 2000 V1.0 Performance
************************************************** **

Tester : &lt;fill this out&gt;

Processor : &lt;fill this out&gt;
Number of CPUs : 2
Physical Memory : &lt;fill this out&gt;
Operating System : &lt;fill this out&gt;

Graphic Card : &lt;fill this out&gt;
Resolution : &lt;fill this out&gt;
Color Depth : &lt;fill this out&gt;

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

Shading (CINEMA 4D) : 8.17 CB
Shading (OpenGL) : 6.36 CB
Raytracing (Single CPU): 10.62 CB
Raytracing (Multiple CPU): 17.86 CB

OpenGL Shading is 1.28 times slower than CINEMA 4D Shading!
2 CPUs are 1.68 times faster than 1 CPU !

[TNproud2b]

Dual Xeons upside tha head.


Cinebench 2000 V1.0 Performance
************************************************** **

Tester : ASUS XG-DLS motherboard

Processor : Pentium3 Xeon 500 1MB L2 cache
Number of CPUs : 2
Physical Memory : 640MB
Operating System : WindowsXP Pro

Graphic Card : nVidia Vanta TNT2 16MB PCI
Resolution : 800x600 (monitor limited)
Color Depth : 32bit (millions)

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

Shading (CINEMA 4D) : 5.81 CB
Shading (OpenGL) : 7.32 CB
Raytracing (Single CPU): 6.21 CB
Raytracing (Multiple CPU): 11.03 CB

OpenGL Shading is 1.26 times faster than CINEMA 4D Shading!
2 CPUs are 1.77 times faster than 1 CPU !

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

[c4zp3rgh0zt]

Some higher end SP cinebench scores here and some DP scores here.

[ 02-14-2002: Message edited by: c4zp3rgh0zt ]

[slipjack]

Originally posted by juanvaldes:
<STRONG>Glad to see your new baby has arrived
Can anyone who knows what this means translate it into english?</STRONG>

Thanks I'm actually already more productive! It's great. My last Macs at home were a G4 350 and a G3 400 Pismo so it's a big jump. (Still have the Pismo) Now I'm stuck at work with a 450DP

[TNproud2b]

ouch

[AlphaQuam]

More scores for more comparison...

Cinebench 2000 V1.0 Performance
************************************************** **

Tester : Tyan S2466

Processor : Athlon MP 1.4GHz
Number of CPUs : 2
Physical Memory : 512MB DDR
Operating System : WinXP Pro
Graphic Card : nVidia GeForce3 Ti 200
Resolution : 1600x1200
Color Depth : 32-bit

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

Shading (CINEMA 4D) : 13.78 CB
Shading (OpenGL) : 24.24 CB
Raytracing (Single CPU): 19.08 CB
Raytracing (Multiple CPU): 31.87 CB

OpenGL Shading is 1.76 times faster than CINEMA 4D Shading!
2 CPUs are 1.67 times faster than 1 CPU !

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

[thanatos]

Has anybody made the test under MacOSX with the dual 1GHz?
Why is Cinema 4D so slow on the Mac? I've seen tests in Lightwave v.7 under MacOSX which is better optimized for the G4 and it really kicks Athlons 1.4GHz ass.

[c4zp3rgh0zt]

Originally posted by thanatos:
<STRONG>Has anybody made the test under MacOSX with the dual 1GHz?
Why is Cinema 4D so slow on the Mac? I've seen tests in Lightwave v.7 under MacOSX which is better optimized for the G4 and it really kicks Athlons 1.4GHz ass.</STRONG>

The Dual G4 1GHz beats the single processor Atlon 1.4GHz in Cinebench2000 too... at Raytracing. The same is true in Lightwave in which the Raytracing benchmark is the one of the battery of tests were the G4 does all right, though it's no where near the top. When you look things like modeling tests (ex: CinemaBench 2000 Shading tests) and other rendering tests like radiosity, zbuffer and textures it's pretty far behind.

Lightwave Raytrace Test
Dual P4 2.2GHz
Dual Athlon XP 1800+
Quad P3 850MHz
Dual G4 1GHz

Lightwave Tracer Radiosity Test
Dual P4 2.2GHz
Dual Athlon MP 2000+
Quad P3 850Mhz
Dual G4 800Mhz

Lightwave Tracer No Radiosity Test
Dual P4 2.2GHz
Dual Athlon MP 2000+
Quad P3 850Mhz
Dual G4 800Mhz

You can find more Lightwave benchmark scores at Chris's Lightwave Benchmarks.

[Max8319]

can someone please translate these scores into english for those lacking in knowledge of computer "lingo" and scores like that?

so, does the dual ghz still lag noticeably behind PCs???

gracias

[slipjack]

Originally posted by Max8319:
<STRONG>can someone please translate these scores into english for those lacking in knowledge of computer "lingo" and scores like that?

so, does the dual ghz still lag noticeably behind PCs???

gracias</STRONG>

Well, the scores show that a dual 1 Ghz is faster than a single AMD 1.4 Ghz... so Mac's are still behind, but who really cares. 1Ghz DP is still the fastest Mac EVER, and that's pretty fast.

On another note, the G4 destroys AMD and Intel on RC5. I get 20 Mkeys/sec compared to less then 10 on a DUAL Xeon setup... check it out here:
http://n0cgi.distributed.net/speed/q...st=all&multi=1

Power PC 7450/7455 G4 1000 2 processors rc5 20,653,051

Intel Xeon 1000 2 processors rc5 5,835,597

[iamnotmad]

Originally posted by c4zp3rgh0zt:
<STRONG>

The Dual G4 1GHz beats the single processor Atlon 1.4GHz in Cinebench2000 too... at Raytracing. The same is true in Lightwave in which the Raytracing benchmark is the one of the battery of tests were the G4 does all right, though it's no where near the top. When you look things like modeling tests (ex: CinemaBench 2000 Shading tests) and other rendering tests like radiosity, zbuffer and textures it's pretty far behind.

Lightwave Raytrace Test
Dual P4 2.2GHz
Dual Athlon XP 1800+
Quad P3 850MHz
Dual G4 1GHz

Lightwave Tracer Radiosity Test
Dual P4 2.2GHz
Dual Athlon MP 2000+
Quad P3 850Mhz
Dual G4 800Mhz

Lightwave Tracer No Radiosity Test
Dual P4 2.2GHz
Dual Athlon MP 2000+
Quad P3 850Mhz
Dual G4 800Mhz

You can find more Lightwave benchmark scores at Chris's Lightwave Benchmarks.</STRONG>

I think the dual macs perform better in LW with 4 threads, rather than 8.

[Eug]

Originally posted by slipjack:
<STRONG>

Well, the scores show that a dual 1 Ghz is faster than a single AMD 1.4 Ghz... so Mac's are still behind, but who really cares. 1Ghz DP is still the fastest Mac EVER, and that's pretty fast.

On another note, the G4 destroys AMD and Intel on RC5. I get 20 Mkeys/sec compared to less then 10 on a DUAL Xeon setup... check it out here:
[URL=http://n0cgi.distributed.net/speed/query.cgi?cputype=all&arch=all&contest=all&multi=1]http://n0cgi.distributed.net/speed/query.cgi?cputype=all&arch=all&contest=all&multi=1[/UR L]

Power PC 7450/7455 G4 1000 2 processors rc5 20,653,051

Intel Xeon 1000 2 processors rc5 5,835,597</STRONG>

Yeah, RC5 screams on the G4. OTOH, the same app that runs RC5 also runs OGR, and the G4 does OK but not great. A G4 933 probably runs about as fast as a 1.2 GHz Athlon in OGR.

Who cares that Macs are behind? Well, actually a lot of people, especially given the cost of the Dual GHz. That's $$$$$ for slower performance in a lot of instances. The Athlon 1.4 GHz is cheap as borscht. The Dualie Mac is a nice machine, but like I've said before Apple still needs some speed boosts, especially with OS X. Competition is good.

Fortunately for me, I don't run these apps.

[KellyHogan]

Originally posted by c4zp3rgh0zt:
<STRONG>


You can find more Lightwave benchmark scores at Chris's Lightwave Benchmarks.</STRONG>

That's my Lightwave benchmark on that site if you noticed!

[driven]

Cinebench 2000 V1.0 Performance
************************************************** **

Tester : Work machine

Processor : Pentium 4 1.8 Ghz SDRAM
Number of CPUs : 1
Physical Memory : 512MB
Operating System : Windows XP

Graphic Card : ATI Rage 128 Pro GL
Resolution : 1024x768
Color Depth : 32 bit

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

Shading (CINEMA 4D) : 11.82 CB
Shading (OpenGL) : 13.20 CB
Raytracing (Single CPU): 15.83 CB
Raytracing (Multiple CPU): --- CB

OpenGL Shading is 1.12 times faster than CINEMA 4D Shading!

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

[c4zp3rgh0zt]

Originally posted by KellyHogan:
<STRONG>


That's my Lightwave benchmark on that site if you noticed! </STRONG>

Yes, I noticed . I remember the thread where you first posted them. If you get the chance to do and submit any of the tests you didn't get too that would be pretty cool.

[Leonis]

Cinebench 2000 V1.0 Performance
************************************************** **


Processor : Power PC G4/450
Number of CPUs : 1
Physical Memory : 1.5GB
Operating System : MacOS 9.1

Graphic Card : ATI Radeon 32MB AGP
Resolution : 1280x1024
Color Depth : 32-bit

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

Shading (CINEMA 4D) : 4.65 CB
Shading (OpenGL) : 5.94 CB
Raytracing (Single CPU): 5.87 CB
Raytracing (Multiple CPU): --- CB

OpenGL Shading is 1.28 times faster than CINEMA 4D Shading!

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


Processor : iBook PowerPC G3 600
Number of CPUs : 1
Physical Memory : 384MB
Operating System : Mac OS 9.21
Graphic Card : ATi Rage Mobility 128
Resolution : 1024x768
Color Depth : 32-bit

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

Shading (CINEMA 4D) : 5.06 CB
Shading (OpenGL) : 6.86 CB
Raytracing (Single CPU): 7.82 CB
Raytracing (Multiple CPU): --- CB

OpenGL Shading is 1.36 times faster than CINEMA 4D Shading!


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


Processor : AMD K7 950Mhz ---- Overclocked to 1.1Ghz
Number of CPUs : 1
Physical Memory : 512MB
Operating System : Win 98 SE

Graphic Card : ATI Rage II+ 4MB
Resolution : 1024 x 768
Color Depth : 32-bit

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

Shading (CINEMA 4D) : 8.49 CB
Shading (OpenGL) : 5.66 CB
Raytracing (Single CPU): 14.74 CB
Raytracing (Multiple CPU): --- CB

OpenGL Shading is 1.50 times slower than CINEMA 4D Shading!


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


Processor : Power PC G4/500 DP
Number of CPUs : 1
Physical Memory : 1.5GB
Operating System : MacOS 9.2.1/ X

Graphic Card : ATI Radeon 32MB AGP
Resolution : 1280x1024
Color Depth : 32-bit

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

Shading (CINEMA 4D) : 4.74 CB
Shading (OpenGL) : 6.22 CB
Raytracing (Single CPU): 5.90 CB
Raytracing (Multiple CPU): 12.1 CB

OpenGL Shading is 1.31 times faster than CINEMA 4D Shading!
2 CPUs are 2.05 times faster than 1 CPU !

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


Processor : 1.2 Ghz PIII
Number of CPUs : 1
Physical Memory : 1 GB
Operating System : Win 2K Pro

Graphic Card : nVidia GeForce 2 MX400, 64MB VRAM
Resolution : 1280x1024
Color Depth : 32-bit
************************************************** **

Shading (CINEMA 4D) : 9.60 CB
Shading (OpenGL) : 17.17 CB
Raytracing (Single CPU): 14.09 CB
Raytracing (Multiple CPU): --- CB

OpenGL Shading is 1.79 times faster than CINEMA 4D Shading!

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


Oh my god. The 1.8Ghz P4 is only slight faster than the 1.2Ghz PIII. Geezzzz. P4 really does stink

[ 02-15-2002: Message edited by: Leonis ]

[c4zp3rgh0zt]

Originally posted by Leonis:
Oh my god. The 1.8Ghz P4 is only slight faster than the 1.2Ghz PIII. Geezzzz. P4 really does stink

Look closely at those results, that's an SDRAM based P4 which cripples its performance. I'd imagine a P4a 2.2GHz w/RDRAM or DDR is right in line with or slightly faster than an Athlon XP 2000+ [EDIT: in Cinebench2000 -- it's definitely faster in Lightwave].

[ 02-15-2002: Message edited by: c4zp3rgh0zt ]

[driven]

Originally posted by c4zp3rgh0zt:
<STRONG>

Look closely at those results, that's an SDRAM based P4 which cripples its performance. I'd imagine a P4a 2.2GHz w/RDRAM or DDR is right in line with or slightly faster than an Athlon XP 2000+ [EDIT: in Cinebench2000 -- it's definitely faster in Lightwave].

[ 02-15-2002: Message edited by: c4zp3rgh0zt ]</STRONG>

It cripples it slightly, but not as much as you'd think. It's still running at a 133 buss speed. (Same as DDR) ... this is a stock HP box.

I'll see if I can post the numbers of the P4 1.8 DDR today when I get into work. It wasn't THAT much faster. (Sadly).

[driven]

Cinebench 2000 V1.0 Performance
************************************************** **

Tester : Home PC

Processor : Athlon XP 1700+ ECS K7S5A
Number of CPUs : 1
Physical Memory : 256MB DDR
Operating System : Windows XP

Graphic Card : nVidia GeForce2 MX (Hercules implementation)
Resolution : 1024x768
Color Depth : 32 bits

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

Shading (CINEMA 4D) : 14.69 CB
Shading (OpenGL) : 25.60 CB
Raytracing (Single CPU): 19.99 CB
Raytracing (Multiple CPU): --- CB

OpenGL Shading is 1.74 times faster than CINEMA 4D Shading!

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

[driven]

Originally posted by driven:
<STRONG>

It cripples it slightly, but not as much as you'd think. It's still running at a 133 buss speed. (Same as DDR) ... this is a stock HP box.

</STRONG>

Just to clarify this .... (before someone mentions it) ... yes ... DDR266 runs at 133Mhz just like SDR.

The difference is that DDR transfers data at the start and the trailing edge of the clock cycle (2X per cycle) unlike SDR which only transfers it once.

But .. the data doesn't move through the rest of the system any faster.

(It's an improvement, but not as dramatic as it could be.)

The only other alternative is RAMBUS ... but then you have latency issues. (And I wouldn't support ANYTHING that RAMBUS produces ... so I keep hoping for a better alternative from another company.)

[ 02-15-2002: Message edited by: driven ]

[driven]

Cinebench 2000 V1.0 Performance
************************************************** **
Tester : Work machine

Processor : Pentium 4 1.8 Ghz DDR
Number of CPUs : 1
Physical Memory : 512MB
Operating System : Windows XP

Graphic Card : ATI Rage 128 Pro GL
Resolution : 1024x768
Color Depth : 32 bit

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

Shading (CINEMA 4D) : 12.02 CB
Shading (OpenGL) : 13.58 CB
Raytracing (Single CPU): 16.13 CB
Raytracing (Multiple CPU): --- CB

OpenGL Shading is 1.13 times faster than CINEMA 4D Shading!

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


(Similar machine as the other P4 1.8, but with DDR ....)

I find it interesting that my home machine (Athlon XP 1700+ -- 1.46Ghz) is MUCH faster than this P2 1.8!

[Leonis]

Confirmed! P 4 sucks

[Turbocharger]

Cinebench 2000 V1.0 Performance
************************************************** **

Tester : P4 1.5MHz

Processor : P4 1.5MHz
Number of CPUs : 1
Physical Memory : 512 RDRAM
Operating System : Win 98SE

Graphic Card : ATI Radeon
Resolution : 1024x768
Color Depth : 32-bit

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

Shading (CINEMA 4D) : 14.77 CB
Shading (OpenGL) : 21.56 CB
Raytracing (Single CPU): 17.35 CB
Raytracing (Multiple CPU): --- CB

OpenGL Shading is 1.46 times faster than CINEMA 4D Shading!

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

RDRAM does help a bit ... notice the clock speed (1.5 MHz)

[driven]

Originally posted by Turbocharger:
<STRONG>Cinebench 2000 V1.0 Performance
************************************************** **

Tester : P4 1.5MHz

Processor : P4 1.5MHz
Number of CPUs : 1
Physical Memory : 512 RDRAM
Operating System : Win 98SE

Graphic Card : ATI Radeon
Resolution : 1024x768
Color Depth : 32-bit

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

Shading (CINEMA 4D) : 14.77 CB
Shading (OpenGL) : 21.56 CB
Raytracing (Single CPU): 17.35 CB
Raytracing (Multiple CPU): --- CB

OpenGL Shading is 1.46 times faster than CINEMA 4D Shading!

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

RDRAM does help a bit ... notice the clock speed (1.5 MHz) </STRONG>

Apparently it helps quite a bit.

(The other thing that is different is that the pair of P4's I tested on used the ATI Rage 128 card while you are using a Radeon.)

[Andreas Niemand]

Cinebench 2000 V1.0 Performance
************************************************** **

Tester : Andreas Niemand

Processor : PPC 604/UMAX Pulsar
Number of CPUs : 1
Physical Memory : 176MB
Operating System : MacOS 9.1

Graphic Card : build in
Resolution : 1024*768
Color Depth : 16bit

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

Shading (CINEMA 4D) : 1.64 CB
Shading (OpenGL) : 1.06 CB
Raytracing (Single CPU): 1.63 CB
Raytracing (Multiple CPU): --- CB

OpenGL Shading is 1.55 times slower than CINEMA 4D Shading!

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

...for the record. anyway, this is my actual computer i am working on- and i even do 3d-graphics on it. it usually takes some nights to render, but....
for the athlon i think it has a very fast fpu- therefore it renders faster than any actual cpu. to sad that different computingprograms require different cpus for best performance. athlon-rendering, g4-video/photoshop.

andreas

[driven]

Cinebench 2000 V1.0 Performance
************************************************** **

Tester : G4 Cube 500

Processor : 500 Mhz G4
Number of CPUs : 1
Physical Memory : 768 MB
Operating System : MacOS 9.2.1

Graphic Card : Rage 128 16MB
Resolution : 1280 X 1024
Color Depth : 32 bit

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

Shading (CINEMA 4D) : 4.51 CB
Shading (OpenGL) : 6.00 CB
Raytracing (Single CPU): 5.57 CB
Raytracing (Multiple CPU): --- CB

OpenGL Shading is 1.33 times faster than CINEMA 4D Shading!

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


Cinebench 2000 V1.0 Performance
************************************************** **

Tester : G4 Cube 500

Processor : 500 Mhz G4 7400
Number of CPUs : 1
Physical Memory : 768 MB
Operating System : OS X 10.1 (Classic)

Graphic Card : Rage 128 16MB
Resolution : 1280x768
Color Depth : 32 Bit

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

Shading (CINEMA 4D) : 4.92 CB
Shading (OpenGL) : --- CB
Raytracing (Single CPU): 6.16 CB
Raytracing (Multiple CPU): --- CB


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

This is interesting. The test ran faster when running in classic mode than it did in native OS9 mode. (I apparently forgot to run the OpenGL test ... I'll rerun it later.)

I am overall though disappointed with how poorly this test runs on the G4. This test really seems to favor the Athlon.

[scarab]

The Athlon has a stronger Floating Point Unit right?

[driven]

Here are the numbers from my family's iMac ...
Sadly some of these numbers are better than on my G4 Cube.

It seems blatently obvious that this test is NOT optimized at ALL for the G4 (or even the Power PC).

Does anyone have any other benchmarks ?


Cinebench 2000 V1.0 Performance
************************************************** **

Tester : iMac DV+ SE 500 (Summer 2000)

Processor : 500 Mhz G3
Number of CPUs : 1
Physical Memory : 384 MB
Operating System : Mac OS 9.0.4

Graphic Card : Rage 128 8MB
Resolution : 1024x768
Color Depth : Thousands of colors

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

Shading (CINEMA 4D) : 4.75 CB
Shading (OpenGL) : 5.22 CB
Raytracing (Single CPU): 6.00 CB
Raytracing (Multiple CPU): --- CB

OpenGL Shading is 1.10 times faster than CINEMA 4D Shading!

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

[TNproud2b]

on the contrary - it IS a cross-platform benchmark.

One of the few ways to compare performance between Mac and PC.

Cinebench is already 'optimized'.

PS, the are few if any true benchmarks where the G4 won't get creamed by the Athlon.

[ 02-16-2002: Message edited by: TNproud2b ]

[driven]

Originally posted by TNproud2b:
<STRONG>on the contrary - it IS a cross-platform benchmark.

One of the few ways to compare performance between Mac and PC.

Cinebench is already 'optimized'.

PS, the are few if any true benchmarks where the G4 won't get creamed by the Athlon.

[ 02-16-2002: Message edited by: TNproud2b ]</STRONG>

But how do you explain why the G3 scored some better numbers than the G4?

It seems that this benchmark takes full advantage of the Athlon while not taking ANY advantage of the Altivec in the G4.

[TNproud2b]

It's not that it isn't optimized for AltiVec - it's because the calculations can't be performed using AltiVec. Not every application (few, actually) can make use of the AltiVec instruction set.

On the x86 side of things, there are MMX, SSE, and SSE2 instruction sets which could be compared to AltiVec.

[driven]

Originally posted by TNproud2b:
<STRONG>It's not that it isn't optimized for AltiVec - it's because the calculations can't be performed using AltiVec. Not every application (few, actually) can make use of the AltiVec instruction set.

On the x86 side of things, there are MMX, SSE, and SSE2 instruction sets which could be compared to AltiVec.</STRONG>

Understand.

Bummer.

(I'm REALLY happen with my Athlon XP though .... it easilly beat a P4 with nearly a 400Mhz advantage!)

[TNproud2b]

The G4's performance is nothing to be ashamed of. Clock-for-clock there isn't anything out there that's faster (for the desktop market, that is).

The problem isn't with the architecture, it's with the manufacturing process which is limiting the core speed to 1GHz or so.

The G4 is depending on raw power to get the job done - smallish pipeline, rather weak FPU, and minimal hardcoded instructions (AltiVec).

The Athlon works very much like the G4, depending mostly on the brute force of its decent FPU, small pipeline (bigger than a G4, tho), and a healthy dose of hardcode (MMX, initially - now SSE2 in the AthlonMPs, as well I believe).

Nobody believes me, but the Pentium4 has a lot of potential. Remember, this is a chip that is supposed to scale to nearly 10GHz in its lifetime. At 2GHz, it's a slouch. At 5GHz it will leave a mark.

The P4 is depending on efficiency to increase performance. Strangely, it looks rather inefficient in benchmarks, doesn't it? Someday when apps are written to take advantage of the SSE2 instructions, you'll see a totally different P4.

Guess what the SSE2 instruction set is supposed to improve?

Exactly the stuff that is useful in a 'digital hub' environment. Exactly the stuff that Apple wants to be good at. DVD playback, 3D gaming, anything related to multimedia.

This year the Athlon is king, next year is anyone's guess.

[ 02-16-2002: Message edited by: TNproud2b ]

[michaelb]

Originally posted by TNproud2b:
<STRONG>on the contrary - it IS a cross-platform benchmark. One of the few ways to compare performance between Mac and PC. Cinebench is already 'optimized'.</STRONG>

Sigh. Won't this pointless thread ever die? Yes, yes, yes, of course an Athlon is better for Cinema4D, for two reasons:

(1) the app is is intended for x86 architecture, the Mac version is a straight cross-compile of the C code. Marketing BS aside, not only is there is no G4 optimization, there is no PPC optimization.

(In candid interviews, the German engineers have admitted as much, and you only have to look at the single-window / common controls interface to realize the validity of the "straight cross-compile" comment.)

(2) the Athlon kicks ass in Floating Point performance.

In contrast, the LightWave 7 engineers have tuned their code to the PowerPC architecture more fully, and their benchmarks reflect this.

Anyway, pointless benchmarks aside, for those interested (although I seriously have my doubts whether many of the people running these "Cinebench" tests have the talent to even model a scene themselves) here is an article on 3D animation film making - how the 'Jimmy Neutron: Boy Genius' film was started using off-the-shelf software, in contrast to 'Shrek' or 'Monsters, Inc'.
����
http://www.latimes.com/entertainment...590feb15.story

Included is a comment from John Lasseter about how great the situation is now where someone with talent can get a Macintosh and use these great off-shelf 3D packages, a big difference to when he started working in 3D.
�
���
���
���

[ 02-16-2002: Message edited by: michaelb ]

[TNproud2b]

If Cinebench is such a bad benchmark to use for comparing x86 and PPC - then name me a few more appropriate benchmarks.

hell...name me one.


I've been hearing about a megahertz myth for years, but I haven't yet seen any hard facts to back it up. Nearly ALL benchmarks place the G4 at the bottom of the pack in performance - but the excuse is always the same...'bad code'.

geez, what good is a processor that depends entirely on 'great' code to operate with reasonable performance? I have about a hundred apps on my puter and in my opinion maybe three of them do not have 'crappy' code. Crappy code is what we all use every day.

Give me a processor that chews through crappy code - so I can get work done today, instead of waiting an eternity for 'good' code.

Seriously, in the last two years I've spent countless hours pouring over benchmark data from all sorts of platforms. Except in the case of 3 Photoshop filters and a RC5 client there ain't no high-scoring PPC anywhere to be found.

The G4's RC5 performance can be equalled with a $7 piece of silicon, as evidenced by custom built key crunching machines, by the way.

so, please, show me where these 'proper' cross-platform benchmark tools are hiding.


[ 02-16-2002: Message edited by: TNproud2b ]

[michaelb]

It's not "bad code" -- Cinema4D is an excellent program. It's just that it's intended mainly for x86 architecture which is 85% of their target market. Can't blaim them for that.

You should definitely select a platform to match your use. If you spend most of your time in Cinema4D - a $2,200 piece of software - get an Athlon MP box to increase your productivity. If you're just using it casually in amongst Final Cut Pro'ing and Photoshop'ing, the comparitively lacklustre performance for 3D on the Mac platform just doesn't matter.

IMHO, most benchmarks *are* pointless. The importance of a "Towers of Hanoi" test in daily life is zero.

What does matter is performance in the smoothness of integration between the software you use every hour and if *that* is fast ... If your speedily filtered Photoshop image is actually color-balanced ... If your workflow can be automated with AppleScript ... If the 3D image you just rendered can be drag & dropped into a Flash project and you know it'll just work ... If that FireWire cable works when you plug it in and how quickly it syncs your backups.

THAT sort of stuff is what should be benchmarked, not the friggin' pointless "My P4 running Cinebench finished 8.51 seconds faster than yours, neh neh! But my OS will probably need a 5 hour virus overhaul afterwards" get-a-life-geekboy nonsense.

Funnily enough, in 10 years *I* haven't seen anyone benchmark useful things! Could it possibly be the benchmark writers are too busy writing "Towers of Hanoi" version 234.5 to actually do any useful work?!

Anyway, you asked for things that PowerMacs kick ass at: you already mentioned Photoshop and RC5 (please join your $7 piece of silicon to Team MacNN, if it can), I'll add: Cleaner 5, Sorenson Video 3, Genentech's BLAST... too lazy to think of more that I don't use anyway.

[TNproud2b]

good reply.

I agree there is no easy way to compare platforms...

and, yes, I'd feel funny using a $2,000 software package on a $450 Athlon rig - just don't seem proper.

[GnOm]

Thanks to michaelb and TNproud2b for making this thread worthwile reading.

Most benchmarks are pointless IMO too, either they are "artifilcial" benchmarks (SPECmark and such) and have not (necessarily) much to do with realworld performance, or they are "realworld" benchmarks (Applications running several tasks) and are not really comparable because of a zillion of factors.
For example: Cinebench is dogslow on G4 but fast on Pentium/Athlon, rc5 (distributed.net) is very fast on G4 and quite slow on Pentium. Photoshop test depends heavily on the filters using AltiVec or not etc. etc.

In the end it�s just a game and I don�t give a **** about any benchmark.


bye.

[TNproud2b]

If you're happy with the machine you have, then benchmark numbers truly are pointless.

Having more power than you currently need is probably a good thing...it will be an asset when newer, hardware-intensive apps and OSes are released. Sure you can do anything with your machine today, but tomorrow when Mac OS_XI hits the shelves you might appreciate those currently 'un-needed' megahertz. Then again, after a year or so every desktop computer is an antique.

Having built and played with literally dozens of different combinations of OS and hardware - I discovered that the fastest machines weren't always my favorite machines.

I like machines that have character. Character and speed would be nice but I haven't seen that yet. Today I'm using a dual P3 Xeon500 on a top of the line motherboard booting XP & Redhat. Last year this technology cost more than my car...this year it's under $500 and considered laughingly slow. I love it because it has character.

It's only fair to mention that I have never owned a Mac. Some day I would like to. After I win the state lottery, maybe. Last month I was one click away from buying a new iMac. Getting me to part with $1500 isn't easy, especially when it's for 800MHz stuffed inside a rather 'girlish-looking' (flame me, but I know you agree) package. But it has character. Enough to make me part with maybe $1200. For that kind of money I could build something 400% faster, but buying character for it might be difficult.

[ 02-17-2002: Message edited by: TNproud2b ]

[cutterjohn]

pipelines: larger pipelines are not necessarily a good thing. In tests like this the difference will not show, as there are few if any branches and they should all be properly predicted, resulting in not having to do a pipeline clear/refill in the case of a branch prediction miss.

The P4 with it's relatively HUGE pipeline will suffer in programs where branches may not always be readily predicted, requiring it to clear the pipeline and then reload with the proper instructions(stall). This consumes CPU cycles. Multimedia apps are relatively easy to handle, so this should not be a problem with graphics rendering and things of that nature, but other programs like matlab, mathematica, custom data analysis software, etc. the P4 will take a hit. Even in these cases it is possible to come up with a data set that will NOT bring to light this deficiency(? I can't really say as I have not done alot of profiling/benchmarking of typical computer program execution paths currently in use, other than in limited cases.)

In any event a suite of benchmarks provided by varying data sets and executing in different problem spaces would provide a truer benchmark of a particular CPU and/or groups of CPUs/families. IMHO single application/problem space benchmarks are worthless.

Also, from looking at the results, I would say that just by that it is fairly obvious that MMX enhancements & x86 optimizations are in place for the x86 version, while little or any optimization is in place for the PPC version. Has anyone checked barefeats? are they still running? any SPECfp, etc. benchmarks?

[driven]

Originally posted by TNproud2b:
<STRONG>If Cinebench is such a bad benchmark to use for comparing x86 and PPC - then name me a few more appropriate benchmarks.

hell...name me one.


I've been hearing about a megahertz myth for years, but I haven't yet seen any hard facts to back it up. Nearly ALL benchmarks place the G4 at the bottom of the pack in performance - but the excuse is always the same...'bad code'.
</STRONG>

The Athlon lays claim to the Megahurts myth fairly well. It consistently outperforms the P4 at higher clock speeds.

To be honest if you were to put a PIII and P4 next to each other at the same clock speed the PIII would win. (Same goes for the P Pro and the P2).

[TNproud2b]

Here's a stunning score...


Cinebench 2000 V1.0 Performance
************************************************** **

Tester : &lt;BeeKs&gt;

Processor : &lt;dual XP1900+ 143fsb&gt;
Number of CPUs : 2
Physical Memory : &lt;512&gt;
Operating System : &lt;2K&gt;

Graphic Card : &lt;GeForce 4 mx440&gt;
Resolution : &lt;800x600&gt;
Color Depth : &lt;32 bit&gt;

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

Shading (CINEMA 4D) : 17.14 CB
Shading (OpenGL) : 25.65 CB
Raytracing (Single CPU): 24.02 CB
Raytracing (Multiple CPU): 39.97 CB

OpenGL Shading is 1.50 times faster than CINEMA 4D Shading!
2 CPUs are 1.66 times faster than 1 CPU !


edited to add: exactly twice as fast as a DP 1000 G4

hmm

[ 03-27-2002: Message edited by: TNproud2b ]

[mrl14]

hahah this is hilarious. All of you are comparing these specs...but the bottom line is this.


My friend just bought a Dell 1.9ghz P4 running Windows XP. Of course he calls me to hook up his printer. Lets see...it took me 1 hour to get a usb lexmark working, not only b/c of plug and pray but because of how slow the damn OS was. I accidently hit the windows update button and while that was working in the bg i couldn't do anything.

My other friend bought a 1.4ghz P4 running windows XP (custom built machine) which I have yet to play around with (I will this weekend), and I'm expecting it to be just as slow.

Now I got a dual 1ghz mac, recently upgrade from an 867mhz g4 and the bottom line is it's fast in everyday tasks. Opening windows, opening programs, surfing the web, adding printers, etc. so what if its 30 secs slower in rendering something, to get to the point being able to start my render process is faster on the mac. Not to mention OS X is only going to get faster which each release.

But I am disappointed that the dual 1ghz doesn't seem much faster than my 867, but that could be because I've got all this crap installed ( all i did was swap HD's).