[terrell]

These are straight from my new 12" powerbook.

867/256/40gig

Results 72.15
System Info
Xbench Version 1.0b6
System Version 10.2.3
Physical RAM 256 MB
Processor PowerPC G4 [867 MHz]
L1 Cache 32K (instruction), 32K (data)
L2 Cache 256K
Bus Frequency 134 MHz
CPU Test 97.62
GCD Recursion 91.12 3.56 Mops/sec
Floating Point Basic 100.11 337.44 Mflop/sec
AltiVec Basic 98.51 5.34 Gflop/sec
Floating Point Library 101.40 4.55 Mops/sec
Thread Test 56.81
Computation 54.92 442.46 Kops/sec, 4 threads
Lock Contention 58.82 738.37 Klocks/sec, 4 threads
Memory Test 83.79
System 76.82
Allocate 104.94 55.20 Kalloc/sec
Fill 87.19 499.98 MB/sec
Copy 55.39 276.95 MB/sec
Stream 92.16
Copy 95.28 416.66 MB/sec [altivec]
Scale 95.25 415.58 MB/sec [altivec]
Add 92.45 420.66 MB/sec [altivec]
Triad 86.26 379.03 MB/sec [altivec]
Quartz Graphics Test 77.07
Line 71.82 1.83 Klines/sec [50% alpha]
Rectangle 82.59 5.81 Krects/sec [50% alpha]
Circle 86.43 1.99 Kcircles/sec [50% alpha]
Bezier 80.71 876.99 beziers/sec [50% alpha]
Text 67.18 1.10 Kchars/sec
OpenGL Graphics Test 102.38
Spinning Squares 102.38 71.64 frames/sec
User Interface Test 58.41
Elements 58.41 19.88 refresh/sec
Disk Test 57.56
Sequential 57.07
Uncached Write 51.44 22.44 MB/sec [4K blocks]
Uncached Write 49.44 20.11 MB/sec [256K blocks]
Uncached Read 97.23 15.31 MB/sec [4K blocks]
Uncached Read 49.67 21.43 MB/sec [256K blocks]
Random 58.07
Uncached Write 70.85 1.07 MB/sec [4K blocks]
Uncached Write 49.26 11.30 MB/sec [256K blocks]
Uncached Read 58.28 0.38 MB/sec [4K blocks]
Uncached Read 57.77 11.32 MB/sec [256K blocks]

[Hozie]

Well, a couple of strange things to note:

1) the results differ from the barefeats test as follows:

CPU Test: BF: 88.0 terrel's: 97.62
Memory test: BF: 96.7 terrel's: 83.79
Thread test: BF: 62.3 terrel's: 56.81
Quartz Graphics: BF: 74.7 terrel's: 77.07
OpenGL Graphics: BF: 73.4 terrel's: 102.38

2) memory in the BF machine was maxed out I think

3) why is the thread test on terrell's lower than the BF one?

4) Graphics don't seem bad at all

[roders]

Can someone put in laymans terms what (each off?) these
results mean, or maybe how these benchmarks translate into real world usability/performance
and/or how they compare to other Mac's?

[morlium]

Originally posted by Hozie:

2) memory in the BF machine was maxed out I think

Yes, BF.com did note that the memory was maxed out, but I don't believe it. None of the machines at MWSF had more than the stock memory config, and that's presumably where they ran their "tests."

[Eug]

Originally posted by roders:
Can someone put in laymans terms what (each off?) these
results mean, or maybe how these benchmarks translate into real world usability/performance
and/or how they compare to other Mac's?

Maybe about 4/5ths the speed overall compared the TiBook SD. Slightly faster than the iBook 800.

By the way, the test is in no way an accurate test, and I should point out that the current version of Xbench is different from what BareFeats.com used.

[Hozie]

Agree that the BF tests are shady.... just wanted to see how they compared to the actual end product

anyway, the graphics test are important for the snappiness of the OS in the sense that the redraws are handled by Quartz extreme, which benefits form a faster video card. In the same manner, games performance will be better if your OpenGL score is high. The CPU test is pretty self-explanatory and the memory test has to do with the speed of the RAM. The thread test measures how fast the processor handles threads (obviously), which are basically the underlying structure of a program.

[Flowbee]

Originally posted by morlium:
. None of the machines at MWSF had more than the stock memory config, and that's presumably where they ran their "tests."

Not true. The 12" I used at Macworld had 640mb installed.

[Hozie]

BTW, thanks for the benchmarks terrell!

I'd like to ask you a couple of other things if you don't mind:

1) Does it get hot?
2) Is it loud (fan-wise)?
3) How's the general snappiness of the little guy?
4) Please feel free to post your comments!!! (I think we're all dying for a short review here)

[Eug]

Originally posted by Hozie:
Agree that the BF tests are shady.... just wanted to see how they compared to the actual end product

Dunno if you were talking to me, but what I was getting at is that the Xbench currently available isn't even the same test. It has been updated a couple of times already since the beginning of Jan. So you can't compare the results directly.

[xyber233]

Do you have any games you can benchmark? Just curious how it does compared to the iBook.

[Hozie]

Dunno if you were talking to me, but what I was getting at is that the Xbench currently available isn't even the same test. It has been updated a couple of times already since the beginning of Jan. So you can't compare the results directly.

Ah, bleh. Didn't know that... I thought they would at least standardize their results so that each version yields the same approximate marks, give or take a few points. But you're right, we can't assume that that's the case, so we're stuck with information that doesn't tell us much. Anyone having a Ti800 (15") or 1ghz care to post their marks with this latest (and preferrably identical) xBench build?

[ima_pseudonym]

Originally posted by Hozie:
Anyone having a Ti800 (15") or 1ghz care to post their marks with this latest (and preferrably identical) xBench build?

Here is my TiBook 867 running the same version (though I didn't tweak anything before running it - like making sure journaling is turned off, etc.)

Results 74.59
System Info
Xbench Version 1.0b6
System Version 10.2.3
Physical RAM 512 MB
Processor PowerPC G4 [867 MHz]
L1 Cache 32K (instruction), 32K (data)
L2 Cache 256K
L3 Cache 1024K
Bus Frequency 134 MHz
CPU Test 92.18
GCD Recursion 86.19 3.37 Mops/sec
Floating Point Basic 95.72 322.63 Mflop/sec
AltiVec Basic 91.87 4.98 Gflop/sec
Floating Point Library 95.60 4.29 Mops/sec
Thread Test 55.37
Computation 53.82 433.57 Kops/sec, 4 threads
Lock Contention 57.01 715.64 Klocks/sec, 4 threads
Memory Test 90.73
System 91.21
Allocate 96.98 51.01 Kalloc/sec
Fill 155.51 891.73 MB/sec
Copy 61.92 309.62 MB/sec
Stream 90.25
Copy 88.82 388.42 MB/sec [altivec]
Scale 95.86 418.26 MB/sec [altivec]
Add 88.39 402.18 MB/sec [altivec]
Triad 88.36 388.25 MB/sec [altivec]
Quartz Graphics Test 103.41
Line 107.48 2.74 Klines/sec [50% alpha]
Rectangle 113.67 8.00 Krects/sec [50% alpha]
Circle 114.10 2.63 Kcircles/sec [50% alpha]
Bezier 99.94 1.09 Kbeziers/sec [50% alpha]
Text 87.10 1.42 Kchars/sec
OpenGL Graphics Test 79.22
Spinning Squares 79.22 55.44 frames/sec
User Interface Test 80.92
Elements 80.92 27.53 refresh/sec
Disk Test 51.90
Sequential 58.21
Uncached Write 49.83 21.74 MB/sec [4K blocks]
Uncached Write 50.93 20.72 MB/sec [256K blocks]
Uncached Read 132.16 20.82 MB/sec [4K blocks]
Uncached Read 46.61 20.11 MB/sec [256K blocks]
Random 46.83
Uncached Write 39.99 0.60 MB/sec [4K blocks]
Uncached Write 44.62 10.24 MB/sec [256K blocks]
Uncached Read 50.74 0.33 MB/sec [4K blocks]
Uncached Read 54.71 10.72 MB/sec [256K blocks]


Not too many surprises, except for the cpu performance - does (unsupported) ddr, but not L3 cache, make a difference on these tests?

[terrell]

It does get a little warm, but compared to other laptops I have used (mostly thinkpads). i wouldn't go so far as to say HOT.

The left hand side does seem to be noticably warmer than the right though (I assume the HD is over there, opposite the battery).

There is almost no noise at all coming from the powerbook. the problems with the Tibooks don't seem to be carried over to the new Alubooks. It is very very quiet.

The general snappiness seems very good. I am very happy with it so far. I'm comparing it to a 466 G4 tower and a 450 cube, which are the two machines I use daily, and it is noticably faster, especially within the finder.

I'm also very impressed with the speakers.... the people who think this is an iBook in a new case should take a listen to the speakers in the powerbook.... it blows the iBook away!

[ebisix]

Originally posted by terrell:
It does get a little warm, but compared to other laptops I have used (mostly thinkpads). i wouldn't go so far as to say HOT.

The left hand side does seem to be noticably warmer than the right though (I assume the HD is over there, opposite the battery).

There is almost no noise at all coming from the powerbook. the problems with the Tibooks don't seem to be carried over to the new Alubooks. It is very very quiet.

The general snappiness seems very good. I am very happy with it so far. I'm comparing it to a 466 G4 tower and a 450 cube, which are the two machines I use daily, and it is noticably faster, especially within the finder.

I'm also very impressed with the speakers.... the people who think this is an iBook in a new case should take a listen to the speakers in the powerbook.... it blows the iBook away!

Excellent. Just wanted to thank you for this info.

[Eug]

Originally posted by Hozie:
Ah, bleh. Didn't know that... I thought they would at least standardize their results so that each version yields the same approximate marks, give or take a few points. But you're right, we can't assume that that's the case, so we're stuck with information that doesn't tell us much. Anyone having a Ti800 (15") or 1ghz care to post their marks with this latest (and preferrably identical) xBench build?

He standardizes it against his own computer, which is a dual 800 with Geforce2 MX I believe. His computer ~ 100 for all the tests. However, he did some changes to make some of the results more consistent.

I don't think the OpenGL test is in any way accurate though. The Radeon 9000 in the TiBooks performs as about as well as older cards in other machines, and I guess about the same as his Geforce2 MX. See this thread. In real life, a Radeon 9000 would destroy a Geforce2 MX.

Also, do a search for the "Let Get Some Benchmarks!!!" thread. Lots of Xbench action (older version) in there.

[Hozie]

cool....

Does anyone else find it strange that the Geforce4Go blows the radeon away in OpenGL tests but majorly lags behind it in Quartz? Hmmm.....

Actually, when you look at the numbers closely, the new 12" beats the old 15" in every test except the 2D graphics. Well, let's hope apple or nVidia come out with a superdeluxe driverpack which fixes this, although I think this proves the Radeon is a better all-round mobile chip.

Still getting a 12" though. Good job terrell!

[Hozie]

I don't think the OpenGL test is in any way accurate though. The Radeon 9000 in the TiBooks performs as about as well as Rage cards in other machines, and I guess about the same as his Geforce2 MX. See this thread. In real life, a Radeon 9000 would destroy a Geforce2 MX.

Also, do a search for the "Let Get Some Benchmarks!!!" thread. Lots of Xbench action (older version) in there.

Ah, ok, so that's why....

[all2ofme]

Great to have some new material to go on!

The things I thought were interesting (I've very little idea of what these things actually *mean*, but it's better than nothing) were:

15.2" / 12"

System: 91.21 / 76.82
Fill: 155.51 / 87.19
Quartz graphics test: 103.41 / 77.07
Line: 107.48 / 71.82
Rectangle: 113.67 / 82.59
Circle: 114.10 / 86.43
Bezier: 99.94 / 80.71
Text: 87.1 / 67.18
OpenGL graphics test: 79.22 / 102.38
Spinning Squares: 79.22 / 102.38
UI test: 80.92 / 58.41
Elements: 80.92 / 58.41

I wonder why the UI test and Elements results are the same...pure coincidence, probably.

Worries me that the ominously named System and UI test results are so much better with the 15" machine. Meaning that I'm trying to talk myself into a 12" of course

Is the UI test a combination of the several tests above it in the list? From Fill down to Spinning Squares I mean...

Thanks a lot to those posting these benchmarks for comparison. Does anyone have a 1GHz machine that they can bench using the same version of XBench?

[Hozie]

Worries me that the ominously named System and UI test results are so much better with the 15" machine.

Yeah I find that strange too. Does anyone know how the UI and system scores are compiled? It seems strange that the 12", which, according to xbench, outperforms the 15" in every aspect except 2D would suffer from such a low UI and system score. I mean, the drives are about as fast and sure, quartz rendering seems slower in the 12" from initial testing, but that shouldn't result in a 15-20 points difference.

Any thoughts?

[Flowbee]

Originally posted by Hozie:
Yeah I find that strange too. Does anyone know how the UI and system scores are compiled? It seems strange that the 12", which, according to xbench, outperforms the 15" in every aspect except 2D would suffer from such a low UI and system score. I mean, the drives are about as fast and sure, quartz rendering seems slower in the 12" from initial testing, but that shouldn't result in a 15-20 points difference.

Any thoughts?

Twice the ram in the Ti? (Just a guess)

[Hozie]

Twice the ram in the Ti? (Just a guess)

Ok, I feel stupid now... Kinda forgot that since I posted a couple of posts above. Gotta get some sleep....

[BigCanoe]

Could someone run the test on a recent model 800mhz iBook too please? I would love to see the results, as these are the two models I am considering.

[iLashes]

Here it is (alas safari in the background and some other stuff running):
Results 52.38
System Info
Xbench Version 1.0b6
System Version 10.2.3
Physical RAM 640 MB
Processor PowerPC 750 [800 MHz]
L1 Cache 32K (instruction), 32K (data)
L2 Cache 512K
Bus Frequency 100 MHz
CPU Test 83.22
GCD Recursion 67.70 2.64 Mops/sec
Floating Point Basic 75.61 254.86 Mflop/sec
Floating Point Library 124.17 5.58 Mops/sec
Thread Test 62.85
Computation 50.29 405.17 Kops/sec, 4 threads
Lock Contention 83.75 1.05 Mlocks/sec, 4 threads
Memory Test 29.31
System 25.50
Allocate 144.96 76.25 Kalloc/sec
Fill 24.78 142.10 MB/sec
Copy 14.21 71.03 MB/sec
Stream 34.45
Copy 34.60 151.30 MB/sec
Scale 34.60 150.97 MB/sec
Add 33.68 153.23 MB/sec
Triad 34.96 153.63 MB/sec
Quartz Graphics Test 83.57
Line 80.98 2.06 Klines/sec [50% alpha]
Rectangle 92.45 6.50 Krects/sec [50% alpha]
Circle 93.52 2.16 Kcircles/sec [50% alpha]
Bezier 86.25 937.23 beziers/sec [50% alpha]
Text 69.55 1.13 Kchars/sec
OpenGL Graphics Test 55.96
Spinning Squares 55.96 39.16 frames/sec
User Interface Test 61.60
Elements 61.60 20.96 refresh/sec
Disk Test 39.20
Sequential 41.94
Uncached Write 35.25 15.38 MB/sec [4K blocks]
Uncached Write 37.64 15.31 MB/sec [256K blocks]
Uncached Read 79.61 12.54 MB/sec [4K blocks]
Uncached Read 35.87 15.47 MB/sec [256K blocks]
Random 36.79
Uncached Write 23.07 0.35 MB/sec [4K blocks]
Uncached Write 38.36 8.80 MB/sec [256K blocks]
Uncached Read 54.08 0.35 MB/sec [4K blocks]
Uncached Read 48.01 9.41 MB/sec [256K blocks]


Sorry for the mess....

[MaxPower2k3]

well the new PB scored considerably better than the iBook, except in 2d graphics. i assume this is due to the Radeon in the iBook, but it's still interesting.

[Altair]

Here are some noticable differences from the ibook 800 and the new powerbook
ibook/powerbook

Results 52.38/72.15
CPU Test 83.22/97.62
GCD Recursion 67.70/91.12
floating point basic 75.61(254.86mflop/s)/100.11(337.44Mflop/s)
" " library 124.17/101.40
Thread 62.85/56.81
memory test 29.31/83.79 *50.17/83.79
allocate 142.10/104.94
fill 24.78/87.19
copy 14.21/55.39
stream 34.45/95.28
more memory tests with similar numbers
Quartz 83.57/77.07
line 80.98/71.82
more like this
bezier 86.25/80.71
OpenGL test 55.96/102.38
UI 61.60/58.41
Disk test 39.20/57.56

*This ibook benchmark was taken from an old thread since i think something was wrong with the 29.31 score:
http://forums.macnn.com/showthread.p...ook+800+xbench

Overall it seems that the powerbook will give a slight speed boost in just about everything. It loses in thread tests and quartz tests. However it is VERY strong in openGL tests, cpu tests, and memory tests. I would also like to see some benchmarks comparing altivec enhanced programs (such as photoshop). Maybe a gaussian blur of a predetermined image or something more strenuous.
p.s. grrr it doesnt seem to like my tabs for some reason.

[icruise]

Is anyone else thinking that maybe these benchmarks aren't worth much? I mean, I thought the point of benchmarking was to get some concrete numbers to go by when comparing system performance (so we wouldn't just have to say, "hmm feels snappy! Maybe not quite as snappy as the other one, but definitely snappier than this one." ) It may be that the tests are not being performed in the way they were intended to be, or that we are interpreting them wrong, but it seems like these numbers are varying so much that making any kind of determination of speed is impossible.

[kcmac]

Looking at the ibook/PB comparison and realizing that the iBook had 640 ram and the PB has 256, I would say I am very happy that I went with the PB!

But then again, the tests could be wrong.

It is so exciting to start hearing real life experiences with the new 'Books. Keep 'em coming!

[Superchicken]

anyone wanna run an iTunes encode at the apple store on an iBook 800 and their nice new Alu book with the same track?

hehehe

That or I could alwyas give you a PNG file to open in Fireworks MX and that'd help me seal my decition on the 12 incher

[iLashes]

Since I posted the iBook 800 benchmark,
I have to say something about these tests:

Before I bought it, I compared a friends iBook 800 to my Dual 450 Cube with radeon, and in the everyday world tasks the iBook was a little faster (finder, copying ...OS X). I even compared altivec enhanced filters in photoshop and here the dual G4 450 was only a tiny bit faster.

With Logic Audio under os 9 the iBook was faster then the cube although logic is g4 optimized.

The bottom line is I think that if you need a g4 the little powerbook is fast. but for everything else i'd get the iBook.
It especially sucks that the 12" Albook does not have DVI, that would have made me get one instead of the iBook. (btw monitor spanning works like a charm on my iBook)

[iLashes]

to the superchicken:

If you are working with Macromedia MX products like flash or fireworks, get an iBook instead of the 12" pb, because The MX line does perform equally bad on both of them (no G4 optimization what so ever....)

-> Open Fireworks, then open Terminal, and enter 'top'. You will see, that fireworks eats up 12 to 18 % cpu just doing nothing !!!!!

[Altair]

I would like to find out how much faster the Powerbook is compared to the ibook in g4 enhanced programs. I'm going to be using a lot of final cut pro and itunes encoding, with a little bit of photoshop editing. I know the g4's are faster in these but I would like to know how much faster to see if its worth the extra money.

[NathanA]

Originally posted by Superchic[k]en:
anyone wanna run an iTunes encode

I'd just like to note that this is an absolutely TERRIBLE metric by which to build a benchmark, for 3 reasons:

1) The benchmark can be limited by the drive speed of the optical drive.

2) Any machine without a G4 CPU is at a TREMENDOUS disadvantage.

3) The actual CD media you're using can greatly affect the results [NOTE THIS ONE]

What I'd like to see is some PS7 Bench scores between the Ti and the new Al PB's. Perhaps some Q3 scores as well, and on that note I find it comical that Apple doesn't post Q3A scores on their site for any model other than the 15" which leads me to believe that it's still the performance leader

-Nathan

-Nathan

[icruise]

Originally posted by NathanA:
I'd just like to note that this is an absolutely TERRIBLE metric by which to build a benchmark, for 3 reasons:

I think people who are serious about benchmarking things with iTunes use an AIFF file and convert that to MP3, since it won't suffer nearly as much from the factors you mention.

And as for machines without G4s being at a disadvantage -- well, yeah, that goes without saying. But I think it is a very valid test, particularly if someone is trying to decide between an ibook and a powerbook. If the powerbook is getting more than double the speeds (as I imagine it will) that might make a difference to some people.

[Dan-the-Man]

To NathanA:

I'd just like to note that this is an absolutely TERRIBLE metric by which to build a benchmark, for 3 reasons:

1) The benchmark can be limited by the drive speed of the optical drive.

2) Any machine without a G4 CPU is at a TREMENDOUS disadvantage.

3) The actual CD media you're using can greatly affect the results [NOTE THIS ONE]

What I'd like to see is some PS7 Bench scores between the Ti and the new Al PB's. Perhaps some Q3 scores as well, and on that note I find it comical that Apple doesn't post Q3A scores on their site for any model other than the 15" which leads me to believe that it's still the performance leader

-Nathan

1) You are correct in assuming that the benchmark of a game can be limited by the optical drive, for instance, if you have a game (or any other program for that matter), and it is constantly accessing the optical drive to retrieve data, a slower optical drive will result in increased "pausing" in the games otherwise smooth animation.

2) I agree with this sentiment entirely, a G4 equipped machine will beat out virtually all other G3 based machines especially when working with OSX which is Altivec aware. As an addendum to this, an L3 cache is EXTREMELY important to realise the full potential of the G4 processor, without it, G3 like performance will result on a number of tasks.

3) I find it interesting that Apple have not posted updates to their powerbook graphics site to indicate benchmarks for the new PowerBooks. However, this does not mean that the 15" TiBooks will outperform the new 17" AlBook. The GeForce4 440 Go w/64Mb DDR competes directly with the Radeon 9000 Mobility w/64Mb DDR. On a number of PC benchmarking sites, they are fairly equal with the nVidia chipset giving stronger performance in some games, while the ATI gives better scores in others. You have to be extremely cautious, however, about translating these back to performance on the Mac. For instance, the nvidia chip does not give full hardware support for Direct X 8 which is the main graphics API for Windows, hence in games that are Direct X 8 aware the nVidia chip doesn't perform as strongly as the ATI chip wehich is fully Direct X 8 compiant through hardware. On the Mac this is a moot point because Direct X isn't used - Open GL is! The only true test will be to compare the machines directly in real world tasks, when they are available.

The GeForce 4 420 used in the 12" PowerBook has only HALF of the memory bandwidth of the 440 in the 17", and it also lacks the 1Mb L3 cache and the 1GHz processor, and it has a slower system bus coupled with slower RAM - all of these will, I would expect, combine to produce a significantly slower gaming system than the 17", which should be overall faster than the 15". By what percentage? We'll have to wait and see.