# PowerMac G5 Tidbits and Benchmarks

I don't make any guarantees but will try to get as much answered as possible.

So, if you have any questions you want answered or benchmarks you'd like to see run leave a message here by Thursday morning at the latest and I'll get the requests to my source.

## Comments

5,768member1. They are running 10.2.7

2. CHUD identifies the G5 as GPUL

3. The Power Mac G5s can clock down the processor, like the PowerBooks, to save power (ie: reduce heat). The 2GHz PowerPC 970 clocks down to 1.3GHz.

4. The Power Macs are a lot bigger (taller) in person, than they look in pictures.

1,401memberOriginally posted by applenutSo, if you have any questions you want answered or benchmarks you'd like to see run leave a message here by Thursday morning at the latest and I'll get the requests to my source.

I've got a question. Whats going on with this Thinksecret comment?

"Another screenshot shows results from Skidmarks GT, a benchmarking utility that's part of Apple's CHUD performance tools. Based on the test we ran, this provides a rough comparison of G5 and G4 performance. The Skidmarks scale has "100" equal a Power Mac G4 at 1GHz. The Dual 2GHz received scores of 172 for integer performance, 270 for floating point performance, and 208 for vector performance."

Only 1.7X better for a dual G5, 1GHz bus, real DDR as compared to a single G4? Are these low numbers due to OS 10.2.7?

I thought EWeek mentioned that the new G5 computers won't see much performance improvement until 10.3 comes out. I hope this is the reason for these low numbers.

5,768member1,164memberOriginally posted by sc_marktOnly 1.7X better for a dual G5, 1GHz bus, real DDR as compared to a single G4?If you don't know what the benchmark measures, the results are meaningless.

5,768memberthat sucks.

oh wellz

963memberOriginally posted by applenuthmm, no questions.

1. Wellz, a clarification of the above mentioned benchmarks would be nice. Inquiring minds want to know why the 970 isn't stomping the G4 so convincingly in this benchmark. What does it really measure?

2. What's Steve's definition of "by the end of the year"? December 31, 2003? Not a G5 question, but can't hurt to ask.

8,254memberOriginally posted by sc_marktOnly 1.7X better for a dual G5, 1GHz bus, real DDR as compared to a single G4? Are these low numbers due to OS 10.2.7?

I thought EWeek mentioned that the new G5 computers won't see much performance improvement until 10.3 comes out. I hope this is the reason for these low numbers.

1) It only measures a single processor.

2) Memory bandwidth is pretty much irrelevant.

You should be impressed by those numbers, the FP ones in particular.

453member7,329member239memberhttp://www.thinksecret.com/cgi-bin/p...gif&p=wwdc03g5

It says that a score of 100 for Integer, Floating Point, and Vector are based on a single 1Ghz G4.

It shows the Dual 2Ghz G5 getting an Integer of 172, FP of 270, Vector of 208.

Correct me if I'm wrong but don't these numbers seem low, considering we are comparing a Single 1Ghz G4 vs. a Dual 2Ghz G5?

Later...

2,674memberOriginally posted by BlizaineCan someone describe the "SkidMark" benchmarks done at ThinkSecret?

http://www.thinksecret.com/cgi-bin/p...gif&p=wwdc03g5

It says that a score of 100 for Integer, Floating Point, and Vector are based on a single 1Ghz G4.

It shows the Dual 2Ghz G5 getting an Integer of 172, FP of 270, Vector of 208.

Correct me if I'm wrong but don't these numbers seem low, considering we are comparing a Single 1Ghz G4 vs. a Dual 2Ghz G5?

Later...

Dude, scroll up like two posts and actually READ the thread!

1,379memberyozaaaa~!

3,226memberOriginally posted by keyboardf12check out page 2 of macrumors for a fractal benchmark.

yozaaaa~!

It's a nice big number, but not particularly impressive.

Let's scale the G4/450 to a comparable GHZ. That would mean multiplying the G4 result by 8.8, and if linearly scaled the G4 would produce a 13.2 GigaFlop result, compared to the 14.0 GigaFlop G5 result.

Of course, a linear scale isn't the way to figure this out, but comparing the numbers against a G4/450 to show big gains isn't, either. Anyone have a Fractal score for a dual G4/1.42 system?

11memberhttp://www.maxon.net/pages/download/cinebench.html

It is a good cross-platform benchmark of 3d render and OpenGL performance.

Blark

51memberOriginally posted by GardenOfEarthlyDelightsWhat's Steve's definition of "by the end of the year"? December 31, 2003? Not a G5 question, but can't hurt to ask.I would guess and shoot for an announcement on Panther availablility at the Apple Expo in Paris in September.

1,379member1,804memberOriginally posted by audiopollutionIt's a nice big number, but not particularly impressive.

Let's scale the G4/450 to a comparable GHZ. That would mean multiplying the G4 result by 8.8, and if linearly scaled the G4 would produce a 13.2 GigaFlop result, compared to the 14.0 GigaFlop G5 result.

Of course, a linear scale isn't the way to figure this out, but comparing the numbers against a G4/450 to show big gains isn't, either. Anyone have a Fractal score for a dual G4/1.42 system?

It's likely largely altivec dependant and not overly reliant on memory bandwidth so seeing it scale close to linearly wouldn't be a surprise.

8,123memberThere is no chance that Apple show benchmarks saying that a dual 2 ghz is not much more efficient than a single G4. Worth if you divide the result by 4 to obtain almost a single 1 ghz G5, or even by a factor 3, the G5 will be slower.

It' s in total contradiction with the Spec benchmarking comparing a G4 to a G5.

3,226memberOriginally posted by PowerdocI don't believe this benchmarks publied by think secret. They said that this info came from Apple employees (and i doubt that Apple give them a chance to try a G5).

There is no chance that Apple show benchmarks saying that a dual 2 ghz is not much more efficient than a single G4. Worth if you divide the result by 4 to obtain almost a single 1 ghz G5, or even by a factor 3, the G5 will be slower.

It' s in total contradiction with the Spec benchmarking comparing a G4 to a G5.

So this:

... is fake?

You can't divide by 4, though, as supposedly Skidmarks GT only uses 1 processor for the comparison, which is why I weighted my results as I did.

Read my bigger explanation here:

http://forums.appleinsider.com/showt...0&pagenumber=3

8,254memberOriginally posted by audiopollutionSo this:

... is fake?

No, it isn't. It's about right. My Dual 1 GHz G4 scores 100/100/100 exactly in Skidmarks GT. I thought I already explained the benchmarks...