Is Geekbench inaccurate for PPC?

[ptdebate]

I've heard this rumored time after time--but is it true?

Here's the reason I ask--my 2.0GHz G4 has a geekbench of about 1000 whereas the G5 I'm currently using pulls a little over twice that. You'd think that two systems scale pretty neatly due to the quotient of number of CPUS/clock speed being a little over twice that of the G4, so that makes sense.

But good god does the G5 perform faster. It feels like a brand new Mac! How is this possible? Even web browsing is dramatically faster than I expected. 10 tabs and 20 programs running at once? No problem. It may be loud, but it is ferociously fast and hasn't shown so much as a spinning beach ball since I've owned it.

How can this only be roughly two times as powerful as my other PowerMac?

 

[redheeler]

Only 2x as powerful? I'd say that's pretty good for the CPU.

As for real-world performance, the G5 has some other things going for it. More and faster RAM, and a faster SATA HDD. The RAM is probably the reason the G5 can handle so many tabs without slowing down, not the CPU.

 

[Beavix]

The benchmarks are pure mathematical tests counting the raw processing power of the CPU and the memory. The way you feel when using a computer is a different thing and depends on other factors.

 

[SuperKerem]

I've heard this rumored time after time--but is it true?

Here's the reason I ask--my 2.0GHz G4 has a geekbench of about 1000 whereas the G5 I'm currently using pulls a little over twice that. You'd think that two systems scale pretty neatly due to the quotient of number of CPUS/clock speed being a little over twice that of the G4, so that makes sense.

But good god does the G5 perform faster. It feels like a brand new Mac! How is this possible? Even web browsing is dramatically faster than I expected. 10 tabs and 20 programs running at once? No problem. It may be loud, but it is ferociously fast and hasn't shown so much as a spinning beach ball since I've owned it.

How can this only be roughly two times as powerful as my other PowerMac?

That score seems about right to me. A score of 1000 is a Power Mac G5 Single 1.8Ghz. (That is what geekbench scores are actually based on!) Sure, it may feel fast after a fresh install, but I highly doubt it can handle '10 tabs and 20 programs' with 'no problem' Even my quad G5 would struggle with that!
How much faster were you expecting it to be? Your G4 may seem slower due to the hard drive?

 

[eyoungren]

I would have to say, use it as a general guide, not anything to measure serious performance by.

My first GB score was 1255 on my Quicksilver when I was running a dual at 1.7Ghz. You say you got 1000 and that was at 2.0Ghz. So, the numbers there can't even be reconciled. The DA is similar in just about all respects to the QS.

As to the G5, well. It's a G5, not a G4. Irrespective of any speed increase you got a more efficient (except for heat) chip that processes information better and faster.

My own G5 (here at work) is only a single 1.8Ghz processor but it handles multiple apps easily. Much more effectively than my QS, although there is a difference of 2.5GB of ram I might add.

 

[Beavix]

My Geekbench scores:

PM G4 "MDD" 2x1.25 GHz: 1101
PM G5 2x1.8 GHz: 1677

 

[redheeler]

Sure, it may feel fast after a fresh install, but I highly doubt it can handle '10 tabs and 20 programs' with 'no problem' Even my quad G5 would struggle with that!

In another thread he showed his G5 having 8 GB RAM, which I think would be enough to accommodate "10 tabs and 20 programs", or at least much more of it than any G4.

 

[SuperKerem]

In another thread he showed his G5 having 8 GB RAM, which I think would be enough to accommodate "10 tabs and 20 programs", or at least much more of it than any G4.

Well, it depends on the tabs and programs. I'm sure a G5 COULD run it but not without slowing down noticeably.

 

[MysticCow]

In another thread he showed his G5 having 8 GB RAM, which I think would be enough to accommodate "10 tabs and 20 programs", or at least much more of it than any G4.

Please, I regularly Cmd-Shift-A, Cmd-A, Cmd-O all the time while in Finder just for jollies. Everyone should do this at least once.

Or you can run Damned, the Applescript from Hell for a similar result. (NOTE: Damned requires access for assistive devices activated in System Preferences)

http://www.insanelymac.com/forum/index.php?app=core&module=attach&section=attach&attach_id=3012

 

[Ih8reno]

Power Mac G4 MDD Dual 1.42 - 1066

 

[Surrat]

In another thread he showed his G5 having 8 GB RAM, which I think would be enough to accommodate "10 tabs and 20 programs", or at least much more of it than any G4.

Actually the 8gig doesnt help much there since G5 apps are 32bit, and no 32bit app can claim more than 2gig of ram. You could run many tabs, but you would have to have more than one seperate instance of the browser running.

The only 64bit OSX app I know for PPC is the Chess thats included with Leopard.. lol

When it comes to ram use, I like to think of my 16gig G5 quad as having eight 2gig 'compartments' for running apps.

 

[lowendlinux]

This is the thing I don't like about benchmarks my z600 Hackintosh scores about 28,000 I don't think my machine is 14 to 28 times faster than yours in actual use.

 

[Dronecatcher]

Geekbench is only ever a vague indication - My G5 2.3 DP score:

3rd Dec 2014....2132
1st April 2015...1888

Same spec, memory, nothing else running....maybe Geekbench knew it was April Fools?

 

[Orizence]

I have never been able to use Geekbench to benchmark any of my computers as it seems to enjoy putting out extremely different scores every-time i try to benchmark using it... Im sure all that has been mentioned already, but in my opinion a better way to benchmark different computers is to do something like encoding a pretty large sized audio file, or find the average fps of your favorite game as these are way more accurate benchmarks in real world performance

 

[bunnspecial]

I have never been able to use Geekbench to benchmark any of my computers as it seems to enjoy putting out extremely different scores every-time i try to benchmark using it... Im sure all that has been mentioned already, but in my opinion a better way to benchmark different computers is to do something like encoding a pretty large sized audio file, or find the average fps of your favorite game as these are way more accurate benchmarks in real world performance

I've been using Mathematica recently to benchmark some OWR 604e(and G3) based systems.

Basically, I just come up with a crazy integral and see how long it takes each computer to calculate it.

Another good one-if you can find one or set one up yourself-is to run a set of Photoshop actions. Apple used to do this in Keynote addresses when demonstrating their current PPC product against competing x86 products.

 

[Orizence]

I've been using Mathematica recently to benchmark some OWR 604e(and G3) based systems.

Basically, I just come up with a crazy integral and see how long it takes each computer to calculate it.

Another good one-if you can find one or set one up yourself-is to run a set of Photoshop actions. Apple used to do this in Keynote addresses when demonstrating their current PPC product against competing x86 products.

The benchmarks I usually do when I get a new computer is recording the time to encode an 10 minute long audio file to a different format (I encode different audio files alot), or sometimes I convert a very long HD movie to x264 and record the time as well. I do these because I work with alot of audio and video for different things like school projects, or converting files for my phone, etc* these just simply allow me to measure performance that is important to me, and compare my different computers.

But yes, the Photoshop test is very similar and is also very good at real world performance compared to "synthetic" benchmarks like Geekbench

 

[Dronecatcher]

The benchmarks I usually do when I get a new computer is recording the time to encode an 10 minute long audio file to a different format (I encode different audio files alot), or sometimes I convert a very long HD movie to x264 and record the time as well. I do these because I work with alot of audio and video for different things like school projects, or converting files for my phone, etc* these just simply allow me to measure performance that is important to me, and compare my different computers.

I'd go along with that, I used to do the same - I don't know whether any of the processing is directed through the graphics card as well though - which would be good as you're using it all the time.

 

[Orizence]

I'd go along with that, I used to do the same - I don't know whether any of the processing is directed through the graphics card as well though - which would be good as you're using it all the time.

No unfortunately, there is no GPU acceleration in any of the apps I use, none that I know of anyway. If I didn't play any games, on my Powermac I would ditch the x1900 and keep using the 6600le, this is just because i am picky about the coil whine I have had on both my x1900's that I tried in my g5. Im considering either buying another x1900 but an OEM card, or a 7800gt as the coil whine is specific to both the cards I own.