Any technical reason why a 2.53GHz MacBook Pro would be faster than a Mac Pro?

[Spanky Deluxe]

I know this is a random question and sounds rather strange but here's the situation I've just discovered:

I'm running a simulation on my Mac Pro. Each output timestep takes about 15 seconds to complete. I've just tried the exact same run on my MacBook Pro and each output timestep takes about 10 seconds.

My Mac Pro is a 2.66GHz Quad Core 2006 Mac Pro with 4GB RAM currently running at 2.9GHz. My MacBook Pro is a 2.53GHz Late 2008 (Unibody) MacBook Pro with 4GB RAM.

It seems a little odd to me. This is a single threaded task so the fact that one's quad core and one's dual core shouldn't make any difference. As far as I can tell, both the chips are Core 2 Duo based. My only thoughts are that maybe the FBDIMMs vs DDR3 are holding up the Mac Pro, still seems strange though. Its not very memory hungry at all - just over 4MB real Memory, 2.6GB Virtual Memory according to the Activity Monitor.

 

[tobyg]

I think you're on the right track. Not only is the 2.53ghz Core 2 Duo chip in the Macbook Pro newer, the memory is standard DDR3 and it's not FB-DIMMS. FB-DIMMS are not the quickest. The Mac Pro you're using is 2006 techonolgy. The Macbook Pro is 2008/2009 technology. 2 years is huge!

The benefit is you can run 4 of these threads on the Mac Pro and finish all 4 in 15 seconds. The Macbook Pro would take 20 seconds to finish 4.

Just curious, how are you timing the operation? When you overclock a machine, it usually messes with the time on the machine so the machine will no longer keep accurate time. So if I were you, I'd use an external clock or stop watch to time the process.

 

[mcavjame]

Other considerations:
- bus speeds of the two machines
- RAM speed of the two machines
- cache levels
- drive speed (RPM)

 

[kabunaru]

2 options:
1. Get a new Mac Pro.
or
2. Hope Snow Leopard will fix some of this issue.

 

[Umbongo]

The Mac Pro is using a 65nm core then Macbook Pro is using 45nm (Penryn).

"Pat Gelsinger cited performance increase indicators for Intel's upcoming 'Penryn' family to the rate of 15 percent for image related applications, 25 percent for 3-D rendering, in excess of 40% for gaming and again, in excess of 40% for video encoding using Intel SSE4 optimised video encoders."

So I guess it depends exactly on what your simulation is. More typical numbers were 5-10% on commonly used general applications.

 

[Consultant]

Computer from 2008 is faster than computer from 2006?

Woah, technology improved in 2 years? You are kidding me... =p

 

[jjahshik32]

No way that the 2.53GHz mbp would run faster than even the 1st gen 2.66GHz 4 core Xeon model that you own.

Either its in your mind or the testings gone haywire.

 

[Ploki]

just like 2.26 nehalem is faster on a single core than the 2.8, so is 2.53 faster on a single core than the 2.66

 

[NoNameBrand]

His 2.66 is really a 2.9, though. For the MacBook Pro to run it in the same time, it would be 14% faster, clock-for-clock. To do it 50% faster - that's 70% faster, clock-for-clock.

 

[Spanky Deluxe]

His 2.66 is really a 2.9, though. For the MacBook Pro to run it in the same time, it would be 14% faster, clock-for-clock. To do it 50% faster - that's 70% faster, clock-for-clock.

Yeah its a bit odd. I would expect clock for clock for single threaded tasks that the Mac Pro and the MacBook Pro would run at about the same speed. Yes I know there's two years difference in 'technology' but these are fundamentally the same processors, Core 2 Duo. Yes they're slightly different inside but not substantially - consider how the speed varies between the 2.0GHz Penryn Mac Mini and the 2.0GHz before.

It has to be the memory speed. Its basically an n-body simulation that I'm running so I guess its pretty memory based. I was just surprised at the difference. Makes me want to go to the Apple Store and run my code on one of the Nehalem 2.26GHz machines.

I'm timing it using the stopwatch facility on my iPhone btw, so the funky timings that you get when overclocking the Mac Pro don't factor in.