Primate Labs Blog

Apple released a lot of new hardware this month, and while I was able to post Geekbench results for the new iMac and Mac mini fairly quickly, it took a little while longer to get Geekbench results for the new Mac Pro.

While other folks, such as Rob-ART over at BareFeats, have done an excellent job compiling benchmark results for the latest Mac Pros, I still wanted to post Geekbench results for not just the latest Mac Pros, but all of the Mac Pros that Apple has produced.

Setup

I’ve gathered Geekbench scores for the all of the Mac Pros that Apple has released. Scores were collected from the Geekbench Result Browser for Mac Pros with standard processors (i.e., no processor upgrades, overclocked processors, or Hackintoshes).

I’ve reported the average overall score for each model and processor combination.

If you haven’t used Geekbench before here’s how it measures performance. Geekbench compares a computer’s performance against a “baseline system” (currently a Power Mac G5 @ 1.6GHz in Geekbench 2). A computer that has the same performance as the baseline system scores 1000 in Geekbench. A computer that’s twice as fast as the baseline scores 2000 while a computer that’s twice as slow scores 500.

Finally, Geekbench 2 only measures processor and memory performance which is why models with the same processors but different video cards have roughly the same score.

Mac Pro Benchmarks

Conclusions

What’s interesting is the performance of the 2008 and 2009 entry-level Mac Pro; both Macs have roughly the same performance, but the newer Mac Pro has a lower processor frequency and fewer cores. The new Nehalem architecture is clearly better than the older Core architecture. While it would be nice if the new entry-level Mac Pro was faster than the old entry-level Mac Pro, the new Mac Pro is cheaper, a change that makes more sense given the state of the economy.

The new mid- and high-end Mac Pros, on the other hand, are insanely fast (and insanely expensive). Currently it’s these models that are dominating most of the top spots in the Geekbench Result Browser. The only computer that can beat the Mac Pros’ score is a 32-core server from IBM.

Bare Feats published a couple of great articles that take a close look at Mac Pro memory performance.

The first article compares the performance of two, four, and eight matched FB-DIMMs in the Harpertown Mac Pro (i.e., the latest Mac Pros). Rob-ART found there’s a significant performance advantage to having all eight memory slots populated on the latest Mac Pro.

The second article compares the performance of 667MHz FB-DIMMs in both the Clovertown and Harpertown Mac Pros against the performance of 800MHz FB-DIMMs in the Harpertown Mac Pros. While it’s obvious that 800MHz is faster than 667MHz, what’s surprising is the significant difference between 667MHz in the Clovertown and 667MHz in the Harpertown. This performance difference really highlights the chipset improvements in the new Mac Pro.

Rob-ART used both Geekbench 2 and a custom-built STREAM benchmark (which I built for him) for these tests.

A couple of weeks ago I blogged about the performance of the new Mac Pros. Unfortunately I only had Geekbench 2 results for the eight-core 2.8GHz Mac Pro. I thought I’d follow up on that post with Geekbench 2 results for all three eight-core Mac Pros.

Setup

• Mac Pro (Early 2008)
  • Two quad-core Intel Xeons @ 2.8GHz, 3.0GHz, or 3.2GHz
  • 2.00 GB 800 MHz DDR2 FB-DIMM
  • Mac OS X 10.5.1 (Build 9B2117)

If you’re not familiar with Geekbench 2, benchmarks are scored against a baseline, where a score of 1000 represents the performance of a Power Mac G5 @ 1.6GHz. Higher is better.

Results

Overall Performance

Integer Performance

Floating Point Performance

Memory Performance

Stream Performance

Conclusions

What’s interesting about the above charts is that the performance difference between the 2.8GHz and 3.2GHz Mac Pro isn’t as great as the difference between running 32-bit code and 64-bit code. In fact, the 2.8GHz Mac Pro running 64-bit code is faster than the 3.2GHz Mac Pro running 32-bit code!

Really, though, if you’re using a new Mac Pro you’re got a ridiculous amount of processing performance at your disposal (even if you’re “stuck” with a 2.8GHz model running 32-bit code).

Earlier this week Apple released updated Mac Pros that use Intel’s new Penryn processors. Also new is the fact that the standard Mac Pro configuration now comes with eight (instead of four) cores.

Of course, what I’ve been wondering (as I sit here and think about getting a new Mac Pro) is how does the new standard eight-core Mac Pro perform compare to the old high-end Mac Pro? I’ve gathered Geekbench 2 results for both Mac Pros to find out.

Setup

• Mac Pro (Early 2008)

  • Intel Xeon E5462 @ 2.80GHz (Eight cores)
  • 2.00 GB 800 MHz DDR2 FB-DIMM
  • Mac OS X 10.5.1 (Build 9B18)

• Mac Pro (8-core)

  • Intel Xeon X5365 @ 3.00GHz (Eight cores)
  • 1.00 GB 667 MHz DDR2 FB-DIMM
  • Mac OS X 10.5.1 (Build 9B18)

It might seem unfair to compare a new Mac Pro with more RAM against an old Mac Pro with less RAM. However, both Mac Pros have two FB-DIMMs installed. Since Mac Pro performance increases dramatically when it’s configured with four FB-DIMMs sticks, I figure it’s more important to make sure the two Mac Pros have the same number of FB-DIMMs installed rather than the same amount of memory installed.

Also, Geekbench itself isn’t incredibly memory hungry (Geekbench prefers faster RAM instead of more RAM) so I doubt the performance of the old Mac Pro will suffer from having less memory installed.

As always, I’m reporting the baseline score, rather than the raw score, for each benchmark (where a score of 1000 is the score a Power Mac G5 1.6GHz would receive). Higher is better.

Results

Overall Performance

Integer Performance

Floating Point Performance

As you can see, floating point performance is off the charts!

Memory Performance

Stream Performance

Conclusions

I’d expected that the new Mac Pro would make a respectable showing, but would still be slightly slower than the old Mac Pro, since the new Mac Pro’s processors run at a lower clock frequency. So I’m surprised that the new Mac Pro is faster in almost every area (floating point performance being the exception) than the old Mac Pro. Of course, the new Mac Pro does have faster memory along with the new Penryn processors (which have a lot of architectural improvements, like more L2 cache) so maybe I shouldn’t be surprised after all.

Plus, when you consider the fact that the new Mac Pro is so much more affordable than the old Mac Pro and it’s hard to call this upgrade anything less than impressive.

I’m excited! Apple’s updated the Mac Pro (and the Xserve) to use Intel’s new quad-core Xeon processors. The updated Mac Pro default configuration now comes with two quad-core Xeons (for eight cores total) for $2799. That’s an insane amount of processing power for the price! Apple’s including more RAM (2GB 800MHz DDR2) and a better video card (ATI HD 2600XT), too.

With this update I think the Mac Pro will continue to dominate the top positions in the Geekbench Result Browser for some time to come.