Leopard Performance (October 2007)

Despite my complaints yesterday, I picked up a copy of Leopard at Carbon Computing today. After making sure both Geekbench and Wiinote work with Leopard (both do!), I thought it'd be interesting to see if performance has changed from Tiger to Leopard using my favorite benchmark. While I was at it, I also checked out 64 bit performance under Leopard.

Setup

• iMac (Late 2006)

  • Intel Core 2 Duo @ 2.0 GHz
  • 2.0 GB 667 MHz DDR2 RAM
  • Geekbench 2.0.7
  • Mac OS X 10.4.10 (Build 8R2232) or
  • Mac OS X 10.5.0 (Build 9A581)

• Power Mac G5

  • PowerPC G5 @ 1.6 GHz
  • 1.25 GB 333 MHz DDR RAM
  • Geekbench 2.0.7
  • Mac OS X 10.4.8 (Build 8L127) or
  • Mac OS X 10.5.0 (Build 9A581)

I’ve reported the average overall score for each model and processor combination, where 1000 is the score a Power Mac G5 @ 1.6 GHz (running Tiger) would receive. Higher scores are better. Keep in mind that Geekbench 2 only measures processor and memory performance; it won't catch any differences that might affect other subsystems (like, say, video drivers).

Update: Some people are wondering about the validity of the benchmarks on a single-processor Power Mac G5. While I would've loved to run the benchmarks on a dual- or quad-processor Power Mac G5, I've only got a single-processor Power Mac G5 at my disposal. I wish I could've included more machines (or more recent machines), but I was only able to use what I had available to me.

That said, I still think the results are valid (and interesting!) because there are a large number of single-processor G5s and G4s still in use; not everybody is running the latest and greatest hardware.

iMac Results

Overall Performance

Integer Performance

Floating Point Performance

Memory Performance

Stream Performance

Power Mac G5 Results

Overall Performance

Integer Performance

Floating Point Performance

Memory Performance

Stream Performance

Conclusions

What's surprising is that performance decreased slightly from Tiger to Leopard. I'm not sure why this is the case, since pre-release Leopard benchmarks showed Leopard performing better than Tiger (at least in some areas). What's not surprising is the 64-bit performance results; x86 64-bit is faster than x86 32-bit (thanks to 64-bit having extra registers), while PowerPC 64-bit is slower than PowerPC 32-bit (thanks to 64-bit instructions being twice the size of 32-bit instructions).

Still, it's entirely possible your system will be faster under Leopard than Tiger since (as I mentioned earlier) since Geekbench only measures processor and memory performance. Changes to other subsystems in Leopard (like, again, video drivers) might markedly improve performance for you. Me, I'm glad I upgraded to Leopard. While my Power Mac doesn't feel quite as snappy under Leopard as it did under Tiger, there are all sorts of interesting features in Leopard that just aren't available in Tiger.