Linked by David W. Kuhn on Fri 10th Jun 2005 16:34 UTC
Was it Palol Rossetti that one said, "People in glass house shouldn't throw stones?" Push away the Intel this, the Pentium-M that, or perhaps the ability to use the Dual Core Pentium 4, Apple has a much bigger challenge ahead of them. For years, they have been throwing down the MHz myth and now? They are sleeping with the "enemy" according to PowerPC zealots. 
Permalink for comment
To read all comments associated with this story, please click here.
To read all comments associated with this story, please click here.
Heh.
To paraphrase the old adage "Lies, damned Lies, and Statistics" - just replace Statistics with Benchmarks.
I generally have never really trusted Apples published benchmark numbers. Generally I've relied on sites like Barefeats (barefeats.com) for more objective benchmarks. Not to mention various video review mags. Needless to say, none of those sources have ever indicated that the Dual G5s were anywhere near 98% or 84% faster for Photoshop or Premier tasks (respectively).
Mostly this comes from Apple "cherry picking" specific tasks which relied on specific operations that the G5 (and previously G4) are good at... Things like arithmetic shift operations. Which while useful for many things are not the only thing. Seriously if these dual G5 systems were so very much faster, you'd think there'd be a little more noise about it.
Needless to say the G5 was Apple's last attempt to ditch the MHz myth. Jobs got hung out to dry by IBM on the 3GHz in 12 months boast (not unreasonably by IBM - Apple's demand for chips represents around 2% of chips sold by IBM, simply not worth the engineering effort). Additionally there has been, and will be no forthcoming G5 or comparably preforming PPC from Freescale for a laptop Mac. Anybody ever look inside one of those dual G5 cases? Wow that sure is a lot of cooling equipment for a chip that supposedly runs so very much cooler than a P4.
Apple is mostly looking at the next generation of processors from Intel - the Pentium M based cores. These run at low speeds with high performance and low heat. In most benchmarks a 1.6GHz Pentium M performs equivalently to a 3.0GHz Pentium 4. The G5 no matter how skewed the benchmarks couldn't honestly make that claim. The race for ever higher clock speeds is mostly over - Intels competition from AMD has left that in no doubt. The Opteron has a slightly better IPC (instructions per clock) than the G5, and when working on non-vectorized code, it's ideal. The move to dual core CPUs and better IPC is underway - along with better vectorization capabilities (witness the Cell and Xenon processors).
Oh yeah, and for the people who are hollering about, "why didn't they use the Cell or Xenon processors?" Keep in mind that these processors have been highly specialized for codec (compression decompression) tasks, and gaming oriented processing. They are somewhat anemic when it comes to general purpose computing involving branchy non-optimized code. Both feature stripped down PPC cores which are only capable of "in order" execution, with minimal or non-existant branch prediction capabilities. This will make them poor at AI and control code. Is that really the CPU you want for day-to-day tasks like browsing, emailing, etc.?