So what's all the hype about, right? Is the G3 really up to twice as fast? Steve Jobs is claiming that the slowest G3 "toasts" the fastest Pentium II. With no disrespect to Apple, it's a fast chip, but the slowest G3 toasting the fastest Pentium II? Test it out for yourself, if you can. Sit down at a G3 Mac, and sit down at a Pentium II based PC, now you can decide. I think you will find both to up to speed. Neither being significantly faster then the other. If you think I'm biased, and simply don't want you to buy a Mac, run some tests of your own. I think you will conclude the same results. This page is dedicated to the facts. I am in no way trying to flame Apple, or keep you from buying a Macintosh. But since Apple has blatantly lied (or highly exaggerated the truth) about the performance of their G3 processors, I thought it would be in the best interest of all Cubase users to understand how Apple came to this conclusion. i.e., "up to twice as fast, etc." I don't want to offend any Mac users, so please note that I think the PPC 750 (G3) is a very fast processor, and that the Macintosh is a great platform for many musicians. "The iMac for $1,299 toasts the fastest PC money can buy at any price-- the Pentium II 400."
-- Steve Jobs, August 13 "There's a lot more cheating this year."
-- PC Magazine technical analyst's response when asked what was different this year in the PC lines project, which tests almost every major computer on the market. Benchmarketing is worse than ever this year. In Steve Jobs G3 benchmark examples he uses the ByteMark benchmark to prove that a 233MHz G3 is 40% faster than a Pentium II 400MHz. Since then, PC Magazine and many others have tried to verify those numbers and concluded very different results. It turns out that the ByteMark benchmark is extremely compiler dependent. Apple used a compiler optimized for its G3 processor to create the executable for the G3, but apparently used an x86 compiler optimized for a almost 10 year old 486 to test the Pentium II. When PC Magazine compiled the same benchmark with a new Intel compiler optimized for the Pentium II architecture, their results were quite different. A Dell desktop (selling for just over $1,000 US dollars) running a 333MHz Celeron based processor (which did not contain any L2 cache) performed at about 50% faster than the a 233MHz G3. i.e., a chip even slower than a Pentium 400, in a machine priced almost the same as Apple's, "toasted" the G3. Even more interesting, when PC Magazine ran a suite of more than 30 application benchmarks, including Quake, PhotoShop, Claris Works, Excel, etc., the Celeron-based machine beat the 233MHz G3 in more than 90% of the tests. You can find their results at PC Magazine. So what's going on here? Well, in these and many other cases, vendors are not comparing apples to apples (pardon the pun). And with all the competition these days, buyers must very carefully inspect any benchmarks quoted by vendors because-- unless you verify the source-- you may well be misled. Here are some tips on cutting through the benchmarketing hype. Only trust disinterested third parties:
Tests from PC Week, PC Magazine, and other impartial labs are the best way to verify vendor claims. Do your own testing: Most benchmarks are available for free. Get the products in and test them yourselves, under your own conditions. Nothing beats testing with your own applications. Real life testing is all that really matters anyway. Make your own decisions. And don't let Apple, ByteMark, SpecBench, Me, or a friend make them for you. So how does this all effect Cubase VST performance? Well, it's hard to say because their are so many factors. I prefer to focus on Floating Point performance since VST's FX engine is almost solely floating point based, and this is where everyone seems to be noticing performance loss in VST. But before you use this as your comparison benchmark, you must calculate the difference between DirectX and VST based plugin's. VST plugin's are not as demanding on the processor as DirectX plugin's are. So if your comparing the Loudness Maximizer for Windows (DirectX plugin), to the Loudness Maximizer for the Mac (VST plugin), the Floating Point test soon becomes less relevant. Never the less... it's a good way to determine which processor is more capable of running VST based plugin's. After calculating the different between the PII and G3's, it's quite obvious that the PII rivals the G3 in floating pointer performance. Now adding the difference between DirectX and VST based plugin's, the difference has leveled out a bit. i.e., both processors perform about equally... with the exception of VST based plugin's which should be a hair faster on the PII. 
| 
