Are you wondering if it's worth it to get a new, low end G4 when you already have a tricked out G3? Is the G4 really that much faster? What effect does the Velocity Engine (AltiVec) really have? It is now time to answer these important questions.

Test System #1:
• G4 466 MHz
• 256 MB RAM
• Mac OS 9.1
• OEM 30 GB 5400 RPM hard drive
• ATI RAGE 128 Pro AGP

Test System #2:
• G3 233 upgrade to 500 MHz
• 288 MB RAM
• Mac OS 9.1
• OEM 4 GB 5400 Hard Drive and Maxtor 27.2 7200 RPM DiamondMaxPlus (startup) hard drives
• VooDoo 3 2000 PCI

The first thing you come across when using a Mac is the startup time, so we'll look at that first. Tests in seconds: Shorter bars are better.

These tests produced very odd results. While the G3 was normal, in that starting up with extensions on was slower than off, the G4 somehow came out with the opposite results. This is very odd, because according to all conventional wisdom, not loading the extensions should be faster than loading them.

---

Next, Everyone seems to like benchmarks, so I ran the three standard suites in MacBench 5.0. Tests relative to G3 300, which gets a 100 score: Longer bars are better.

According to MacBench, the RAW CPU is slower in the G4 (which goes along with the MHz). The FPU is a little faster in the G4, which is along the lines of the normal results. But what was weird is that the G4's hard drive tested faster than the G3's. This is weird because the G3's hard drive rotates 33% faster than the G4's. I guess the faster IDE bus really does help quite a bit.

---

Then, I made a bitmap file in Photoshop, and open it with QuickTime's picture viewer. Tests are in seconds: Shorter bars are better.

This file was 103 MB in size, and because of the bitmap format, normally does that longer to open that other image formats. That means this test shows the power of the CPU, and the hard disk, combined into one. The G4's combination turned out to be about 40% faster than the G3's combination.

---

Finally, I used a program called AltiVec Fractal Carbon 1.0. Tests are in MegaFlops: Longer bars are better.

This program will harness the power of multiple processors, Mac OS X's carbon, and the G4's AltiVec, to calculate how fast the CPU really is. Because of the nature of this program, it is very accurate, and is a great way to benchmark CPU Speed.

That said, it turns out that the CPU with AltiVec in the G4, is about 5 times faster than the G3. This really shows what can happen when a program harnesses the power of the Velocity engine. For reference, the numbers on the graph (in the X/Y format) is how the application was set for that test. X=Color Speed setting, and Y= Maximum Count Setting.

---

I guess the numbers say it all. The G4 466 MHz, with slower components, is actually faster in many repsects than a G3 500 MHz, that has been tweaked for years. Iw ould have never thought this to be true, but now I know it to be true.