[KDrifter340]

What Intel chip can the Mac G3 300 mhz be compared to? For speed wise?

[aliasfox]

All else being equal, a PII 300-400, depending on the system the chip is in and the amount/speed of the L2 cache. Are you looking at a Blue and White G3, Beige, or upgrade card?

[DaBeav]

Well, considering that the G3 is RISC, I would say it could hold its own over a 450-500MHz PIII. Wasn't until the SSE2 on the P4 (similar to AltiVec on a G4) and much much faster clock speeds that Intel started to pull ahead. That and not only did the G4 sit at low clock speeds, the architecture didn't exactly scale very well as the chips got faster...

[aliasfox]

It depends. A G3/300 with 512k half speed cache sitting in a 66 MHz bus wouldn't be too much faster than a PII running at the same clock on a 100 MHz bus. On the other hand, a G3/300 with 1 meg of full speed L2 sitting in a 100 MHz bus should be able to hold its own on most operations against a PII 400, possibly a little bit higher. However, there are some applications, such as games, where megahertz matters more than design. In addition, Pentium IIIs had SSE support, which enhanced performance in applications such as Photoshop. Therefore, there's very little chance that a G3 at 300 can keep up with a PIII at 450- remember, that's a 50% MHz gap.

[Bob_Blaylock]

Some time ago, I performed a simple benchmark, comparing my 300 MHz Beige G3 with my wife's 550 Mhz Pentium III. At this particular benchmark, my G3 was almost exactly twice as fast as the Pentium III. Click here to see my posting about this benchmark.

[ViolentGreen]

It's subjective. Different ISAs do different tasks for efficiently. Benchmarks don't really mean a whole lot across ISAs. I assume you are awaer of that saying, "Lies, Damned lies, and benchmarks."
I am curious about your code you put though though. Especially since you did not use the same compilers on each machine. If I can remember, I will do a similar test using gcc compilers on my 1GH PB with 768 RAM and my 1.7 GH Athlon2100 desktop with 512 RAM. I do know that the compiler is optomized for my Athlon. I assume the PPC version is optomized as well. I also know that pentiums are much better at number crunching then their AMD counterparts but I don't have a pentium.
I know nothing about the PPC architecture. I did a little studying on the P4 about two years ago (note this is before the hyperthreading and bus speed increases.) At this time the speed of the P4 was actually a hinderence. It ran entirely too fast to keep the pipelines full; the resulting pipeline bubbles hurt the performance. In this case the high speed was not a help.

[aliasfox]

How big was the program? Your G3 has twice the cache of the PIII, does it not? If the program was say... 550 K compiled, then it would fit into the G3's L2 (1 meg) but not the PIIIs (512k), giving the G3 a significant advantage.
Of course, PPCs have always been renowned for robust FPU performance (look at ByteMarks from a few years ago).

[blackwind]

Specifically, the PowerPC 601, 601+, 604, 604e, and 604ev were renowned for their robust floating-point performance. The original version of the G3 (PowerPC 750) significantly boosted the PowerPC's integer performance by having twin superscalar integer units. Unfortunately, the G3 also had a weaker floating-point unit than the 604 varieties since the G3 was based on the inexpensive 603.
The G4 restored the 604's floating-point implementation, but by then, the Athlon and Pentium III were around, and they were not push-overs (especially the Athlon). As such, for the last little while, Macs were not renowned for their floating-point performance.
Of course, now with the G5, things are pretty even with top-end PC's, albeit the G5 still has inferior integer performance, as confirmed in Macworld's lab tests. (Thankfully, the G5 has Altivec to balance things out, and floating-point is pretty consistent from processor to processor with a few applications being particularly fast on a particular processor.)
As for BYTEMarks, the G3 performed particularly well only on one particular subtest. On the other tests, the G3 was still faster than the Pentium II, but it was by a far smaller margin.
Back to the original question, I would put a 300-MHz iMac at about the speed of a 400-MHz Pentium II (rough ballpark figure).

[ViolentGreen]

550KB seems WAY high to me. I have not done much programming on the mac so I can't say much about the compile size. On the PC I would expect that program to be closer to 10-15 KB, if not less.

[Nobody]

Well let me insert my 2 cents here, I have OS X Panther running on 233 G3 rev b imac, you try running win xp on a 500 anything and let me know how it goes. A 300 G3 is at least as fast as 600 or 650 Pentium for most tasks and if you have OS 9 installed on that 300 G3 then you are going to need a pretty fast PC for windows 98, NT, or 2000 to feel as fast as OS 9. OS 9 flies on those older machines.

[ViolentGreen]

I don't see what the point is here. You are talking about two different OSs running on two different ISAs. How can you make a comparison?
And for the record, my mom has XP running on a 450MH PIII(or 2 maybe) with 384 RAM. It runs fine.

[Nobody]

My point was try running windows anything on a 233 and tell me if it is even useable. Running a brand new OS on a 6 year old machine is impressive to say the least, The G3 is damn fine chip. XP would choke to death on a 233 if you could even install it.

[ViolentGreen]

Well I ran the code and this is what I got:
MAC (1GHz PB, 768 Ram)
gcc version: 3.3
Shortest Time: 19.48 s
File Size: 20 kb
PC : XP2100 (1.7 GHz, 512 MB RAM, Running gentoo)
gcc version: 3.2.3
Shortest Time:
console: 14.04 s
gnome: 14.06
kde : 14.08
File Size: 10 kb
It suprises me a little as I expected my powerbook to preform a little better. Keep in mind this is a notebook compared with a desktop. I would hope a desktop would perform better. I am fairly confidant that a dual g4 would do much better.