So can anyone explain why the G5 system run circles around the old Pentium 4 computers? I had a Pentium 4 system that I was running as a backup computer when my first gen AMD Athlon 64 x2 based computer died, my back up was a Northwood core and was kinda slow, but I'm seeing evn a G4 being faster. What's up with this?
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Power Mac G5 vs Pentium 4?
- Thread starter gryffinwings
- Start date
- Sort by reaction score
The answer to this is in the RISC (reduced instruction set computing) vs. CISC (complex instruction set computing).
From how I understand it the RISC chip has fewer instructions to perform within a clock cycle and generally processed more information per cycle. Therefore if a RISC and CISC chip both have the same clock speed, say 1.5 GHz, the RISC chip will generally perform faster because it is achieving more per cycle. Even in CISC chips that had a much higher clock speed than the RISC chip could still not outperform it.
Obviously I'm being very vague in these descriptions but this is a general answer to your question.
From how I understand it the RISC chip has fewer instructions to perform within a clock cycle and generally processed more information per cycle. Therefore if a RISC and CISC chip both have the same clock speed, say 1.5 GHz, the RISC chip will generally perform faster because it is achieving more per cycle. Even in CISC chips that had a much higher clock speed than the RISC chip could still not outperform it.
Obviously I'm being very vague in these descriptions but this is a general answer to your question.
Its mostly due to the Pentium 4's NetBurst architecture which is a complete joke. They ran stupid hot and Intel got way too caught up in adding MHz/GHz and no improving how the chip handled data. That has a lot to due with why the Core architecture is based more on the Pentium III.
It's not so much RISC vs CISC. Practically, every modern processor is kind of a crossover of those two techniques. The Pentium 4 architecture scaled well if high clock speeds were achieved (which it did in the end), but power issues were too grave by then. I still own a hp xw6000 workstation with Dual 3,06GHz Xeons, and they are still okay. The G5s are certainly not running circles around them if you have a look at benchmarks from back then.
It's not just the differences in the CPU's themselves. Another crucial factor in PowerPC performance is that there are only 7 stages in the pipeline vs. 14+ in x86 hardware.
This means a P4 board had to go through 14 different stages to crunch a process vs. 7 for G4/G5.
Also, PowerPC is only one type of RISC chip. There are several different ones. Here is the list of different RISC CPU from wiki:
"Well known RISC families include DEC Alpha, AMD 29k, ARC, ARM, Atmel AVR, Blackfin, MIPS, PA-RISC, Power (including PowerPC), SuperH, and SPARC. In the 21st century, the use of ARM architecture processors in smart phones and tablet computers such as the iPad provided a wide user base for RISC-based systems. RISC processors are also used in supercomputers such as the K computer the fastest on the TOP500 list in 2011."
http://en.wikipedia.org/wiki/Reduced_instruction_set_computing
This means a P4 board had to go through 14 different stages to crunch a process vs. 7 for G4/G5.
Also, PowerPC is only one type of RISC chip. There are several different ones. Here is the list of different RISC CPU from wiki:
"Well known RISC families include DEC Alpha, AMD 29k, ARC, ARM, Atmel AVR, Blackfin, MIPS, PA-RISC, Power (including PowerPC), SuperH, and SPARC. In the 21st century, the use of ARM architecture processors in smart phones and tablet computers such as the iPad provided a wide user base for RISC-based systems. RISC processors are also used in supercomputers such as the K computer the fastest on the TOP500 list in 2011."
http://en.wikipedia.org/wiki/Reduced_instruction_set_computing
Actually, Willamette and Northwood had 20 stages, not just 14. But it gets better! The ultimate P4 chips with the Prescott and later cores had 31(!) stages in their instruction pipelines.
In fairness, the double-pumped ALUs in the P4 chips is a pretty clever thing, but at the same time, it's difficult to optimize code for non P4 i386 chips when taking full advantage of that and the rest of the integer performance enhancing technology. Clearly the use of PIII tech in the genesis of the Core line of Intel procs shows the deficiencies of the P4 architecture, but the sheer weirdness of it in order to maximize reported GHz fascinates me to no end.
Due to the timeframe and how Lackluster the Pentium 4 was ( shame really, the Pentium, Pentium Pro, Pentium 2, and 3 were GREAT chips with awesome performance for the time ), I would rather compare a G5 to a higher end AMD chip from the day, or maybe even a dual opetron depending on what model G5 we are talking.
Too add to the good pipeline explanation Zen gave, there is another thing to take into consideration. Every so often in the pipeline there is a dependency branch "bubble" in the pipeline. This causes the pipeline to drain and it would need to fill with data again. With a longer pipeline, this would cause the data to take more time traveling through the pipe. With a shorter pipeline, data does not have to travel far after a error.
An accumulation of these "bubbles" would take valuable CPU cycles. This is where RISC computing gets an advantage.
In summary, it's faster to get the data to the cores on a RISC chip than an x86 chip.
Edit: why the down votes?
An accumulation of these "bubbles" would take valuable CPU cycles. This is where RISC computing gets an advantage.
In summary, it's faster to get the data to the cores on a RISC chip than an x86 chip.
Edit: why the down votes?
Last edited:
Northwoods are kind of slow.
However, I got to work on the Pentium 4 Apple Developer systems (alongside the G5s of that time) which were Prescott based, and I can say Prescott Pentium 4s were faster than the G5 running the exact same OS/software. They also ran cooler.
Yeah, I know, horrible thing to say. But I tells them like I sees them.
Are you referring to a G5 running windows virtually? If so, of course the Mac will run slower.
No. As a developer I worked on both early Intel Macs with Pentium 4s running OS X and G5s running OS X side by side. Both running OS X natively.
Same systems noted here, with other developers noting the same thing:
http://www.appleinsider.com/article...ple_intel_dev_systems_impress_developers.html
Pentium 4 was faster at running OS X. Most my time was spent getting my software working on both Intel and PowerPC so I spent a lot of time comparing performance.