In the add for the Mac G4 Cube, they say that it comes equipped with a 450Mhz G4 chip with built in Velocity Engine. With this, they say it can compute 3 billion calculations per second. Now I know a hertz is the measure of complete cycles per second, so how can a chip that has 450Mhz do 3 billion calcs a second?
Although one is roughly correlated with the other to the extent a faster Mhz CPU will (generally) be able to process instructions faster than the same hardware/software architecture at a lower clock speed, they are not directly related.
The number of raw calculations a machine can do is largely a reflection of the combination of the efficiency and power of the overall chip design and it’s wired arithmetic (ie internal FPU) logic among lots of other things.
Example:
The sun ultrasparc II at 250 MHz performs about 4 instructions per clock cycle.
4 ops/cycle * 250,000,000 cycles/second = roughly 1,000,000,000 instructions per second.
The Intel Pentium III at 1000 MHz uses about 3 clock cycles per instruction.
1/3 ops/cycle * 1,000,000,000 cycles/second = roughly 333,000,000 instructions per second.
So the 250 MHz Sun Ultrasparc is about 3 times faster than the 1000 MHz Pentium III.
Get it? The clock speed is just how fast the chip’s internal timer oscillates, not how fast the computer can actually work.
In other words: if I run 100 meters and time it with a watch that only ticks every second, but somebody else runs it with a stopwatch that ticks 1000 times per second, that doesn’t mean they are running it faster. Instructions per second is the statistic that matters, but also the one they never publicize.
The last two answers pretty well sumarize it. The point is that accross architectures (x86 and PPC) simple clock speed means very little. For complex floating point operations, a 500MHz PPC is equivelent to about a 1.2GHz x86.
More exotic architectures like Alpha and SPARC are even more efficient.
While we’re talking about computer speeds, I have two questions I hope somebody can answer:
-
Since Mhz isn’t the only factor of how fast a computer runs, can anyone tell me the REAL comparison of the speeds of Intel’s 1.18 Ghz Pentium III processor and Apple’s 500 Mhz processor? I’ve noticed that Apple’s best processors have been at 500 Mhz for months now and I was wondering if their computers are still getting faster (A while ago I heard that Apple’s 500 Mhz processor was about three times faster than Intel’s 500 Mhz processor).
-
Apple’s new desktop G4’s use two 500 Mhz processors. What kind of impact do dual processors have on the computer’s speed? Does it double the effective speed (I doubt it)? Can it only do two tasks at 500 Mhz, or can the two processors share the workload?
Saw this the other day. Might be of interest to mac users.
http://www.wired.com/news/technology/0,1282,38101,00.html
1.2 gigs. Is that fast, or what?
Peace,
mangeorge
Well, they’re not all up-to-date on the latest numbers, but here’s a listing of a whole lot of links at mac speed zone.
You can scroll down to look at different sections, too. There’s some benchmark comparisons there.
And the answers to Max’s questions are sort of related. Basically, it depends on what you’re doing; plus, the software itself matters. Processors are using more and more specialized sections of the chip to accomplish different tasks. So chips from different manufacturers often have quite a different approach to processing instructions. The PPC 7400 (G4) excels when working on certain tasks, but other processors do well on others. Having two or more processors is in a sense like just having more of those different units; although some tasks only need one processor; unless you can split your tasks up, it won’t help.
The special features of the newer processors often require that new software is compiled (or even written) to take advantage of their capabilities. This is why you may hear better and better performance for the G4 systems as time goes on; it just means that the software is being made to take advantage of these features. Dual (or more) processors will require more work done in compiling the software to take advantage of the software. Work has been done on thousand-processor systems that require a very intricate method of using all of that processing power.
If you’re interested in the technical details of all this, visit http://www.arstechnica.com - for a specific article relating to this subject, follow the link at the beginning of this post, then go about a page down to the ‘G4 vs. K7’ link.
And as a plug for the design work of Motorola & Apple, I’d like to point out that the G4 cubes do not use a CPU fan. (One reason the PPC 7400 won awards for design is that it has a low transistor count and thus low power consumption.)
Originally posted by Max the Immortal
While we’re talking about computer speeds, I have two questions I hope somebody can answer:
- Since Mhz isn’t the only factor of how fast a computer runs, can anyone tell me the REAL comparison of the speeds of Intel’s 1.18 Ghz Pentium III processor and Apple’s 500 Mhz processor? I’ve noticed that Apple’s best processors have been at 500 Mhz for months now and I was wondering if their computers are still getting faster (A while ago I heard that Apple’s 500 Mhz processor was about three times faster than Intel’s 500 Mhz processor).
I doubt there is a 300% differential in efficiency between the two architectures. Generally Intel i86 and Apple chip (PPC) performance relative to speed has been roughly comparable in the past with Apple’s being somewhat more efficient for a given clock speed (not 300% tho) relative to the i86 platform.
The real difference is that clockspeed for clockspeed Apple usually trumps Intel running comparable apps, but not by much, with the exception of the bet the business multimedia, graphic and publishing applications where Apple usually dusts Intel badly using identical clock speeds. Apple puts a lot of effort into optimizing these programs for the PPC and it shows.
The problem is that you can usually buy an Intel based unit with twice the amount of storage, memory and CPU clockspeed for the same price as the more elegant Apple machine. So… although Apple may have a slight to healthy lead in clock to clock comparisions, it’s really sort of meaningless when you can get an Intel machine with a gigahertz CPU - 256 megs of RAM and 40 gig 7200 RPM harddisk etc etc vs an Apple Unit with half that amount of candy and have a considerably more powerful machine overall than the Apple unit.
It’s at this point that Steve Jobs starts screaming that the point is being missed that Apple is a much finer and better designed piece of hardware. Well… maybe so but unless I’m desktop publishing or multimedia producing, the rationale for purchasing an Apple is hard to justify on a purely esthetic basis when I need to get real work done and I can get a lot more useful computing horsepower with my Intel dollar than my equivalent Apple dollar/.
- Apple’s new desktop G4’s use two 500 Mhz processors. What kind of impact do dual processors have on the computer’s speed? Does it double the effective speed (I doubt it)? Can it only do two tasks at 500 Mhz, or can the two processors share the workload?
Joe_Cool wrote:
So the Sparc can execute more instructions per second than the Pentium. But are the instructions themselves comparable? The Pentium is a Complex Instruction Set Computer (CISC). The Sparc is a Reduced Instruction Set Computer (RISC), IIRC, whose instructions are simpler. In a RISC computer, you have to do the more complex operations with software, instead of letting the processor do it an a few clock cycles.
So comparing which is faster depends heavily on the task at hand. RISC computers were invented about 15 years ago, on the idea that the time spent executing most simple instructions much faster would more than make up for the increased time software spends doing what a CISC computer could do in hardware. Benchmarking them is not simple.
The Mac might be in theory faster than an intel, but I have yet to see any software I use (mostly games) see them any faster… as far as I am concerned for practicle usage its a bunch of smoke.
However, if you want to play a hardcore game of head to head photoshop, then hey, the mac’s your machine.
My .02c that will surely get rebuttled by some macheads here
I have both Mac and PC. I don’t use games at all, except for Solitaire!
My 333 MHz PC was much slower opening Netscape than my 266 MHz iMac, for instance. From my own personal experience, I believe that my 333 MHz PC is overall noticably slower than my iMac in most things. However, this does not include games, since I don’t use them. I don’t doubt that the PC would be faster in some games…but like I don’t care! I’m a Photoshop person!
My PC is getting an upgrade to 450 MHz. I am curious to know how it’ll then compare to the iMac. My hunch is that they will be simular in speed, but I don’t really know what to expect. I suspect (though I’ll wait and see) that the iMac will still be faster with Photoshop, though.
bernse:
OK, you’ve convinced me. When I go out looking for a game computer, I’ll look at an Intel… I suspect it won’t hold a candle to the PlayStation 2, in terms of performance or price, but I’ll give it look-see anyway…
In the mean time, since I have to get real work done on a computer, I’ll stick with my Mac most of the time (though I do use a Wintel machine from time-to-time, but not by choice).
Max:
The only acid test is benchmarking. Unfortunately, the only benchmarks that mean anything are the ones that closely approximate the way you use a computer. I have access to both Macs and PCs - Macs because that’s what I spend my hard earned cash on and PCs because that’s what my employer wants me to use. I run a lot of the same software on both and find that, for what I do, the Macs are much faster. I prefer to use my 300MHz G3 over my 650MHz Pentium III.