Sign in with
Sign up | Sign in
Your question

How do new clock speeds compare with old?

Tags:
Last response: in CPUs
Share
September 13, 2006 6:20:04 PM

Excuse the noobish question, but I want to know how the new CPU clock speeds compare to older clock speeds.

For example, my old system used an AMD XP2400+ at 2.0GHz where my new system is using the ever popular E6600 at 2.4Ghz. Now, beyond the E6600 having two cores, even one at 2.4 would smack my old CPU at 2.0, what's the difference?

Or another example, HL2: Lost Coast complains I only have a 2.4Ghz processor (recommended 2.6Ghz) but again the E6600 can easily handle Lost coast, so what accounts for the difference in the numbers?

Thanks.

More about : clock speeds compare

September 13, 2006 6:42:35 PM

The short and simple analogy would be, dig a hole using a shovel, and then dig another hole of the same size with a spoon. Same amount of dirt, but it will take you a whole lot more motions with the spoon than the shovel.

The new CPU's do more with less. Same reason why the AMD processors have been out performing the pentiums with less clock cycles for the past several years.
September 13, 2006 6:42:59 PM

I am very glad someone brought this up. I wish I could answer your question but i'm interested myself in knowing the specifics.

I know processor architecture (the design of the chip) plays a huge role but beyond that, I have no idea.

Anyone care to enlighten us?
Related resources
September 13, 2006 6:50:23 PM

You can't really. You cannot compare direct clock speed. This was pretty much always the case when comparing AMD vs intel, right? That's why AMD launched their new numbering system way back when w/ the 4400+'s and stuff.

This is becoming a bigger issue i think. I'm seeing more and more system req's say P4 2.4Ghz or equivalent. Keyword there is equivalent. That should mean "equivalent performance".

But there isn't any across the board numbering system that says X CPU is better than Y CPU. AMD and Intel would never agree to that because everyone would want the top ones, right? (Intel would probably agree to that now since they're on top - heh).

The CPU's architecture matters sooooo much in determining the performance of the CPU. Take prescott (P4) - high clock speeds, but it got wasted by the lower clocked XPs. Now take conroe (core 2 duo) which wastes the XPs at and below the same clock speed.

Seriously, there's nothing better than using benchmarks to determine which CPU you should get. If you want to get a feel for that, use Tom's CPU Charts and punch up a game benchmark.

Benchmarks are the only thing that we consumers can depend on when comparing AMD vs Intel CPUs.

I wonder if AMD will redo their model numbers.
September 13, 2006 7:01:34 PM

That's a good point about the measure of performance, you can't really tell anymore what "P4 2.4GHz or equivalent" means. My guess is that on things like PC games, publishers will adopt the Vista performance rating tool for minimum/recommended specs, so instead of saying P4 2.4GHz or whatever, it might simply be replaced by a vista CPU rating of 3 for example.
September 13, 2006 7:42:54 PM

I highly doubt they will adopt the Vista rating system. 1 thru 5 isn't a broad enough spectrum to catagorize many different cpu's small steps from one speed to the next. At some point when cpu's get fast enough they'll all be rated a 5....then how do you choose?
September 13, 2006 7:54:55 PM

I think they're supposed to have numbers higher than 5 as well for future higher performing parts.

I dunno, I can see alot of people fighting MS's ratings. But hopefully that'll just spur them to make some consortium that'll make their own! :p 
September 14, 2006 12:22:37 PM

Yeah, but remember that you have a decimal point to play around with too - so you could have 3.5 for example, and I imagine it would be easy enough for MS to raise the cap on the scores as time goes on.

Maybe there will be opposition to the rating system since it's made by MS, but afaik, it has been developed by futuremark, which might give it some credibility. Would make sense to use it though provided it provides an accurate reflection of CPU power, since it's one tool you know that everyone who has Vista will have access to. Failing that, we do need some kind of standard performance indexing system, which both AMD and Intel agree to using.
a c 472 à CPUs
September 14, 2006 1:54:17 PM

It's basically all about how the CPU is designed. The architecture determines how well instructions are processed and the end result would be the data required. On aspect is the CPU have pipelines that are kinda like high speed highways. The instructions races thru the pipelines for very fast processing, the longer the pipeline the less time it takes to process the instructions. The Pentium 4 has long pipelines, say 32 (guessing), while AMD uses shorter pipelines, say 16, for highspeed processing.

The kicker is that if the resulting data is "bad" then the whole process has to start from the beginning of the pipeline all over again. The longer the pipeline, the more likely "bad" data would occur. Basically speaking of course. There are ways to reduce the likelihood of "bad" data using branch prediction algorithms. But that's a different topic....

Core 2 Duo has fewer pipeline stages than the Pentium 4; don't know how many off the top of my head. But it can also process 4 instructions per clock cycle as opposed to only 3 instructions per clock cycle of the Athlon 64/XP and Pentium 4 series. Simply stated if you can process 4 instructions per clock cycle, then you do not need work as fast as someone else who can only process 3 instructions per clock cycle.
!