Sorry if this has been asked before (I did search, really!), but I couldn’t find an existing thread on this:
How do AMD CPU numbers compare to their speed ? I know an AMD 2200 does not run at 2.2 GHz, but I don’t know what speed it does run. Can anybody help with a link or a way to do a direct comparison with an Intel processor (i.e. "an AMD 2200 is roughly the same speed as an Intel 1.5GHz P4) ?
Thanks.
Here are a few examples from AMD’s web site. An Athlon 3000+ runs at 2167MHz. I believe it will be roughly as fast in processing terms as a P4 3.0 GHz.
Just to clarify, the Athlon numbers are put there to allow consumers compare AMD processors directly with Intel. An Athlon wxyz+ should perform roughly the same as a P4 wxyz MHz. In your example the AMD2200 will be roughly equivalent to a P4 2.2GHz.
OK - I wasn’t aware of that. That’s what I was wanting to know. Thanks!
Unfortunately, it’s not wholly correct to say the Athlon 2200 is as fast as a Pentium 2.2 GHz. That’s certainly what AMD wants you to believe. Independent chip testing companies often find Athlon chips perform slower than the Pentium chip they are supposedly equivalent to.
Googling for a cite…
Here you go, according to these performance results from the Standard Performance Evaluation Corporation, which is an independent CPU test consortium, the best peak score AMD submitted for an Athlon XP 2200 (671), is much lower than the best peak score Intel submitted for a 2.2GHz Pentium 4 (808).
Similar results for other Athlon vs. Pentium chips.
That is interesting, though it goes against what I have been led to believe: if anything an Athlon 2200 will outperform a Pentium 2.2GHz.
This is what Tom’s Hardware has to say about some similar chips:
I do not understand why the results from the two groups are so different though. Anybody?
Depends on what software/test is used. One CPU-A might be faster than CPU-B on test X, but B might be faster than A on test Y. They’re not designed the same, so one may be optimized for certain functions and the other optimized for other functions.
Technically, the rating numbers given by AMD are supposed to correlate with how fast an old Thunderbird core would be when clocked to those speeds - For example, the AthonXP 2000+ I have is clocked at 1.67ghz, but according AMD, should perform as well as a 2ghz Thunderbird Athlon. Of course, everybody uses the number to compare to Pentium 4’s anyways.
Generally, on the lower end of AthlonXP’s, the number rating does correlate quite favorably with the Pentium 4’s in real life benchmarks -Look at these older benchmarks - an AthlonXP 2000+ vs 2.2ghz Pentium 4. , though the AthlonXP line does slip once you start getting past the 2400 model. To be honest, I think AMD should have had the 3000+ as a 2800, and the 3200 as a 3000. The Athlon64 processors compare very well against high clocked speed Pentium 4s though, and run much cooler to boot.
Also note that AMD & Intel Processors tend to do better in different things. Pentium 4s tend to be very good 3d rendering & media encoding (though Athlon64’s take media encoding when you use a 64bit OS) and Athlon 64’s better in games. For example, in Doom 3 an Athlon 64 2800+ has no problems holding its own against a 3.2ghz Pentium 4.
Aragowyn was posting the results from a single synthetic benchmark; the benchmarks you posted from Tom’s Hardware & the similar benchmarks I posted from **The Tech Report ** are taken from several different synthetic benchmarks, as well as several different real world test. Athlons and Pentiums and each faster in specific areas, and do poorly in others. You can’t make a single benchmark that will tell you which processor would be better for your needs.
Here’s an analogy to illustrate the difference between the two types of processors. Picture Intels as narrow-diameter high-speed water pumps, and AMDs as wide-diameter low(er)-speed water pumps. The AMD pump obviously turns at lower rpm’s when the two pumps are moving the same amount of water, abstracting away effects of intake and output. However, if we pay attention to what the water intake and output conditions are, one or the other pump (moving at speeds which nominally make for identical flow) might have the advantage. For example, if the output flow is limited at the pump output in speed (not cubic feet per minute, but actual velocity), then the AMD pump can move more water. You can invent constraints on either end of the pump that will result in either pump moving more water. So it is with processors.
That said, AMD wxyz+ and Intel w.x GHz processors are closer in performance than AMD and Intel processors with identical clock speeds. An AMD running at x GHz will absolutely blow the doors off an Intel running at x GHz, no matter what the application. Course, it’ll cost a lot more, too. Hence AMD’s marketing scheme.
Actually, the Centrino processors Intel has for laptops can match/are somtimes faster than AMD’s processors at the same clockspeed. Of course, these chips are NOT based on the Pentium 4 design, but are closer to a beefed-up Pentium 3 - the Pentium 4 design was heavily influenced by marketing, which wanted really high clockspeeds, even if the performance wasn’t increased by much. Thus a 1.4ghz Pentium 3 would easily beat a 1.4ghz Pentium 4 in benchmarks, but it was possible to ramp up the Pentium 4 clockspeeds more quickly.
Of course, Intel is now running into some serious heat problems - many people are reporting that if you use the stock cooler with the latest Prescott core Pentium 4s (3.4 & 3.6ghz) the chips will quickly overheat if you try to run them at full load, and the chips will have to throttle back, killing performance. Prescott’s are currently using well over a 100 watts of power; the max power for the Athlon64 family is 85 watts(and that includes chips like the 4000+ & FX55 that haven’t been released yet) and AMD has stated that even its dual-core Opterons will use no more than 95 watts. (If you are wondering what dual-core means, it is basically putting two whole CPUs on a single chip. AMD & Intel are both planning releasing dual-core chips next year.)
That is why Intel will eventually be using the Centrino design as their desktop processor. Of course, Intel’s own marketing, that mhz=performance, has come back to bite them in the butt. I remember having to explain to my father a couple months ago that yes, that Centrino laptop is probably at least as fast as the Pentium 4 one, even though the Centrino is at 1.5ghz, and the Pentium4 was at 2.4ghz. Intel’s budget line the Celeron’s have caused problems too- one wonders how many people bought a 2.8ghz Celeron, thinking it would be just as fast as a 2.8ghz Pentium 4.
This sort of thing is why Intel has introduced its own model scheme, that seems to be quite similar to BMW’s car model numbers. 3xx models are for the Celerons, 5xx are for the Pentium 4’s, and 7xx are for Centrinos & Pentium 4 Extreme Editions. Of course, I wonder how long it will be before people start complaining about high end chips in one bracket (say a 590) are outperforming a low end chip in another bracket (like a 710). I say this system will confuse people who don’t have time to look at benchmarks.
AMD also uses a 3 digit model number system for its Opteron chips(designed for server & workstation use) - 1xx chips are for single CPU systems, 2xx chips for dual CPU, and 8xx for 4 and 8 CPU systems. I guess AMD realizes that people who need to buy this class of chip are able to examine various benchmarks to see what they need, while keeping a higher single number = better system for the normal consumers, who are used to judging performance by a single number.
I’m currently running a PIII 1.4GhzS cpu which is running PC150 memory(which is very rare stuff ideed) this means I have to overclock to 1.575Ghz.
This particular cpu was intended as a server chip, the Tualatin core was based upon a manufacturing process that ran cpus at well over 2Ghz so the overclock is no problem at all.
Point is, when I run it against other P4 cpus, it does well until speeds exceed 2Ghz, and this is against CPUs with 266 memory.
So the actual clock rating of a CPU is not the only thing to take into account, sure it makes for an easy selling point, especially to those who are even less well computer savvy than myself, which is perhaps the majority(dont hold me to that though it wont stand up in court!)
I can play all the recent games, but I’ve not tried out DOOMIII yet and I think maybe I won’t since the online multiplayer isnt very good.