Popular Posts:
Recent Posts:
The AMD and Intel MHz Controversy
I’ve been into computers since I was like 10 years old, now 25. Ever since then I always thought that Intel jacked up their clock speed to make it seem faster than AMD processors to the normal computer user. Until a few years ago that is still what I though, that Intel was trying to fool everyone. I now know what the difference is and that Intel was not lying about their clock speeds they just had different factors that made it higher.
This post is directed at people who have the same belief and to set the record straight. This is by no means meant to be an in-depth article into pipelines and how cpu’s work etc. This is just to give you the basic idea behind why there is a difference between a AMD MHz and Intel MHz.
First off a MHz is a MHz no matter what, there is no difference in any MHz. a MHz is a unit of measure similar to an inch, foot etc. A MHz is one million hertz, MHz is short for megahertz. MHz are used to measure the transmission of data on electronic devices. Basically 1 MHz means the device can process the data one million times per second. Now lets move on to how AMD and Intel are different.
A computer program runs on your computer and send instructions to the CPU to be carried out. The program send out the instructions in order like 1,2,3,4,5 but the CPU may execute the instruction like so 3,4,1,2,5, this is called “out of order execution”. The CPU breaks each instruction down into pieces like 1a,1b,1c and so on, but the CPU may execute 3a before 1a and 2a last because it is waiting for data from 1c. It gets very complicated and I could never explain how all of it works. This is just so you understand that CPU are very complex and the execution of programs is very complex.
Basically this is where Intel and AMD differ. The speed difference comes from how fast the CPU can execute instructions out of order and how fast the pieces can be executed. Imagine you need to build a brick garage and you have 22 or 31 small trucks to carry the bricks to the construction site. Then imagine at another site you have 12 larger trucks to carry the bricks. Which do you think will get done faster? If you chose the 12 larger trucks your right and this is the path that AMD chose. Intel chose the other path which turned out to be the wrong one (The information on the pipeline is referring to the older Intel and AMD processors).
What happens here is that in order for an Intel CPU to complete a simple task it has to do 22 or 31 things and a AMD CPU only has to do 12. What this means is that you have an AMD CPU running at 3.0GHz and a Intel running at 4.0GHz, the Intel chip is doing less work per clock cycle and more cycles. Basically the Intel chip is running faster in MHz but doing much more work for the same result.
This is why Intel and AMD MHz don’t match because there is a lot of other factors that go into figuring the clock rate. Just because an Intel chip sports a faster clock rate does not mean that it’s faster in real world applications than a slower clock rate AMD chip.
Since Intel dumped the Net Burst model for their cpu’s now they use a 14 stage pipeline which is much better. Using more pipelines and running at high clock rates was causing overheating and other problems for them anyways. These new Intel cpu’s require less power and can operate at cooler temperatures. So here’s one to Intel, now just tell AMD they were right all along and I will be happy. AMD has stuck with the 12 stage pipeline in their Phenom processors.
Again, this article was just to put to rest some of the misunderstanding of how AMD and Intel figured clock rates and why a slower clock rate AMD CPU could out perform a fast clock Intel CPU. This is true of the older AMD and Intel cpu’s, AMD is losing the speed battle now that Intel has cut the number of pipelines down. If you find any of my information inaccurate please leave a comment or shoot me an email using my contact form.
10 Responses to “The AMD and Intel MHz Controversy”
Leave a Reply
Current Poll:
Loading ...Latest Comments:
- Dragged Design: Photoshop on the top spot? I thought it was Windows XP on my head. Thanks for the great information...
- Wheelwright: 1. It’s Free You get what you pay for 2. Normally Better Software See #1 3. Better Community...
- Vadim P.: Might want to try http://gufw.tuxfamily.org for a better firewall UI.
- Mundo_Infinito: I highly recomend the free Spybot Search and Destroy for anti-malware. Also, Zone Labs has a...
- ad4m: They just broke ground and have not started construction yet. The will not start accepting applications this...
I think this is misleading. You didn’t go into any detail about pipelining, just a generalization that 12 pipeline stages is faster than 31.
That’s not neccesarily true, each pipeline stage doesn’t take the same amount of time to execute. Probably every individual stage of the Intel pipeline were considerably faster than any single stage in the AMD pipline. However, having more stages means there is more overhead for each stage.
When doing a lot of instructions in a row that don’t require branch instructions the 31 stage pipeline should be faster. However, when there are branch instructions that require the pipeline to be cleared is when the longer pipeline suffers in performance. Since the 31 stage pipline has to throw away 31 stages of operations as opposed to 12.
As I stated at the beginning of the post this wasn’t supposed to be an in depth article on pipelines etc. I understand what you are saying. Also this article was more geared towards the average PC user, not someone as knowledgeable as yourself.
[...] sixdoubleo: [...]
I didn’t realize there were people out there confused about this.
I think pretty much the population of the world that are not computer engineers are confused about this. I for one appreciate it.
This is misleading, MHz in no way refers to “processing” of a data unit. It’s informally defined as the number of events occurring per second [wiki]. The clock speed drives your circuit, some flip-flops and latches have different response time, some are edge triggered some are not. Clock is used for synchronization of parts of the circuit. It’s stupid to compare the pipelines or order of execution in reference to clock speed.
You example of trucks is biased in a sense of you are saying that in order to complete the same operation it’s unnecessary to split the operation in small parts. But let’s go ahead with it and further extend.
Say you need parts with different sizes. Some small and some large. But smaller bigger ones are needed for foundation first. So even if you deliver small ones at the same time as you deliver big ones [you have larger trucks to carry them] you will still be waiting till the big ones make the foundation. In other words you are referring to throughput of a data bus.
Now imagine you have 2 building to make instead of 1. And you have multiple smaller truck each one enough to fit enough big blocks for foundation 1 building. You load all the big block for first building on truck 1 send it, while it’s driving you load the second truck with big blocks to 2nd building. So at this point the first truck arrives offloads and goes back for small blocks. then the 2nd one arrives, offloads and heads back. the first building foundation by the time 2nd truck arrived is done, so construction on 2nd building is starting. So basically you have 2 “threads of execution” [i use it rather loosely here] or 2 things happening in a pipeline. This is the benefit of a pipeline, don’t wait for other things to finish first if you can do some other work at the same time.
Keep in mind right or wrong cpu design is defined by it’s success on market and at what is cpu designed to do. For example consider supercomputers [not clustered regular computers together] some of them are good at specific type of operations. For example vector processor that is able to run mathematical operations on multiple data elements simultaneously in vectors. But these cpus would be horrible at other things. Good example is a CELL processor. It’s good at what it is designed to do but will suffer horrible problems while doing other things.
Clock speed vs CPU design are related but yet different things.
Like Ian said, I also didnt realize people were STILL confused about this. This is about the Pentium 4 era. I agree with the article and if anyone needs an in depth, technical explanation of this, just head to anandtech or some hardcore tech site and search through their old articles.
Strange “analysis”…
Actually, all Intel’s CPUs are faster than AMD’s operating on same “frequency”. More expensive, but faster.
A simple article to explain things in laymens terms. It doesn’t go into deep detail, and I don’t think the author intended to. For someone who doesn’t know much about computers this is a good explanation.
[...] [...]