Just what is the deal with AMD and Intel?

How is it that Intel provide processors at one frequency, and Amd provide them at a lower frequency - and claim them to be equivalent products…?

It seems insane, however there are many hundreds of ways different processors can outshine eachother. For example all processors have an onboard cache of fast access memory on-board, cryptically called an on board cache. This stores instructions, input values and a stack of results which the processor or processors can process and store to, among other things.

Back in the old days it was fairly simple, how big was the cache? How fast can it access its contents? and how fast can the core read and write to this cache? The main limitations were how quick memory could be chunked around, loaded into the cache for detailed processing, and spat back out. Later on porcessors became capable of directly accessing much bigger chunks of memory - your system RAM. Once games could directly “blit” large chunks of graphics around the screen memory, things really began to take off.

Now as you can imagine - I am not a CPU engineer - and anyone who claims to fully understand the intricacies of a modern day CPU because they’ve read a few magazines articles, its quite possibly, full of it. I am basing my comments here on what I know - which was MCGA/Xmode graphics programming, back in the dark days of DOS, and 286 processors. Xmode was a special VGA graphics mode which ran 320 pixels wide, 200 high and could display 256 colours. It was special because instead of being a linear store of pixels in the graphics memory, the data was stored across four bitplanes. Technical geekery aside - this meant that you could not only do hardware, instantaneous page flipping (which hitherto had required enormous copies and pastes of graphics memory per frame), but you could also access and change the contents faster than ever before.

The end result was a successful way of transferring the same graphical information, and processes, in a very much faster and more efficient fashion, using exactly the same hardware.

My analogy follows to processors with regard to their pipelines. Initially people thought “two processors doing stuff at the same time! - Genius!” - this could be compared to putting two graphics cards into one pc when they couldnt actually talk to each other properly. Software engineers began to frown, as they understood the grim truth: Software needs to be specifically coded as a threaded application with dependants, and precedents in order to take advantage of this extra processor. Otherwise the extra processor was just waiting for the other one to finish what it was doing. It died a death rather quickly with regard to personal computing. It has considerable use in a Server application however. But then I dont have a server. I want to play Half Life Episode 2.

So along came intels “Hyper threading”, which, sort of tried to do the same thing in a half-assed way with just one processor. Needless to say the entire theory was rather flawed and ultimately limited as you would expect. It died a death - although I have one in my lounge - a P4 3gig Hyperthreaded beast which doesnt perform much, if any better than any other 3gig P4. Overall I really havent been able to find much difference ebtween that and my 2.6 gig P4 in my study - upon which I am writing. So much so that my non-hyperthreaded PC is actually my gaming computer - and I’m not concerned about the drop in frame rate.

In summary - that all sucked. Bring on the new fish - DualCore!

This took everything that might have worked about Dual Processors, and instead merged the two processors - or “cores” onto one CPU die. So effectively you have two brains on one chip. The difference is that instead of having two brains with pipelines to coordinate together - these two cores have a communal pipeline, which keep everything in order. One of the fast-devleoping techniques is branded “superscalar” pipelining - and it is used to predict what will be needed and when - allowing both cores to work simultaneously on different things- wihtout the computer code having to be defined and refined specifically for the processor application.

Now - somehow - AMD reckons that their chip is faster per Mhz than Intels chip. All odwn to the design of the cache, pipeline and internal buses - For whatever technical what not reason they give. I wish to test this in the only way I know how……..Benchmarks. In short- the word is - that this all kicks ass. I’m not so sure. It might well be a whole lot of buzz to sell more over-powered processors.

Its all jibber jabber until its been tested. At work I have two almost identically specced processors one of them an Intel DualCore at 2 x 1.76Mhz, with 2 gigs of DDR2 ram.

The other is an AMD X2 4200+, which is similiarly 2 x 1.83mhz cores with 2Gb of DDR2 ram. The front side buses are the same - 1000Mhz. Both have on board graphics cards which I will use for the test.

2 rounds of 3dMark06 in software only mode will sort the men from the boys, and tell me for certain if intels 4200+ claim is worth any salt.

Then ill pit them both up gainst my 3ghz Hyperthreaded beasty, and my 2.6ghz non-hyperthreaded beasty.

I dont expect my PCs to beat either of them - i ust want to see what the difference really is.

Watch this space for the results in the next few days…….

written by Pensive