The “Break it Down” Blog

Long time reader and all around great guy Gordon Walter wrote in and asked us about making heads or tails of the computer-offerings (especially the quad-core) computer offerings showing up online this holiday season.

His questions were good ones and we thought it best to answer them for everyone to read. See our answers inlined below (in Green):

OK, while exploring the BB Cyber Monday offerings, I encountered an old problem I have figuring out what the real world speed of processors is. I picked the top-rated (ah, by customers….) desktops at BB and then selected the three below to compare.

It seems to me that the cheaper Gateway (on the left) should be about as fast as the more expensive HP on the right, but I am just multiplying 4 times 2.4 to arrive at the speed… Is that basically correct?

Sort of… fortunately each new family of processor that is released by Intel and AMD is much more optimized than the previous generation in predicting and ordering instruction execution. The CPUs also use additional strategies like Hyper Threading to improve performance. So a straight GHz-to-GHz comparison is no longer an accurate predictor between chips.

The reality is of course a bit cloudier than that. I would never say do a GHz-to-GHz across Intel and AMD, the chip performance is quite a bit different (if you are curious, Intel is stomping all over AMD in the last few years). And beyond that, I wouldn’t do a GHz-to-GHz comparison outside of a family of chips on either platform, because each new family (like the Intel Core 2 and Intel Core i7) can offer sometimes significant jumps in performance for the same GHz speed.

Now, specifically to your question above, the Q6600 is a favorite of the last-gen Intel Core 2 chips. It’s one of the most overclockable chips Intel has released in a long time and was the cheapest Quad-Core CPU for over a year. If I were building a quad-core machine today however, I would look very hard and long at the Core i7 (NOTE: That would not be a value computer… it would be $2k+ at the least).

As for the performance of the Phenom X4 chips from AMD, I cannot speak to that. I stopped following AMD chips 2 or 3 years ago when they decided they could release a significantly slower chip and charge approximately the same as Intel.

Suppose that I am using one program–and only one… does the quad 2.4 processor actually make a single program work faster than would a single 2.4 processor or a duo 2.4 processor?

It depends. If the program is (mostly) written to be single-threaded, as most normal applications will be, you won’t notice any difference between a single core, dual core or quad core machine (We are of course assuming that you don’t have any heavy background tasks running like a virus scanner that the Quad-Core could speed up for you).

If you are using a scientific application, video application or advanced photo editing software, most of those are now predominately multi-threaded applications and you will see a significant boost in speed especially in CPU-intensive tasks (like rendering the video or applying a filter to an image).

Most “normal” desktop applications are single-threaded though… designing multi-threaded software is easy in theory and hard as hell in practice to get right, so it’s not the sort of thing a lot of places are investing time and money into doing yet.

So in your case, you probably won’t notice a boost here… but do notice what I mentioned about background tasks. Everything thing your computer is doing is a process that can be given execution time on the processor. So if you are the type of computer user that has a real-time virus scanner, real-time spyware scanner, email running in the background, chat going and likes playing music, these are all background tasks that can be sped up with a multi-core processor while your main focused application that you are working on wouldn’t need to sacrifice any CPU time for them.

As I understand it, the quad will process four programs at once, all at a 2.4 speed–more or less.

Yea… more or less. Literally it will execute 4 parallel threads of instructions at the same time, that could be 4 threads from the same program and none from the operating system… 2 from the program and 2 from the OS, etc. It basically has 4 hands that can do something at once, whatever that means for the software on your computer.

If you have 2 GB of RAM in a single 2.4 processor, do you then need 8 GB or RAM in a quad 2.4, if you want the quad to work four times faster than the single?

Fortunately not. RAM is RAM (and dirt cheap), your system will use whatever RAM is available, no need to match it to the CPU.

I’d suggest 3GB because it’s the most ram you can get in a 32-bit Windows machine and not need to move to a 64-bit install of Windows… which still has quirks and I would never recommend it to anyone yet (Possibly with Windows 7). This recommendation has a lot more to do with 3rd party application and driver support not being up to snuff yet and a lot less to do with Microsoft doing a shoddy job.

Does the system bus speed have anything to do with how fast the processor works?  I find it a little strange that the more expensive Gateway (in the middle) has a slower system bus than does the cheaper Gateway.  Or is “speed” even a relevant term when talking about the system bus?

In theory bus speed is very important, in practice, it’s the last thing you need to worry about because your bottleneck in a system with a slow bus speed will lie elsewhere. Also AMD has always had insanely high bus speeds because they build the memory controller right onto the CPU itself; Intel hasn’t started following suite until recently with the Core i7 CPUs. That all sounds very impressive, but in reality, what does that mean to you? Barely anything… you’ll get a fast processor and your computer will chug along quite a bit faster than it was.

This is a lot like going to the car dealer and having them tell you there is a Super-Flux Differential in all the cars on the lot. It’s not that the differential isn’t important… it’s very important, but it’s just so obvious that the differential you get in the car you are looking at is going to be good enough to drive that car around just fine. If you have any qualms with that car (e.g. slow/no pickup) your problems won’t be with the differential, they will be with something else (like not a big enough engine).

So it’s one of those “don’t worry about it unless you have to” things.

And that wraps it up, I want to thank Gordon for sending in those excellent questions. That is not the first time folks have asked us about that, but it’s always been in an offline fashion and it didn’t dawn on us until now just how damn confusing these computer-comparison charts are for shoppers.

We hope that helped!