How can low spec PCs outperform higher spec ones?

  Solgoth 21:58 11 Dec 2005
Locked

Whilst researching buying/building a new PC, I came across an article in Computer Shopper magazine that confuses me. I had thought that the most important, crucial determinants of a system's speed were the processor, the motherboard, the graphics card and the amount of RAM. Following from this theory, it would seem logical that those computers with faster processors and better graphics cards, good motherboards and more ram should be measurably faster than those with worse specifications in those regards. However, this month's computer shopper confused me on this point. In their round up of a variety of £650 PC systems, there appeared to be little correspondence between a system's specifications and its performance in the tests they gave it. I'm aware that these tests and benchmarks are hardly necessarily an accurate reflection of actual real life performance, but I'm assuming they are accurately reflecting *some* sort of difference. What is causing the variations in performance between machines? It seems to me it just can't be the components.

Eg, in their image editing test, a PC with an AMD Sempron 3400, Ati 200G *integrated* graphics, and 2gb ram was allegedly the fastest?! Bottom of the list of 9 which included AMD 64 3000+ systems with £70 graphics cards being beaten by semprons with none, was a system with an AMD 3200+, 512mb RAM and Ati 200 integrated graphics. Why the difference? The bottom machine had integrated graphics as did the top machine, and yet the bottom machine has the best AMD processor of the machines on the test?!

Plus, adding to my confusion, there were two PCs with almost identical specifications yet different performances on the tests. A Mesh and a Carrera both had AMD 64 3000+ processors and GeForce 6600 128mb Pci-express graphics cards, and maxtor SATA 150 hard drives. The Mesh had twice the ram at 1gb and a better motherboard, yet in the PCMark04 benchmark test the Carrera is the fastest out of the whole lot and the Mesh third bottom, underneath a couple of AMD semprom 3400s with worse graphics cards and only 512mb RAM?

Why should AMD sempron 3400s with worse graphics cards and 512mb ram outperform AMD 64 3000+ systems with good graphics cards, like the mesh and Carrera and others? In the same issue, they also test the individual processors. In the individual processor tests, the Amd semprons come off pretty badly and are beaten by AMD 64s of all speeds at everything. So how can those computers with bad spec outperform those with better spec, if it's accepted that the tests mean anything at all?

I was thinking that I would be getting a decent machine if I went for an AMD 64 3200+, with the A8N-E motherboard the Mesh in the test had, and a slightly better graphics card than the Mesh and Carrera (ATi X700 256mb or GeForce 6600 256mb). But the fact that there seems to be so much variation between systems that seems unrelated to their specifications makes me doubt whether getting a higher specification machine is worth it. What accounts for differences in speed between systems? How can the results seem so unrelated to the specifications?

  DieSse 22:05 11 Dec 2005

"So how can those computers with bad spec outperform those with better spec, if it's accepted that the tests mean anything at all?"

Well you hit the nail on the head there. The validity of the tests is everything. It all depends on what is the limiting piece of hardware, in any test of a system.

So my personal conclusion would be that the tests are not well designed (rubbish in other words).

  Djohn 22:30 11 Dec 2005

in what he says, but I will differ slightly in saying that the test are not well designed. Lower specked machines will outperform higher ones if everything is matched correctly.[Machine is only as good as weakest link]

My system due for upgrade shortly is 3 years old with AMD1800xp+ CPU 1Gb memory 2 hard drives, 2 optical drives, separate sound and graphic card, PSU, but all are well matched with each other and the motherboard.

Just a short while back tonight, I had "Word" open, 5 websites, Messenger, email application checking every minute for mail yet I burned a DVD movie at full speed while carrying on with other work, took 6 minutes to burn with PCA's recommended DVD burner. LG GSA 4163B. Superb piece of equipment by the way.

  Solgoth 23:28 11 Dec 2005

Thanks for the responses. The tests would have worried me less if it wasn't for the fact that they tested individual processors in the same issue according to the same criteria, so there was something to compare it to. With the individual processors, results were a bit more consistent in that AMD processors tended to do better or worse than each other according to their spec.

Assuming that the tests are the same with the individual systems, I don't see what the 'weakest link' in these machines could be to bring them below the level of much lower spec machines. The Mesh, for example, had the highest spec motherboard. and more ram.. don't get how it could be slower than others with lower spec.

How can I tell if the components I choose for a system are 'well matched'? I was considering getting a system quite similar to the Mesh used in the test regarding motherboard and graphics card so it worries me that there seems to be a weak part in the system when nothing is visibly wrong with it at all.

My friend has been warning me that I would be just as well buying something lower spec as the differences in performance are not that great.

  ashdav 23:58 11 Dec 2005

Just to put things into perspective..
4 to 5 years ago a 500MHz computer was pretty good.
A 2GHz computer is now pretty much the norm.
That's a 4x improvement.
If you go to a 3GHz computer that's only 1.5x improvement.
I doubt the extra money is worth the "improvement".

This thread is now locked and can not be replied to.

Elsewhere on IDG sites

Galaxy Note 8 vs iPhone X

This is what design agencies will look like in 2032

How to update iOS on iPhone or iPad

WhatsApp : comment lire vos messages sans que l’expéditeur le sache