800mhz vs. 533mhz — What’s The Difference?

With the advent of the new 800mhz frontside bus (FSB) on Intel’s latest processors, the automatic assumption is that they must be faster than their 533FSB cousins. In many cases this is true, but let’s take a closer look at why and how.

First we’ll gauge processor performance; pound for pound, can the faster FSB get more work done? The test system was an Intel D865PERL motherboard, two sticks of Apacer PC3200 CL3 RAM, a Maxtor Diamondmax Plus 9 120GB hard drive and an MSI GeForce4 TI4200 4X 64MB video card. If you’d like to find out what the scores below mean, click here. I strongly suggest visiting this link before reading further because it will explain what is being tested and why the differences are the way they are.

CPU Hyper-Threading IntegeriSSE2 (it/s) Floating-PointiSSE2 (it/s) Dhrystone ALU(MIPS) Whetstone FPU/issE2(MFLOPS)
2.4B/533 No 9385 11966 6868 1360/3067
2.4C/800 Yes 10991 17307 7299 2052/4531
2.66B/533 No 10441 13306 6961 1514/3465
2.6C/800 Yes 11904 18573 7920 2229/4944
2.8C/800 Yes 12833 20072 8546 2400/4547
3.06B/533 Yes 14062 21995 9365 2630/5873
3.0C/800 Yes 13761 21459 9168 2573/5713

Next, how is memory bandwidth affected by RAM frequency and the frontside bus speed? For this test I again used SiSoft Sandra 2003 Pro.

CPU Memory bandwidth (MB/sec)
2.4B/533 3224
2.4C/800 4202
2.66B/533 3192
2.6C/800 4277
2.8C/800 4341
3.06B/533 3207
3.0C/800 4334

As you can see, the faster 800FSB can use about 30% more memory than the 533FSB can. Increased memory bandwidth leads to more efficient CPU-intensive calculations and higher framerates in 3D games. The more RAM bandwidth the CPU has to work with, the more speedy its operations will be. Considering the fact that all of the 800FSB processors have Hyper-Threading Technology built-in, the increased memory bandwidth is even more important because of the extended processing capabilities that HT provides. While it doesn’t exactly make one processor do the work of two, it does have the capacity to execute two instructions simultaneously. The more memory bandwidth available to the processor, the more the advantage of HT is shown.

Let’s take a closer look at just how 3D gaming is affected by the 800FSB and HT Technology. There are a lot of Internet rumors about negative gaming performance from HT-enabled processors. Unfortunately these rumors seem to stem from hardware review and editorial/gossip sites and that gives them more credence. As you can see below, HT Technology has little or no effect on in-game benchmarks and 3DMark.

CPU 3DMark 2001 Quake3 Arena FPS Quake3 Arena time (seconds) UT2003 Antalus Flyby (average FPS) UT2003 Asbestos Flyby (average FPS) UT2003 Antalus Botmatch (average FPS) UT2003 Asbestos Botmatch (average FPS)
2.4B/533 11207 259.5 5.2 114.01 178.19 55.75 72.34
2.4C/800 11295 265.4 5.1 114.38 180.00 56.13 73.51
2.66B/533 11332 267.6 5 114.62 180.59 57.89 75.69
2.6C/800 11567 279.6 4.8 114.89 182.51 58.18 77.69
2.8C/800 11858 292.9 4.6 115.22 184.36 59.45 81.93
3.06B/533 11766 290.2 4.6 115.32 184.19 59.92 82.33
3.0C/800HT enabled 12115 306.5 4.4 115.66 185.51 60.48 85.56
3.0C/800HT disabled 12073 307.5 4.4 115.64 185.25 60.40 85.33

So while there is a significant and noticeable difference between the 2.4B and the 3.0C, there is very little difference between the 2.4B and the 2.4C in terms of gaming performance. The 3.06B seems to be in a rough spot — it’s outperformed by the 2.8C in most areas due to the 800FSB’s higher memory bandwidth. There is a measurable difference between the 3.0 with HT enabled and the same CPU with HT disabled, but the difference is so small that it would be impossible to notice it in a real game. This testing procedure doesn’t take background processes into account; in other words, if you have antivirus software or other applications running in the background you’ll see HT make a positive difference in framerates. Newer games such as Aquanox 2 and Spellforce and others that are built on the Krass engine are designed with HT Technology specifically in mind and will get much better performance with a Hyper-Threaded CPU. So no matter what the other hardware sites say, my advice is to leave Hyper-Threading support enabled even if all you do is play 3D games; overall system performance will be significantly reduced in Windows XP as you can see here in the Sandra CPU tests:

CPU Hyper-Threading IntegeriSSE2 (it/s) Floating-PointiSSE2 (it/s) Dhrystone ALU(MIPS) Whetstone FPU/issE2(MFLOPS)
3.0C/800 Yes 13761 21459 9168 2573/5713
3.0C/800 No 11789 15005 8639 1702/3909
2.6C/800 Yes 11904 18573 7920 2229/4944

So in effect, disabling HT on a 3.0C processor is almost like replacing it with a 2.6C.

Finally let’s take a look at how price factors into the equation. The following prices were taken from a wholesaler’s dealer price list as of 6/16/03:

CPU Retail Box (USD)
2.4B/533 $161.25
2.4C/800 $199.50
2.53B/533 $189.00
2.66B/533 $189.50
2.6C/800 $227.00
2.8B/533 $250.75
2.8C/800 $279.00
3.06B/533 $366.50
3.0C/800 $412.00

In conclusion we can safely say that the 800FSB does make a significant difference in computing power and performance, and the small price difference makes it a great bargain considering what you’re getting for what you’re paying. For a new Intel system build don’t even consider the 533FSB — go for an 800FSB CPU and DDR400 memory. If you have any questions about RAM, feel free to join our moderated forums by clicking the Forums link on the left.

Discuss this article or get technical support on our forum.

Copyright 2003 Jem Matzan. Verbatim copying and redistribution of this entire article are permitted without royalty in any medium provided this notice is preserved.

About the Author:

Leave A Reply

Your email address will not be published. Required fields are marked *