How Fast is my Processor?

How Fast (or slow) is your x86 Processor?

Introduction

Lately, it seems as though everywhere you turn on the Internet someone has a new set of performance figures for a processor, memory, motherboards or an entirely new computer. If you’ve been searching for answers as to what is right for you only to find yourself sitting at your present computer with a blank stare, take heart, you’re not alone. It is becoming increasingly obvious to anyone reading all these statistics that they are just that, statistics. As we read through the multitudes of these articles and assessments, it became clear to us that while much of this material was well prepared, there was a great deal of it founded in pure fiction.

While there are large numbers of people who want their computer to be the best of the best, the fastest of the fast, there are even greater number who just want their computers to do what “they” want it to do, and do it without problems. In this section of our Web site, we decided that it was time that we assembled a review of some of these performance articles in an effort to provide you with a “real world” look at some of these performance improvements as well as some of the fiction incorporated in some of the articles you will find. We have provided brief and concise answers to the realities of some of these advances in technology, along with some of our own recommendations in various categories. Amid all of this, we suggest that you try and focus on what your “real world” environment is and then try and determine how these technology improvements may apply to you.

How fast or slow is your x86 Compatible CPU?

That’s a reasonable question, but not one easily answered. There’s two ways this can be measured, one testing just the performance of the processor and another testing a given processor in a ¹ specific hardware environment running real world application. And yes, there is a difference. Testing just processor performance provides you with nothing more than the raw horsepower of the processor, not how that processor will perform in an application intensive environment.

The first chart noted below represents the data provided by published benchmarks for the ² Ziff-Davis Benchmark standard tests for low-level CPU performance , CPUmark99 and real world performance of business applications via the Winstone99 Business test. Note: Some numbers were measured on different systems and cannot be directly compared and should only be referred to as a loose guide. Note also, the AMD^® Athlon and similar clock speed Intel^® Pentium III Coppermine chips, provide close comparative figures.

As a general “Rule of Thumb”, you should upgrade only when your applications require more speed or are more processor and memory intensive. Obviously, moving from a Pentium 200 to a Pentium II 450 or to a Pentium III 600 will result in a notable speed increase. Given the drop in processor prices over the last six months, we recommend nothing slower than a Pentium III 600 processor. The following chart coupled with your experience with your present system will help you determine when you need to move to a faster system. While we do not use the Ziff-Davis Benchmark standard tests to perform “Real World” testing, (we feel that they are not updated frequently enough to take advantage of the fast paced technology changes), we do feel that sufficient comparatives have been done from which you can obtain basic performance information. Higher scores are better.

1. Specific hardware environment, in this instance, means that a given processor is installed in a test machine with a specific test bed motherboard with a specific amount of memory and then tested against another processor in the exact same environment.
2. Source: CPUMark99 PC Magazine September 1999, December 1999, March 2000 (Windows 98), June 6, 2000 (Win NT); Anandtech.com September 22, 1999, October 20, 1999, Intel website. Italics represent extrapolated estimates. Pentium III E indicates Coppermine CPU, B indicates 133 Mhz Bus. Higher numbers are better. ** indicates figures unavailable.

 

Model	Speed	Winstone99	CPUmark99
Intel Pentium	200	**	11.6
Intel Pentium MMX	233	10	10
Intel Pentium Pro	200	**	18.20
Intel P2	233	**	18.80
Intel P2	266	**	21.30
Intel P2	400	22.2	31.6
Intel P2	450	23.9	34.1
Intel P3	450	23.1	34.6
Intel P3	500	25.1	38.2
Intel P3 E	500	26.4 (27.2 i815)	47.8
Intel P3 EB (i815)	533	28.1	52
Intel P3	550	25.9	40.6
Intel P3 Xeon 512	550	27	45.4
Intel P3 E	550	27.3	51.9-52
Intel P3	600	27	44.1
Intel P3 B	600	**	44.4
Intel P3 E	600	28.7 (29.5 i815)	55-56
Intel P3 EB (i815)	600	30.3	58
Intel P3 E	650	29.3 (35 NT) (30.4 i815)	59-60
Intel P3 EB	667 (i820)	29.9 (31.3 i815)	61.3
Intel P3 E	700	29-30 (31.3 i815)	62-64
Intel P3 EB	733 (i820)	31.3 (32.2 i815)	66.3 (69 i815)
Intel P3 EB	750/100	40 (32.1 i815)	66-67.5 (67 i815)
Intel P3 E	800/100 (i815)	32.8	71
Intel P3 EB	800/133 (i820)	31-32 (39-42 NT, 35 NT SDRAM)	70-72.1 (66 SDRAM) (74 i815)
Intel P3	850 (i815)	33.4
Intel P3	866 (i820)	32.6 (37-42 NT, 35 NT SDRAM) (34 i815)	74-77 (69 SDRAM) (80 i815)
Intel P3	933 (i820)	34-35 (33-34 SDRAM) (35 i815)	81-84 (79-80 SDRAM) (85 i815)
Intel P3	1000 (i820)	34.5-35.6 (42 NT)	85-86
Intel Celeron	266	**	12.2
Intel Celeron	333	19.3	27.0
Intel Celeron	400	21	30.9
Intel Celeron	433	21.7	33.1
Intel Celeron	466	22.4	34.8
Intel Celeron	500	20.2-24	36.1
Intel Celeron	533	20.8	37.9
Intel Celeron	566	21.1	39.8
Intel Celeron	600	21.6-25.5 (26.4 NT)	41-43
Intel Celeron	633	21.9	42
Intel Celeron	667	22.1	44
Intel Celeron	700	22.6	45
AMD K5	PR75	**	7.15
AMD K6	PR200	**	14.3
AMD K6	266	**	17.6
AMD K6 3D	333	**	23.8
AMD K6-2	400	20.3	28
AMD K6-2	450	20.6	28.6
AMD K6-2	475	21.1	28.3
AMD K6-2	500	21.2
AMD K6-3	400	23.3	43.4
AMD K6-3	450	24.1	48.1
AMD K6-2+ Mobile	500	**	45.4
AMD K6-3+ Mobile	500	**	53.8
AMD Athlon	500	25.1
AMD Athlon	550	26.2	51.9
AMD Athlon	600	28.1	56
AMD Athlon	650	30	60
AMD Athlon	700	31.3	63.8
AMD Athlon	750	31.6	67.9
AMD Athlon	800	28-30.8	65-72
AMD Athlon	800 (T Bird)	(45.9 NT)	75.7 (76.4 NT)
AMD Athlon	850	29-31 (36-39 NT)	70-72 (73-74 NT)
AMD Athlon	950	**	74
AMD Athlon	1000	31-32 (41 NT)	79 (81 NT)
AMD Athlon	1000 (T-Bird)	(48.9 NT)	91 (91.6 NT)
AMD Duron	600	25.5 (34.8 NT)	52.9 (53.7 NT)
AMD Duron	700	26.9 (36.2 NT)	58.9 (59.8 NT)
Cyrix	200+	**	15.5
Cyrix M2	400	**	25.6

As you can see from the numbers provided above, there are real gains to be had when moving up from the older Pentium processors, including moving up from the veritable Pentium Pro to the Pentium II processor (Pentium II 400-450 specifically). You can see substantial gains when moving up to the new Pentium III Coppermine, especially in the 600 to 733 Mhz range. The momentum seems to stop though above 733 Mhz until you get into the 1000 Mhz range (One Gigahertz).

Let’s look at Laptop performance.

How fast or slow is your Laptop Intel x86 Compatible CPU?

The chart noted below, very much like the first one, represents the data provided by published benchmarks for the ¹Ziff-Davis Benchmark standard tests for low-level CPU performance , CPUmark99 and real world performance of business applications via the Winstone99 Business test. Note: Some numbers were measured on different systems and cannot be directly compared and should only be referred to only as a loose guide. The following chart coupled with your experience with your present system will help you determine when you need to move to a faster system. The Celeron 400A, 450, and 500 processors are merely Pentium III’s with half the cache disabled. Higher number are better.

1. Source: CPUMark99, PC Magazine September 21, 1999, December 1, 1999 (Windows 98), Intel website, AMD Website. Higher numbers better. Italic numbers are extrapolated estimates.

 

Model	Speed	Winstone99	CPUmark99
Intel Pentium MMX	233	9	13
Intel Pentium MMX	266	11	12-13
Intel P2 PE	266	14-15	26
Intel P2 PE	300	15-16	27-2
Intel P2	333	16-18	29
Intel P2	366	16-19	31-32
Intel P2	400	18-20	34
Intel P3	400	20-22	37
Intel P3	450	23	41
Intel P3	500	23.7	47
Intel P3	550	24	51
Intel P3	600	25	55
Intel P3	650	25.9	59
Intel P3	700	26.5	63
Intel P3	750	27	66
Intel Celeron	266	13-14	22-23
Intel Celeron	300	14-15	24-25
Intel Celeron	366	15-16	27-28
Intel Celeron	400	16-17	29-30
Intel Celeron A	400	17-18	30-31
Intel Celeron	450	18-19	31-33
Intel Celeron	466	16-17	33
Intel Celeron	500	18-20	33-35
Intel Celeron	550	19-21	**
Intel Celeron	600	20-22	**
Intel Celeron	650	21-23	**
AMD K6-2	300	11	28
AMD K6-2	400	16	28
AMD K6-2+	500	**	45.4
AMD K6-3P	400	**	46.6
AMD K6-3P	450	**	50.7
AMD K6-3+	500	**	53.8

How does the Celeron II compare to the Pentium III?

The Intel Celeron 533A and faster are modified Pentium III chips, which have their bus speed fixed at 66 Mhz. Intel also reduced or disabled half the cache, and modified it to be slower than that of the Pentium III. A comparison was made and published April 11, 2000 by Sharkyextreme that compared the two processors and found the Celeron 600 to be about 20% slower on the ZD CPUMark 99 than a P3-600, 10% slower in Content Creation 99, 15% slower on Business Winstone 99.

Processor Bus Speed

An often asked question, can performance be gained by moving from a 66 Mhz bus to a faster 100 Mhz Bus?

Over the last year or so, many technology news sites have referred to an article published by Tom’s Hardware Guide on April 16, 1998 that showed that moving from a 66 Mhz bus speed to a 100 Mhz bus speed with the Intel 440BX chipset gave only a 1-2% speed improvement. Referring to this article though is entirely misleading, as the testing was done on what is now old technology motherboards and relied on Pentium 200 MMX, Pentium II 300 Mhz, Celeron 300 Mhz, AMD K6 300 and Cyrix/IBM 6x86MX 200 Mhz processors for testing. Obviously, processors, motherboards and memory have made great technological strides since this article was written 2 ½ years ago.

Therefore, in answer to this question, can performance be gained by making the move from a 66 Mhz bus to a 100 Mhz bus, in a word, yes! Just remember that performance improvements are collaborative, not singular. To reach performance peaks, motherboards, processors and memory should be mated and assembled based upon the performance desired.

Is there a performance gain when moving from a 100 Bus to a faster 133 Mhz Bus?

As was the case with moving from a 66 Mhz bus to a 100 Mhz bus, over the last year or so, the technology news pundits have been saying that moving from 100 Mhz bus to 133 Mhz bus speed on the Intel 440BX chipset gives no speed improvement for normal business applications, and only shows a small gain for applications such Photoshop and Netshow encoding. To substantiate their point, these pundits refer to another article released by Tom’s Hardware on October 5, 1999. Relying on this article, they (the pundits) state rather emphatically that the new 133 Mhz bus motherboard chipsets utilize faster but “more expensive memory technologies called PC133 and PC800 RAMBUS”, and that there is no reason to upgrade.

We wonder sometimes if these people actually read what they publish. In fairness to Tom’s Hardware, the article referred to was a review of Intel’s i820 chipset, which happens to be one of the more problematic chipsets ever released by Intel, and a comparison with the older Intel BX chipset and VIA’s Apollo Pro 133. The article was not a discussion of the 133 Mhz bus, PC-133 memory or Rambus.

So, as was the case with a move from a 66 Mhz bus to a 100 Mhz bus, so is the case with a move from the 100 Mhz bus to 133 Mhz, is there a performance gain, by all means, yes! As you travel throughout the performance sections of our Web site, you will see how these bus speed increases effect overall system performance.

Summary

Improving system performance is not an issue of purchasing a faster processor, purchasing a new motherboard that supports either a 100 Mhz or 133 Mhz bus speed or even replacing your old 66 Mhz or 100 Mhz memory with PC-133, Rambus or even the latest and greatest, DDR memory. It’s an issue of all three! To realize noticeable performance gains, the motherboard, processor and memory (not withstanding the hard drives) should be properly mated to achieve the gains or performance desired.

Many people spend hundreds of dollars on upgrades and new computers based purely upon presumption, and most times they are disappointed with the end result. Be wary of the pundits that post several pages of theory about new motherboard, processor and memory combinations that produce phenomenal figures when they run games such as Quake. Performance tests running Quake and other similar games prove nothing more than the fact that the chosen combination plays Quake well, but says nothing about how well WordPerfect 2000 or Office 2000 will function. First choose what it is you want to do with your new system, then build it with the components that will produce the desired results.