Pentium 4's FP ability is lower than Athlon64 1 time

  • Thread starter Thread starter yu_wing_sin
  • Start date Start date
  • Tags Tags
    Time
Click For Summary

Discussion Overview

The discussion revolves around the floating-point (FP) performance of the Pentium 4 compared to the Athlon 64 and other processors. Participants share their experiences with CPU performance testing, particularly focusing on FP capabilities, architectural differences, and the implications for various applications.

Discussion Character

  • Exploratory
  • Technical explanation
  • Debate/contested
  • Meta-discussion

Main Points Raised

  • One participant reports that their Pentium 4 1.6GHz has lower FP performance compared to a Pentium 3 1GHz and an Athlon 4 1.33GHz, suggesting the Athlon series is a better choice.
  • Another participant attributes the Pentium 4's lower efficiency to Intel's design choices that prioritized clock speed over processor efficiency, noting that subsequent updates have improved its performance.
  • Some participants discuss the architectural differences, such as the number of pipelines, with one claiming the Pentium 4 has 20 pipelines, which is later corrected to refer to its 20-level pipeline structure.
  • There are claims that the Pentium 4 can achieve better performance per MHz in specific applications like Prime95, but this requires utilizing SSE2 instructions.
  • One participant questions whether the performance benefits from SSE2 instructions are primarily due to the compiler or require programmer intervention.
  • Another participant argues that while the Pentium 4's SSE2 performance is better for certain tasks, it does not significantly enhance basic floating-point operations, which are critical in scientific applications and gaming.
  • There is mention of a class action lawsuit against Intel regarding the performance claims of the Pentium 4 compared to the Pentium 3, with a request for more information about the lawsuit.

Areas of Agreement / Disagreement

Participants express a mix of agreement and disagreement regarding the performance characteristics of the Pentium 4 and Athlon 64. While some agree on the Pentium 4's inefficiencies, others defend its performance in specific contexts. The discussion remains unresolved regarding the overall superiority of one processor over the other.

Contextual Notes

Participants reference specific performance metrics and architectural features without consensus on their implications. There are also discussions about the impact of SSE2 instructions that remain conditional on application context.

Who May Find This Useful

This discussion may be useful for individuals interested in CPU performance comparisons, particularly those evaluating the Pentium 4 and Athlon 64 for gaming or computational tasks.

yu_wing_sin
Messages
73
Reaction score
0
My computer is Pentium 4 1.6Ghz, I found some interesting discoveries recently. I used a CPU tested software Standard 2001 to test my P4 1.6Ghz's FP ability. I found that my P4's FP ability is not good as the P3 1 Ghz.

P3 1Ghz has 13xx millions FP/s , Athlon4 1.33Ghz has 18xx millions FP/s, but my P4 1.6 Ghz only has 8xx millions FP/s!

P4 is very slow and very bad on FP efficacy.

How do you think about this? I think the Athlon series are the better platform's choice.
 
Computer science news on Phys.org
Intel made a conscious decision with the P4 to chase clock speed at the cost of processor efficiency. The decision was partly due to some engineering constraints that made it difficult to push the p3 core any higher. Updates to the p4 core since have made it roughly equal to the p3 again in efficiency.

However, new technologies such as dual core (two processors on the same chip) mean all of this is becoming irrelevant: For the first time since the P3 and Athlon were released, there has been a sudden and massive increase in processing power - with AMD coming out way ahead.
 
Yes! I agree with you.

P4 has 20 pipelines, but P3 and Athlon64 only have 12 pipelines. Therefore the Athlon64's branch prediction efficacy is better, and it has more workload per period, so the Athlon64's FP mark is higher. I rathan than to choose Athlon64. I am going to upgrade to Athlon64.
 
yu_wing_sin said:
P4 has 20 pipelines

I am quite certain that the p4 does not have 20 pipelines. Perhaps you mean stages?
 
Townsend said:
I am quite certain that the p4 does not have 20 pipelines. Perhaps you mean stages?

Sorry, I made a wrong. It should be P4 has a 20-level pipeline constructure.
 
P4 has a low efficiency unless you take advantage of the SSE2 instructions.
There are applications (like Prime95) where the P4 gives better performance/MHz than any other CPU, by a large margin (about 150% compared to an Athlon running at the same clock speed).
 
flava said:
P4 has a low efficiency unless you take advantage of the SSE2 instructions.
There are applications (like Prime95) where the P4 gives better performance/MHz than any other CPU, by a large margin (about 150% compared to an Athlon running at the same clock speed).

As far as taking advantage of the SSE2 instructions, is this accomplished by the compiler or is it something the programmer has to do? Or I imagine it's probably some combination of the two...

Anyone know?
 
Its mostly a function of the compiler, but I'm sure the programmer has to have a little to do with it.
 
flava said:
P4 has a low efficiency unless you take advantage of the SSE2 instructions.
There are applications (like Prime95) where the P4 gives better performance/MHz than any other CPU, by a large margin (about 150% compared to an Athlon running at the same clock speed).

Yes, you are right. The p4's SSE2 efficacy is better than Athlon 64, but SSE2 is only clearly to improve the efficacy on graphics and sounds, but SSE series and MMX can not benefit the basic float points' operations. This difference between Athlon64 and P4 is clearly to appear in science study, smart robots, handwriting input, etc., also games. But the float points' operation proportion in games is only occupying a few. So the games' performance in P4 sometimes will exceed Athlon64.
And I admit the P4's dual buses operate in simultaneity, it can has better performance. But you may be wrong in a point, "about 150% compared to an Athlon running at the same clock speed", the Athlon64's speed is PR value.
 
  • #10
Ei, I want to upgrade my computer to Athlon 64 3200+ or 3000+ very much, they are very suitable for me to view DVDs, VCDs and make websites. Only the woman is still hasn't come, I am very discontent, I have waited for many monthes, I really don't know what slow transport she is using. I will not content her intentions if she is too late. I know much about this, she has selfishness.
When I got the money, I will spend much money to make a powerful computer by DIY, a Pocket PC, a smartphone, open a shop for selling computers and digi products... And the most important is I can have money to spend for amorism with my girlfriend.
 
Last edited:
  • #11
The Pentium M is a pentium III with a great big cache and outperforms the
pentium IV with a lower clock, and with much less power consumption. The
Pentium M is a good machine, comperable to the Athlons. I own them both.
 
  • #12
flava said:
P4 has a low efficiency unless you take advantage of the SSE2 instructions.

AMD Athlon64 and Pentium BOTH have up to SSE3 multimedia instructions... and in my opinion AMD is greater than Intel. at least for gaming.
 
  • #13
yu_wing_sin said:
My computer is Pentium 4 1.6Ghz, I found some interesting discoveries recently. I used a CPU tested software Standard 2001 to test my P4 1.6Ghz's FP ability. I found that my P4's FP ability is not good as the P3 1 Ghz.

P3 1Ghz has 13xx millions FP/s , Athlon4 1.33Ghz has 18xx millions FP/s, but my P4 1.6 Ghz only has 8xx millions FP/s!

P4 is very slow and very bad on FP efficacy.

How do you think about this? I think the Athlon series are the better platform's choice.

There was a class action lawsuit filed against Intel for this very reason. Intel touted the P4's as "more powerful" when the reality was the P3's outperformed the P4's for quite some time. Ealry adopters of the P4 got screwed.
 
  • #14
There was really a class action suit aginst Intel for that? I didn't hear about that one... Do you have a link?