How much each mgz help the score in benchmark

[juin]

I have take in reference Toms benchmark with P4 1.8 and T-bird 1.4.Benchmark Quake 640*400*16 bit and sysmark overall.

Quake
P4 1.8 236.4
P4 1.7 229.7

T-bird 1.4 198.3
T-bird 1.2 185.8

So P4 gain 6.7 FPS vs 6.25 for T-bird for each 100 mghz

Sysmark

P4 1.8 179
P4 1.7 167

T-bird 1.2 134
T-bird 1.4 157

So P4 gain 12 point for each 100 mghz vs 11.5 for T-bird.

There some missing point for T-bird or each mgz increase in P4 have more impact or it just the precision of the benchmark.In the benchmark from X-bit lab (photoshop) T-bird dont seen to increase performance with faster core.

Maybe i get this part wrong.

T-bird do more instruction clock for clock.
(fiction number)If we tell that Tbird do 2.3 instruction per cycle and P4 2 instruction.So in rought theory t-bird suppost to be faster by 15% for each 100mghz.Ex: T-bird do 2.3 instruction per cycle 2.3*100=230 P4 2*100=200.230-200= 30 so T-bird do 30 instruction more that P4 be each 100 mghz.In real life P4 increase faster with each 100 mghz in quake and sysmark and maybe more or it just the case of Quake ans sysmark.

There some question and theory.If you know a part that i have miss tell me.

 

[juin]

**

 

[jollygrinch]

i seem to remember the atlhon beating the p4 in all quake three benches, and at that low res the graphics card should be the limiting factor....so, if i'm wrong, please provide a link to the specific article you are refering to so that i might look at what you are citing.

---Cold pizza, cold coffee, and the THG forum...oh the heartburn.---

 

[juin]

look at all benchmark i have never see a P4 lose at Quake 3.

never even the P3 is near the score of T-bird.

 

[jollygrinch]

ah yes, my bad, i was thinking of unreal

---Cold pizza, cold coffee, and the THG forum...oh the heartburn.---

 

[Makaveli]

Pentium 4 is the champ in Quake 3 there is no doubt about it. But a win by frames per second @ 640x480!
I don't know anyone that plays quake 3 @ that res, so once again Video card is the bottle neck not the Cpu. Today's cpu are more than powerful enough to handle all of today's games. Besides Quake 3 Athlon does do better in most games due to it's more powerful FPU.

I want to see Some quake 3 benchs on the offical A4 when's it's released. I know quake 3 benifits more from SSE than 3DNOW! In the end it doesn't matter cause 200fps from 185fps makes no difference to the human eye. So anyone buying a P4 just for Quake 3 is quite retarded! I'm interested in seeing P4 with DDR. Other than that Current P4 is not worth the money in my opinon. I'm not gonna say anything towards northwood considering it's not out!

CPU choice is by preference all the Battling over which cpu is better is pointless.

 

[jollygrinch]

compairing the fps in q3a of the p4 to the athlon is not realistic however, the article you used shows an athlon using 133sdram, and the p4 (of course) using 800 rdram. the performance of the p4 is saved solely by the speed and bandwidth of the rdram, if you look at the p4 with pc133sdram <A HREF="http://www6.tomshardware.com/mainboard/01q3/010702/pentium4-10.html" target="_new">here</A> (hey a link, what a concept), you will see that when the pentium does not have the aid of it's super (and super expensive) ram, it is soundly thumped by the athlon.

---Cold pizza, cold coffee, and the THG forum...oh the heartburn.---

 

[juin]

You have miss the main point try again

 

[jollygrinch]

perhaps i worded my post poorly as you have missed the point, not me. you're post compares a bench of q3a scores of two procesors, however, my link proves that the procesor of the p4 system doesn't make the difference in the scores, the ram does...try clicking on the link i've provided.

---Cold pizza, cold coffee, and the THG forum...oh the heartburn.---

 

[jollygrinch]

sigh, i meant "your" not "you're" sorry for the grammer.

---Cold pizza, cold coffee, and the THG forum...oh the heartburn.---

 

[FatBurger]

No, I think you missed the point, juin. P4 users always talk about how the P4 wins in real-world applications, but you benchmark at 640x480x16? Nobody plays at that resolution, as pointed out.
Why not use software rendering, and run it at the highest possible resolution and color depth? Say, 1280x1024x32 (so you can do a wider range of games). Then see who wins.

-----------------
Whoever thinks up a good sig for me gets a prize