Pentium 4 a POS ?

[Muddy]

I just got a 2.4 GHz P4 and ran some tests against my old 1 GHz P3 using a processor intensive application.

Surprisingly, the P4 is only 38% faster. I was SHOCKED. I ran similar tests back when i migrated from P1 to P2 and then from P2 to P3. Each time the processor was more than 10 times as fast!

I would have expected to more than double my performance based on clock speed alone.

*** is going on with the P4? is it just a piece of crap?

[siyan]

what test?

[JungleMan]

Originally posted by Muddy
I just got a 2.4 GHz P4 and ran some tests against my old 1 GHz P3 using a processor intensive application.

Surprisingly, the P4 is only 38% faster. I was SHOCKED. I ran similar tests back when i migrated from P1 to P2 and then from P2 to P3. Each time the processor was more than 10 times as fast!

I would have expected to more than double my performance based on clock speed alone.

*** is going on with the P4? is it just a piece of crap?

P4 is less efficient than the P3, clock-per-clock, but 38% seems quite low to me. what app is it?

I, for one, would have bought an Athlon system :)

[Sastraxi]

Justin: You've said that so many times, it's implied :p

[JungleMan]

I can always put it in my signature :p

[Yonatan]

You made sure to not get a motherboard that uses SDRAM, right? :eek:
P4 = must use DDR, or RDRAM if you're rich - SDRAM drastically lowers its performance.

[Zaei]

If its some sort of loop benchmark, its quite obvious why the p4 doesnt perform as expected. Its pipeline is so much longer then a p3's, its not even funny. Each time you execute one of the various flow control operands on the chip, the pipeline gets flushed, and you start all over again. Applications optimized specifically for a p4 will perform FAR better then they would otherwise.

Originally posted by Jungle-Man
I can always put it in my signature :p
I think ill put this in my sig ;)

Z.

[Muddy]

Originally posted by Yonatan
You made sure to not get a motherboard that uses SDRAM, right? :eek:
P4 = must use DDR, or RDRAM if you're rich - SDRAM drastically lowers its performance.

it has ddr

[Muddy]

ok,

i made a really simple loop test:


Private Declare Function GetTickCount& Lib "kernel32" ()



Private Sub Command1_Click()

a = GetTickCount
For i = 1 To 100000000
j = i
Next i
b = GetTickCount
MsgBox b - a
End Sub


and the P4 (2.4 GHz) runs about 2.2 times as fast as the P3 (1 GHz).

So there seems to be something less efficient than simple looping in the P4.

Ive got some ADO stuff in the app that showed only 1.38 performance increment ... wonder if thats it ...

[Muddy]

Originally posted by Zaei
If its some sort of loop benchmark, its quite obvious why the p4 doesnt perform as expected. Its pipeline is so much longer then a p3's, its not even funny. Each time you execute one of the various flow control operands on the chip, the pipeline gets flushed, and you start all over again. Applications optimized specifically for a p4 will perform FAR better then they would otherwise.


I think ill put this in my sig ;)

Z.

how do you optimize for the P4??

[Zaei]

Originally posted by Muddy
how do you optimize for the P4??
Your compiler would do it, mostly.

However, take a read through this:
http://www.cfxweb.net/modules.php?name=News&file=article&sid=630

Z.

[Zaei]

Be advised however, that VB hides so much from you that youll never know if your optimizations are even doing anything.

Z.

[Yonatan]

A simple loop test, especially one written in VB, is a very weak test, and will not reveal the true power of your system.
I suggest a professional benchmark program, such as SiSoftware Sandra (http://www.sisoftware.co.uk/), to get real results. (The "Standard" edition of this program is a freeware.)
If you still have your old computer(s), run the benchmarks on them as well, so you can compare. If not, compare against one of Sandra's "reference" results.

I believe the latest version is optimized for P4.