[Pifast]350 vs 725 vs 740 @ 1.3 ghz

[Thorry]

I just thought it would be nice to see the different Dothans out there on their stock FSBs compared at the same clock speed to see cache impact and FSB impact.

So I got my P4P800-SE, one of my CT-479 and put in the 350 running at 1.3 ghz (which happens to be stock for this CPU) since the celly doesn't allow any MP changes this was the speed to go with.

I popped in the 725 to see what's the difference between 1.3/100/1MB and 1.3/100/2MB in Pifast.

Then I popped in the 740 to see what's the difference between 1.3/100/2MB and 1.3/133/2MB in Pifast.

So it's not a question of which CPU clocks the best but just what exactly is the difference in performance between the different blends.

Hopefully this is usefull to somebody.

Simply put: the 1 MB extra cache gives you 2 secs Pifast @ 1.3 ghz, the 33 mhz extra FSB also gives you 2 secs Pifast @ 1.3 ghz.

http://www.thorry.net/pm/celeron350.GIF
http://www.thorry.net/pm/pentium725.GIF
http://www.thorry.net/pm/pentium740.GIF

[vtx_]

nice comparison.. are you running these in a desktop or laptop?

[Disposibleteen]

i would say desktop, changing out the chip on a lappy would be too much work.

[Thorry]

P4P800-SE as stated, so desktop.

Changing the CPU in my notebook is absolutely no problem, it takes me about 1 minute to remove the current CPU and install another.

Most Centrino notebooks have a socket 479 on the motherboard allowing for easy switching of CPU's. However my notebook is locked @ 100 mhz FSB so it is completely useless to install another CPU. Also there is no PCI/AGP lock so the FSB cannot be raised over 5 mhz or the VGA freaks out.

I tried several pin mods and even modified the pins on the clockgen in my notebook, but nothing worked, it simply always runs @ 100 mhz FSB, or it doesn't run at all.

I will try to bench later today @ 1.6 ghz (raising the FSB across the board) and check how the difference holds up at higher clock speeds.

[caater]

is it possible to run also superpi and 3dmark2001 for comparison?

[toledo]

is it possible to run also superpi and 3dmark2001 for comparison?

Yes , would be good :D .

[Thorry]

As promised higher clocks.

Please note the 740 has a disadvantage because of the non-existing MP of 10.5

SuperPi will be the same difference as PiFast. Also 3DMark will have to wait till I'm back from vacantion.

Pics:
http://www.thorry.net/pm/celeron350-133.GIF
http://www.thorry.net/pm/pentium725-133.GIF
http://www.thorry.net/pm/pentium740-166.GIF

[Pongi]

No difference clock/clock between a B1 Dothan and a C0 Dothan, do you happen to have a Banias Celeron M? Like a 320? (1.3Ghz). Or a Banias Pentium M?

I have a C-M 310 (1.2Ghz) I can compare at 1.6/133FSB.

[Mike@Portugal]

In PCMark 04, 05 and 3Dmark there is absolutly no perceptible diference between C-M and P-M at the same clock, i have already tested both of them a 2.4Ghz.

In SuperPI there is a 6/7secs diference and in Pifast a diference of 7secs.

Both C-M and P-M where tested on the same board (Asus P4P800SE) with the same memory at similar speed and timings.

Bottom line is: C-M is as good as P-M for daily use at the same clock speed... more... it as good as A64 in the same clock TOO!!

[Pongi]

In PCMark 04, 05 and 3Dmark there is absolutly no perceptible diference between C-M and P-M at the same clock, i have already tested both of them a 2.4Ghz.

In SuperPI there is a 6/7secs diference and in Pifast a diference of 7secs.

Both C-M and P-M where tested on the same board (Asus P4P800SE) with the same memory at similar speed and timings.

Bottom line is: C-M is as good as P-M for daily use at the same clock speed... more... it as good as A64 in the same clock TOO!!

The Only reason I'd consider a Pentium M C0 stepping is for hitting 200Mhz FSB or more, Some folks around here seem to still have problems with the C0 stepping Celeron M's getting over 160FSB on the CT-479.

Wish I could spare $85 on a 350J to test it for myself though.

[Mike@Portugal]

The Only reason I'd consider a Pentium M C0 stepping is for hitting 200Mhz FSB or more, Some folks around here seem to still have problems with the C0 stepping Celeron M's getting over 160FSB on the CT-479.

Wish I could spare $85 on a 350J to test it for myself though.

Yes, my Celeron 370 (c0 stepping) didn't do more than 160x15 (2400Mhz) stable, and i don't believe that any celeron-m c0 does more than 2400Mhz.

That's why i bought a P-M, not because it's faster for benching in SuperPI or PiFast, i bought it for it's better overclocking potencial.