2.8b @ 3.6 - benches & piccys within

[PiLsY]

Easy as pie...

These 2.8 C1s are damn good cpus. However, im realising now why intel have upped the power specs for the 3gig chip. This IT7 cant really take it over 3.5ghz. Voltages start going a bit wonky and I get huge low dips. Also everything on the board is overheating so heatsinks everywhere :).

Anyway, it went straight in at 3.5ghz default volts. Because of voltage issues I had to go up to 1.675v to maintain 3.6ghz stability. Anything over that is no go no matter what voltage - theres just too many spikes going through the power lines and too much heat coming off the board.

Pish to all that, time to cut to the chase.

2.8b
FPO L228A584
Step - SL6HL
Pack Date : 08/26/02

Ill have to spread the screenies out due to a lack of any FTP on my behalf :).

First off cpu id....

[PiLsY]

Next some sandra....

[PiLsY]

Some more sandra....

[PiLsY]

A bit of western digital SE raid power ;). Dont forget this is at 33mhz fixed PCI - at 44mhz itll be a lot higher :p.

[PiLsY]

And some Pi results...

[PiLsY]

And finally to round things off a little bit of a before and after comparison.

173/230mhz 2/7/3/2/Enhance and 2950mhz (2.26b).

3D Mark (standard XP SP1 Ti4600 drivers - det 29.85 I think - no tweaks and no overclocking).
1024 res - 13246 3dmarks
640 res - 16445 3dmarks

171/228mhz 2/7/3/2/Enhance and 3601mhz (2.8b).

3D Mark (standard XP SP1 Ti4600 drivers - det 29.85 I think - no tweaks and no overclocking).
1024 res - 14013 3dmarks
640 res - 18041 3dmarks


Same as above for PCMark 2002

2950mhz 2.26b -

7301 CPU
7053 Memory
1580 HDD (2 x maxtor 20gb Raid 0)

3600mhz 2.8b

8915 CPU
7695 Memory
1756 HDD

The PC Mark Raid bug surfaced again, ignore the HDD scores. Overall pretty happy. Not as much of an improvement at 1024 res as Id have liked, but the 640 res test went WAY up. All these benches above were run one after another without a reboot inbetween and no tweaking whatsoever.

Im going to have a bash with the 40.52 dets and see what theyre like now. Didnt like 40.41 at all, hopefully these will proove a bit better in games and for DVD Image Quality. Should get a nice 3D Mark improvement as well.

PiLsY.

[Marci]

Madonion.com & guru3d.com showing Det 40.71...
I've just run em thru at 1.9, 200x9.5 blah de blah (ya know my specs by now)... compare url: HERE (http://service.madonion.com/compare?2k1=4591290)

[PiLsY]

Well, 40.52 and my first sight of 20k - even if it was only at 640 res ;).

Still, I just got the #2 Ti4600 score on the ORB (only Opp above me) without even clocking or tweaking the gfx card :).

Click me (http://service.madonion.com/compare?2k1=4592009)

Chuffed...

The 1024 res score was a particularly dull 15741 3D Marks. Not bothered upping it, not gonna get anywhere with that. Im always crap at 1024 benches :(.

Ill give those new dets a shot marci. Be interesting to give a head to head compare.

PiLsY.

[majormav]

nice goin pilsy ,,,,,, get mine tomorrow hope to get same batch etc as you looks like them western dig hdd are hot property been doing a bit of prepping on the board as well in anticipation some 2.8 goodness with my sapphire rad 9700 pro should be interesting

[Jupiler]

Those are some impressive numbers, congrats.
Those WD's SE really are fast.

[SKATAN]

Pilsy

what cooling are you using ?


btw nice OC ;) :toast:

[Pedro Rocha]

Originally posted by PiLsY
..
3600mhz 2.8b

8915 CPU
7695 Memory
1756 HDD

..


Hello PiLsY

That is quite strange I get much better memory scores with the DDR at just 440 (see scream capture bellow)

Are you sure that you are using the 3:4 memory ratio ?

What sticks do you have ?

(my is Corsair XMS3200 v.2.1 - Cas 2.5)

[PiLsY]

Thats because youve got a radeon 9700 m8 :). It doesnt just test the memory - it does the graphics as well. The radeon is worth about another 1000 points on top of a ti4600 score on the memory test in PC Mark.

Im convinced somethings up with my pifast result though - it looks about 30 seconds too fast :p. Just mailed someone in the know about it to see if theyve got any ideas whats happened.

Memory is at 228mhz cas 2 tras 7 trcd 3 trp 2 Enhance on btw.

PiLsY.

[Edit] - running Apacer PCB with Winbond 5ns BH chips @ 2.8v atm. Just about to try 2.9v and 172fsb 3:4

[Pedro Rocha]

Originally posted by PiLsY
Thats because youve got a radeon 9700 m8 :). It doesnt just test the memory - it does the graphics as well. The radeon is worth about another 1000 points on top of a ti4600 score on the memory test in PC Mark.

Im convinced somethings up with my pifast result though - it looks about 30 seconds too fast :p. Just mailed someone in the know about it to see if theyve got any ideas whats happened.

Memory is at 228mhz cas 2 tras 7 trcd 3 trp 2 Enhance on btw.

PiLsY.

[Edit] - running Apacer PCB with Winbond 5ns BH chips @ 2.8v atm. Just about to try 2.9v and 172fsb 3:4

Yes, your PIfast result is not correct ..for compare, check this http://213.232.90.30/showthread.php?s=&threadid=206939&pagenumber=3
I'm the Numero UNO there, of course..;)

You can download it here (http://img.hexus.net/hexus_pifast.rar)
Simply unrar and run the hexus_pifast.bat file and wait.

[gandalf027]

[I'm the Numero UNO there, of course..;)

[/B]

Humbleness is a very good quality that some people do not have :D

[mike.elmes]

PiLsY,

Why does your system score 8129mips@3.6ghz as stated above when Pedro Rocha is getting only 7181mips with his @3.62ghz ??

Are they 2 different versions of Sandra? The difference is almost 1000 mips...what gives??:confused:

[Creative]

You impress me more and more each day PiLsY

:banana:

[PiLsY]

Possibly down to memory timings mike - not sure in all honesty.

Cant get that hexus pifast to run. It quits just after the huge divisor result and the log states :
"Program : PiFast version 4.1, by Xavier Gourdon
Computation of 10000000 digits of Pi
Method used : Chudnovsky
Size of FFT : 1024 K
Physical memory used : ~ 61163 K
Disk memory used : ~ 0.00 Meg

==================================
= ERROR =
==================================

Wed Aug 28 10:53:22 2002


P does not have the right sign in GiantStep

==================================
= DEBUG INFORMATION =
==================================

--------------------------------
Giant step is on, value=1
n0=333751, n1=335652
DoComputeA = 1"

No idea what any of that means. Im still getting the odd voltage drop. May have to look into getting one of those PC Power PSUs imported.

PiLsY.

[PiLsY]

Originally posted by Pedro Rocha


I'm the Numero UNO there, of course..;)



Not any more youre not ;).

[Raptor]

great going pilsy...

i'm waiting for a 2.5A later on, and i got a borrowed Prometeia just to play with it! :P

[PiLsY]

I was debating the 2.5a, but in all honesty I need to be in the 170 - 174mhz area to get the most from my memory. Between 228 and 232 is a seriously sweet spot on this stick as I can run TRP 2 at these speeds. The 2.5a would most likely top out in the 160 - 165mhz area which would cost me heavily on mem bandwidth with the IT7. A 4:6 divider would make this another story alltogether, but ill have to wait for my next board before I get that :).

PiLsY.

[Raptor]

you're right, in the "DDR point of view"... with the P4T533-C (witch is a wonderfull motherboard!) 156MHz of FSB is her maximum... that's why i choose the 2.5a...

[DDA]

unless these C1s are really different, I find yr PI result out of reality

will u pls repeat for us the bench ???

tks

[macci]

nice going Pilsy :)

DDA, which pi result do you mean?

[DDA]

Originally posted by macci
nice going Pilsy :)

DDA, which pi result do you mean?

super pi, 19 iterations, 43s

[macci]

C1 is quite a bit faster than the old NW steppin in that particular test. A stock 2.8Gigger on a RDRAM mobo completes the test in 51secs :)

[DDA]

Originally posted by macci
C1 is quite a bit faster than the old NW steppin in that particular test. A stock 2.8Gigger on a RDRAM mobo completes the test in 51secs :)

tks Macci

[Pedro Rocha]

Originally posted by DDA


super pi, 19 iterations, 43s

I'm getting 44s Super PI, 19 iterations, with DDR a 450 2-7-3-2 - and CPU at 3500mhz

Of course the result is worst if use 1:1 ratio and CPU at around 3.700mhz (my maximum)

:toast:

[DDA]

doing 52s nw 2.53 (B0)@3486 with it7, mem 3:4 (DDR 488), I was
... in doubt

[zakelwe]

Since the C1 came out and fixed the " unknown " flaws in the P4 chip the Super Pi scores are very very impressive.

I like the bit in the help document where it says :-

" By using single CPU of HITAC S-3800/480 vector supercomputer at the computer centre, University of Tokyo can generate 1 Million decimal digits of pi within 5 seconds and 33.55 million decimal digits of pi within 4 minutes. How about your personal computer time? "

Well, we're getting there.

Regards

Andy

[PiLsY]

Pedro - I can do 3.5ghz at default voltage here m8 (1.47v actual under load) :). The problem is my IT7 cant handle the power the chip is pulling. Everything is getting ultra hot and its basically randomly unstable at any speed over 3.5ghz.

Its acting a lot like the P4T 533 actually....

Vcore dips and rises constantly - most of the time too fast for mbm or the bios monitor to pick up, but occaisionally you see whats truly happening. Im talking about dips and rises of up to 0.15v here. 3497 is 100% stable and voltages are fine. 3518 and voltages are a little shaky. 3540 and above and all hell breaks loose.

I need a board with some VERY solid voltage regulation. Im considering the new Soltek P4X400 board at the moment, but I quite like the look of the Leadtek i845E board thats just come out. Decisions decisions.

Any suggestions anyone?

PiLsY.

[Charles Wirth]

DDA, can you please remove your "Nature" bugged score from page 1. It is obvious that it is nature bugged. I can post one just like it with a nature way beyond my capabilities too.

I can bug out the nature to 138+ FPS too (beyond my capabilities) but I refuse to post scores with knowing that nature bugged out.

[Pedro Rocha]

Originally posted by PiLsY
Pedro - I can do 3.5ghz at default voltage here m8 (1.47v actual under load) :). The problem is my IT7 cant handle the power the chip is pulling. Everything is getting ultra hot and its basically randomly unstable at any speed over 3.5ghz.

Its acting a lot like the P4T 533 actually....

Vcore dips and rises constantly - most of the time too fast for mbm or the bios monitor to pick up, but occaisionally you see whats truly happening. Im talking about dips and rises of up to 0.15v here. 3497 is 100% stable and voltages are fine. 3518 and voltages are a little shaky. 3540 and above and all hell breaks loose.

I need a board with some VERY solid voltage regulation. Im considering the new Soltek P4X400 board at the moment, but I quite like the look of the Leadtek i845E board thats just come out. Decisions decisions.

Any suggestions anyone?

PiLsY.

Yes m8, I experience similar problems, but olny after 3.577 (170FSB).

I experience a good evolution in terms v.core stability from IT7 to IT7-2, but still no perfect at very high speeds.

Im my case, a very well regulated Antec PSU help a lot, is crucial that the 12v never ever go bellow 12.0 on full load (my is 12.28 on ilde).

You can find a good guide for regulate the Antec pots See this MBM scream capture, a compare with your vcore, 12v and 5v line:

:cool:

[mike.elmes]

Originally posted by Pedro Rocha


Yes m8, I experience similar problems, but olny after 3.577 (170FSB).

I experience a good evolution in terms v.core stability from IT7 to IT7-2, but still no perfect at very high speeds.

Im my case, a very well regulated Antec PSU help a lot, is crucial that the 12v never ever go bellow 12.0 on full load (my is 12.28 on ilde).

You can find a good guide for regulate the Antec pots See this MBM scream capture, a compare with your vcore, 12v and 5v line:

:cool:
Pedro,
I have an IT7 and prometeia with a 2.66 arriving next week.My psu is an ANTEC 550 truepower , will I need to pull it apart
and fix the +12 volt rail too. On my currant system a Vapochill, I am only getting around 11.7 volts, but my 5 v = 5.48
This Vapo has never in 2 years had a 12v above 11.7
:(

Its looking alot like the new P4's over 3 ghz will have to have a mobo that fixes the voltage requirements. I guess this is why intel has requested changes to mobo manufacturers power handling specs.

[gandalf027]

The new Antec TruPower don't have pots for you to regulate. I don't know if yours is from the latest generation that have the active PFC .. mine 480w have. :mad:

[PiLsY]

Its not the power supply pedro :). 460watt enhance 0246 rev 2.1 - never seen lower than 5.03 and 11.98 on any system AMD or intel at any speed :). I can comfortably power an 80watt TEC as well as my whole system at 3.5ghz without the voltages going any lower at all. The only thing thats unstable is the vcore. Its a board thing, not PSU. No amount of POT tweaking or modding will solve it, only a new board with better power regulation will sort it out.

[Raptor]

wow, that really is a lot of fluctuation... but it seems that the IT7-Max2 is a lot better in what matters with the Vcore...

At least my P4T533-C raises the Vcore, doesn't lower's it! :)

[Pedro Rocha]

Originally posted by PiLsY
Its not the power supply pedro :). 460watt enhance 0246 rev 2.1 - never seen lower than 5.03 and 11.98 on any system AMD or intel at any speed :). I can comfortably power an 80watt TEC as well as my whole system at 3.5ghz without the voltages going any lower at all. The only thing thats unstable is the vcore. Its a board thing, not PSU. No amount of POT tweaking or modding will solve it, only a new board with better power regulation will sort it out.

Yes my friend PiLsY, you PSU values are very good..

The vCore prob only can be solved my motherboard redesign, I hope Abit solve this in the next revision of IT7 - the IT7II Version 2.0 - that should be out in a few weeks this ABIT Revised its Pentium 4 Mainboard Plans (http://www.xbitlabs.com/news/story.html?id=1033165631)

Also you temperatures values are very similir to mine, but how can be possible to have a low of -105 ºC, maybe a bogus sensor reading.

[PiLsY]

The IT7 sensor cant read below -24c Pedro - as soon as it hits goes below that the sensor reports -105c.

You can confirm this by setting you FSb to 100mhz and you vcore to say 1.3v (dont worry it works fine ;). My record is 1.13v actual at 2.1ghz STABLE for pifast, prime, super pi etc etc :P. Quite nice actually being able to run that low voltage. Would be handy on air for a super silent system.

FYI this cpu will run 2.8ghz stable at 1.24v actual. Quite impressive huh?

PiLsY.

[OPPAINTER]

Originally posted by PiLsY
The IT7 sensor cant read below -24c Pedro - as soon as it hits goes below that the sensor reports -105c.

You can confirm this by setting you FSb to 100mhz and you vcore to say 1.3v (dont worry it works fine ;). My record is 1.13v actual at 2.1ghz STABLE for pifast, prime, super pi etc etc :P. Quite nice actually being able to run that low voltage. Would be handy on air for a super silent system.

FYI this cpu will run 2.8ghz stable at 1.24v actual. Quite impressive huh?

PiLsY.

PiLsY,

How cold does your evaporator get when you get -24 cpu temps??

OPP

[PiLsY]

Evap temp at idle (depending on voltage) is between -47c and -42c. Under load ive seen as low as -31c before, but that was on a warm day at 3.75ghz and 1.82v actual running the super pi 16m test.

PiLsY.

[EDIT] Idles lower than -24c all the way up to 3.65ghz @ 1.67v actual. Load temps are quite poor though comparitively. can get as high as -3c @ 3.77ghz / 1.8v actual under 100% load (pifast test 0).

[Pedro Rocha]

Originally posted by PiLsY
Evap temp at idle (depending on voltage) is between -47c and -42c. Under load ive seen as low as -31c before, but that was on a warm day at 3.75ghz and 1.82v actual running the super pi 16m test.

PiLsY.

[EDIT] Idles lower than -24c all the way up to 3.65ghz @ 1.67v actual. Load temps are quite poor though comparitively. can get as high as -3c @ 3.77ghz / 1.8v actual under 100% load (pifast test 0).

More or less the same in the temps Evaporator - idle at 1.6v around -43 and -36 at full load

But the CPU temperatures in IT-Max2 never went up of -8 with 1.725 or -13 with 1.6v. Idle always -24 no matter how much voltage I use (was playing UT2003 a few minutes ago and the case temp went up to +45, but the CPU never pass -13)

On the IT7 that I have before this IT7-2, the temperatures at 1.7v came close to 0 at full load and -18c at idle.

I don't know why the difference, but I suspect there is a slight improvement on mobo sensors.

Now, on the IT7-2, I don't like the motherboard components high temperatures, at 3600 I measured +53 with thermal probe - that’s a lot!!


:flame:

[Creative]

On my temporary p4b533 and 2.66 @ 3.2 with ram at 213 2-6-3-3 I get 52 seconds.

I cant get the screenie as its not on this pc and my rig is down for the next week....:(

But lookin at these results here that seems not right?
hmmmmm

[macci]

Yeah that seems very slow.
even my 2.5G@3.1G (124FSB! LOL :D) does it in 48 secs (http://www.solidhardware.com/macci/nw/25G_spi_3102_stock.gif).

[DDA]

Originally posted by macci
Yeah that seems very slow.
even my 2.5G@3.1G (124FSB! LOL :D) does it in 48 secs (http://www.solidhardware.com/macci/nw/25G_spi_3102_stock.gif).

your sSpec ????

i see yrs says stepping 7 and mine 4

[Raptor]

of course DDA, that's a C1 stepping!

[macci]

its a C1 (SL6GT) 2.5GHz (25x100)

[toofast]

Hi PiLsY heres a piccie of my 2.66b @ 1.26v-core over 3 gigs :D http://www.boomspeed.com/toofast/temps2.JPG

This is the lowest i`ve seen my evaporator temps .
http://www.boomspeed.com/toofast/lowevaptemp.JPG
:D

[majormav]

well just got the prometeia today and started to play with it and wow cant believe it . Ive played with some stuff in my time but this takes the biscuit . I just cant get used to how low a voltage it takes for massive overclock still going lower as well .. ive also been informed after a conversation with a good freind (pilsy) from xtreme forums that the temps and voltages could be lower than they are so i cant have a good a contact as i should .. however some very early results below .. However the voltages are still all over the place with a 431w enermax (looks like thats the next upgrade )

http://www.majormav.pwp.blueyonder.co.uk/pwp/promsand.gif

[Bushboy]

LOL look at that multimedia score! 43783/17516!!!

And Pilsy your CPU arithmatic benchmarks is extremely high! All other 2.8s I've seen OCed ge around 7100 and you got 8100:eek: Did you run it again and see if you get similar results again?

[sysfailur]

Originally posted by Bushboy
LOL look at that multimedia score! 43783/17516!!!

And Pilsy your CPU arithmatic benchmarks is extremely high! All other 2.8s I've seen OCed ge around 7100 and you got 8100:eek: Did you run it again and see if you get similar results again?

BWHAHAHAH!!!! I noticed that! Awesome!

[PiLsY]

The Arithemtic didnt stand out as high so ive not run it again since. Will do in a bit though (could do with waking up a bit first thought :D).

Nice multimedia mav ;). Gotta love sandra when it does that :D. PCMark can do the same thing occaisionally - ive seen a cpu score of well over 50000 before :p.

PiLsY.