Overclocking a P4 1.8A w/Pack date: 8/29/02...

[TenaciousPee]

Same 'ol cliche: "Long time listener, first time caller..."

I've been watching everyone else's insane 1.6A/1.8A overclocks for quite some time now. Finally, it's my turn...


SYSTEM SPECS
P4 1.8A 400 mhz FSB
Pack date: 8/29/02, Philippines
PC: xxxxxxxxxxSL63X
Version: xxxxxx-004
Abit IT7-MAX (bios 1.00)
Thermaltake Volcano 7+ /Medium Speed/AS2
1 x 512 MB Corsair XMS3200C2 running@2-3-3-6 unless otherwise listed
RAIDMAX 350W
GeForce4 4400 (300/300) 30.82
Win2k SP3

All tests were preceded by 30 minute burn-in loop w/SiSoft Sandra CPU arithmetic test (Highest Priority, 100%util)
update!
- Went back and took SupermanCK's advice and used UT2003bench instead. Thanks for the tip.

Systemp is 31c, idle temp is 42c, load temp (so far) is 50.5c (while running UT2003bench) per Windbond




3DMARK2001SE
1800 mhz 100fsb 266DDR = 9726(1.5v, 1.45 actual)
2168 mhz 120fsb 322DDR = 10980
2400 mhz 133fsb 356DDR = 11444
2520 mhz 140fsb 373DDR = 11821
2610 mhz 145fsb 388DDR = 12028
2710 mhz 150fsb 402DDR = 12231
2800 mhz 155fsb 412DDR = 12456 voltage to 1.55, (1.50v actual), failed memtest86, adjusted memory timings to 2-3-3-7, passed memtest86.
2880 mhz 160fsb 426DDR = failed, adjusted memory timings to 2.5-3-3-7, failed 3dmark
2880 mhz 160fsb 322DDR = failed 3dmark
2880 mhz 160fsb 322DDR(2-2-2-6) = 12262, voltage to 1.575v(1.54v actual)
2970 mhz 165fsb 330DDR = failed memtest86
2970 mhz 165fsb 330DDR = adusted memory timings to 2.5-3-3-7 passed memtest86,failed to boot into Win2k
2970 mhz 165fsb 330DDR= 12298, voltage to 1.60v(1.57v actual)
3010 mhz 167fsb 330DDR= failed memtest86
3010 mhz 167fsb 221DDR (4/3)= 12008, left Prime95 running all night, just passed 10 minutes ago.





More updates in a few hours; I gotta take my girl out to dinner...
update!
- Dinner done, she's drunk and asleep...
Me: "Honey, I'm going to go benchmark the computer now..."
Her: "Mmmmm, that's nice..."

 

[SupermanCK]

TenaciousPee: as you go up higher in FSB...you should break it up even more to like 5mhz increment and then to 1mhz...
another work of advise...do the burn in later...get the benchmarks until system starting to become unstable...
use tests like UT2003 demo...I am very sure this game is stressing your system
keep it up...looks very promising...

 

[TenaciousPee]

From what I've seen in the oc'ing forum in the past few weeks, the fact that I'm still at default voltage with such a high clock speed is a good thing, correct? However, it looks like I'm about to run out of room with the RAM. My question is, should I just go to 1:1 and accept the diminishing returns, or tweak it down to 2.5-3-3-7/8 and try to get it to 426ddr or 445ddr? Anyone else out there getting their XMS3200c2 to run@426-445?

 

[Bobbyeyes]

"- Dinner done, she's drunk and asleep...
Me: "Honey, I'm going to go benchmark the computer now..."
Her: "Mmmmm, that's nice...""

you see... overclocking can be a babe magnet...
it really does drive the girls crazy!

looks like you have a great chip there.. you are undervolted
w/ a little more juice you mite get a great overclock


Superman.. get back to work .. he he

 

[CrazySaint]

You have it @ 2.8 still on default voltage?? Dude, you're lookin' 3GHz straight in the eye! Yes, absolutely reduce the mem ratio to 1:1 to see how high the chip will go, worry about how high the RAM will go later.

 

[adlep]

The spects on that chip look similarly to my PIV 1.8....
It is version -004 like mine, and it has been made in Philippines as well.
I can do on it 2.52Ghz = 18x140Mhz fsb at default v.
Later on computer hangs.... but it is because my RAM - CRUCIAL PC 2100 is nowhere near the stick you have.
But it seems to me that the 004 has a lot of o/c potential, your post just confirms that...
It was a good investment wasn't it?
Welcome to the forums.....

BTW: I have Volcano 7+/ASII combo too...
I run it at low settings though.....
With this case it is an adequate setting.

 

[NewSc2]

Originally posted by: TenaciousPee
From what I've seen in the oc'ing forum in the past few weeks, the fact that I'm still at default voltage with such a high clock speed is a good thing, correct? However, it looks like I'm about to run out of room with the RAM. My question is, should I just go to 1:1 and accept the diminishing returns, or tweak it down to 2.5-3-3-7/8 and try to get it to 426ddr or 445ddr? Anyone else out there getting their XMS3200c2 to run@426-445?

I'd tweak it down to less aggressive timings. I'm currently running my Samsung PC2700 on 135mhz fsb (4:6 cpu-mem ratio), and it's on default settings 2.5-4-4-7-1. I'd have to go down to the 4:5 multiplier to get 2.5-3-3-7-1, and obviously, the performance hit is noticeable.

 

[CrazySaint]

Come on, TP, how's the OC going?? You can't just leave us hanging like this!

 

[THUGSROOK]

really nice chip so far :Q

 

[TenaciousPee]

Sorry for not updating for a while, guys. My weekend filled up faster then expected, leaving me with little time for any more testing. I'll try to get some more done later today. FWIW, I ended up bumping up the voltage to 1.55 (1.5 actual) after failing SiSoft Sandra Burn-in. Since then, no crashes, no errors and passing all stability tests. I gotta admit though that I'm a little dissappointed, despite the fact that technically I am now running at "default voltage"; I kinda liked the idea of hitting 2.8 ghz at 1.5 volts (1.45 actual).

 

[THUGSROOK]

you need to run Prime95.......

 

[SupermanCK]

yup...don't forget memtest86 also...that will double your stability security...

 

[Dolemite1]

where did you purchase the chip and is it oem or retail. I'm about to buy a 1.8A myself to use with corsair xms3200 and Asus p4b533E

 

[TenaciousPee]

Ran prime 95 for 5 hours (thanks for the advice, Thug) while I was away at class and it passed w/flying colors.

Load temp was 57c, idle was 43c, systemp was 38c, and room temp was 38.5c. My take is that the load temp is a little high, but is also to be expected considering the environmental conditions present. Memtest86 failed(!), with only one test, test#5, causing any errors. I then bumped the memory to 2-3-3-7 and it passed.


Dolomite1: Don't laugh, but I bought it at Fry's. I've never had a problem taking burnt processors back to Fry's, or burnt RAM, or burnt video cards or... etc. etc. etc. Oh, and the chip is retail.




Up next, 160 fsb...

 

[THUGSROOK]

very nice

 

[TenaciousPee]

2.88 scores are up.

The RAM is holding me back. Or perhaps I'm starting to hit the limitation Thugsrook mentioned in the XMS3000 benchmark thread; something about the 854E chipset not being able to run 166+/- at the 3:4 memory divider...

 

[TenaciousPee]

3 ghz scores are up.:Q



Unless someone can convince me otherwise, I'm not going to push it any further. I hit 3.0 ghz, my goal from the start, but until someone makes a motherboard that can handle the 3:4 divider at 166 FSB, and the RAM that can tolerate the same, I'll just keep the chip running at 2.7-2.8ghz.

 

[SupermanCK]

wow...but your 3ghz is slower than 2.8ghz...and with more vcore...so i guess 2.8 is the sweet spot kinda like mine...

 

[Spicedaddy]

You're only at 1.6V?? Set it to 1.7V and see how high it'll go... If you can get it a little higher, it'll compensate for the slower memory speed.

 

[TenaciousPee]

Not that I claim to be an expert or anything- I'm simply repeating what I've already read in the forums: the main thing that's keeping the chip from performing better at 3.0 ghz vs 2.7-2.8 ghz is the A) the lack of bandwith provided by the RAM simply because it can't go any faster, and B) the limitations of the chipset itself and the fact that it isn't able to handle an FSB greater than 166-170 mhz. Notice how as I got closer to 3.0 ghz, I had more problems with the RAM; I had to run it at 221 mhz DDR just to get it to pass memtest. Go back in the benchmarks and see that I had no problems getting it to run at 333+DDR in previous benchmarks. To me, that says that it's a chipset limitation. Keep in mind that the I845E doesn't officially support 533 mhz FSB, let alone a 664 mhz FSB. Of course, I'm open to other alternative theories as to why my 3.0ghz < 2.8 ghz... but I'm sure Thugsrook could probably explain all this stuff much more eloquently than I.

 

[Spicedaddy]

i845E has official support for 533MHz FSB, that the only difference with the previous i845D (except for the new south bridge). I hadn't noticed you had dropped the mem from 1:1 to 4:3, so you're right, it could be the chipset, but then again, people running 2.26B CPUs around 3GHz are running even higher FSBs. (BTW, at 167FSB and 4:3, your mem is running DDR251, not 221) It might be your power supply also, so upping the ram voltage by .1 or .2 could help.


If it's stable, the only reason it's slower at 3GHz vs 2.8GHz is your memory bandwidth (DDR413 vs DDR251). DDR251 is very slow...


Anyways, I'd try it at 1.7V and see how high it'll go just for fun...

 

[CrazySaint]

Man, that is one awesome chip! :Q Just too bad you don't have a board that can keep up with it

 

[SupermanCK]

how could you ram don't pass memtest at 1:1.....
my pc2700c2 would pass 161fsb @ 2-2-2-5 setting.....
i found out that memtest also stress the CPU...so by upping your vcore can reduce the error created by the memory...try it...i did and i get no more errors

 

[Mikki]

darn nice oc, mate, congrats!