Completed rock solid @2736Mhz 171fsb 342ddr 2-2-2-6 1.71v...got another 1.6A to play with

[SupermanCK]

SYSTEM
P4 1.6A 400mhz bus 16x multiplier
Costa Rica 05/??/02 SL684
3204A717-0181 - Week 4 of 02
Asus p4b533 i845E (bios v1008) revision 1.03
Swiftech MCX 462U with AS3
Corsair XMS2700 1x 512mb - highest my memory could do is 418ddr @ 2.5-3-3-7
Enermax 550W (FMA) / slowest speed possible
ATi Radeon 9700 Pro - Catalyst 2.3 - 6193
Win2000 Pro SP3
* indicates memory-to-fsb is 1:1


3DMARK2001SE default >>> Vcore >>> Temp room/cpu/mobo (degree C)
*1600mhz 100fsb 200DDR = 9900 1.58 >>> 23.5/37/30 *default voltage*
1600mhz 100fsb 266DDR = 10156 1.58 >>> 22.5/37/30 *default voltage*
2133mhz 133fsb 356DDR = 12904 1.58 >>> 23.0/41/30 *default voltage*
2240mhz 140fsb 373DDR = 13381 1.58 >>> 23.5/39/30 *default voltage*
2320mhz 145fsb 387DDR = 13663 1.58 >>> 21.5/39/30 *default voltage*
2400mhz 150fsb 400DDR = 14084 1.58 >>> 21.5/42/31 *default voltage*
2480mhz 155fsb 413DDR = 14340 1.58 >>> 22.5/41/30 *default voltage*
2560mhz 160fsb 427DDR = 14662 1.62 >>> 23.0/45/31 upped 0.025v in bios
2640mhz 165fsb 440DDR = 14940 1.62 >>> 22.5/45/31 upped 0.025v in bios
*2720mhz 170fsb 340DDR = 14563 1.65 >>> 23.5/43/31 upped 0.075v in bios
*2800mhz 175fsb 350DDR = 14991 1.73 >>> 24.0/48/32 upped 0.125v in bios (not completely stable)


UT2003 D3D w/ HardOCP Benchmark UltraHigh Quality 1024x768 antalus > asbestos > citadel >>> 1280x1024 ant > asb > cit >>> 1600x1200 ant > asb > cit
*1600mhz 100fsb 200DDR = 102.91 > 132.58 > 83.41 >>> 99.93 > 127.88 > 82.63 >>> 86.66 > 115.81 > 82.18
1600mhz 100fsb 266DDR = 105.48 > 137.73 > 86.77 >>> 102.12 > 133.18 > 85.95 >>> 87.61 > 118.06 > 85.03
2133mhz 133fsb 356DDR = 139.59 > 181.79 > 115.86 >>> 120.87 > 158.69 > 112.47 >>> 91.08 > 124.61 > 101.89
2240mhz 140fsb 373DDR = 143.95 > 188.90 > 120.42 >>> 122.42 > 160.80 > 115.90 >>> 90.93 > 124.46 > 103.03
2320mhz 145fsb 387DDR = 148.21 > 192.76 > 124.09 >>> 123.05 > 161.74 > 118.31 >>> 90.99 > 124.51 > 103.70
2400mhz 150fsb 400DDR = 154.42 > 198.95 > 128.52 >>> 124.32 > 163.45 > 120.97 >>> 90.97 > 124.51 > 104.81
2480mhz 155fsb 413DDR = 156.85 > 204.65 > 132.75 >>> 125.22 > 165.29 > 123.57 >>> 90.96 > 124.54 > 105.34
2560mhz 160fsb 427DDR = 160.44 > 209.69 > 137.21 >>> 126.19 > 166.27 > 126.15 >>> 90.96 > 124.61 > 106.10
2640mhz 165fsb 440DDR = 163.92 > 213.78 > 140.99 >>> 126.85 > 167.14 > 127.83 >>> 91.02 > 124.57 > 106.30
*2720mhz 170fsb 340DDR = 162.54 > 208.95 > 136.74 >>> 126.55 > 166.69 > 126.27 >>> 90.81 > 124.54 > 106.14
*2800mhz 175fsb 350DDR = 166.85 > 215.55 > 142.63 >>> 127.81 > 168.08 > 129.06 >>> 90.84 > 124.47 > 106.60


SCISOFT SANDRA memory >>> cpu >>> media || memory time settings (cas#, Ras-to-Cas, Ras#, Cycle Time)
*1600mhz 100fsb 200DDR = 1512 / 1510 >>> 3156 / 845 / 1968 >>> 6371 / 7982 2-2-2-5
1600mhz 100fsb 266DDR = 2004 / 2003 >>> 3115 / 845 / 1968 >>> 6371 / 7991 2-3-3-6
2133mhz 133fsb 356DDR = 2696 / 2690 >>> 4181 / 1127 / 2622 >>> 8497 / 10656 2-3-3-6
2240mhz 140fsb 373DDR = 2808 / 2798 >>> 4293 / 1172 / 2725 >>> 8834 / 11081 2-3-3-6
2320mhz 145fsb 387DDR = 2909 / 2901 >>> 4466 / 1214 / 2823 >>> 9143 / 11475 2.5-3-3-7
2400mhz 150fsb 400DDR = 3012 / 3001 >>> 4709 / 1256 / 2922 >>> 9465 / 11854 2.5-3-3-7
2480mhz 155fsb 413DDR = 3110 / 3104 >>> 4785 / 1297 / 3015 >>> 9784 / 12255 2.5-3-3-7
2560mhz 160fsb 427DDR = 3213 / 3205 >>> 4944 / 1340 / 3118 >>> 10107 / 12661 2.5-3-3-7
2640mhz 165fsb 440DDR = 3325 / 3313 >>> 5170 / 1382 / 3210 >>> 10409 / 13068 2.5-3-3-7
*2720mhz 170fsb 340DDR = 2545 / 2542 >>> 5335 / 1423 / 3310 >>> 10729 / 13445 2-3-2-6
*2800mhz 175fsb 350DDR = 2625 / 2624 >>> 5383 / 1465 / 3407 >>> 11047 / 13845 2.5-3-2-6


HD Tach Random Access Time (ms) > Average Read Speed (Mps) > CPU utilization (%)
Seagate Ultra-3 Wide SCSI U160mb/s, 3.9ms avg access time, 15k rpm
*1600mhz 100fsb 200DDR = 6.6 > 53.45 > 5.6 ***RAID-0 (2) WD 120 SE*** 14.1 > 45.05 > 5.5
1600mhz 100fsb 266DDR = 6.4 > 53.30 > 5.8 ***RAID-0 (2) WD 120 SE*** 14.2 > 45.02 > 5.5
2133mhz 133fsb 356DDR = 6.1 > 53.62 > 4.8
2240mhz 140fsb 373DDR = 6.0 > 53.65 > 4.7
2320mhz 145fsb 387DDR = 6.0 > 53.39 > 4.4
2400mhz 150fsb 400DDR = 6.5 > 53.69 > 4.1
2480mhz 155fsb 413DDR = 5.9 > 53.23 > 4.2
2560mhz 160fsb 427DDR = 5.9 > 53.35 > 3.7
2640mhz 165fsb 440DDR = 6.0 > 53.29 > 4.1
*2720mhz 170fsb 340DDR = 6.7 > 53.64 > 4.0
*2800mhz 175fsb 350DDR = 6.0 > 53.63 > 3.9


STRESS TESTS
Memtest86 all 11 tests 6 loops = passed upto 2512mhz 157fsb 3:4 418ddr
Prime95 (High Priority) looping for 5 hours = passed upto 2720mhz 170fsb 1:1 340ddr 1.65v


P4B533 BIOS SETTINGS
Halt On = No Errors
DDR Voltage = 2.6v
AGP Voltage = 1.5v
AGP/PCI Frequency Lock = 72/36
USB Legacy Support = Enabled
SDram Config = By Spd
SDram Idle Timer = Infinite
Graphics Aperture Size = 128mb
Video Memory Cache Mode = UC
Delay Transaction = Disabled
Serial/Parallel Ports = Disabled
PCI Latency Timer = 64
USB Controllers = 3 Controllers
USB 2.0 = Disabled


Format Proudly STOLEN from THUGSROOK

 

[THUGSROOK]

(notice he started this thread @ 4:13am ~ hardcore overclocker)

 

[SupermanCK]

in the east where the sun will rise...but here in the west is where the sun will fall...
anyway...it's only 1:28AM over here...
see you are not asleep yet at 4:28AM EST - look who's talking

 

[Deskstar]

The "sun" you say? Is that the big yellow ball in the sky that moves slowly around the sky?

By the way there are 182 Forum members on at 3:39am...we are not alone.

 

[SupermanCK]

YELLOW???
i thought the sun is the white thingy ma jiggie circle slowly in the dark black sky

 

[THUGSROOK]

the funny thing is youll still be here @ 4:13am your time running tests

 

[Duvie]

Don't play around...jump to 2.4ghz with 300mhz ddr...All that other stuff is a waste of time...

Also why do you test your oc by ocing the memory and cpu at same time??? You would be better off leaving mem to 1:1 and push the cpu to find its limit first....

My chip was a costa rica 5/16/02...

 

[THUGSROOK]

hehe Duvie ~ as gamers, we want to see the benchies asap!
he knows the limit of his ram and will revert to 1:1 to go higher.

but yea dont waste so much time there SCK ~ go straight to 133:355

 

[SupermanCK]

i like to make u guys wait...good things come to those who wait...

 

[SupermanCK]

day1 finished...me tired...must go bed :Q:frown:

 

[THUGSROOK]

im a little confused here....

"RAID-0 (2) WD 120 SE 14.1ms > 45.05 avg > 5.5%"

im using a single WD 80 SE ...8.7ms > 48.38 avg > 5.0%

? ?

 

[Bobbyeyes]

pci latency value?
you have shown that 96 is good for the p4b533

 

[SupermanCK]

i don't know...14.1 seems right on par...
did you use advance check disk option?
i will mess around with the bios later...
i got bunch of stuff on the RAID-0 now...MP3, Movies, ehem PORNO, etc...
and i think the 80GB is suppose to be slightly faster...since it only has 2 platters...but i don't think that would matter...
did you guys see the review of the new 200GB JBs over at Storage Review

 

[THUGSROOK]

i donno....

im on win98se so i have to use HD Tach v2.52 ~ but still, someones results/program aint right ???
i use a PCi latency of 64 ~ 96 causes Winamp to pause under heavy file transfers.
also ~ i dont use IAA garbage.

? ?

its not very important as you are just running tests to make sure your HDs arent being effected by OCing ~ but i do find it strange. maybe Mikki can shed some light on this as i know hes done tests on both OSes.

 

[SupermanCK]

DONE!
final results
anywhere btw 160-170fsb...to be determine

 

[THUGSROOK]

very nice

howd ya get the RAM to cooperate?

 

[LastRide]

wow,good work SupermanCK.Thats a real nice chip you have there

 

[SupermanCK]

i didn't...you know you can still run some benchmarks...that's why i am adjusting to a lower fsb for memory...
conclusion:
3:4 mem:fsb ratio would give about the same performance increase as ~10mhz in fsb (ie. 150 w/ 400ddr = 160 w/ 320ddr)
if you look at UT2003demo benchmarks carfefully, you will see that the Radeon9700 Pro has a limit...just look at 1600x1200 and 1280x1024 on citadel, even increasing the mhz didn't get anywhere after 2133mhz...it was pretty much constant from 2133-2800...although there were still a slight increase in performance under 1024x768
HD Tach also shows that increasing fsb didn't do jack to the SCSI set up...
last...every 5mhz fsb gives approximately ~300+ pts in 3dmark2k1

 

[Deskstar]

Very well done, indeed.
Congrats on a great job.

 

[THUGSROOK]

UT is very poorly coded ~ then again so was the old UT wasnt it?

 

[Mikki]

maybe Mikki can shed some light on this as i know hes done tests on both OSes.

Yeah, but not enough to offer anything, and my wife threw a bunch of my papers away a while ago "mistakingly", which included a bunch of my results.

So I'm basically useless at this point....

 

[Mikki]

Jeez, I forgot....SCK, that's an awesome oc, well done!

 

[THUGSROOK]

gotta love them 1.6As

looks like SCK will be running at least 166fsb in 1:1 mode for Granite Bay ~ nice job!

 

[SupermanCK]

hee hee...only if you sell me that stick of 512mb ram for cheap

 

[SupermanCK]

prime stress testing now...
so far 170fsb 1:1 340ddr vcore=1.68 (might lower more if possible) with currently 2.5-3-3-6...i tried 2.5-3-2-6...but that failed quickly...i might even try to do 2-3-3-6 which is pretty close to rated speed 333ddr 2-3-3-6