Dothan @ CT-479 - howto and fixes.

Printable View

Show 100 post(s) from this thread on one page

08-14-2005, 03:40 AM
gaddster

Quote:

Originally Posted by Junez

Ok. My 730 boots whit 12x203FBS and its stable whit that. but thats all. i cant set any higher or its not stable anymore. i had used setfsb program to set clocks in windows. temps are 33C idle and 46 full load whit 1.6V.. i had tryed whit lover mems setting. still :banana::banana::banana::banana:ty clocking. any advices?

You could have reached the limit of the cpu. Unlucky if you have. Are you running turbo mode ? Try running standard if you are, it seems to help mine clock more reliably. Is it just locking up ?
08-14-2005, 04:15 AM
Junez

Quote:

Originally Posted by gaddster

You could have reached the limit of the cpu. Unlucky if you have. Are you running turbo mode ? Try running standard if you are, it seems to help mine clock more reliably. Is it just locking up ?

no turbo mode. all standard. looks that im extreme unlucky whit overcloking cpu:s... yeah. it just locks up. looks that 1.5->1.6vcore dosnt have any effect. looks that im going to sell this and get new one and hope the best
08-14-2005, 04:23 AM
bcn

try higher fsb and lower multi :)
08-14-2005, 04:40 AM
Junez

Quote:

Originally Posted by bcn

try higher fsb and lower multi :)

tryed.

and it runs something like this. 9:250. 10:225. 11:215... so its almoust about same speed as 12x201.

i had done vdroop and vdim mod. so vcore is stable... looks that i just dont have luck whit overcloking. or im doing something wrong.
08-14-2005, 04:49 AM
jrw

Try finding the 1M breakingpoint with every multi. That should give you a fair idea of the CPUs max.

Maybe the 875 board doesnt like TCCD memory (which i presume you have).
08-14-2005, 05:32 AM
gaddster

Quote:

Originally Posted by stryg

Try finding the 1M breakingpoint with every multi. That should give you a fair idea of the CPUs max.

Maybe the 875 board doesnt like TCCD memory (which i presume you have).

God don't say that ! I've just bought some TCCD for my p4c800.. I thought it was good stuff for the 875 chipset?
08-14-2005, 05:46 AM
jrw

Quote:

Originally Posted by gaddster

God don't say that ! I've just bought some TCCD for my p4c800.. I thought it was good stuff for the 875 chipset?

I dont know that for a fact, I simply suggested it.

Oldschool BH5 and 865/875 simply loves each other like mad. Thats a fact.
08-14-2005, 06:22 AM
Junez

29s whit 1M is the fastest what i can get. 9x270 gives that. 10x250 givest that too. but those are not stable clocks... so i think that i stay whit 12x203 and wait for water block for northbridge and check does that help at all... "passive cooler is extreme hot"...

and my mems are kingston hyperX 460ddr 2-3-3-7... not BH5. dunno what chips those have. and vdim mod is for 3.3v

Edit: i noticed weird thing. when setting 204-205FBS. my wlan "usb" goes really slow. downloads drops 100kb/s -> 10-15kb/s... when i set it to 206 its normal again....
08-19-2005, 04:15 PM
alexio

Quote:

Originally Posted by ThePredator

Hi there , recently got my P-M 750 , and it's damn fast but i have a very very strange problem with it

ok first of all i'll put my rig that is:

Dothan 750 + CT-479
P4C800 E- Deluxe
2x256 old bh-5 + booster
cooled by vapochill LS

made the u-wire mod and flashed with last bios 1023 , vcore is set in bios at 1.6v , also the temp in bios displays -20ºC

i managed to achieve 3030 Mhz (12x152) at 2-2-2-5 3.4v

so here goes the rare issue , i could get 24s in super pi 1M , thats ok , but when i pass 2M or 8M the performance drop , this example should explain better

Dothan at 3000 (250x12)

1M result--> 25s
2M result --> 1'04s
8M result --> 5'07s

San Diego 3700+ @ 3230 Mhz (320x10)

1M result --> 25.500s
2M result --> 58s (@3216 Mhz)
8M result --> 4'47.500s (@ 3210Mhz)

Dothan at 3050 Mhz

1M -->24s
2M --> 1'02s
8M --> 5'01s

i don't get this.... best overclock on San Diego and i can hardly get 25.433s in 1M , i've got to relax clocks to pass 2M and get 58s , why Dothan is faster in 1M and slower at 2M and 8M ???

the temp is always the same according to speedfan , even on 1M or 8M , what could be wrong??

saludos and forgive my bad English

During the small calculation of 1m most/all data fits in the 2mb L2 cache of the dothan which makes it faster because it has twice the cache of the San Diego. During longer calculation the dothan loses it advantage because it has to use the system memory more during the calculation giving the San Diego an advantage with it's bigger raw calculation power and ondie memory controller giving lower latencies.
08-20-2005, 01:45 AM
dqniel

Quote:

Originally Posted by alexio

During the small calculation of 1m most/all data fits in the 2mb L2 cache of the dothan which makes it faster because it has twice the cache of the San Diego. During longer calculation the dothan loses it advantage because it has to use the system memory more during the calculation giving the San Diego an advantage with it's bigger raw calculation power and ondie memory controller giving lower latencies.

Good insight, I was wondering what caused that issue myself and that seems logical. My 730 hits a wall around 213x12 :(. That's not TOO bad but i was hoping for at least 2.6ghz. Best SuperPi I got was using 2-2-2-5 @ 208x12 which gave me 30.8 sec or so.
08-20-2005, 08:25 AM
wwwww

Quote:

Originally Posted by alexio

During the small calculation of 1m most/all data fits in the 2mb L2 cache of the dothan which makes it faster because it has twice the cache of the San Diego. During longer calculation the dothan loses it advantage because it has to use the system memory more during the calculation giving the San Diego an advantage with it's bigger raw calculation power and ondie memory controller giving lower latencies.

I agree to some extent, the Dothan holds about a quarter of super pi 1M in it's cache whereas the Deigo holds 1/8 so that gives the dothan a major advantage, but in more memory intensive tests (such as 2M, etc) the low latencies from the ondie controller allow it to pull ahead. However, what I disagree with is the statement of bigger raw calculation power. I'm inclined to believe the pM has a better raw calculation power than the Deigo at any given clock (I mean as in both at the same clocks). It has a 9 stage pipe, compared with a 12 stage on the Deigo. It's not jsut the cache that makes it faster. I mean, take a series 6xx vs a deigo at 3GHz. The 6xx can fit double the amount of the test in its cache yet the Deigo will probably do it twice as faster regardless 'cause it has a much shorter pipeline. That's just my 2c.
08-20-2005, 11:23 AM
alexio

Quote:

Originally Posted by wwwww

I agree to some extent, the Dothan holds about a quarter of super pi 1M in it's cache whereas the Deigo holds 1/8 so that gives the dothan a major advantage, but in more memory intensive tests (such as 2M, etc) the low latencies from the ondie controller allow it to pull ahead. However, what I disagree with is the statement of bigger raw calculation power. I'm inclined to believe the pM has a better raw calculation power than the Deigo at any given clock (I mean as in both at the same clocks). It has a 9 stage pipe, compared with a 12 stage on the Deigo. It's not jsut the cache that makes it faster. I mean, take a series 6xx vs a deigo at 3GHz. The 6xx can fit double the amount of the test in its cache yet the Deigo will probably do it twice as faster regardless 'cause it has a much shorter pipeline. That's just my 2c.

Yes I agree that clock for clock the Dothan has probably slightly more raw power. But this was a comparation between a 3ghz Dothan and 3.2ghz San diego. I think a SD at 3.2ghz has slightly more raw calculation power than a Dothan at a 200mhz lower clock.
08-21-2005, 08:17 AM
mrlobber

stryg, please add some SuperPi 1M 26.844 secs on unmodded P4GD1/740/BH-5/stock air here: http://www.xtremesystems.org/forums/...ad.php?t=72164 :D
08-22-2005, 05:58 AM
jrw

Quote:

Originally Posted by mrlobber

stryg, please add some SuperPi 1M 26.844 secs on unmodded P4GD1/740/BH-5/stock air here: http://www.xtremesystems.org/forums/...ad.php?t=72164 :D

Will do asap. I just got back from Berlin but will add it later this afternoon.

BTW, I just received my P4GD1 today! Now I just need a GFX card. :D:D
08-22-2005, 01:12 PM
The_Dare_Devil

1 Attachment(s)

Nice with this thread :D..
I am just waiting for my 750er, and i need som info - do anyone knows the distancepieces, which can make it possible to mount a watercoolinghead through the 4 std. mountingholes?

I have i picture, but i need some dimensions! Does anyone have got these?
08-23-2005, 01:11 AM
Neller

Hi men. I have tvwo qustions:

1.what is bootblock "trick"?
2. Can I use Dothan CPU on ASUS P4C800 DELUXE rev. 1.03 (non P4C800-E deluxe)
08-23-2005, 01:59 PM
Jericho

Quote:

Originally Posted by Neller

Hi men. I have tvwo qustions:

1.what is bootblock "trick"?
2. Can I use Dothan CPU on ASUS P4C800 DELUXE rev. 1.03 (non P4C800-E deluxe)

1: Use the search function as this has been covered in depth.

2: Yes the none -e supports dothan, but will not allow multiply adjustments from BIOS>
08-23-2005, 09:27 PM
bhoward

When overclocking the PM, do the jumpers on the CT-479 matter? I OC'ed my 750 to 2.85g. I am getting ready to replace the stock cooler with a Zalman 7000. I removed the jumper just for fun ... did not seem to make any difference???? Can they be left off??

SuperPi @ 2.80 is 25.82sec.
08-24-2005, 06:53 PM
rivers

Quote:

Originally Posted by wwwww

I don't get why everyone's temps are sooooo high. I get under 30C @ 1.6V with a MCX159 chipset air cooler and only a few degrees above ambient on water.

Maybe the b1s run cooler than the c0s...

*checks*

well Intel says the C0s arelike 27W whereas the B1s are 21W, don't know if that effects it too much once o/ced.

The TDP is more a guideline for sys integrators than a real indication. And is the same for all the 533 family (from 1.6 to 2.26).
IMHO there isn't a different at all in dissipation, after all the voltage is the same as B0. Probably this different results are just dependent on case cooling, Tambient and overvolting (well, and termistors).
08-24-2005, 06:56 PM
rivers

Quote:

I removed the jumper just for fun ... did not seem to make any difference???? Can they be left off??

Umh.. interesting found.
easier to mod in this way, just cut the jumper pins, no solderings or low profile minijumpers.
08-25-2005, 02:49 AM
Unseen

isnt it beautifull???

http://img394.imageshack.us/img394/7845/bolts5yf.jpg
08-25-2005, 04:45 AM
DrJay

Quote:

Originally Posted by wwwww

I agree to some extent, the Dothan holds about a quarter of super pi 1M in it's cache whereas the Deigo holds 1/8 so that gives the dothan a major advantage, but in more memory intensive tests (such as 2M, etc) the low latencies from the ondie controller allow it to pull ahead. However, what I disagree with is the statement of bigger raw calculation power. I'm inclined to believe the pM has a better raw calculation power than the Deigo at any given clock (I mean as in both at the same clocks). It has a 9 stage pipe, compared with a 12 stage on the Deigo. It's not jsut the cache that makes it faster. I mean, take a series 6xx vs a deigo at 3GHz. The 6xx can fit double the amount of the test in its cache yet the Deigo will probably do it twice as faster regardless 'cause it has a much shorter pipeline. That's just my 2c.

The Dothan has a comparatively weak FPU,....after a certain point (Pi @ 2 meg), it's larger cache will lose out to the A64 cpus.
08-25-2005, 05:28 AM
toledo

Quote:

Originally Posted by DrJay

The Dothan has a comparatively weak FPU,....after a certain point (Pi @ 2 meg), it's larger cache will lose out to the A64 cpus.

It makes sense, but not explains why it still kicks A64 in games. I guess games use the same type of calculation .
08-25-2005, 06:08 AM
DrJay

Maybe I was doing something wrong but my Dothan @ 2.6 (260x10) was way slower in 3dmark01 than a Venice I was running between 2836 and 2862....especially in nature. Of course if I had a better Dothan that would run at 2.8+, I'm sure it would win. The substantially lower nature fps I got with my Dothan seemed to bear out the weaker FPU thing. I should have ran the Venice at 2.6, as well and just compared the '01 nature fps.
Did get good CL fps though. :D
08-25-2005, 06:18 AM
toledo

Quote:

Originally Posted by DrJay

Maybe I was doing something wrong but my Dothan @ 2.6 (260x10) was way slower in 3dmark01 than a Venice I was running between 2836 and 2862....especially in nature. Of course if I had a better Dothan that would run at 2.8+, I'm sure it would win. The substantially lower nature fps I got with my Dothan seemed to bear out the weaker FPU thing. I should have ran the Venice at 2.6, as well and just compared the '01 nature fps.
Did get good CL fps though. :D

hum , it is quite strange because Bigtoes running a P4C800 and a Dothan at 2500Mhz /220@2-2-2-x seems to be much faster than that . Here is a test of a Aopen board at 200fsb running a Dothan at 2.1 (slower than an Asus ) . :confused:

http://geocities.yahoo.com.br/marcel...oledo/d3-b.JPG

Show 100 post(s) from this thread on one page