PDA

View Full Version : Centon DDR2 667 w/ Micron 3.7ns



Charles Wirth
01-17-2005, 04:57 PM
New ES stuff came in and pulling a huge Mhz over everything I had before.

I dropped it in a new DFI 925X-T2 and astounded myself with booting into windows at 298FSB 3:4 @ 4.2Ghz on air 4-4-4-12

This is not an XE chipset...

I backed off on cas to try and break 300FSB and the motherboard doesnt want to do it.

I will try again on the Fatal1ty board later and post some benchmarks. In the mean time, here is some grinding from memtest86

Sorry about the flash, timings different as I was trying different things.

http://www.xtremesystems.org/forums/attachment.php?attachmentid=22839

MaxxxRacer
01-17-2005, 05:27 PM
wow that is some pretty high clocks you got there. granted the timmings are not that amazing, but none the less, ddr 788 is pretty amazing.

MaRtIe
01-17-2005, 05:29 PM
yea thats some sick DDR clocks :D thats the reason ime not a fan off DDR2 is timings but hmmm those speeds convince me

Charles Wirth
01-17-2005, 06:29 PM
The AA8 will not do that speed
The AA8XE Fatal1ty will not do that speed

I need to update the bios on my P5AD2 Premium to support 570J
The DFI 925X-T2 fires up with 570J ES and does not fire up with retail 570J

Yes these are smaller chip with solderpads showing. 2x2 pads on top and bottom.

Charles Wirth
01-17-2005, 07:00 PM
I need to check out the P5AD2-E and another Fatal1ty. Mine is a fatality being severly FSB limited. I am yet to see a "special" bios.

The T2 is a good board and rock solid as far as stability but it lacks an XE chipset. I need to run some benchmarks on it tomorrow.

eva2000
01-17-2005, 08:11 PM
woah nice!

Onepagebook
01-17-2005, 08:44 PM
Friend's DDR 814... :slobber:
http://img58.exs.cx/img58/3650/8146ie.jpg

Skip
01-17-2005, 09:29 PM
damn OPB, you and your friends get some amazing clocks!!!!!!

when dfi comes out with an XE chipset its going to fly, their regular x chipset doing 298!

too bad i don't have enough money for computers to convince myself to pay so much for DDR2.

varzmaster
01-17-2005, 09:55 PM
Friend's DDR 814... :slobber: There's a whole lot of difference between your friend's DDR800 (SuperPI stable) and Fugger's DDR800 Memtest86 stable... don't you think? ;) Not to mention Fugger's tighter timings...

Onepagebook
01-17-2005, 09:58 PM
There's a whole lot of difference between your friend's DDR800 (SuperPI stable) and Fugger's DDR800 Memtest86 stable... don't you think? ;) Not to mention Fugger's tighter timings...


Yup, I think Fugger's is the better one. ;)
Totally agree! :D

Supertim0r
01-22-2005, 05:01 PM
wow nice stuff
i could get this board @ supplier for 231$ canadian
sould i give em a try ? :slobber:

Supertim0r
01-22-2005, 05:20 PM
lucky man with this "ES" stuff !! :banana4:

Supertim0r
01-22-2005, 05:41 PM
is winbond will produce ddr2 chips ? :D

Charles Wirth
01-22-2005, 05:45 PM
Not sure on time frame for winbond DDR2 but I am sure they will switch over at some point.

4263MB/s is higher than my Asus P5AD2-E can produce(so far).

The Stilt
01-23-2005, 04:09 AM
DDR740 3-2-2-5-2 (http://www.akiba-pc.com/570J/DDR740-3225.PNG)

Thats with 2.2V only :)
Second stick broke down, so I had to run in single channel.
16x380 1:1 3-2-2-1-2 could be quite fast ;)

Bennah
01-23-2005, 04:33 AM
oh, great result on the mem speed and timmings there The Stilt.

FUGGER, 394fsb memtest stable, props to you :icon_beer

Kanavit
01-23-2005, 05:26 AM
very nice Fugger, it seems DDR2 doesn't seem to bad as I as thought. even with those high CAS timings, they still peform quite well.

Supertim0r
01-23-2005, 11:06 AM
DDR740 3-2-2-5-2 (http://www.akiba-pc.com/570J/DDR740-3225.PNG)

Thats with 2.2V only :)
Second stick broke down, so I had to run in single channel.
16x380 1:1 3-2-2-1-2 could be quite fast ;)

WOAH !!! DDR740 :slobber:
nice timings :eek:

Charles Wirth
01-23-2005, 06:14 PM
It runs a little hot and could use a fan.

The DFI board is delivering more MB/s than the Asus P5AD2-E

MaRtIe
01-23-2005, 06:34 PM
just got my dfi 925x t2 installed..ill see what my centon ddr2 can do..

fugger did your chipset run hotter than a mother..mine is realy hot and with only 1.7v

OMG something in your computer is actually hot? :D

Supertim0r
01-23-2005, 09:02 PM
OMG something in your computer is actually hot? :D
ROFL !!

The Stilt
01-24-2005, 01:19 PM
:D

DDR800 3-2-2-1-2 (http://www.akiba-pc.com/570J/OCZ-4200EB.PNG)

P4 3.8E E0 (@ 14x multi)
Asus P5AD2-E Premium (VDDR modded, Bios 1003 Final)
2x 512MB OCZ PC2 4200 EB Platinum (Micron 3.7ns) 3-2-2-1-2 / 2.32V

For God sake :slobber:

:D

The Stilt
01-24-2005, 02:12 PM
and its still only 30s pi.

Yep, but it isn´t patched like the one showed in OPB post ;)

althes
01-24-2005, 02:41 PM
Nice runs, looks like I may have to go spend some money.

quicksilverXP
01-24-2005, 03:09 PM
Man.... I wish we can see more Sandra scores of these near DDR800 shots. I can get to near DDR800, like 290 MHz FSB with a 3:4 divider (so memory is around 375) but Sandra reports much lower bandwidth and efficiency as well. This is on a P5AD2-E....using the same chips as PCice and Fugger...

Rancidelephant
01-26-2005, 05:15 PM
and its still only 30s pi.

lol cant quite believe im posting this but it is *ONLY* 4.2Ghz :p

Charles Wirth
01-26-2005, 06:08 PM
Once you go past 4.3Ghz sandra and everest fall victim to the 2^32 bug

Here is a few screen shots for ya.

299FSB 3:4 Read

http://www.xtremesystems.org/forums/attachment.php?attachmentid=23304

299FSB 3:4 Write

http://www.xtremesystems.org/forums/attachment.php?attachmentid=23305

299FSB 3:4 Unbuffered (notice bus bandwidth)

http://www.xtremesystems.org/forums/attachment.php?attachmentid=23307

This is a 1:1 screen shot, CPUZ reads incorrectly. I could just say yea DDR900 but that would be BS.

http://www.xtremesystems.org/forums/attachment.php?attachmentid=23308

Here is a 8M running 5.1Ghz 1:1 340FSB 4-4-4-11 (SPD) stock voltage on DDR2 and CPU -100c

http://www.xtremesystems.org/forums/attachment.php?attachmentid=23306

Xassius
01-28-2005, 04:28 PM
:D

DDR800 3-2-2-1-2 (http://www.akiba-pc.com/570J/OCZ-4200EB.PNG)

P4 3.8E E0 (@ 14x multi)
Asus P5AD2-E Premium (VDDR modded, Bios 1003 Final)
2x 512MB OCZ PC2 4200 EB Platinum (Micron 3.7ns) 3-2-2-1-2 / 2.32V

For God sake :slobber:

:D

VDDR mod eh? Care to show me how do it? :)

hipro5
01-29-2005, 03:11 AM
Not sure on time frame for winbond DDR2 but I am sure they will switch over at some point.

4263MB/s is higher than my Asus P5AD2-E can produce(so far).

Those DDR2 looks promising........BUT I have almost 4385MB/s at 267MHz and 2-5-2-2 with 2 x 512MB Twinmos BH-5 sticks.......SO I think that it's not time yet for me to "change".......... ;) :D

hipro5
01-29-2005, 03:20 AM
Once you go past 4.3Ghz sandra and everest fall victim to the 2^32 bug

Here is a few screen shots for ya.

299FSB 3:4 Read

299FSB 3:4 Write

299FSB 3:4 Unbuffered (notice bus bandwidth)

This is a 1:1 screen shot, CPUZ reads incorrectly. I could just say yea DDR900 but that would be BS.

Here is a 8M running 5.1Ghz 1:1 340FSB 4-4-4-11 (SPD) stock voltage on DDR2 and CPU -100c


Now they're some GOOD results there......... :D

The Stilt
01-29-2005, 10:10 AM
VDDR mod eh? Care to show me how do it? :)

Solder a 15k ohm variable resistor between RT9202 chip sixth leg and ground. This will give you almost same value that you set in bios. Trimming down the resistance will give you higher vddr. Watch out because the voltage will raise EXTREMELY fast after 2.3V... You can find this chip under those yellow VDDR capasitors.

saaya
01-29-2005, 10:25 AM
hmmmm no more 3-2-2? :(

The Stilt
01-30-2005, 12:21 PM
It´s a shame that with current Asus boards cpu won´t overclock as high ad DFI or Abit boards do. This is probably caused by too weak cpu power regulation. Adding couple capasitors to cpu circuit increased overclockability from 4200MHz to 4460MHz with 570J cpu.

I have noticed same thing when I used LN2 with P5P800. No matter what the temperatures and voltages are, any of my chips havent´t run stable over 5.2GHz, even 6GHz+ shots are possible.

Maybe I should add some more capasitor, and see if it helps ;)

Overall a great board, but that shutdown-restart issue is simply ridiculous.

btw. DDR840 with 3-2-2-1-2 settings was quite easy.
That took only 2.34V :) After that motherboard start to limit fsb, but Vchip mod should fix it ;)

saaya
01-30-2005, 05:11 PM
wow! ddr840 with 3-2-2? :eek: can you post some benches?

hmmm did you meassure vcore while the cpus are under load?
hmmm a friend of mine told me he could build some kind of special psu with incredibly stable rails to increase vcore stability, if you want i can ask him about it again, hes building tesla coils and all kind of that stuff

Xassius
02-01-2005, 04:39 AM
Solder a 15k ohm variable resistor between RT9202 chip sixth leg and ground. This will give you almost same value that you set in bios. Trimming down the resistance will give you higher vddr. Watch out because the voltage will raise EXTREMELY fast after 2.3V... You can find this chip under those yellow VDDR capasitors.

*me happliy gets our solder pencil :p: *

Micron DDR2 is really voltage sensitive....but my old one died at 2.48v on my AA8 Duramax :mad:

I'll be careful this time :)

saaya
02-01-2005, 10:49 AM
so your sure the mod wasnt the problem?
meassure vdimm and vref under load and make sure it doesnt fluctuate too much, if it does the mod might be the problem...

Xassius
02-01-2005, 10:57 PM
Solder a 15k ohm variable resistor between RT9202 chip sixth leg and ground. This will give you almost same value that you set in bios. Trimming down the resistance will give you higher vddr. Watch out because the voltage will raise EXTREMELY fast after 2.3V... You can find this chip under those yellow VDDR capasitors.

A few questions
1. I realise that you are running 2.34v for your DDR2 - how much resistance for the resistor between RT9202 pin 6 and ground? 14k? 13k?
Also, where to measure vDimm?

2. vChip mod, can you suggest what kind of capacitors (I have heaps :p: ) and where to solder?

3. Couldn't you solder parallel mosfets (same part) to the existing mosfets?

4. I notice that vCore drops from 1.69v to 1.60v (only reading from bios though) from idle to load on my Gallatin (not even prescott)...will this hamper overclocks? What about a vDroop mod?

Xassius
02-10-2005, 02:27 AM
bump...please? :)

PicaXuu
02-10-2005, 07:15 PM
Look at my results please:

http://www.cyclopnet.pt/~nsl69249/316Sandra1.Jpg

http://www.cyclopnet.pt/~nsl69249/316Sandra2.Jpg

http://www.cyclopnet.pt/~nsl69249/317Sandra1.Jpg

http://www.cyclopnet.pt/~nsl69249/318Sandra1.Jpg

What do you think of the bandwith ??? :confused:

My CPU is a 560j E0 watercooled...

I also would like to know which is the max Vcore it is safe to use to run 24/7 ???

Charles Wirth
02-10-2005, 07:22 PM
Very nice bandwidth

What memory and mobo are you running picaxuu?

PicaXuu
02-11-2005, 02:17 AM
Asus P5AD2-E with 2x512 Corsair XMS2 pc5400...

And my question about the CPU voltage please ???

Xassius
02-11-2005, 02:23 AM
Asus P5AD2-E with 2x512 Corsair XMS2 pc5400...

And my question about the CPU voltage please ???

hmm...

I think with decent watercooling, (MCW6002 or above...etc..etc with dual 12cm rad)....1.5v is pretty/very safe, with limits of 1.65v for 24/7

PicaXuu
02-11-2005, 02:58 PM
I'm actually running at 1,5v !!! :)

But I can go without any problem to 1,65v for 24/7 ??? :confused:
Are you sure ???

It would be a really really good news to me...

A pic of my rig with 3 rads:

http://www.cyclopnet.pt/~nsl69249/Rad-Li87.jpg

iboomalot
02-11-2005, 04:06 PM
with low timings like that maybe AMD needs to start looking at DDR2 now its coming around.

Xassius
02-11-2005, 04:17 PM
with low timings like that maybe AMD needs to start looking at DDR2 now its coming around.

low timings of 3-2-2 on ddr2 is old!!!

as I said before, 1.65v would be pushing the limit for 24/7...I personally would go for a max of 1.6v

PicaXuu
02-11-2005, 04:50 PM
OK Xassius and thanks ! :)

I will try 1,6v and we'll see what I can do... ;)

iboomalot
02-11-2005, 05:17 PM
never said its was new but at those speeds it sure is