DDR2 vs DDR3 - ASUS Blitz Formula vs Blitz Extreme...CellShock Rams involved...

[filmbot]

Few tweaks included

What kind of tweaks help with bandwith?

[saaya]

It doesn't matter if one chip is cool or not when we talk about HIGH voltage/current.....EVEN 1sec of heavy load could damage it no matter what....In 1sec it will not even get hot....
You can't say what is safe and what not in such occasions and here's why:
You may feed it with - let's say - 2.3Vdimm and you're "pussing" it too hard OR it has a "by manufacturing problem" (one chip at a million/billion)....IF ONE internal Mos-Fet of the entire SINGLE chip gets damaged, then the whole ram module IS damaged.....and we are talking about hundrents/thousands/millions Mos-Fets into ONE chip.....and Ram modules have 8 or 16 of them....

i know, i know, but you know what i mean :P
a voltage that 99% of the chips can take without degrading is what id like to know... what everybody would like to know i think.

victor said that he is now noticing some degradition on his blue kit from running 2.28v on it... for a few weeks i guess... so 2.25 already seems to be risky for long term use. but in the end thats good news as 2.25v is pretty high, and if there is some degradation after weeks then it makes me think 2v and maybe even 2.1v are safe for 24/7. more or less at least.
this should be good news for corsair and ocz etc who are already selling micron ddr3 rated to 2v or even higher ^^

GOOD question....I had the same Voltage for 1T or 2T.......BUT we have a problem.....and HERE IS THE PROBLEM.....and a BIG ONE....
Feeding a P35 Northbridge chipset with HIGH Voltage IT GETS DAMAGED......YES YOU READ IT WELL.....IT GETS DAMAGED.....
I used to play with 1.74 to 1.81Volts on this mobo and I was OK....BUT after some crushings on 3Ds, it NEEDS 2.01Volts now to work my rams high....
First I though that was the rams produsing errors and crushings so I fed them with 2.34, 2.36, 2.38, 2.40, 2.42, 2.45Vdimm with NO LUCK at all.....
I then fed them with 2.34Vdimm and UPED the P35s Voltage to 1.91V.....I then was OK.....BUT NOT FOR LONG.... Keep upping the P35s voltage so to stabilize my rams, now I NEED 2.01Volts to do so.....DUMN this is NOT a good sign at all.....and we are talkin about 2 - 3 days period of time....

interesting... i noticed something similar...
im not 100% sure but i think i could run more than 1825 dc stable on the old p5k3 i have here. its one of the first boards, we bought it right after it was available. well now it only does 1825 stable with 1.7v..

i assume you put some heavy high rpm and high air pressure fan your chipset, right george? hehehe so heat shouldnt be an issue...

Some S-Sandra now........Actually it reads that I have 12GB of ram..... I don't like Sandra any more coz it aint accurate and it writes whatever it wants......BUT I'll upload it coz......it is full of colours.....

[saaya]

Nice job there George

Since my board can´t do over 1900MHz DRAM CLK in dual channel I tried single channel instead.

I wonder how high the Everest score would be with the same clocks at dual channel.

Few tweaks included

SINGLE CHANNEL

hey stilt!
long time no see man, how are you doing?
try different memory slots, it can make a huge difference.
on one kit the first slots for cha and chb were 200mhz worse than the second slots!

the difference between single channel and dual channel is pretty low for ddr3, i guess its because the fsb a huge bottleneck already, even at 500fsb+.
so running one or two sticks only slightly increases the data you can push through the bus. i think for ddr2 the difference between sc and dc was already rather small, percentage wise, and i think with ddr3 its worse.

in sc you should be able to reach really high speeds.
even the worst board i have here can run above 2ghz stable in single channel and i think cmd 1t.

oh and dont forget, with micron ddr3 less volts sometimes gets you higher, did you try with less vdimm?

[saaya]

What kind of tweaks help with bandwith?

lower trfc, i know george will disagree but imo everything else is chicken sht hehehe

[hipro5]

i assume you put some heavy high rpm and high air pressure fan your chipset, right george? hehehe so heat shouldnt be an issue...

Not warm at all......I have placed an Enzo copper heatsink on it and a DELTA fun 4x4cm some thousands rpms blowing it....

Thanks guys....

[The Stilt]

hey stilt!
long time no see man, how are you doing?
try different memory slots, it can make a huge difference.
on one kit the first slots for cha and chb were 200mhz worse than the second slots!

the difference between single channel and dual channel is pretty low for ddr3, i guess its because the fsb a huge bottleneck already, even at 500fsb+.
so running one or two sticks only slightly increases the data you can push through the bus. i think for ddr2 the difference between sc and dc was already rather small, percentage wise, and i think with ddr3 its worse.

in sc you should be able to reach really high speeds.
even the worst board i have here can run above 2ghz stable in single channel and i think cmd 1t.

oh and dont forget, with micron ddr3 less volts sometimes gets you higher, did you try with less vdimm?

I´m good, thanks for asking

The MCH is simply weak since when the other module is plugged in the DRAM is failing thru the whole address range.

After 1900MHz it is a struggle to make it work with CAS latency under 10 CLK.
The voltage does help at all, but rewriting certain register will help a bit

After rewriting the register there is no problem using lower tCL latency.

The MCH might be damaged like the other George said.
When MCH voltage is set to 1.7V from bios it actually measures 1.82V.

[saaya]

Enzo copper heatsink
DELTA 4x4cm
some thousands rpms

HA! i knew it ^^

I´m good, thanks for asking

The MCH is simply weak since when the other module is plugged in the DRAM is failing thru the whole address range.

After 1900MHz it is a struggle to make it work with CAS latency under 10 CLK.
The voltage does help at all, but rewriting certain register will help a bit

After rewriting the register there is no problem using lower tCL latency.

The MCH might be damaged like the other George said.
When MCH voltage is set to 1.7V from bios it actually measures 1.82V.

hmmmmm
is that on a p5k3?
1.7v is actually 1.8? hmmmm

well on the boards that do 1950+ here i didnt notice any degradation and im running them with 1.7v vmch for 2+ weeks now.... hmmm
i think only few boards can run 2ghz stable anyways, OR you need really high vmch for that, which makes it stable but... not stable for long, lol as it appears more than 1.8 is damaging the chipset

[dinos22]

The MCH might be damaged like the other George said.
When MCH voltage is set to 1.7V from bios it actually measures 1.82V.

wow big difference there

outch

[metro.cl]

I´m good, thanks for asking

The MCH is simply weak since when the other module is plugged in the DRAM is failing thru the whole address range.

After 1900MHz it is a struggle to make it work with CAS latency under 10 CLK.
The voltage does help at all, but rewriting certain register will help a bit

After rewriting the register there is no problem using lower tCL latency.

The MCH might be damaged like the other George said.
When MCH voltage is set to 1.7V from bios it actually measures 1.82V.

Hello Stilt,

Could you please post some measure points? also what registers are you talking about?

Regards

[The Stilt]

Hello Stilt,

Could you please post some measure points? also what registers are you talking about?

Regards

You can measure the MCH voltage from various places around the MCH, but the easiest one is the empty capacitor slot under the MCH.

The DRAM is also being overvolted on my board.
Setting 2.1V from the bios measures 2.18V from the slot.

I have noticed similar action on Asus Commando motherboard too.
The memory is being overvolted by 0.1V compared to the bios setting.

Some of the Commando boards actually undervolted the dram, so I guess this varies between the inviduals.

Maybe Asus is using resistors with too high tolerance, who knows.

It seems that motherboards with "PCB Made In China" print on board have tendency to overvolt the MCH & DRAM unlike the boards with "PCB Made In Taiwan" print.

Atleast the components (coils, caps) are different between Made In China and Taiwan boards.

[saaya]

so its really a different board revision then?
hmmm

[Infa]

My Commando overvolt almost everything...
when set 1.60 etc on MCH it goes 1.65, SB also 0,05
Vdimm +0.05 , And my vcore are all over the place
when i set 1.75 it is 1.70 1.76 is over 1.80 ...really strange board, but it overclocks well... but im glad to switch to x38 maaany new settings and dividers, yum yum

[The Stilt]

Got the board working with DRAM at 2000MHz in dual channel.

ODT was the key

[Infa]

What is ODT?

[3oh6]

nice comparison George...one thing though.

you use the Luna theme for some benches and the Classic theme for others. Luna theme can have as much as 5+ second effect on 32M times...depending on OS setup. so your DDR3 times could possibly be even better using Luna theme which is scary as all hell.

your OS setup might not show as a big a benefit but it might be worth trying...i have always had Luna theme run faster on XP/2K3 and depending on other tweaks it can be a very large difference.

[saaya]

Got the board working with DRAM at 2000MHz in dual channel.

ODT was the key

oi! so it gave you a what, 100mhz boost?
what mem? can you see the batch of the chips?

What is ODT?

on die termination i guess, i didnt know it was possible to adjust, i thought its selfadjusting... interesting

[Don_Dan]

Got the board working with DRAM at 2000MHz in dual channel.

ODT was the key

I assume ODT means on die termination, but: how can you change the setting? Or can you do something else with it? Is it only possible on DDR3? or only on P35 motherboards with a new tool?

[dinos22]

The Stilt drops bombshells every once in a while

now this is a good one

[saaya]

well he only hinted at things so i think asming questions is futile, he would have posted details if he could, im sure. he probably has access to some dev tool to adjust the odt, and yeah, sounds damn nice!

a 100mhz+ boost is very sweet...
i wonder if adjusting odt helps to get the orange slots to work better, so maybe its just an interference problem after all and not the voltage...

[hipro5]

nice comparison George...one thing though.

you use the Luna theme for some benches and the Classic theme for others. Luna theme can have as much as 5+ second effect on 32M times...depending on OS setup. so your DDR3 times could possibly be even better using Luna theme which is scary as all hell.

your OS setup might not show as a big a benefit but it might be worth trying...i have always had Luna theme run faster on XP/2K3 and depending on other tweaks it can be a very large difference.

Thanks M8.....

I do experiment in XP themes but I got almost the same resault by changing them....Maybe it has something to do with the XP or something.....

I'll bench it with the Luna theme and I'll report back.....

[dinos22]

i use Luna Silver as well thanks to before who mentioned this a long time ago over at www.i4memory.com >> don't be shy to visit boys

[eva2000]

Yeah Luna silver for me.. but now with new nlite made winxp pro sp2 disk i'll try luna royale blue and black as well

[saaya]

i thought disabling the themes service all together would help to get some tiny bits of extra performance?

[dinos22]

i thought disabling the themes service all together would help to get some tiny bits of extra performance?

nah they are actually used for Pi strangely enough

[JMKS]

To continue my DDR2 vs. DDR3 comparision (post #28):

I played with DDR3 "as promised" . And the result is as predicted - 4:5 4-4-4-4 i965 is equal with 1:2 8-6-6-x 1T P35 DDR3.


P35 DDR3 1:2 8-6-6-18 1T 6-56-094 [the same subtimings as used by hipro5]


i965 4:5 4-4-4-4 6-56-328 {1,65V NB, probably ~1,50V will do the same}

or


i965 4:5 4-4-4-4 6-56-203 modded1333 [boot @ 401, SetFSB => 400, PMem 10, 24, 03, 06, 09 and run; 1,25V NB]

514,6 FSB 4:5 4-4-4-4 with i965 is possible, unfortunately with worse "modding" [10, 24, 03, 04, 08 - like this, I'm not 100% sure] - the performance is like 4:5 P35 5-4-4-4 only [tested @ 500 4:5 4-4-4-4 16M], a bit more modding or sth and i965 will fly


DDR3 is good [better than DDR2] but only with really high FSB (>450 minimum, to exclude 2:3 5-4-4-4 DDR2 I think ) - that's my opinion .