3-3-3-2-1T DDR2 is worth it!

[DEVIL K-ce]

Some AMD systems with CR 1T

Who have better latency

[Zeus]

Integrated memcontroller helps a bit.

[marauder16]

Integrated memcontroller helps a bit.

yeah, a 'bit'

oh, and I noticed that I get like cca 50MB/s more Read in Vista x64 than in XP x86, that's very nice

[3oh6]

Some AMD systems with CR 1T

DDR3 on Intel can beat that



just need a little more CPU power than normal...520FSB PL6 doesn't hurt either
and no...this isn't 24/7 stable. i am not claiming it is so i don't want to hear those guys start whining about that, i am just having a little fun.

[The0men]

LOL I love that. Damn if anything I thought DD3 would have been worse. Well then I guess I'm just waiting for prices to take a fall.

[3oh6]

LOL I love that. Damn if anything I thought DD3 would have been worse. Well then I guess I'm just waiting for prices to take a fall.

yeah, DDR3 is so much fun to play with. here is another decent Everest bench at a reasonable CPU frequency, still mighty nice numbers. i am just waiting on an X38/X48 motherboard to really unleash some of the DDR3 i have

[Spyrus]

Performance level on P35 makes the difference.Not the timings only.
Try lower your dividers some of you to see the differences

[The0men]

Yer but we are talking about the latency before. And the 680i is wicked for getting low latency, usually much better than P35. I did'nt think DDR3 even on a P35 would have so little latency. The read and write can be done with DDR2 on a 680i but not with that low latency.

And when I lower deviders yer my read, write and copy go up, but I get latency up to 60ns.

[Rattle]

Hello!

I am very pleased with the 3-3-3-2-1T DDR2 times...
This is from an old 680i chipset.
Only 420MHz RAM, on a 3.6GHz C2Q / 450 FSB

Everest MEM


I bet, on the X38 (that can handle +600FSB), a DDR2 running @ only ~480MHz 3-3-3-2-1T will perform just like (or better) then the 700MHz CL5 that I have seen here...

Can anyone, with that kind of setup, try it?

[]'s
Simps

your write beats me but read is SLOW also my latency is much better, is it because I have a dual? I am at 3-3-3-12 2T mind you.

[KTE]

Might pick up 680i to see bandwidth with these DDR2 modules stock voltage. It's known to be very low anyway, much lower than P35 can get.

3-2-2-1 1T 450MHz should be good. Phenom gets plus 11k read DDR2-1066 5-5-5-15 CR2 with ganged mode in Sandra (EVEREST doesn't support it properly) which is double what an Intel CPU/chipset would get at that spec.

[The0men]

Might pick up 680i to see bandwidth with these DDR2 modules stock voltage. It's known to be very low anyway, much lower than P35 can get.

3-2-2-1 1T 450MHz should be good. Phenom gets plus 11k read DDR2-1066 5-5-5-15 CR2 with ganged mode in Sandra (EVEREST doesn't support it properly) which is double what an Intel CPU/chipset would get at that spec.

Why bother with stock volts, we all know how wicked awsome the 680i overvolts RAM lol. You can get +3.6v out of them.

[KTE]

Why bother with stock volts, we all know how wicked awsome the 680i overvolts RAM lol. You can get +3.6v out of them.

On P35 2.3V will get me 1350 5-5-5-15/18, 2.35V 1350 5-5-5-5 and 2.2V will get me 1200 4-4-4-4 just fine. That's good enough for me to check bandwidth possibilities to compare with other chipsets. Not higher because I killed two of them at just 2.3V and 2.35V a few weeks ago. One died with just 20 mins at 2.35V BIOS/2.31V real even while DIMM temp was 11C and stock it is usually 41-44C.
DIMMs are great but there's nowt to gain for me to push them. Next customer can do that in about 2 weeks..

[The0men]

O ok, I have no worries running +2.5v for benching. Normally the 680i boards overvolts dimms anyway, and dont droop or change, if anything they raise under load, which is probably why they clock betterwhen comparing a P35 against 680i at 2.3 volts bios You would have to put bios to about 2.225V with a 680i to run 2.3v real. It will depend on how far vender differes from reference design though.

[Alexxx650]

[marauder16]

here's what I've got in like 10 mins with my Cellshock DDR1000 CAS4, using 2.25V:

this chipset doesn't like 1T, whatever I do I can't even boot to windows with it at 1T, and I'm not giving my RAM more than 4 Driving strength

EDIT: your nforce680 result is kicking my board's butt, Rattle

[railmeat]

here's what I've got in like 10 mins with my Cellshock DDR1000 CAS4, using 2.25V:

this chipset doesn't like 1T, whatever I do I can't even boot to windows with it at 1T, and I'm not giving my RAM more than 4 Driving strength

EDIT: your nforce680 result is kicking my board's butt, Rattle

i have both DFI ICFX3200 and dfi 680i,and my findings with both boards running my Cellshock DDR1000 CAS4, using 2.25V also.the 680i can hit 1t on any speed to about ddr900 without flaking out without crazy voltage hence not realistic for long term usage.but the 680i seems to stably shine (10k+ read) running 2.2v running under ddr 850 1t for 24\7 use and runs AMAZING.but the dfi 680i has some issues including no boot issues sometimes and dreaded code 88 rendering the board unbootble without bios chip swap.shame cause the board could be so much more if some bugs were ironed out plus its sli.

onto the DFI ICFX3200.the board def does not score as well ram wise but it will hit just about anything(speed\timings)as the 680i does 1t (timings\speed) wise all day,weird.as far issues with the board i have found none so far,it runs so much cooler without any volt drop to cpu,nb,or ram.unlike the 680i,s very erratic under\overvoltage issues.the rd600 is %100 stable with anything i throw at it and completely stable at that.it runs cooler,clocks the same and has a overall better feel to a overclock vs the 680i without question.i really wish the 680i did not have the voltage issues im thinking that cause this hidden instabilty ppl have with it vs the rd600.my 680i is in my closet now with a replaced bios chip i have yet to throw back in my rig but i really like the rd600 and im not in no hurry to swap the boards again.there was nothing like getting a great clock 3.6 and good ram timings\speeds(4\3\4\10 1t 800=10k read) only to have my 680i one day just code 88 me out of nowere..

[marauder16]

can you tell me your settings for that 1T setup on -the ICFX? cause I just can't make it run 1T.
on settings I mean: driving strength, voltages, strap, etc.
thnx

[railmeat]

can you tell me your settings for that 1T setup on -the ICFX? cause I just can't make it run 1T.
on settings I mean: driving strength, voltages, strap, etc.
thnx

driving strength ------weak
voltages ----2.26vdim ,1.475 nb
strap---------- 1-1(sync) always or pointless

cellshock ddr1000 5-5-5-15\\\ 2x1 gig sticks d9ghk

ram running at 4-3-3-9 ddr800 1t rd600(icfx)
on dfi 680i 4-3-4-10 ddr 800 1t same settings


things to try-
1st-- lower that overclock! your "E6600 L632A407 w/ Storm Rev.2 @ 3.7GHz 1.536V" is not a healthy overclock to begin with(1.536V?)yet your trying to 1t your ram with it?try my cfg 400x9 with 1.45 vcore,1.475nb, then 1to1 your ram at true 800 1t.i tired 3.73ghz and it was wayyyyyy to squirley for 24\7 gaming use,let alone prime stable.

2nd---once your lower your cpu clock a lil bit(3.6) then try running 5\5\5\15 1t then widdle your timings down.jack your squirley cpu clock back up and watch your ram fail again.

[marauder16]

driving strength ------weak
voltages ----2.26vdim ,1.475 nb
strap---------- 1-1(sync) always or pointless

cellshock ddr1000 5-5-5-15\\\ 2x1 gig sticks d9ghk

ram running at 4-3-3-9 ddr800 1t rd600(icfx)


on dfi 680i 4-3-4-10 ddr 800 1t same settings

thnx, will try that

[railmeat]

please lemme know what happens for you,im glad i could hopefully help u.im
%95 percent sure that higher voltage cpu clocks,along with raised fsb on the ragged edge for stabiltys sake on a water system like mine with your same cpu\mobo that 3.73ghz at 1to1 would put your ram at 824 or so.so right from jump your trying 824 on the ram at 1t?this is another reason to lower your clock at first and see how high ram can run at 1t period for you.try the 400x9 and set EVERYTHING in timings to AUTO except the 1t option and voltage to 2.2.6v and it should boot right up.1.45for nb NO MORE is needed.and i need 1.475 for my e6600 to be ROCK solid at 3.6ghz on my icfx rd600.see what happens at 3.6!!u can fry your 6600 at 1.56v at 3.73ghz later,lol....

[marauder16]

yeah, will try at 3.6GHz, but you can imagine how it's fried @ 3.83GHz with 1.6V (stable in all but orthos, very strange, but can only do 3.7GHz +maybe 2 more FSB to be orthos stable, and it's not the chipset because with x8 multi it goes to 440MHz without any effort)

[marauder16]

hmm, this is strange, it won't pass the code 95 with 1T, no matter I throw at it, 3.6GHz, even with driving strength 4, timmings 5-5-5-15 (and that was desperate), even if I raise the NB + related voltages, I've tried everything, no go.
It seems I get best results with 535MHz CAS4 tight timmings, read = 8820MB/s 13.250s 1M Pi, will tweak that for now.

[railmeat]

hmm, this is strange, it won't pass the code 95 with 1T, no matter I throw at it, 3.6GHz, even with driving strength 4, timmings 5-5-5-15 (and that was desperate), even if I raise the NB + related voltages, I've tried everything, no go.
It seems I get best results with 535MHz CAS4 tight timmings, read = 8820MB/s 13.250s 1M Pi, will tweak that for now.

try BONEE STOCK cpu speed(no o\c!)and then try and see if 1t is possible on your sticksa period,cause your chipset is...

[marauder16]

I will, but somehow I think it's not the sticks

[eva2000]

my 2 cents for 680i chipset