Multipliers & Stability E6600 + mem timings explaination

morgan · 23 Jan 2007 at 10:00

this may be a noob question but which should be more stable / which is better to use if the system is happy?

3.2ghz by:

9 x multiplier, 355 fsb / 2 x system multiplier (mem @ 710Mhz)

or

8 x multiplier, 400 fsb / 2 x system multiplier (mem @ 800Mhz)

this is with e6600 / ds4 / geil PC6400C4

which is a better alternitive given the choice?

also can anyone explain exactly what the CAS 4-4-4-12 numbers mean?

for memory in order to overclock more what can I do with these timings to get more Mhz? I've been searching the forum and it seems that its the last number (ie the 12) that can be changed. ie if I reduce it to 8 the memory will be faster, but more unstable, or increasing it to 15 will slow it down....can anyone explain?

regards

moogle · 23 Jan 2007 at 11:27

im not a pc expert in ram, but i think the 2nd one 800mhz mem seems good, if its stable. as for this cas thing, i've read about it and i still dont understand it.

heres a link http://en.wikipedia.org/wiki/Cas_latency

afaik the first 3 numbers should stay low as possible and for the last one the higher it is the better.

Cob · 23 Jan 2007 at 19:44

Tbh lowering the timings will give very little benefit outside benchmarks.

Keep them at 4-4-4-12 at 8x400. If going near 450x8 you *may* need to raise them to 5-5-5-15 or thereabouts. But in that case you may be better trying 9x400.

Spetznaz · 23 Jan 2007 at 19:57

I'm using an ASUS P5B m/b and cant get it quite stable @ 400 / 8, works fine @ 355 / 9.

Why?...

I really dont know, I have a theory that the memory controller might be being pushed too hard at 400fsb but its hard to explain,see here.

I'm not sure whether this applies to other 965 chipset boards but it could well do.

Junk · 23 Jan 2007 at 20:00

Spetznaz said:
I'm using an ASUS P5B m/b and cant get it quite stable @ 400 / 8, works fine @ 355 / 9.

Why?...

I really dont know, I have a theory that the memory controller might be being pushed too hard at 400fsb but its hard to explain,see here.

I'm not sure whether this applies to other 965 chipset boards but it could well do.

slacken your ram timings a little, or up the ram voltage a little, or both if necessary

Spetznaz · 23 Jan 2007 at 20:05

Junk said:
slacken your ram timings a little, or up the ram voltage a little, or both if necessary

ram is OCZ platinum PC6400 running 5-5-5-15, would have thought this should be fine at 400fsb?, unless somewhere along the line it is being overclocked..

:confused:

Cob · 23 Jan 2007 at 20:14

Yea it should be fine at that. But it probably needs to be running at 2.1v.

Spudgun · 23 Jan 2007 at 21:54

My Asus p5wDH is the same, 9x355 is fine, but it won't have 8x400 at all

Cob · 23 Jan 2007 at 22:12

Have you tried 8x404?

Spudgun · 23 Jan 2007 at 22:15

No, will give it a go now. OUt of interest, though, why should that be any different?

Cob · 23 Jan 2007 at 22:16

Some boards have an fsb hole at 400.

Spudgun · 23 Jan 2007 at 22:26

Won't even post at 8x404. Tried slackening off memory timings to 5-5-5-15 at 2.1v and upping CPU voltage to 1.4 and i get nothing at all.

[email protected] (1.34 actual) works absolutely fine, however

Spetznaz · 24 Jan 2007 at 01:31

Spudgun said:
Won't even post at 8x404. Tried slackening off memory timings to 5-5-5-15 at 2.1v and upping CPU voltage to 1.4 and i get nothing at all.

[email protected] (1.34 actual) works absolutely fine, however

At least I'm not alone it seems.. although it's near stable at 402fsb, but will error after several hours of Priming. What ram are you using?.

morgan · 24 Jan 2007 at 08:54

cheers for the replies guys.

last night had a tinker.

Basically the quickest the cpu would go is 3.20 Ghz stable (@ 1.45V). I did this by trying 2 methods:

1). 9 x 356fsb = 3204Mhz, Mem @ 890Mhz (2.5 multiplier)

2) 8 x 400fsb = 3200Mhz, Mem @ 800Mhz (2 multiplier)

this is a DS4, E6600, G.Skill HZ mem @ 2.1V, +0.1 to chipset and fsb volts.

I think 3.2 is the limit of my chip, whereas the mem could go faster than 890.

Did benchmarks (pcmark) between method 1 & 2 and they were 8204 & 8232 respectively - ie pretty much the same....strange as I thought option 1 would be more responsive?....

any tips to get faster?...the chip temp doesnt really get past 54/56 deg and system temps are <30 deg...

cheers

Spudgun · 24 Jan 2007 at 18:35

I'm using the gskill hz 6400 memory also.

i need 1.425 volts to get 8x400 orthos stable and also need to adjust chipset voltages, but 9x356 is stable at 1.375 so i think that i'm going to stick with that

morgan · 25 Jan 2007 at 08:42

Spudgun said:
I'm using the gskill hz 6400 memory also.

i need 1.425 volts to get 8x400 orthos stable and also need to adjust chipset voltages, but 9x356 is stable at 1.375 so i think that i'm going to stick with that

so the same setup as me? Wierd that these E6300 and E6400 seem to clock better than the E6600...

as for choosing 9x356 over 8x400...is your reason so you can run lower fan speeds/less noise or another reason?

what have you got your HZ memory running @ 2x=712mhz or 2.5x=890 mhz?

Spudgun · 25 Jan 2007 at 18:48

Only difference between our setups is the MB, as i'm on a P5W DH Deluxe. I stuck with 9 x 356 due to the slightly lower temps as i'm a big lover of very, very quiet computing