quiekMew - Core 2 Quad Q9550 @ 4550.2MHz - 515 cb Cinebench - R15

[quiekMew]

Testing new board. Thanks to ground1556 for the ram even though it isnt behaving as well as hoped

[bolc]

nice score mate
hwbot seems down but am curious about those ddr2 ram not behaving as you want. we can chat avbout ddr2 timing if you want (and if you have d9gmh khx chips)

[quiekMew]

Hey,

thank you  

i borrowed the sticks from @ground1556. in his p5q deluxe they did 690MHz and 720MHz 5-5-5-15 at 2.3 Volt in single channel respectively . 

I tried to run them in dual channel with several boards (P5Q Pro, EP45 UD3R) but when the cpu is close to maxxed (still stable, just high FSB, around the 500 mark) the sticks couldnt go over 590MHz.

When lowering FSB from ~490FSB to ~420FSB they suddenly could do 690MHz 5-5-5-15 2.3V SPi8M stable in dual channel.

I then tried to up FSb to 500, drop the mem multi down so itd be at 666MHz with same timings and voltage and the system didnt even post. 

After that i tried to see with what FSB each stick can boot on its own and i stopped testing at ~520 FSB. i dropped the multi down so the mem ran at 520MHz 

so, my conclusion is, that (this) ddr2 cant clock as high with a higher fsb

 

have you observed similar things? 

[GeorgeStorm]

On 775 fsb:mem strap can make a big difference on what clocks can be run

[quiekMew]

i definitely used the 400MHz strap to do 690MHz and the 200Mhz one when attempting to do 666MHz IIRC

[bolc]

11 hours ago, quiekMew said:

i definitely used the 400MHz strap to do 690MHz and the 200Mhz one when attempting to do 666MHz IIRC

the ram will do what it can.
imo, not achieving high ram frequency has to do with the chipset not handling high fsb and ram frequency but not the rams.
 

otherwise i had a ep45 ud3p which would not do x 2.4 B (strap333) cas5 with fsb of 540+ ish so had to do 540+ cas4, but another board has no pb doing x2.4B cas5...
which mch, mch ref and ram ref voltages do you use ?

690/720 cas5 (tight?) at 2.3 V is sweet ! really. see if they can do spi32m, nonetheless, very impressive.

[bolc]

for the time being, your adata don t handle 540+ cas4 ?
won t be as good as 650+ cas5 but not far...

Ps: do you play with clock skew for cpu and mch. there are some magic numbers
same with mch ref and cpu gtl ref

[ground]

15 minutes ago, bolc said:

690/720 cas5 (tight?) at 2.3 V is sweet ! really. see if they can do spi32m, nonetheless, very impressive.

Got really lucky with a couple sets of Adata Vitesta, 6 out of 10 sticks did at least 675 5-5-5 (loose) at 2.3V on my P5Q Deluxe for 8M, loaned the best pair of them to Quiekmew for a while as I'm currently not feeling like benching DDR2 stuff . I'll do more testing once I get a new board and the sticks back.

[ObscureParadox_3XS]

As above, get to testing different GTL levels, ram timings, performance levels and straps etc to find what effect it has on your board. This will be a chipset issue and not a ram one. Good luck

[quiekMew]

Ill do more testing with the ep45-ud3r once im done messing with some G-Die sticks on x79  

if the tips you gave me help to fix this issue, ill post the results here ?