Memory OC -- Greatly reduced in Dual Channel

[Mullzy]

I had one 512MB stick Kingston HyperX DDR400 running at DDR440 for several weeks (220 FSB 1:1 on an Abit IC7). Since slapping in another 512MB stick of the same memory and moving to dual channel I only seem to be able to hit DDR420 (210 FSB 1:1).

I haven't done extensive testing, but dropping out of Dual Channel mode seems to make a difference. Before I spend several hours testing all the scenario's with Memtest86, I thought I'd see if this is a common occurance. Both sticks seem capable of running DDR440 individually, but together in DC it's fails the first or second pass during test 5.

Does this sort of thing happen often when increasing RAM and switching to Dual Channel? Will dropping my timings down to something like 2.5-3-3-6 make any difference?

Thanks for your input.

 

[yourdeardaniel]

you can try running at SPD, or switching memory banks, and increase vdimm

 

[babyelf]

er... which KHX3200 have u got?

some maybe bh-5 and some not.. if u're lucky u might have gotten one and the other one is holding u back..

but at 220 FSB only on the IC7 i think u have the CH-5

Check ur RAM and see what it says on the stickers

if it says AOO then it's BH-5
if it says AO1 then it's CH-5 (like mine) got confirmation from Kingston itself

 

[tamps]

CH-5 aint much worse than BH-5. i've seen 'em CH-5 going up to 265@2-2-2-11 on 3.2v

beside winbond chips, Kingston might also have used other chips on HyperX series.

 

[Mullzy]

- I've had vdimm up to 2.8 (as high as I can go)
- I've tried banks 1/3 & 2/4 (dual channel) and each does basically the same thing

I didn't try to push above DDR440 with my first stick... I was pretty happy with that and it was stable.

I definately have 2 different KHX3200's - Sandra shows the description of one different from the other. I can't remember the exact details but my first stick is described as requiring looser timings to run at DDR400 than the second stick.

SPD... not sure what I'm going to do there. I believe both detect differently (which could be the cause of my problems come to think of it). I think my first stick reports as 3-3-3-8 or something lame in my bios, but it was supposed to be 2-3-3-6 so I set it to that. I'm not sure that this new one will detect at... but perhaps I'll set it to SPD and then run some more Memtest86 tests.

Thanks.

 

[tamps]

maybe bios update solves some compatibility issues. not a bad thing to try out.

 

[Stormgiant]

Originally posted by: tamps
CH-5 aint much worse than BH-5. i've seen 'em CH-5 going up to 265@2-2-2-11 on 3.2v

Those were some SUPER CH-5 chips... Never seen that much high at 3.2V...

 

[Stormgiant]

Originally posted by: Mullzy

I definately have 2 different KHX3200's - Sandra shows the description of one different from the other. I can't remember the exact details but my first stick is described as requiring looser timings to run at DDR400 than the second stick.

SPD... not sure what I'm going to do there. I believe both detect differently (which could be the cause of my problems come to think of it). I think my first stick reports as 3-3-3-8 or something lame in my bios, but it was supposed to be 2-3-3-6 so I set it to that. I'm not sure that this new one will detect at... but perhaps I'll set it to SPD and then run some more Memtest86 tests.

Thanks.

Everybody knows that running two sticks takes a little bit of OC to the memories than running single.
I would start by setting the two dimm to 3-3-3-8 and then chopping off the timmings until they get to the max.

Or, you could exchange those for a better KIT already dual-channel garantied to run at the same speed

 

[babyelf]

u definitely can't go higher than a single stick if u have a mismatch of chipsets in the ram.. my advice.. go to www.outpost.com get the 3200llpt running at 2-2-2-6 .. not the 2-3-2-6 ones.. they are the BH-5s i reckon.. so u'll see huge improvement.. 3200 speed rated at 2.5-2.6V so at 2.8 u can expect around 220-230 already

edit: i can't be certain.. but from the research i made when i'm looking for my BH-5 rams those are. there's 2 versions.. just make sure u got the 2-2-2-6 ones
and better still.. ask them to check on the sticker and see if it says 3200LLPT rev1.1
then congrats.. u found urself some bh-5s

 

[Avalon]

I don't think it would be BH-5. That stuff is only found on eBay and antique shops now ;D
Anyway...yes, it's normal for your OCing potential to be slightly reduced once you go from one stick of single channel to two sticks of dual channel. When I had a single stick of my Geil PC3200, I could run it at 216mhz 2-2-2-5 timings. Now in dual channel, I can't get timings any lower than 2-3-3-6 at 216mhz. Not too big of a deal for me though.

 

[Mullzy]

I decided to let my timings to be set by SPD for now which turns out to be 2.5-3-3-8.

I was able to do a FSB of 215 (DDR430) easily with these settings and I'll try 220 next. My CPU can handle a 235Mhz FSB (2.8c @ 3290Mhz) but I'm assuming my ram will top out around 220-225 with loose timings if I'm lucky.

Thanks for your help so far.