TwinMos Winbond Oldschool BH-5 problems

[kaze]

I recently bought 2 more oldschool twinmos winbond BH-5 512mb sticks. In single mode they will do 250+ with 3.2v in test5. Both of the new sticks will do about the same in single mode actually, but they dont seem to run at all at higher speeds in dual channel.
SD 3700+
DFI LanParty UT NF3 Ultra-D BIOS 711 (ive heard of a modded BIOS with 4v vdimm but I dont seem to find it, or maybe its just bs)
OCZ DDR Booster
I was aiming for 300fsb with the 166 divider 3Ghz, DDR 500 but I dont seem to come very close to that.
Are there any settings that ive missed, does that BIOS that I was talking about exist? If so, has anyone tested it ?

[|3ourne]

max Async = 8ns
read preamble = 5ns or 6ns
tras = 6 or 7
idle cycle = 16
tref = 4708 ( 166mhz , 1.95us )
Trc = 7
Trfc = 14 or 16
Trtw = 2

[largon]

Read Write Queue Bypass -> 16x
Bypass Max -> 7x
32 Byte Granularity -> disable

BH-5 does tRFC 12/13 easily, no point upping it.

[Revv23]

can anyone tell me anything about 32 byte granularity?

like whats it do? does it effect bandwidth? is it an OC killer or is it just better to leave off?

[kaze]

Read Write Queue Bypass -> 16x
Bypass Max -> 7x
32 Byte Granularity -> disable

BH-5 does tRFC 12/13 easily, no point upping it.

tnx neighbour x)
I cant set Bypass max in BIOS and I didnt "have time" to change it before I took the screenshot.
What about DDR DQ Drive Strenght ? The values are, "Not reduction" "15% reduction" "30% reduction" "50% reduction" What do you guys use ?

[largon]

Revv23,
32 byte granularity enabled means bandwidth hit of about 200MB/s.

kaze,
"DDR DQ Drive Strenght" would propably be better left "No reduction", Winbond chips tend to like strong drive strenghts.

[kaze]

Revv23,
32 byte granularity enabled means bandwidth hit of about 200MB/s.

kaze,
"DDR DQ Drive Strenght" would propably be better left "No reduction", Winbond chips tend to like strong drive strenghts.

Ok, will try to set it to not reduction then. running dual channel now @ DDR 470 , 235fsb no problems so far, passed 32M (I know, bad results for BH-5) I will try 273fsb with 183 divider 11xMulti, that should be ~230?

*edit* Set to not reduction and 183 divider, not reduction doesnt seem to be the problem thou

[largon]

273MHz x 11 = 3003MHz
(183 divider)

->

Dram frequency:
3003MHz / 12 = 250.25MHz

[kaze]

273MHz x 11 = 3003MHz
(183 divider)

->

Dram frequency:
3003MHz / 12 = 250.25MHz

multi 11 :stick: Well CPU-z said ~231, but im sure youre right :toast:

*edit* Just woke up and decided to try with 3.56vdimm, result: :)

Im getting kinda bored , could I possibly missed something huge?

[s7e9h3n]

273MHz x 11 = 3003MHz
(183 divider)

->

Dram frequency:
3003MHz / 12 = 250.25MHz
Here's my old-skool Twinmos Bh5 2x512 sticks and my FX55 (in my sig)....Who needs dividers? :p: :toast:

http://img437.imageshack.us/img437/43/11x2731mtwk50li.jpg

[kaze]

Here's my old-skool Twinmos Bh5 2x512 sticks and my FX55 (in my sig)....Who needs dividers? :p: :toast:

http://img437.imageshack.us/img437/43/11x2731mtwk50li.jpg


Nice x) Well I made "some" progress, right now im trying to change the TREF, first I used 3120, playing around with A64 Tweaker now. With 3120 I didnt pass loop 1 in 8M, found one that were a little bit better, "all the way" to loop 11