9NDA3 + San Diego 3700+

[buzz_lips]

OK this chip runs niceley at 250x11 in my board

http://server2.uploadit.org/files/bu...desktopnew.jpg

The only trouble i'm having is running my Gskill in Duall Channel mode, Single channel is fine.

Memtest give errors under test 8 in Dual Channel Mode but each stick is fine when tested in isolation. This is also the case if using Mem Banks 1&2, or 3&4. The rig won't post if i put sticks in 1&3 or 2&4.

I'm running everything at stock volts for the mo and wonder it might help if I inc the mem volts maybe.

I'm running the 01/04/05 Bios from the American site, latest i could find for UK was 02/02/05 but guess that shouldn't be problem.

Any help solving my dual channel problem would be apprecited.

[rE-vErb]

i have bought a venice myself and have similar problems on my 9nda3j.
considering the facts that the venice and san diego cores are identical except for the l2 cache size, and the 9nda3+ and 9nda3j are identical except for a few additional features, i'm getting a feeling that there is a memory timing problem with memory chips(i'm using a pair of 512MB corsair pc4400c25) that did fine on older cores.

Try synchronzing them in 2T command rate, they should do fine, just that 1T was fine on winnies and not on the new rev.E cores.

Searching through the web, i have found out that the popular favorite DFI nforce4 LanParty models had the same problem(not being able to set 1T command rate on the same FSB after switching from winnies to rev.E cores), but it was fixed through one of the new modded(?) beta bioses which had a different RAM table used.

The bios that enabled the lanparty nf4 series to enable 1T command rate on rev.E cores has the following notes:

Date Code 2005/04/14
File N4D414-3.zip
Size 390,190 bytes
Description:
1. BIOS 4.14-1 with 3.10 RAM table.(TCCD)
2. Add support for AMD Venice , San Diego Core cpu Cool n quiet
3. Prevent Jp 17 jumper problem cause vdimm still stay at over 3.3v when jp17 set back to 1-2.
4. DFI recommand cumstomer need to set to 3.3v AND clear CMOS before set the jp17 back to 1-2 default position.


now i'm not sure if such a mod can be easily done with the 9nda3 series, but i am pretty sure that it is a timing issue specific to rev.E cores which can be solved through a new bios release. The description #1 on the information above states using a 3.10 RAM table(not sure if that means the RAM table used in the march tenth release of the lanparty nf4 series, or an independent versioning specific to RAM tables that are used for 939 cpus). If using a similar RAM table stated is possible on the 9nda3 boards, then we might have hope on enabling 1T command rate on rev.E cores.

I am not a bios modder so there is a big possibility that what i've said can be very wrong, they are just my own thoughts and a possible fix for this issue.

If someone with bios modding knowledge passes by this thread, maybe he/she can enlighten us more in detail on this problem.

[danrok]

1T is a no go for me. I think my chips are BH5 not TCCD.

[bonzer]

How would one formally request EPOX to look into and perhaps address this issue at the BIOS level????

[rE-vErb]

well, waiting an official bios which solves this problem can take a very very long time...
i'm just hoping that someone with the knowledge would be kind enough to mod one for us.

[rE-vErb]

Originally Posted by danrok

1T is a no go for me. I think my chips are BH5 not TCCD.

it may not be a tccd specific issue, my adata pc4000(not tccd) pair also could not do 1T, not even 2T on 250Mhz which right on spec with the new rev.E cores. i'm carefully assuming that it is a timing setting that is not up to date in the bios, relating to the new memory controllers on the rev.E cores.

[danrok]

Originally Posted by bonzer

How would one formally request EPOX to look into and perhaps address this issue at the BIOS level????

By buying a controlling share in Epox?

[buzz_lips]

OK i'm going to test using 2t. See if I can get further using dual channel. Back in a tick lol

[buzz_lips]

Originally Posted by buzz_lips

OK i'm going to test using 2t. See if I can get further using dual channel. Back in a tick lol

Yup Dual Channel is no go even at 2t. It just posted in safe mode so I had to load optimised defaults and now i'm back at stock speeds

Does Mr Epox Tech work today?? lol

I did wonder if it was worth buying some other memoery but if others are having issues I guess it's not worth the outlay

[GrahamGarside]

Have you increased the DDR voltage at all? an increase to 2.7-2.8V might help

[danrok]

Originally Posted by buzz_lips

I did wonder if it was worth buying some other memoery but if others are having issues I guess it's not worth the outlay

Runs fine for me in dual-channel with my Corsair modules.

Can you borrow some other modules to try? Just to see if the problem relates to your modules in particular.

I've never had these modules running at 1T. There is an element of 'luck of the draw' when it comes to running at 1T.

[bonzer]

Originally Posted by danrok

1T is a no go for me. I think my chips are BH5 not TCCD.

I'm confused, are these different modules you speak of????? .....nevermind I thought you tried a second set


reverb couldnt get 2 sets of modules adata and top-o-the-line (tccd) corsair's to run at 1T


buzz cant get his very good GSkill LC's (tccd) to run 1T


I dont expect my GSkill ZX econo, I believe are (tcc5), will run 1T come this moday when my venice shows up

[buzz_lips]

Originally Posted by danrok

Runs fine for me in dual-channel with my Corsair modules.

Can you borrow some other modules to try? Just to see if the problem relates to your modules in particular.

I've never had these modules running at 1T. There is an element of 'luck of the draw' when it comes to running at 1T.

Unless you could lend me those Corsair modules mate it's a no go!!! lol

Might try some other Gskill mods maybe LA. At least I can send em back if it doesn't sort my problem, plus the package is small emough to sneak past the Mrs

I really don't want to have to get another mobo as that'll mean a move to PCI express and another GPU

@ Graham,

Yeah mate i've tried upping the volts to the mem doesn't seem to make any difference!

[bonzer]

buzz,

the only bios check I can think of that may help is the 'system performance' make sure its at normal instead of aggresive

[buzz_lips]

Originally Posted by bonzer

buzz,

the only bios check I can think of that may help is the 'system performance' make sure its at normal instead of aggresive

tis set to normal mate

[bonzer]

buzz,

do you have a + or J board??

make sure and use the latest bios for Rev E. support

try stretching out your timings... say 2.5-4-4-8 and 2.5-4-4-10 and lowering your vdimm, your LC tccd's might like 2.6-2.7-ish range better as many do.

you might need to use A64 tweaker to do some timing changes that the epox bios doesnt allow or show...

It looks to me that the standard newcastle/winnie ram settings are not gonna fly with the Vinnie/SD updated mem controller.... and its gonna take some trial and error to figure out what they like.

my 3200 vinnie retail (boxed) should be here monday so I'm flyin by the seat of my pants but have read alot of feedback. It also appears (from other forums) that the SanDiego controller is more sensitive than the venice.

I'm quite sure that with the proper feedback to this forum and ET, the folks writing the bios code will make the appropriate changes in the next bios release.

[buzz_lips]

Originally Posted by bonzer

buzz,

do you have a + or J board??

make sure and use the latest bios for Rev E. support

try stretching out your timings... say 2.5-4-4-8 and 2.5-4-4-10 and lowering your vdimm, your LC tccd's might like 2.6-2.7-ish range better as many do.

you might need to use A64 tweaker to do some timing changes that the epox bios doesnt allow or show...

It looks to me that the standard newcastle/winnie ram settings are not gonna fly with the Vinnie/SD updated mem controller.... and its gonna take some trial and error to figure out what they like.

my 3200 vinnie retail (boxed) should be here monday so I'm flyin by the seat of my pants but have read alot of feedback. It also appears (from other forums) that the SanDiego controller is more sensitive than the venice.

I'm quite sure that with the proper feedback to this forum and ET, the folks writing the bios code will make the appropriate changes in the next bios release.

I've actually managed to get this working in Dual channel now. It seems a HTt of 3x if preferable to 4x. Gaming and general rig useage have met with no probs for about 12 hours now which is good.
All my mem setting are at auto for the mo. I'm going to try and tighten these timings a bit later. I'll try 1t first, there no reason why the mem shouldn't be able to do this - tis gskill after all The bios seems ok, it idenifies my chip correctly.

I'll let you know how I get on

[danrok]

Glad to hear it.

So, you're running at 3x HTT, but 3x what? The HTT should be able to run at atleast 1000MHz.

[buzz_lips]

Originally Posted by danrok

Glad to hear it.

So, you're running at 3x HTT, but 3x what? The HTT should be able to run at atleast 1000MHz.

250ht X11 multi mem @ 1:1

3xhtt = 750. Many people seem to report that these Venice/San Diego Cores don't like going past 1000htt.

The board just goes skitszoid when i put it to 4x

I've no idea why... i'm hoping Epox Tech might be able to shed some light

[rE-vErb]

buzz_lips:

does 1t command rate work? i'm also running on 3x HTT, sorry i forgot to mention that; could have helped you out much earlier.

i remember being able to set 1t on the default 200Mhz FSB, but not much above that: which wasn't the case with winnies. Currently running 270 at 2T 1:1.