MAX and two 128MB DIMMs, problems

[mkaresh]

See my system below. I have been running very stably at 133fsb, AGP 2x. I can even run 3dmark 75% of the time at 144fsb, still AGP2x.

The problem is I'm trying to add a second DIMM, to make Dragon Naturally Speaking work better while multitasking (it likes 100+MB all by itself), and it's not going well. I'm pretty sure the new DIMM itself is not healthy. But aside from that, stability with both DIMMs is quite a bit worse than just with either DIMM alone. Even at 115fsb it's shaky. Which makes me think I might also have a motherboard problem.

Symptoms:
3dmark2000 and 99max go poof and disappear
CC Winsteon 2000 and Winstone 99 lead to BSODs sometimes, and milder crashes more often. The offender in G400drv or something like that with the really bad crashes.
Finally, when 3dmark2000 ran for a few seconds at 133 it was all grey and yellow, as if 8 bit color.

So, here's the question:
Is there a chance that with two DIMMs and a Max I'm running out of power at higher bus speeds? My board supplies 3.5v on the 3.3v channel. I cannot adjust this. My power supply is 300w, which should be plenty.

What else might be going on here?

Anyone out there overclocking a Max on an overclocked Soyo Sy6BA+IV with two PC133 129MB DIMMs?

Thanks

------------------
PIII 500e @667
Soyo SY6BA+IV
Iwill Slotket II
G400 32MB Max
Nokia 445Xi 21"
and POS 14"
Aureal Vortex Superquad
128MB PC133
IBM DPTA 20.5GB 7200rpm
Toshiba 8/40 DVD

[Choi]

Are you sure that you are using pc133 ram?

If you overclock ram , you will have a crash

sometime...

I have a pc100 which can works over 140 fsb

but when I put it with another module, it

makes problems.

There is another posibility.

If you use double side ram which take two

ram bank, It will make problems when you put

the ram without skipping a space


dimm 1 ------------------------ ram

dimm 2 you shouldn't put

dimm 3 ------------------------ ram

dimm 4 you shouldn't put

[mkaresh]

Definitely PC133, so the RAM is not overclocked. But the board is.

It is doublesided. I tried 1 and 3 earlier, but haven't recently. I'm in one and two now. I'll give 1 and 3 a shot again.

Anyone else know about this skipping a slot thing?

Thanks for the tip.

[LAMFDTK]

Hey why not post the identity of the ram you have. Post the markings on the chips itself. Not this "I have pc 133 ram", stuff.

------------------
PIII-450@600, 128 HDSRAM, Asus P3BF, G400/32, SBLive!,Brand stinkin' new Sony G400 19", AOPEN DVD-1040 10x slot,PLEXTOR 8x4x32 ATAPI CD-RW,PromiseULtra66,various maxtor HD's,W98se,W2k pro, and some fish,




[This message has been edited by LAMFDTK (edited 06 March 2000).]

[mkaresh]

I thought I had written a book as it is. Also, sent that material to the Soyo forum and Kingmax tech support. But if you're interested, here goes:I have two 128MB modules. They differ a little:

One, purchased from multiwave, has chips that say:
Kingmax
KSV884T4F0A-07

The other, purchased from usamemory, has chips that say:
KTI
TSV884T4A1A-07

KTI is for Kingmax Technology, Inc., by the way. These two modules look exactly the same except for the markings on the chips.

I should also say that the more I use the new module the better it gets. I'm currently typing this at 133mhz (3-2-2). But it's still very flaky. Even at 100mhz (2-2-2) crashes still occur that don't with the first stick, and the higher the mhz the worse it gets. But when I first got it I had nasty crashes even at 100mhz 3-3-3. How long can "burn in" take?

[IceStorm]

I have the same MB and had the same problem.

I've got a PIII-450 OCed to 540 on the board, using 7.5ns PC133 128MB ECC SDRAM DIMMs, CAS3 latency (all identical DIMMS from www.crucial.com).

First, I had the system working fine with a 120FSB and a DIMM in Slot4.

I added a second DIMM. Anywhere from 20 minutes to hours later, the screen would freeze.

Underclocking to a 112 FSB didn't help.

Then I remembered - the 440BX doesn't handle 4 DIMMs stock. It needs extra electronics added to handle the fourth DIMM (buffers, I believe).

I moved the DIMMs from slots 4 and 3 to slots 3 and 2. Slot 1 is blocked by the Alpha.

Walla! No more freezes.

Just my $0.02


[This message has been edited by IceStorm (edited 06 March 2000).]

[mkaresh]

Thanks for the tip. 2 and 3 and 2 and 4 are the only combinations I haven't tried yet. Right now I'm in 1 and 3.

The weird thing is stability keeps improving, but still has a way to go. How long can RAM take to "burn in?"

[zippy]

The ram that says Kingmax (not KTI) is 7.5ns despite the "-07". I think this is the value PC133 from them. I think...I could be wrong...but I know that there is one kind of their TinyBGA that is 7ns and one kind that is Value TinyBGA that is 7.5ns. Still though, the 7.5 ns should work at 133 easy.

[mkaresh]

Do you have a good source for this info?

According to Kingmax tech support, KTI and Kingmax are the same chip, just different markings. The reason for the KTI is so Kingmax won't be called for Tech Support, the vendor will. And because some vendors prefer the KTI markings. The other difference between the chips I have are my Kingmax were made in a US fab, while my KTI were from a Taiwan fab. But they are supposedly the same exact design.

The weird thing is that in some ways the "bad" KTI stick ran better, it just wasn't stable. It would boot at 133 at 2-2-2, and completed all benchmarks (both Winstones and 3dmark2000) at 124 at 2-2-2, while my "good" stick required 3-2-2 above 117 to even boot (the 3 being RAS to CAS).

Anyone have any idea why this could be? That the "bad" stick could handle 2-2-2 much better? Maybe the KTI chips really are better. Of course running well sometimes at 2-2-2 doesn't help if everything I ran on it eventually crashed...even at 117.

The "good" stick is very stable at 3-2-2 up to and including 133 (where I run it 99.9% of the time), and then needs 3-3-3 up to 150. At 150 it boots fine every time and runs Word, AOL, etc., just don't throw anything 3D at it. At 140 3dmark freezes the system every time, but at 144 it's OK 3/4 of the time. Go figure. System won't boot at 155, but that could be for any of a number of reasons.

One thing I haven't played with is the AGP. Even at 150 I have 2x and sidebanding enabled at boot. As I said, it won't handle 3D there, but usually will at 144. It will handle 3D at 100fsb and 100AGP though--runs 3dmark2000 reliably. This seems to often be the case. Any ideas why?

[This message has been edited by mkaresh (edited 08 March 2000).]

[This message has been edited by mkaresh (edited 08 March 2000).]

[mkaresh]

My curiousity piqued, I reopened the box in which the bad stick was packed for shipping. I ran the Motherboard Manager SPD reader on both chips. Same readings. So it appears that they are the same chips, or that they have the same SPD data at any rate.

Isn't the module ns always a little longer than the chip ns? (7.3125 vs. 7 in this case)

[Lucian]

I have same kind of problems with a different configuration. G400(not max)DH on a Tyan S1590S AT VIA MVP3 with AMD K6-III 450. Nothing overclocked (FSB 100 & AGP 66). I can't get the system to work with more than 128MB. I tried 2x128, 1x128+1x64, 1x128+1x32, 3x64, 2x64+1x32. I called Matrox and they said they KNOW about this kind of problems but they can't reproduce them in their labs. I sent them my complete configuration (hardware and software). I wait now for their reply (and I'm expecting a loooong wait). I guess there's a problem with the G400 trying to access the system RAM using the AGP bus. I tried to get some information about AGP tweaks and/or G400 tweaks (Bus Mastering, AGP1x, Sideband, etc.). Anyone knows some good links on this topic?
TIA,
Lucian

[mkaresh]

My problem's not quite so serious. I got it stable at 100mhz with both sticks. Somewhat stable at 115. Above that not stable.

Kingmax tech support said many motherboards have trouble dealing with sticks that aren't exacty identical in their timing, especially when oc'd. They figured the timing of one of the sticks was off. I'm hoping a new stick works well.

[Lucian]

IS FIXED!
(I hope.)
After a ton of painful (and expensive) tries my system finally works with G400 DH and more than 128MB (256MB in this case). The tries included a million BIOS tweaks; multiple reinstallation of multiple VIA-Matrox-Windows drivers; registry hacks, buying two new Corsair DIMM-s ($155 each), moving the DIMM-s around, etc.
Here's the fix: I don't use the onboard IDE controllers, I use instead Promise Fasttrak but I installed the VIA drivers anyway. The problem seems to be that the VIA IDE Bus Master drivers don't like it when you try to install on disabled controllers (I know about the MS VIA patch - didn't work for me, same as many others i.e. shutdown patch ). This is the "winning" sequence:
1. Uninstall Matrox drivers
2. Uninstall VIA Bus Master drivers
3. Enable BOTH IDE controllers in BIOS
4. Reinstall VIA Bus Master drivers
5. Reinstall Matrox drivers
6. Disable IDE controllers *
7. Have fun!

Lucian

* - to free IRQ-s; System/Drivers shows yellow "!" for VIA Bus Master IDE controller as it did before starting anything

[mkaresh]

Lucian, what symptoms did you have? Did the system not boot, not recognize both sticks, or run but was not stable? I'm #3.

[Lucian]

My system was stable with 128MB. It was locking up (on black screen with hourglass or on color garbage screen) when I installed more than 128MB. It did work with 2 sticks of 64MB. The BIOS recognized every single combination. The system ran fine in any configuration as long as I uninstalled the Matrox drivers.

CLARIFICATION: I read my message again and it seems confusing, so: The fix is not to use Promise Fasttrak. I used that to begin with. The fix was "the winning sequence".