** OFFICIAL ** Intel DX38BT & DX48BT2 BoneTrail Thread

[Sean-E-Boy]

Im laughing . . (on the inside) LOL
Gotta love Intel at times!

Heres some other things Ive gleaned if anyone wants to grade these with a True / False:

XMP only works with 2 RAM sticks in the black slots Not positive, but going with false i recall having XMP working with 4 dimms, but had cpu host clock underclocked, so.... XMP is a joke anyway...tuning the RAM yourself will get you much better results with many fewer headaches.

RAM wont run at 1600 stable on BIOS 2000 and higher False running @ 1616 right now with BIOS 2006 and have been through BIOS revisions since this time last year.

This is one of those places where we tend to get caught up with the numbers game and insist that we get the ram to 1600, or our timings to 4-4-4....In reality, CPU FSB (host clock frequency) has one of if not the largest impact on RAM performance, with overall CPU frequency probably taking cake.

while faster RAM frequency can reduce overall latency, if FSB performance doesnt increase with it, large memory transfers can suffer. Play around sometime and run some tests with your memory @ 1:2 and cpu @ say 333x8. next drop your memory to 5:3, your multi to 6, and increase the FSB to 360. run some more memory tests and compare. you may be surprised to see that "slower" memory can fair quite well with the "faster" when FSB increases. This is especially important to remember as you push the FSB beyond 400 when 1:2 ram = greater than 1600 and problems occur.

Intel wont let the RAM run at a bus frequency higher than the FSB Sad, but true

Sean

[Manel]

Yes, Intel has some restrictions on mem configurations. XMP is supposed to work only with 2 slots populated and specific ones too.
I thought we would be seeing new LAN drivers soon because their page changed but now back to what it was.
I hope, if we get to see a new BIOS version, we don't get the new color scheme as it was used in the latest BIOS for the DP55KG Kingsberg board, black and white as opposed to current blue and gray. The menu configurations has changed as well, it seems more logical now I guess. Still, for that board it's way too Taiwanese for me, with the skull lighting and all, even in BIOS they used characters to draw the skull. Intel is going south as far as I'm concerned.

[Manel]

It seems that Larrabee was "postponed", it's now only a software project. Seemed too good to come true. Guess some of it will eventually go into Itanium 2. Man, Itanium should have replaced x86 ages ago, we keep tied to compatibility and development is only made upon crap, ages old hardware and software.
Anyway, nVidia Fermi has also some specs out and I really don't like the mem bus widths. I know someone must have studied real hard required bandwidth and probably those number of channels are what was needed but still I rather see 512 and 256 bit wide buses, like AMD still uses. Maybe I'm just too picky.
Still impressive though.

[crazysn2ke]

Hi there,

First off i will say thanks to all the people who have contributed to this thread. I have been a lurker here since i got my DX48BT2 in Feb, so thanks for letting me benefiet from your tireless hours of effort.
Anyways, I do have a question to ask, I have 8GB of corsair XMS3 DHX memory (2x 4GB pairs, part # TW3X4G1333C9DHX) and i cannot get them to run stable together at 1333. WIth 4GB, each respective pair or any combination of DIMMs, my PC is rock solid and stable at 1333 with default timings, but not with both pairs of DIMMs. (i only mixed the DIMM pairs as part of my troubleshooting, for normal use, i do keep each DIMM pair together in their respective channels) At 1333 with default timings, the closest i can get to stable is to bump the RAM voltage to 1.62 (Will crash at any other voltage between 1.62 and 1.9) and slightly raise the MCH, currently at 1.3 (have tried votlages between 1.25 and 1.4 while testing RAM voltages, but i am keeping that one as low as possible) At those voltages i am somewhat stable, i can play games and generally do most things, until i have a bucket load of browser tabs open or run a virus scan, windows defender or even large file transfer, it will just reset after about 10 mins. No BSOD, just system reset with 'Disruptive Shutdown' event in event viewer.
Memtest is fine to 3 full passes and prime95 for 90 mins dosn't even flinch. A few months back, i bought a 40mm fan for the MCH, which has helped both stability and ambient tempreture, and a couple of months after that i bought a Corsair Dominator AirFlow Xtreme (part # CMXAF1) which has dropped my ambient temp by 2 - 4 degrees and my RAM temp buy 1 - 2 degress, but hasn't done anything for stability. The Corsair forums have several people with 8GB stability problems with this RAM and Board and Corsair recommends dropping the speed down to 1066, which i have done and it has been stable, so maybe i am asking too much from this RAM and Board. But as this thread is the definitave resource for the DX38BT / DX48BT2, and some people in this thread have done some pretty amazing things with it, i just thought i would ask for anybody's thoughts, particualy HighTest and GasPipeJimmy, if either of you guys are still around, your RAM seems to be the same as mine, and as you are running at 1333 i would be interested in your timings or any info you may have.

Sorry for the long post, i just wanted to give as much info as i could, and thanks to everybody again for a very informative thread.

[GasPipeJimmy]

Hi crazysn2ke!
My RAM timings are in my sig line.

Take a look at your memory timings in your BIOS and make sure your "Command Rate" is set at "2T" and not "Auto" or "1T".

Next, look at the revision number of your RAM; are they IDENTICAL?
If not, that is likely your problem. If they are identical, then I would talk to Corsair and ask them if you could send them the RAM and if they could check them out for you, to determine if they are functioning correctly, and/or if they can do some more through checking to see if perhaps they used different modules on the same revision number. They have the hardware and the documentation to find that out.

If all else fails, E-Bay all your RAM and get an actual 8GB matched set.

Sometimes non-matched RAM will work with four sticks, but I have seen it not work (completely) more often that I have seen it work.

Finally, don't bump the voltage on your Northbridge past 1.3v.
With a NB fan, you should be safe at that voltage, but anything higher is likely pushing your luck.

OK, that's all I think I know on this subject!

[crazysn2ke]

Thanks for the reply Jimmy,
Yeah, they are differnt revisions, i bought them 3 months apart, i learned after i bought them that Corsair doesn't reccommend mixing revisions. I'll write that off to a flaw in the pre build research. Down the track i will eventually end up getting an 8GB kit, but depending on which flavour i go for i'll be looking at around 350 - 450 smackeroos down here, so i'm at 1066 until i find new homes for my 2 4GB kits first. Thanks also for conformation on MCH volts, i must admit i didnt like having it at 1.3, but now that my mem speed is lower i've lowerd that too.

[GasPipeJimmy]

I'm glad I could help.
What you could do is sell both sets on E-Bay and then when they sell, buy your 8GB then.
I think if you sold them as you bought them (in 4GB lots), you could eliminate or at least severely limit your down time.

After the first set sells, you still have 4GB in your box and can still use it until the second set sells.

Since the price of RAM has increased recently, you might not lose any money on that side of the deal.

My 8GB was $150.00 (U.S.D.) back at the end of October and now I see them selling for $250.

I'm just saying that with a little bit of planning, you could economically do it without taking a big financial hit.

Let us know how it goes, and good luck!

[Manel]

Anyone tried 4GB modules yet? Not even sure the mem controller supports them anyway. Board specs mention 8GB max but that was back in the days when only 2GB modules were available.
Imagine 16 G installed!!! Sweet. Not that one would normally need that much mem but still it would make you wow.

[BlackKnight00]

I recently changed video cards from a 4870x2 to a 5870 when I installed the new card the board wouldn't post. After a few try's it eventually booted up just fine.

I decided to sell my 5870 and go back to the 4870x2 but after I re installed it, the board refuses to boot up. It gives me the bios no post error message and when I choose y to go into bios, the bios is frozen and I can't move around in it. After trying all the usual power off on tricks I decided to move the jumper pins and it went into the bios just fine but when I powered off and moved the pins back I get the same problem. Any idea's on how to solve this? Thanks in advance.

[BlackKnight00]

Quick update. I was able to get the system working again by pressing F9 in the frozen bios. Now system boots fine but the bios is still frozen so I can't make any changes.

[GasPipeJimmy]

Have you tried the manual jumper reset again?
Then the remove the CMOS battery and give it an hour or so trick?

[GasPipeJimmy]

I just bought a Q9650 to replace my Q6600.
The Q6600 is going back into my old BadAxe2, which is going back into my old Coolermaster Praetorian chassis to be given away to a worthy person that can't afford a decent computer.
'Tis the Season!

[Sean-E-Boy]

The BIOS freezing up is something that I have been experiencing for a while...does it in the recovery BIOS too (jumper moved). All I can say is to know what it is that you want to change and do 1 modification at a time and then save and exit.

sometimes it freezes up in just a few seconds and yo cant get anything done, other times I can navigate around without any issue.

on a different note, has anyone noticed any functionality changes with this board and windows 7? I had no issues with the beta and RC versions, but now when I restart from within windows, the system will actually power down and then back on, rather than a soft reset as expected.

sean

[GasPipeJimmy]

on a different note, has anyone noticed any functionality changes with this board and windows 7? I had no issues with the beta and RC versions, but now when I restart from within windows, the system will actually power down and then back on, rather than a soft reset as expected.

sean

I noticed the same thing in mine. Feature and not a bug, maybe? I liked it as it gives what I want in a full shutdown but a lot quicker than waiting for it to go black and then manually restarting it.

[Manel]

Anyone running an nVidia card is also unable to open PhysiX Control Panel Applet? It reports missing NvCplUI.exe missing. Still, you can access it from the nVidia Control Panel, which for me is the right way to do it, but why the PhysiX icon in the CP then? Drivers v18x and before would work fine, although I didn't find the use of having a direct link to PhysiX in CP just to enable or disable it. Anyway, driver 191.07 didn't work at all for me but now 195.62 is working, only this PhysiX issue. There is a newer version of the PhysiX driver out but haven't downloaded it yet to see if it fixes this.
Also funny, GPUZ reports no OpenCL support for my 8800GT while nVidia states that series 8 and above will support OpenCL with this driver. Direct Compute is also unmarked but that is because support is only for Win 7, maybe OpenCL also?
There is also LAN driver v14.8 out at Intel.

[BlackKnight00]

I finally got everything working fine again. I should know by now that this board really doesn't like hardware changes.

I was able to change settings with the jumpers in password reset mode and once the voltages were bumped back up, I was able to get into the regular bios and slowly ramp up all my settings.

[Manel]

Hey BK, glad everythings fine now. Did you keep the 4870x2 or the 5870?
Had any more BIOS freezes?

[Sean-E-Boy]

Chipset 9.1.1.1025 out on Intel site.

Sean

[JohnZS]

Is anyone here aware of a bug in the X38 chipset which effects the PCI-E function?
Since April nVidia have dropped Gen 2.0 PCIE support from their drivers if you are using an x38 motherboard.
If anyone reading works for Intel please PM me.
Tested on 8800 GTS, GTX 280, 285 and other gen 2.0 products.
ALL FORCEWARE drivers 190.xx or newer run cards in legacy PCIE 1.1 compatibility mode.
Thank you

Now I am no longer on phone I can explain in more detail.

I have been having issues with nVidia hardware (G92 or newer) not running in PCI-E Gen2.0 mode on my ASUS P5E3 motherboard and have also reproduced this issue on Rampage Forumula and P5E3 Premium. Naturally I reported this issue to ASUS as graphics cards which experienced this issue ranged from a 8800 GTS 512 (G92) to a GTX 28(0)/5... all PCI-E 2.0 cards

ASUS technical support managed to reproduce the issue and subsiquently reported it to nVidia.
nVidia eventually sent ASUS an official response which was forwarded to me.

Gen2 has been disabled on X38 for quite some time now, since around last year April to be exact. The reason behind this is due to a bug in X38 and the new SBIOS from Intel still shows problem. So our decision is to disable gen2 on X38

If anyone from Intel is reading, please can you liase with nVidia to resolve this BIOS issue.
I have noticed the DX38BT motherboard is now officially on the legacy products list, it would be nice to resolve an issue which we have all paid for (The reason I chose the X38 over the P35 was FOR Gen 2.0 PCI-E and potential crossfire at these speeds)

This issue MIGHT be resolved by the July 2009 BIOS for DX38BT, however I would need people to test this (Forceware 191.07 or newer would be required). and BIOS 200 or 2006 would be required on your motherboard.

Thanks

John

[Manel]

John, aren't you the party spoiler?! Just kidding. Hope you're wrong bout this but it seems you have it more or less solid.
Never heard of this issue before. Millions of questions coming up: what does AMD say about this? Is it only nVidia that can't handle the issue? How come o one else ever reported it? Is it fixed in X48?
Funny, nVidia drivers 191.07 didn't work for me, never detected my card. Now, 195.62 have no issues.
My DX38BT is running BIOS 2006 and GPUZ reports PCI-E 16x v2.0, no sure this is accurate though cause this is what he vid cad reports as supporting. How did you notice it, running only v1.1? How can you test?
Man, this being true I guess I quit the computing world. I'm sick of nVidia with their SLI crap, Intel with these issues, not in the mood for AMD...
Just say what needs to be done.
Aren't you the John that was helping Intel a while back with the mem timings tweaking in BIOS on this board?

[Manel]

Have any specifics on what the issue actually is? Any info available?

[JohnZS]

Hi Manel

I do not know exactly what the specifics were, but I and a few members on the official ASUS forums were experiencing issues with our nVidia cards not always working @ PCI-E Gen 2.0. on P5E3 (x38) based motherboards. Sometimes our cards would work @ PCI-E Gen 2.0 and other times they would not.

At the time I was also able to reproduce the issue on a DX38BT (BIOS 1197)
Eventually I had enough reproduceable results from people with all sorts of graphics cards (G92, G200, G200b, G92b) to prove that this issue was somehow linked to the motherboard as opposed to the graphics card. (as P45 based boards had no issues whatsover).

A reboot usually resolved the issue, that was until the first forceware 190 drivers were release. Then we lost all PCI-E Gen 2.0 capabilities. Not only does GPU-Z, Everest and nVidia control panel report our cards as PCI-Express 1.1 cards, but also Scores in 3dmark were reduced and WEI reduced on Windows 7.

Further testing was done with 191.07, 195.62WHQL and 195.81. However using older forceware drivers from the 18x series brought back the intemitant "sometimes it works, sometimes it defaults back to 1.1" issue.

After several bug reports and discussing the issue with ASUS, I received a statement from ASUS forwarded to them from nVidia quoting that a bug in the Intel BIOS meant nVidia had to disable PCI-E Gen 2.0 support on the X38 platform.

I have a hunch that intel resolved this issue in SMBIOS 2.4? boards (would this be July 2009 BIOS update).

I have read (and relayed on information to ASUS and nVidia) that BIOS 2006 release notes under the "2000" release section, list this fix.

"Updated support for high power PCIe adapters."

Presumably this resolves the issue with the BIOS and Gen 2.0 support. (I think it is a slot capability detection issue, but do not quote me on this).

A screen cap of GPU-Z would be most appreciated as ASUS and nVidia are in "denial" that Intel have fixed this issue.

Thanks

John

[Manel]

John
Both GPUZ and DXDiag report PCI-E x16 v2.0, so I guess all is ok. Never noticed before that it ever operated otherwise though, but I wasn't checking for it, it might have slipped.
Attached I'm sending an image of my desktop with GPUZ, CPUZ, DXDiag and IDCC open so that you can check for yourself how my system is working.

[JohnZS]

Thank you very much Manel
I have passed this onto ASUS and nVidia as it appears the issue is now isolated to ALL ASUS X38/X48 motherboards (regardless of BIOS revision), but only Intel X38/48 motherboards using BIOS 1197 or earlier.
With ATi cards the issue is slightly different. The card Reports that it supports Gen 2.0 x16, yet it only runs at Gen 2.0 x4
nVidia cards run @ Gen 1.1 x16 and report that they only support Gen 1.1 x16 support, but this is intentional as the driver prevents a BIOS bug causing issues.
John

[Manel]

Do you think it's Asus that's kicking it down instead of nVidia? Funny ATI taking a different route here.
If this is a power issue, I'm not sure throtlling down PCI-E will help much, a little maybe since lower traffic but GPU will still be crunching away full speed with same power draw.
Anyone has it running at 1.1? Sean? Anyone?
It seems X48 has it too?