** OFFICIAL ** Intel DX38BT & DX48BT2 BoneTrail Thread

[gtj]

I'm finishing up the Linux driver for the monitoring chip this weekend. It'll give you fans, voltages and temps that the coretemp driver doesn't.

[vipertwo]

I'm finishing up the Linux driver for the monitoring chip this weekend. It'll give you fans, voltages and temps that the coretemp driver doesn't.

gtj I'd be very interested in trying out your Linux driver. Is it available anywhere?

[gtj]

gtj I'd be very interested in trying out your Linux driver. Is it available anywhere?

I'll upload it and post the link when I get home from work tonight.

[GJW]

First I want to apologize for how long this post is going to be. First the highlights of my system;
Intel DX38BT motherboard with bios version 1478
QX9650 CPU (I had a QX6850 in it and all the problems I’m having I had with that CPU as well)
4 gig of Corsair WIN3X40961600C7DHXIN memory
2 ATI HD3870 Video Cards
Watercooled CPU

I have several problems that I cannot seem to resolve. The first problem is more of an annoyance but maybe someone knows what to do. Sometimes when it boots, just about 100% of the time after it sits for a bit (I travel quite a bit) and you turn it on, it turns itself on and off several times and then eventually boots. It will also do this after changes to the bios. During the changes to the bios sometimes it will boot, sometimes it won’t. When it won’t it just keeps turning on and off and eventually if I shut off the power supplies and give it about 5 minutes, it will boot ok, even if you change nothing in the bios. Even if I leave the bios at default settings it will do this after it sits for a few days and you turn it on. I contacted Intel and they told me it was the memory because the board is only designed for 1.5 volt memory and that memory runs at 1.9. I also tried some of the same memory but not Intel “certified” and had the same problem. (I also had this issue with another DX38BT board. The initial board I received did this, but the chipset drivers were messed up and it would not recognize the USB ports. Something when very wrong in the bios recovery (probably the fact that the optical drives were not really recognized during boot) so I sent that board back).

Second problem.

I have an issue, or really several issues overclocking the CPU. The main problem, I think, is the fact that I can’t seem to be able to actually change the CPU core voltage. I can change it in the bios, but when looking at CPU-Z it doesn’t show any change in voltage. In addition to this the CPU core temp agrees with the fact that voltage is not changing. When running at 100% load I do not see any change in core temp (averages about 100 F at full load no matter what CPU voltage is set at). However when I use the 300 mV offset there is clearly a significant increase in CPU core temp at 100% load (CPU core temp goes to about 140F with the offset at full load). Both the QX6850 and the QX9650 exhibited these characteristics.

From an overclocking standpoint with stock voltage I can get it about 3.65 GHz. With the 300 mV offset I can actually go to 4.42 GHz. No matter what combination of voltages, bus speeds, CPU multiplier, etc, I cannot set the bus speed faster than about 360 MHz or it is not stable. From a memory standpoint I normally just use XMP Profile 1, which is automatic memory overclocking. But I’ve had the same problem when I use my own memory timings.

Of course I talked to Intel about this issue as well and they won’t even talk to you if you are overclocking. I don’t want their help, I just want to know if there an issue with their hardware or something with my system.

I hope there is something easy that I have messed up or am overloocking. If anyone can offer help I would really appreciate it.

jerry

[GJW]

Oh, I can't use the Intel IDCC because I'm running XP Pro 64 bit and they don't support this OS. So I can't use that to verify anything.

j

[gtj]

Oh, I can't use the Intel IDCC because I'm running XP Pro 64 bit and they don't support this OS. So I can't use that to verify anything.

j

IDCC is also useless. Get Everest. It does everything you need.

http://www.lavalys.com/forum/

[GJW]

Ok, thanks for the help so far.

I have Everest but I just have not installed it on that machine yet. I'm not running the 300 mV offset, I just tried it to see what would happen. I do have a sub zero cooling system ordered but it won't be here for about a month (Cooler Express 2009). Goals for the board, that's a good question. To be totally honest it's a toy for me. The only thing I've done well consistantly in my life is make money so I decided I just wanted to build something really fast and cost was/is not an issue. I do play a few games with it (CoD4, Halo 1&2 and a few others), but it's mainly just for messing around with. I just want it to be around 4 gig, that's my personal goal.

I had a D975XBX2 with a QX6700 and had pretty good luck with it, which is what made me buy this board.

I won't be able to play with it again until this weekend, as usual I am going out of town to Dallas tomorrow morning. I will install Everest and play some more this weekend and report back. I'm gone again all next week as well, so I only have two days to play with it over the next 10-12 days.

Thanks for the help so far.....

jerry

[s0lid]

IDCC is also useless. Get Everest. It does everything you need.

http://www.lavalys.com/forum/

IDCC is USELESS, U CAN only RAISE FSB 10MHz per BOOT
Use newest SetFSB and use this setting: SLG505YC256BT IF u want to OC from windows.

[GJW]

IDCC is also useless. Get Everest. It does everything you need.[/url]

In installed Everest this morning, set the CPU voltage to 1.3 and sure enough it shows voltage as 1.238. Same problem I had before.....Stuck on stock voltage for the CPU no matter what the settings.

jerry

[gtj]

The first problem is more of an annoyance but maybe someone knows what to do. Sometimes when it boots, just about 100% of the time after it sits for a bit (I travel quite a bit) and you turn it on, it turns itself on and off several times and then eventually boots. It will also do this after changes to the bios. During the changes to the bios sometimes it will boot, sometimes it won’t. When it won’t it just keeps turning on and off and eventually if I shut off the power supplies and give it about 5 minutes, it will boot ok, even if you change nothing in the bios. Even if I leave the bios at default settings it will do this after it sits for a few days and you turn it on. I contacted Intel and they told me it was the memory because the board is only designed for 1.5 volt memory and that memory runs at 1.9. I also tried some of the same memory but not Intel “certified” and had the same problem. (I also had this issue with another DX38BT board. The initial board I received did this, but the chipset drivers were messed up and it would not recognize the USB ports. Something when very wrong in the bios recovery (probably the fact that the optical drives were not really recognized during boot) so I sent that board back).

Don't listen to Intel.

I have the same problem. It's an annoyance I don't deal with by just leaving it on and crunching for WCG.

It's got something to do with the board resetting the voltage regulators.

Second problem.

I have an issue, or really several issues overclocking the CPU. The main problem, I think, is the fact that I can’t seem to be able to actually change the CPU core voltage. I can change it in the bios, but when looking at CPU-Z it doesn’t show any change in voltage. In addition to this the CPU core temp agrees with the fact that voltage is not changing. When running at 100% load I do not see any change in core temp (averages about 100 F at full load no matter what CPU voltage is set at). However when I use the 300 mV offset there is clearly a significant increase in CPU core temp at 100% load (CPU core temp goes to about 140F with the offset at full load). Both the QX6850 and the QX9650 exhibited these characteristics.

From an overclocking standpoint with stock voltage I can get it about 3.65 GHz. With the 300 mV offset I can actually go to 4.42 GHz. No matter what combination of voltages, bus speeds, CPU multiplier, etc, I cannot set the bus speed faster than about 360 MHz or it is not stable. From a memory standpoint I normally just use XMP Profile 1, which is automatic memory overclocking. But I’ve had the same problem when I use my own memory timings.

Of course I talked to Intel about this issue as well and they won’t even talk to you if you are overclocking. I don’t want their help, I just want to know if there an issue with their hardware or something with my system.

I hope there is something easy that I have messed up or am overloocking. If anyone can offer help I would really appreciate it.

jerry

Don't listen to Intel.

CPU-Z and CoreTemp report the VID that's requested, not the actual voltage. Get Everest. It'll show you the actual voltages as reported by the sensor chip. The guide in my sig talks about this and even though it's for the D975XBX2, 99% applies to the DX38BT.

STAY AWAY FROM THE 300MV OFFSET UNLESS YOU ARE BENCHING WITH SUB-ZERO COOLING. Think about it...If you have voltage set at 1.5000 and set the 300mv offset, you're pumping 1.8v to the processor. Not good.

You have to manually set the memory timings or you won't be successful. Use the calculator in my signature to see what results from various FSB, Reference Frequency and Memory Frequency settings. Also DDR3 memory is VERY sensitive to voltage, both under and over. You really have to take your time with this board.

What are you goals for the board? It'll help us give you some guidance.

Oh yeah,

[Mr Steveo]

Don't listen to Intel.

I have the same problem. It's an annoyance I don't deal with by just leaving it on and crunching for WCG.

It's got something to do with the board resetting the voltage regulators.

I was chalking it up to a temporary "cold boot" bug. In the past few years I've seen Intel fix these with BIOS updates - like on Dragontail Peak and Blue Mountain2 mobos.

But from what you are saying, maybe with this mobo there is nothing TO fix here? (A normal stage stepping operation of the VRs?)

[vipertwo]

I'll upload it and post the link when I get home from work tonight.

Thanks, looking forward to giving it a go.

[clark_carty]

New IDCC with bug fixes here: https://download.desktopboards.com/idcc/

This one holds my customized fan settings between reboots now. It claims it needs BIOS 1478, but works fine with 1379 since I'm not updating due to the S3 Standby Failure in 1478.

[gtj]

Thanks, looking forward to giving it a go.

And of course I totally forgot to upload it so I'll do that now and post the link in about 30 minutes.

OK, posted at http://www.peakin.com/x38/asc76xx.zip
Unzip it in an empty directory and read README and INSTALL

[vipertwo]

And of course I totally forgot to upload it so I'll do that now and post the link in about 30 minutes.

OK, posted at http://www.peakin.com/x38/asc76xx.zip
Unzip it in an empty directory and read README and INSTALL

Thank you very much!

At first I just saw this error in dmesg:

[ 4558.898570] i2c-adapter i2c-1: asc76xx not found at 1,0x2e.

The chip seems to be an aSC7621a (with an a on the end), i.e. COMPANY=0x61 and VERSTEP=0x6d. I was
not able to make your kernel module work unless I patched it like this. The kernel is linux version 2.6.24.

--- a/asc76xx.c 2008-02-22 11:23:17.000000000 +0000
+++ b/asc76xx.c 2008-02-22 14:17:07.000000000 +0000
@@ -37,7 +37,7 @@
 };

 /* Insmod parameters */
-I2C_CLIENT_INSMOD_2(asc7611, asc7621);
+I2C_CLIENT_INSMOD_3(asc7611, asc7621, asc7621a);


 #define INTERVAL_HIGH  (HZ + HZ / 2)
@@ -46,7 +46,7 @@
 #define PRIORITY_LOW 1
 #define PRIORITY_HIGH 2
 #define FIRST_CHIP asc7611
-#define LAST_CHIP asc7621
+#define LAST_CHIP asc7621a
 #define MASK_ASC7611 0x01
 #define MASK_ASC7621 0x02

@@ -80,7 +80,7 @@
         .addresses = {0x2c, 0x2d, 0x2e, 0x00},
         },
         {
-         .name = "asc7621a",.chip_type = asc7621,
+         .name = "asc7621a",.chip_type = asc7621a,
          .chip_mask = MASK_ASC7621,
          .company_reg = 0x3e,.company_id = 0x61,
          .verstep_reg = 0x3f,.verstep_id = 0x6d,

Then after

modprobe asc76xx
dmesg

I see the following (with debugging messages turned on):

[  267.141938] i2c-adapter i2c-1: Autodetecting device at 1,0x2e ...
[  267.157915] i2c-adapter i2c-1: Checking 0x2e,0x61,0x6d against 0x61,0x69
[  267.157918] i2c-adapter i2c-1:   Didn't match asc7611
[  267.173897] i2c-adapter i2c-1: Checking 0x2e,0x61,0x6d against 0x61,0x6c
[  267.173900] i2c-adapter i2c-1:   Didn't match asc7621
[  267.189888] i2c-adapter i2c-1: Checking 0x2e,0x61,0x6d against 0x61,0x6d
[  267.189891] i2c-adapter i2c-1:   Matched asc7621a
[  267.189893] i2c-adapter i2c-1: Detected asc7621a device at 1,0x2e with COMPANY: 0x61 and VERSTEP: 0x6d

Which looks ok. Then

sensors

gives the following

asc7621a-i2c-1-2e
Adapter: SMBus I801 adapter at 2000
in0:         +1.21 V  (min =  +0.00 V, max =  +3.31 V)
in1:         +1.01 V  (min =  +0.00 V, max =  +2.99 V)
in2:         +3.24 V  (min =  +0.00 V, max =  +4.36 V)
in3:         +4.99 V  (min =  +0.00 V, max =  +6.61 V)
in4:        +11.84 V  (min =  +0.00 V, max = +15.94 V)
fan1:          0 RPM  (min =    0 RPM)
fan2:          0 RPM  (min =    0 RPM)
fan3:          0 RPM  (min =    0 RPM)
fan4:          0 RPM  (min =    0 RPM)
temp1:       -62.0°C  (low  = -127.0°C, high =  +0.0°C)
                      (crit = -21.0°C)                  sensor = diode
temp2:       +44.2°C  (low  = -127.0°C, high = +127.0°C)
                      (crit = +65.0°C)                  sensor = thermal diode
temp3:       +46.5°C  (low  = -127.0°C, high = +127.0°C)
                      (crit = +65.0°C)                  sensor = transistor
temp4:       +41.8°C  (low  = -127.0°C, high = +127.0°C)
                      (crit = -128.0°C)                  sensor = Intel PECI
temp5:       -62.0°C                                    sensor = Intel PECI
temp6:       -85.0°C                                    sensor = Intel PECI
temp7:       +96.0°C                                    sensor = Intel PECI
temp8:      +118.0°C                                    sensor = Intel PECI

I only have a fan connected to the MCH fan header on the motherboard so I would have
expected to see one non-zero fan speed -- but actually I don't see it in
the bios (v1397) either.

Does that look ok to you? Do you know how to interpret the figures -- i.e. which fans and temps are which? The CPU is a QX9650.

[clark_carty]

I only have a fan connected to the MCH fan header on the motherboard so I would have
expected to see one non-zero fan speed -- but actually I don't see it in
the bios (v1397) either.

Latest IDCC docs state that MCH header RPMs are not monitored.

[gtj]

Thank you very much!

sensors

gives the following

asc7621a-i2c-1-2e
Adapter: SMBus I801 adapter at 2000
in0:         +1.21 V  (min =  +0.00 V, max =  +3.31 V)
in1:         +1.01 V  (min =  +0.00 V, max =  +2.99 V)
in2:         +3.24 V  (min =  +0.00 V, max =  +4.36 V)
in3:         +4.99 V  (min =  +0.00 V, max =  +6.61 V)
in4:        +11.84 V  (min =  +0.00 V, max = +15.94 V)
fan1:          0 RPM  (min =    0 RPM)
fan2:          0 RPM  (min =    0 RPM)
fan3:          0 RPM  (min =    0 RPM)
fan4:          0 RPM  (min =    0 RPM)
temp1:       -62.0°C  (low  = -127.0°C, high =  +0.0°C)
                      (crit = -21.0°C)                  sensor = diode
temp2:       +44.2°C  (low  = -127.0°C, high = +127.0°C)
                      (crit = +65.0°C)                  sensor = thermal diode
temp3:       +46.5°C  (low  = -127.0°C, high = +127.0°C)
                      (crit = +65.0°C)                  sensor = transistor
temp4:       +41.8°C  (low  = -127.0°C, high = +127.0°C)
                      (crit = -128.0°C)                  sensor = Intel PECI
temp5:       -62.0°C                                    sensor = Intel PECI
temp6:       -85.0°C                                    sensor = Intel PECI
temp7:       +96.0°C                                    sensor = Intel PECI
temp8:      +118.0°C                                    sensor = Intel PECI

I only have a fan connected to the MCH fan header on the motherboard so I would have
expected to see one non-zero fan speed -- but actually I don't see it in
the bios (v1397) either.

Does that look ok to you? Do you know how to interpret the figures -- i.e. which fans and temps are which? The CPU is a QX9650.

Hmmm. I think I posted an interim version by mistake. You shouldn't have needed any patches. Sorry.

Clark is correct about the mch fan header. It isn't monitored.

in0 = vMCH
in1 = vCORE
It looks like sensors is scaling them incorrectly though.

I'll have to dig out my notes on the temp and fan mappings. I don't have them with me at work so I'll grab them tonight.