Jump to content

6970m and original M17x (a.k.a. R1)


Recommended Posts

Okay. I got my "pulled from new computer 100% tested" motherboard from eBay. The LVDS connector had been repaired. Brightness is stuck on low. Checked for the brightness PWM signal from the EC to the panel, nothing. So whatever took out the LVDS on this board also took out the EC's PWM. So the EC is bad and I will have to send it back. However, it still works so I did some testing while I had it.

I flashed my MXM-modified SBIOS, but no change.

According to the MXM software spec (which has disappeared from the internet at the moment) the MXM card should respond on SMBus before any other signals such as power_good become valid. So I hooked up my logic analyzer to the MXM SMBus line.

My LA doesn't have a good sample depth because it has a high sample rate, so I couldn't get a good picture of no MXM module vs. a working MXM module. I needed to get the whole sequence to compare, but my LA only has a 256k buffer so I can only sample for a second or two. The bus runs at about 67kHz so I need 500kHz sample rate for the I2C decoding to work properly.

An interesting thing I found was that the board is built to prevent the MXM module from receiving SMBus communication. It's buffered through a switching transistor to only allow the MXM module to write to the bus. I tried shorting the drain/source of the transistor so the module could receive replies but it evidently doesn't because nothing changed.

What I did figure out with my 6970m is that the SMbus lines are in some sort of tri-state condition. They don't switch with power on and stay around 1.7V. This is an invalid bus condition. Since SMBus is the first thing that is supposed to happen, things don't look good.

I don't think I will pursue this any farther as I have no way to verify that the 6970m is actually working. I also don't have a logic analyzer at the moment with sufficient sample depth to get a clear picture of the SMBus activity on startup. AMD ignored my request to join their development group so I don't have any data for the card. I will probably put it back together when I get a working MB and sell it for a newer release.

  • Thumbs Up 1
Link to comment
Share on other sites

Can anyone comment on putting an R2 motherboard in an R1? I just had a look at the R2 motherboard, and it looks pretty much identical except the MCP79 has its own heat exchanger, it's not part of the CPU heat exchanger. All of the other connectors appear to be in the exact same spot as the R1. Maybe I'll have the eBay folks exchange me an R2 MB instead.

Link to comment
Share on other sites

Well the R2 MOBO has already been proven to even run the latest of GPU's so i don't see why it would not work. I believe the board is slightly different in design if i remember reading correctly from a post that tried to plug a R2 board in an R1 case. I would of loved seeing the R1 (original MOBO) with a breath of new life.

[Guide] Installing AMD Radeon Mobility 6970 / 6990 in your M17x R2 (Single GPU and CrossfireX)

Link to comment
Share on other sites

Okay, I got my R2 motherboard and an 840QM. I have the same problem with the R1 board and the 6970m, fans spin constantly and nothing happens. I did some more research and figured out that it is probably hanging the system by not asserting power_good. Power_good is when all rails are within spec, before anything happens and before signals are applied to the bus. If the system is following the MXM spec then it shouldn't do anything until power_good is asserted. I figured out the other stuff I was looking at (SMBus) is probably just thermal info. The system reads EDID and other info from the LVDS DDC/Aux connection. Haven't sniffed that for the panel's EDID yet.

So probably this card is dead. Nice $300 paperweight, since I have no information on the card I can't even try to fix it. I think I might check out the eBay buyer protection on this one.

Anyway, putting the 260m in the R2 motherboard and trying to boot gives me a flashing num/caps. I've searched Google but have no idea what this means. I thought it might have something to do with the LCD, since it's the only non-R2 part (everything else is disconnected). I verified this by checking the MXM system information in the BIOS. The R2 is requesting a 24-bit LVDS bus where the R1 LCD is an 18-bit bus. I modified the MXM info to fix this, updated the checksum, and flashed it. I had to pull the chip since it wont even POST. No fix though. Still flashing num/caps.

Side note: the EC and BIOS code is split in the R2. The old flash location is now a 2Mb part that apparently just holds boot code for the EC. The first 2Mb of the ROM is mapped to the EC flash. Then there is a 32Mb flash in between the CPU/SB that holds the rest of the code. It's mostly offset from the 2Mb of the EC code, but there is some stuff in the first 2Mb. So with the R2 you can't simply flash the ROM image to the chip, it's split and there's some unknown code in the first 256k.

My R1 MB is supposed to be here tomorrow, hopefully this one actually works because the shipping from these guys is RETARDED slow. However I really think the 6970m is junk so I don't know if I'll put it in or keep plugging away at getting the R2 board working. I would much rather have the i7 over the C2Q, if nothing else than for the hyperthreading.

I'd appreciate if anyone could tell me what flashing caps/num is supposed to mean, and if there is a place to potentially repair my 6970 (besides Dell of course, can't imagine what they charge).

Link to comment
Share on other sites

SUCCESS! Finally something is working. I found this: Alienware

Apparently it wont POST without RAM. That's a first for me.

The R2 board is now running in my R1. I have an R2 motherboard, 840QM CPU, and R2 heat exchanger since the CPU/SB are in physically different locations on the R2 board.

Now I'm wondering if the video worked without my MXM system info BIOS mod. I hate to pull the chips again just to reflash and find out. Anyone know if the R1 LCD works natively on an R2 board?

Testing will continue! I'm not sure if the audio board will work or not, there are different part numbers for R1/R2 audio boards. More later.

  • Thumbs Up 3
Link to comment
Share on other sites

More good news. I went back to my hunch that lack of a power_good signal from the 6970m was holding up POST. Turns out I was right. I checked the two main regulators for what I think is the memory and for the core. The memory regulator was spot on, but its POK signal was being pulled low. So I checked the core regulator. Unfortunately it is so tiny and on the back side of the card that I can't test it with power on. I also can't see what the voltage is supposed to be because the VIDx lines that set the voltage are only valid with power on. However I did measure the core voltage at 1.04V, which is not unreasonable. So I decided to take a risk and cut the ISL62883C's PGOOD trace, since I knew vcore wasn't going to damage the core at 1.04V. Sure enough, it powered up and booted. I could only run it for a few seconds since the heatsink is removed.

So, the core voltage regulator appears to have drifted enough that the regulator is saying it's out of limits. I'm not sure I can do much about it really, since the card is a mess of tiny traces and vias. I think what I might do is get it running and measure vcore against the commanded vore in software, see how far off it is and probably just run it. Maybe undervolt it a bit to be on the safe side.

So my R1 board is supposed to come today, I can FINALLY test the 6970m in it with my modified BIOS, which was the whole point four pages ago. Sheesh.

  • Thumbs Up 3
Link to comment
Share on other sites

More good news. I went back to my hunch that lack of a power_good signal from the 6970m was holding up POST. Turns out I was right. I checked the two main regulators for what I think is the memory and for the core. The memory regulator was spot on, but its POK signal was being pulled low. So I checked the core regulator. Unfortunately it is so tiny and on the back side of the card that I can't test it with power on. I also can't see what the voltage is supposed to be because the VIDx lines that set the voltage are only valid with power on. However I did measure the core voltage at 1.04V, which is not unreasonable. So I decided to take a risk and cut the ISL62883C's PGOOD trace, since I knew vcore wasn't going to damage the core at 1.04V. Sure enough, it powered up and booted. I could only run it for a few seconds since the heatsink is removed.

So, the core voltage regulator appears to have drifted enough that the regulator is saying it's out of limits. I'm not sure I can do much about it really, since the card is a mess of tiny traces and vias. I think what I might do is get it running and measure vcore against the commanded vore in software, see how far off it is and probably just run it. Maybe undervolt it a bit to be on the safe side.

So my R1 board is supposed to come today, I can FINALLY test the 6970m in it with my modified BIOS, which was the whole point four pages ago. Sheesh.

Goodluck bro! Ive been following your endeavor and its an awesome project. Im glad the R2 motherboard works now and i hope you can get that gpu into a useable condition.

Link to comment
Share on other sites

@mw86: Thanks.

Unfortunately I was not successful at getting the 6970m running on the R1. It would have helped if the card was working when I was doing the majority of my testing and experimentation. It sucks that the card was dead so I was effectively wasting my time.... I am really hoping eBay's buyer protection will cover this, but I'm not holding my breath.

After I hacked the card and got it working, all I did was test it in the R1 MB with a system BIOS that I modified the MXM info on. It wasn't enough to get the card to POST, there is still something else in the BIOS that is holding it up. Since I successfully got an R2 motherboard in my machine, I don't really have the motivation to tear it apart again and keep messing with the R1 board. Actually what I want to do is get 7970ms for it, maybe get the 300W power supply working in it too.

The 6970 works with the vBIOS that it shipped with, which is not the Clevo vBIOS, however it locks up anytime the screen goes off. If I install the Clevo vBIOS it shuts down properly but I get bad screen tearing.

Something else that I can't figure out at the moment is ever since I replaced my original R1 board the LCD brightness has been stuck on low. I've tried two different R1 boards and now the R2 board and they all do the same thing. Windows can't control it either. It's so dim I can hardly see the screen. I'm about ready to hack the inverter so I can have brightness again.

  • Thumbs Up 2
Link to comment
Share on other sites

Okay I'm finishing this project up; I have to admit failure at getting the R1 and 6970 working together. Having the R2 mobo working means I don't really want to tear my computer apart for weeks on end again.

I wrote two articles on my website, one for the 6970 and another with some more details on the conversion to the R2 motherboard. That post also includes files for download to dump / program the two flash memory chips on the R2 motherboard. Thanks to everyone for the help.

  • Thumbs Up 2
Link to comment
Share on other sites

Im sorry to hear that it didnt work out. Im also sorry to hear you had wasted some money. I hope ebay can help you. The 7970m and the 300watt psu sounds like agreat addition. You could go dual R2 psu in parallel like StamatisX did :) he wrote the a really wonderful guide on it and its even hosted on our Frontpage. :) googluck in your endeavors. Feel free to discuss further projects here, I would be glad to hear about them.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use. We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.