Alienware M17xR4 with PNY based 680m from HP-Z1 design.

Stats:
Alienware M17x R4 w/A13 stock bios, all other parts stock and nothing is overclocked or modded.
3720qm at stock speed, 16GB Samsung ram (8x2) also stock from Dell
Intel iGPU of course HD4000 and the standard 60hz 1920x1080 display (all parts standard)
7970M - died (as per recommendations in a post from over a month ago)
Windows 7 64bit - fresh install, followed install order and stopped at graphics (after a failed Nvidia attempt to install), on a HDD non-UEFI.

Bought a 2GB 680m from Compeve on Amazon. The card has the nvidia emblem on it, but Compeve says it is a PNY OEM card. PNY says all their cards are stamped with PNY, however another person says they don't make it even when I gave him the product code for the PNY HP 680m for their HP-Z1 workstation. Company says PNY provided these to them.

Problem, I can't get the card to accept any nvidia drivers, plain, mod, or anything.

Bios/Vendor/DevID on card:
GPU z states bios 8.04.33.00.24 6/4/2012 Vendor Dell (1028), Revision A2, DevID 10DE-11A0, Samsung GDDR5
Kepler Bios Tweaker (using exported rom file via GPU-Z) says 80.04.33.00.24 6/4/2012 Vendor MSI (1462), Revision (sku 002), no memory ID (says memory clock is 1800)

HardwareID via Device Manager
PCI\VEN\_10DE&DEV_11A0&SUBSYS_05511028&REV_A1

680m Board has green not blue PCB and verified Samsung memory.

@Raidriar, @j95, @Mr. Fox
You three appear to be the resident experts on all things Alienware and GPU. I cannot figure out what to do to get this card to work. Modified Video Bios, Modified System Bios or something else including the actual driver version. I have learned a few things today about what not to do once Optimus is working. I would prefer to keep Optimus capability.

Here is a copy of my bios
https://www.dropbox.com/s/yot0sq36rha47hj/680M Bios as of July 26 2016.rom?dl=0

 

Step 1: download svl7's modded vBIOS which is supposedly universal, and flash that with a DOS usb bootable using nvflash. Google the proper command syntax, its not difficult I believe. His vbios is on TI: https://www.techinferno.com/index.p...king-editions-modified-clocks-voltage-tweaks/

2. Make sure you grab j95's optimus modded INF, not the one for PEG if you really want to keep optimus. Download his latest INF for the latest nVidia driver

3. Extract the nvidia driver setup, but don't execute (will not recognize your hardware due to missing IDs in the INF)

4. Copy j95's INF in the display directory of the extracted driver files, you should be overwriting 2 files

3. Run DDU before you run the nvidia driver installer.

Report back when you've done all of the above or if you have difficulty on any of these steps.

PS: I wish you had heeded our advice when you asked for replacement cards a while back. Rule of thumb, always grab Dell or Clevo MXM cards only. Some machines are more picky than others. I've never heard of a PNY manufactured MXM card, but it's not impossible. However, it's safe to say nobody on this forum has ever used one. If you're still in a return window, return the card and purchase a Dell or Clevo card instead. Less headaches that way

 

Thanks, @Raidriar, you were the one who convinced me to stop wasting time on my dead 7970m, glad I listened. Dell wouldn't talk to me at all since my warranty expired, only about 2 weeks ago I finally got the parts people to help me and both there parts and warranty replacements were out on 7970m and 680m, as I was going to return this one back to compeve.

Okay I have these I downloaded as a more than 5 post member, which one is recommended, especially since I don't know if the PNY is 2 or 4gb still, 680m - 80.04.5B.00.02_'OCedition'_revised_00.zip Dell 680m - 80.04.33.00.32__'OCedition'_revised_00 - OV 1000v.zip Dell 680m - 80.04.33.00.32__'OCedition'_revised_00 - OV 1025v.zip Dell 680m - 80.04.33.00.32__'OCedition'_revised_00 - OV 1050v.zip Dell 680m - 80.04.33.00.32__'OCedition'_revised_00.zip

2. Will have to look if I did grab that, I can't get anything else till Aug 18 2016.

3. Will report back if I have the drivers downloaded, I have so many files I have downloaded since the first problem in May 2016, it could take a lifetime just to sort.

4. When I run the DDU, should I also remove Intel HD Drivers as well or leave them (they are the originals from Dell, but are only marked version wise for Dx9 - per Intel naming scheme, I have 8.15, should use 8.17. - I assume update via device manager and not install), assume same install route, Intel, Nvidia.

I thought I was buying a Dell on Amazon, because the ad seemed to hint it was a Dell based on my reading and also seeing the price ($329 plus picked up from Square Trade (they do Office Depot and Staples extended policies) a 3 year policy for under $30)

It is actually finally starting to cool down enough for me to try to get sleep will get started on this in morning when cool.

Once again you have no idea how greatly appreciative and thankful for you help. If you saw my bookmarks for Alienware Laptop Issue, it would look more like those old phone books, as I had to literally try to figure out which site might be more valid than another.

 

Its all good man, that's what an enthusiast community is for.

1. As far as I know, only the blue PCB clevo cards are 4GB, so I assume everything else is 2GB. I would go with the first vBIOS you listed, since it operates on stock voltage. If you want to tinker with overvoltage later, thats fine (not a fan though). For starters though, just flash 680m - 80.04.5B.00.02_'OCedition'_revised_00.zip.

2. At time of writing, 369.00 is latest. j95's thread is here: https://www.techinferno.com/index.p...t-upgrades-driver-support-modded-inf/&page=54

3. At time of writing, latest nvidia driver is 369.00. Grab it directly from nVidia's website

4. If you're running stock intel hd 4000 drivers, DDU them as well and download the latest intel hd 4000 drivers directly from Intel's website

 

@Raidriar
Ran into problem I have seen everytime I have tried to flash the vbios.

Reference this link when reading the --version section (reason it seems to match my --version)
http://forum.notebookreview.com/thr...t-70-w-gtx-680m-owner’s-lounge.678611/page-74

==================================================
Checking for matches between display adapter(s) and Image(s)...

Adapter: Geforce GTX 680M (10DE,11A0,1028,0551) H:--:NRM B:01,PCI,D:00,F:00

WARNING: Firmware image Board ID (E10D) does not match adapter Board ID (E10E)

ERROR: Board ID mismatch

==================================================

Since the -4 has been removed from the most recent version of nvflash for dos, what should I replace it with.


------------------------copy of --version command--------------
*** NVFLASH LOG FILE ***
Version 5.163


Command: version
FindDevices:
Device:00:00:00=8086:0154:1028:057B
Device:00:01:00=8086:0151:1028:057B
Device:00:02:00=8086:0166:1028:0551 GPU
Device:00:14:00=8086:1E31:1028:057B
Device:00:16:00=8086:1E3A:1028:057B
Device:00:1A:00=8086:1E2D:1028:057B
Device:00:1B:00=8086:1E20:1028:057B
Device:00:1C:00=8086:1E10:1028:057B
Device:00:1C:03=8086:1E16:1028:057B
Device:00:1D:00=8086:1E26:1028:057B
Device:00:1F:00=8086:1E57:1028:057B
Device:00:1F:02=8086:1E03:1028:057B
Device:01:00:00=10DE:11A0:1028:0551 GPU
Device:01:00:01=10DE:0E0A:1028:0551
Device:07:00:00=1969:1083:1028:057B
Device:08:00:00=10EC:5209:1028:057B
Device:08:00:01=10EC:5209:1028:057B
Action: Display version.
Action: Identify EEPROM.
Atmel Serial EEPROM Identify.
Reported Manufacturer Code (JEDEC): C2
Reported Device Code (JEDEC): 2012
EEPROM ID (C2,2012) : MX MX25L2005 2.7-3.6V 2048Kx1S, page
Action: Read EEPROM image.
Reading EEPROM bytes.
Start: 0x00000000
Count: 262144
Read complete.
IFR Data Size : 372 bytes
IFR CRC32 : 2A32E0C5
IFR Image Size : 512 bytes
IFR Image CRC32 : 3FB2F8E4
IFR Subsystem ID : 1462-10BC
Image Size : 89088 bytes
Version : 80.04.33.00.24
~CRC32 : 028FC3E1
OEM String : NVIDIA
Vendor Name : NVIDIA Corporation
Product Name : GK104 Board - 2051b002
Product Revision : Chip Rev
Device Name(s) : GeForce GTX 680M
Board ID : E10E (GK104 P2051 SKU 2)
PCI ID : 10DE-11A0
Subsystem ID : 1462-10BC
Hierarchy ID : Normal Board
Chip SKU : 700
Project : 2051-0002
CDP : N/A
Coproc Build : NO
Build Date : 05/23/12
Modification Date : 06/04/12
UEFI Support : No
UEFI Version : N/A
UEFI Signer(s) : Unsigned
Sign-On Message : GK104 P2051B SKU 002 VGA BIOS (HWDIAG)

 

Since this is a M17x R4, there is no need for the -i flash to specify which PCI port so I think the proper syntax is:
nvflash -6 vbiosname.rom

 

Since this is a M17x R4, there is no need for the -i flash to specify which PCI port so I think the proper syntax is:
nvflash -6 vbiosname.rom

 

( @Raidriar )

Tried
nvflash -6 vbiosname.rom, error remains the same

Just in case this matters, I named the .rom file 80045B.rom (I am assuming it doesn't but just in case)

Board ID Mistmatch (E10D) does not match (E10E). I don't know how much the Firmware Board ID is taken as the absolute most accurate or its role per se when doing a modded vbios.

I could not find a PNY Firmware Board ID (something tells me I am misunderstanding Firmware Board ID with Vendor ID. I could not find who E10D who it belongs to. PNY itself isn't listed. E10E on googling shows MSI.

To prevent complications I will wait for new instructions, since if I try things, it makes it more complicated and even
"You did what! Well that is whats wrong" :

Edit: Jul 28, 2016
Did some legwork, and got the md5 and sha1 values of the rom file that was backed up before attempting to put the modified bios on, used microsoft File Checksum Integrity Verifier version 2.05 (dos based) with -md5 for the md5 check and with -sha1 for the sha1 check
Values
MD5: 76c3a2595b8564561f80d47e09777c1c
SHA1: fbd94c28c2bd4bd88cd4b49622895568fef05b9c

If the vga bios at techpowerup is to be taken as valid, then the only card that has both entries (no other card had either 1 on its own, so this is the only card that had either or both of the values - both in this case)
MSI GTX 680M 4096 MB BIOS

I don't know if that means anything, but figured I could do some research as well.

 

Its possible the seller has no idea what they are selling. If your vBIOS matches the MSi 4GB 680M MD5 signatures on TechPowerUp, then 100% it is a MSI card.

the nvflash command I gave you should have worked, did it not prompt you for option to overwrite, press 'y' to confirm? When flashing prema's vBIOS to my 980M, it gave this prompt and it worked fine, overriding the ID mismatch. Hopefully somebody else will chime in, this is all I can think of at the moment, other than trying this alternate command:

nvflash -4 -5 -6 vbiosname.rom

if it prompts you to ignore ID and flash anyway, press 'y' to confirm

 

@Raidriar

I wanted to wait till you gave me the go on the -4 -5 -6 (even though -4 isn't listed anymore in list of commands) before I did it.

It beeps but won't give you the response of typing Y for yes. I have seen others post screen shots and my nvflash -5 did not provide even the option to type the letters YES as a final warning before flashing.

I am going to do -4 -5 -6 nvflash vbioname.rom now that I have heard back.

The new MD5/SH1 result of doing a backup after flashing
md5 7048ed912f73565553f132c0ce2d8bfa
sha1 140cr4e0fab200f27c71f066277d093fc7b3c0e82

Once again just to do some work, I googled both. The MD5 matched an unverified Dell GTX 680m 2048 MB bios ( https://www.techpowerup.com/vgabios/142349/142349 ) however it didn't match the sha1 value on it, as it was missing the r after 140c.

If this info is not helpful (md5/sha1) please let me know, just trying to be as detailed as possible in any response.

Flash was completed without any issue. Will update the next state after attempted.

 

I hope it is okay, I got delayed tonight and I didn't want to rush the driver install part and then try to fix it when I should be dreaming of what my laptop would be if it works - I am compulsive when it comes to fixing issues (stayed up 36 hours once fixing an issue - don't like to give up - working on stopping that, getting older means I can't pull those all day marathons as much anymore.

I will follow up in about 16 hours from now. I hope I can dream/sleep just because of thinking of what it will finally be like to have my laptop back to full readiness, rather than limping across the finish line on the HD 4000.

Thanks again @Raidriar

 

So did you manage to flash the vBIOS? Kinda lost me back there, not sure if you were successful or not

 

I don't know if j95 takes any comments about his instructions, but he says to delete the folders below, then he says keeps these folders only both in the same space, if you know what I mean. I wonder if he original was showing the folders to remove and then changed or went with keep folders only. When I just read it again (I read it dozens of times but never paid that much attention)

 

Do me a favor and just install this, see what happens:
http://international.download.nvidi...notebook-win8-win7-64bit-international.hf.exe

Just download and install it as is, don't modify anything

 

(This was typed before you new message posted so this is not the results from the new link just provided above this).

Hmm, Nvidia says Failed to Install.

I failed to install drivers for HD Audio and the Display Driver, Physx was the only thing to complete the install process successfully per the end of the install info screen. I opened up Device Manager and now there is only the Intel HD 4000 listed, there is no second graphics adapter.

Rebooted, windows says detecting a new hardware, installs Standard VGA Adapter, wants computer restarted. Standard VGA Adapter is now back with the Intel 4000 under the Display Adapters in Device Manager. Hmm.

UPDATE #1:
Currently in safe mode to have DDU remove the Nvidia drivers and all the content of that.

UPDATE #2:
Now installing, All installed (obviously left the Geforce Experience out) rebooting

LASTEST UPDATE ON DRIVER YOU REQUESTED @Raidriar
nVIDIA Geforce GTX 680m - still has exclamation in that little yellow triangle (inside the device in device manager windows has stopped this device..... (code 43)

 

That's a hardware fault. Your card is busted. the driver I linked was a hotfix driver that installs regardless of device and vendor ID, and should work for any nvidia card. I would contact your seller and tell them they sold you a borked card.

out of curiosity, what OS are you running?

 

Windows 7 64 bit Pro Edition (without Service Pack 1 installed, since I am doing a fresh install and following the install order that has patches/updates/packs installed much later in the install.

 

yeah ok, just was checking. Almost 100% certain your card is a dud, hope you can get a refund

 

Would the service pack play any role? I have not installed any updates from WU. I did some research and supposedly you must have the sp on for the more current nvidia drivers, but is that true? @Raidriar

Paid with Amex through Amazon, so at least two ways that I should get a refund. Especially since I can claim fraud just based on the offer.

They have a new one ready to ship with a pre-paid label, but you responded so I told them to hold off, till I could confirm this think was borked (if it did work, I could have got stuck with the return shipping on a working card)

 

Just to cover all bases here, just download and install windows 7 sp1, re-install the hotfix driver I linked you, and see what happens. Code 43 tends to be a hardware fault though, 8 times out of 10

 

Man I wish MXM was actually PCI-E, so I can pop it into my desktop and make it quicker to see if dead or alive. Okay sp1 applied, disabled all items in startup, disabled essentially all non-microsoft in services (like google update, Intel MEI, etc), disabled driver signing.

Driver installation complete. Firewall disabled. Restarting.

UPDATE: @Raidriar Nope, still error 43. I would assume switching to discrete only via the fn-f7 will be useless? I will finish doing as many of the Windows Updates I can and then will post a the final test results and then will either be jumping up and down or looking for a very tall building to drop my laptop off of. I will get that new card on its way and this one sent back at no cost to me.

 

@Raidriar
To save you the time of looking back in, if the card is a failure, I will just notate it failed after Windows Update has updated mostly everything (except for the optional stuff) probably take a day or two because of how Win7 WU seems to take forever these days to find updates. If it is successful or something really odd happens I will post a note so you know you have solved it or there is something new that happened after all the patches.

If it fails, I won't leave comment, instead I will await the arrival of the replacement unit. Hopefully they include new thermal pads and paste so I don't have to go order some or run down to my local best buy/frys. If it ships like it did last time it will take about 7 business days (so they won't ship out on Sat. so Monday would be the earliest (August 1st) and latest arrival should be by (August 12th a Friday). Once I have got the card seated, thermal pads and paste on, if you are still willing to deal with another go around, I will contact you (if there are any special instructions you want me to follow before I put the card in the system and before I do the bios flash, let me know if you can that way I can be ready to move through the steps faster for you - I have listed the steps I can take in advance of if you will be of assistance again - feel free to say no, I wouldn't be upset, you have done more than your share)

I will post the unflashed bios saved rom first with the printout from nvflash (hopefully not just a repeat of what we have been through) and then wait till you say go ahead and use same mod bios or change to a different one.

I assuming the following for the new card
1. First go to safe mode, ddu the nvidia drivers and intel drivers, reset after each one, do nvidia last and choose clean and shutdown.
2. Yank card out (gently).
3. Reset bios - take out batterys (both), ac adapter, memory, and hold down power button 30-60 seconds.
3.5 - Do a fresh reinstall of Windows 7 64bit (to clear at debris from this past attempt) and should it be brought up to date fully (sp1 and all updates on WU)?
4. Restart system (without the new nvidia). Install latest Intel drivers (2015 circa)
5. Insert Nvidia but do not flash and do not modify at all. Launch GPUZ and save the rom. Then launch into dos via usb stick and run nvflash to get the version/list. Then see if it is a match to the previous or not and also try to find the bios in the vga database and just let you know if it matches anyone of them via the sha1/md5.
6. Then I wait till you have free time to catch up (if still assisting) on the post and tell me what to do next in terms of bios to flash, driver to use, etc.

So hope you have a good 2 plus weeks (just think no me nagging you - j/k). Seriously though I appreciate your help. I wouldn't have been able to get to this state of knowing if the card is good or not with time to return it without your help. Thank your very much for all of the help you have given and might have to give again. Enjoy your August, hopefully you aren't in one of the hot spots right now.

Thanks,

Tim

 

@Raidriar
The card arrived today, they apparently sent it by ground 3 day. Before I put any thermal tape, heat sink, or anything on the card I thought I would throw some information from the chip and PCB that can't be seen once the heat sink is put on. I will hold off on even putting it in the system so in case you see something of concern.

Thanks.

On the chip itself
A Taiwan 1310A2
NFF474.01
N13E-GTX-A2

Information found about the labeling
N13E-GTX-A2 - Techpowerup says it is an Nvidia 680m (no branding) https://www.techpowerup.com/gpudb/344/geforce-gtx-680m
Some say that it is a Quadro 680m (of course I know Quadro is a different product line from the 680m)
1310A2 - seems to refer to several Nvidia models
NFF474.01 I can't see to find it to match to anything


Memory Chips
Samsung 319
K4G20325FD-FC04
Samsung manual says K4G20325FD-FC04 319
Samsung DDR5 2G
16K/32ms
Organization: x32-16 banks
POD_15 (1.5V,1.5V) (VDD, VDDQ)
5th Generation
Package Type: GDDR5 : 170FBGA (Halogen Free & Lead Free)
Temp and Power (VDD): Consumer Normal
Speed: 0.40ns (5000Mbps)

On the PCB
Above the GPU chip socket:
MS-1W091
Ver 1.1

Once again hard to find out who or what, so it could be a general reference.

This first image is about the only image I could find on the net quickly that looks like mine, since I have never taken pictures of a video card before so my images aren't as sharp or neat)
1. From web


2. These are four images from a Samsung Galaxy S5 Cell Phone Camera

HTML:

<a href="http://www.imagebam.com/image/f6a2e8498553047" target="_blank"><img src="http://thumbnails116.imagebam.com/49856/f6a2e8498553047.jpg" alt="imagebam.com"></a> <a href="http://www.imagebam.com/image/a89cf0498553058" target="_blank"><img src="http://thumbnails115.imagebam.com/49856/a89cf0498553058.jpg" alt="imagebam.com"></a> <a href="http://www.imagebam.com/image/7cd850498553065" target="_blank"><img src="http://thumbnails116.imagebam.com/49856/7cd850498553065.jpg" alt="imagebam.com"></a> <a href="http://www.imagebam.com/image/188b25498553077" target="_blank"><img src="http://thumbnails116.imagebam.com/49856/188b25498553077.jpg" alt="imagebam.com"></a> 

 

MS-1W091 is dead giveaway to being a MSI card. Only MSI uses that type of internal codenaming. Did you see if the computer posts with the card installed?

 

@Raidriar
No I did absolute no install or anything, since in case you needed to know something from a chip, a number, a certain location of something, etc. Easier to look it over before the thermal paste is applied than after it has been applied.

I will go ahead and get it installed. One question. When I was screwing down the previous graphic card with the Nvidia X Bracket, I had to push down on the screws to tighten them (more force than you normally would do for screwing things tight. The bracket is correct, but the screws I assume are from the 660m (I don't know if the installer who put in the 7970m at my home changed the screws or not). From what I have read it is the X bracket that is the most important vs the screws, just want to make sure.

Once installed, I will save the bios first, then post it here in case anything is off before applying the modified bios, but only after a full clean install of Windows 7 (just to insure nothing is left over anywhere in the reg or win dir from the previous). Will take me about a day to get the new fresh install of Win 7 installed and then fight to get the updates from Windows Update as something is seriously wrong with the update site (even my brother has run into the windows update just searching and searching forever and not listing updates)

Thanks Again!

 

Oh I hope you didn't push down too hard....x-bracket comes with different length of posts for the screws to get in. If the post is shorter, you must compensate with a longer screw or change the x-bracket to a Dell style bracket. Push down too hard, and you crack the die.

 

Hmm. So I assume don't apply too much force then? The X bracket they sent me, the screw holes on the X bracket are shorter than the xbracket from the AMD 7970m. The AMD 7970m the die appears to be slightly higher/taller than the Nvidia 680m die/chip. When I put the bracket back on the 680m (old one), they holes to be screwed into don't really rise beyond the pcb, they are almost at the same level as the pcb.

Should I bend the legs of the heat sink X to get them closer? If I need longer screws, I assume they can't be replaced by just any screws, would have to be specially fitted size and height wise for just that heatsink and nvidia die.

 

Do not use the 7970M x-bracket, it will not match the component arrangement on the back of the card. You need to get a hold of a Dell nvidia x-bracket. Cheap on ebay, i'm sure some forum users have spares for sale from dead nvidia cards. Do not over tighten or apply pressure, you'll kill the GPU very easily. I've done this a few times myself, and it's an awful feeling. You should not have to use any extra force whatsoever.

 

@Raidriar

When I test out the x bracket that came with the card (original), I have to press the legs of the heatsink with the screws down a little to make contact with the pcb. Is that normal.

Do you know where I can find a picture of how the dell x bracket should look when put on, hopefully a view from the pcb level to see if they match what I have and then of course how the heat sink should compare?

If pics on aliexpress are right, then I don't have the correct x bracket

http://www.aliexpress.com/item/For-...Vidia-GeForce-675M-680m-580m/32344212453.html

Do you have a spare (you deserve first shot at sale obviously) if not who would you absolutely trust and are they located in the US and what method of payment do they take (prefer not to just send a check, but I know will have to be flexible)

I saw this on ebay, can't tell if they rise much
http://www.ebay.com/itm/X-bracket-w...371926?hash=item3f6447e116:g:EV4AAOSwvg9XVRmY (didn't notice from China)

Will a X bracket from the 660m Dell model work? Since I assume sellers on the net might put a premium price on the part since the 680m dell model was likewise premium priced.

Special Note: I wonder if I should find someone who would be willing to test my 7970m card to see if it works in their Alienware M17x R4. Could still be a mobo issue, I would hypoth, but then if the Intel GPU works, it would seem then the 680m should work since the mobo did recognize just couldn't load drivers while the 7970m everytime I tried to get into windows with the amd drivers, the system would crash (now that could be a driver/amd/nvidia issue however in terms of how the hardware interacts with windows)

 

Just post in the WTB section of the forum, I'm sure somebody got spares laying around, but I do not. There should be absolutely no bending involved, so you definitely have the wrong bracket on there.I would try to get a refund for the 680M if possible and go buy a genuine Dell card.

as for the 7970M, it is likely core failure. They have a reputation of doing that. MXM slot go bad...unlikely.

 

Any recommendations of who to buy from? More important how does things or should things work in a wtb offer.

 

@maxslo typically has a nice cache of alienware parts laying around, as does @Rengsey R. H. Jr.

Ask them if they have spare nVidia x-brackets laying around that would work in Alienware systems

 

Okay. Sending message to maxslo now and then Rengsey R. H. Jr. Curious as why they have spares laying around or are they an unofficial parties that have parts for the Alienware laptop series.

Well once I have secured a proper X bracket I will be back in touch in this forum. Thanks again.

Also do you know what PNY ID is, there is someone asking about it in another forum.
http://forum.notebookreview.com/threads/hp-z1-mxm-cards-for-laptops.794664/#post-10310287

 

@Raidriar Is this the one dell 3mhht
http://www.ebay.com/itm/Genuine-OEM...44d08dc&pid=100623&rk=2&rkt=4&sd=231066978800

safe to buy from them?

 

I would think so. US Seller, highly rated, looks to be the right part. Just be careful prying the old one off not to damage components or the PCB. I usually try to push the screw posts out then gently pry with a flathead screwdriver with a paper towel piece acting as a buffer between the screwdriver and the PCB.

 

When you say prying, all I did was undo the screws, then it had tabs on all four corners of the X and it came off with ease, now you have me worried there is more to this process and I am missing something in terms of putting the heat sink on or taking it off (I know the x bracket on the 7970m is on the card as if they used super glue to put it in place.). I assume any used model the adhesive will be weak or not at all on the xbracket itself.

I might go with that because it has 30 day back free shipping, I just pay for return shipping. The next problem would be the thermal pads. I have seen so many mm this and mm that. One person said you don't even need really to have that extra piece of metal with pads to cool the memory chips on that underside. I have seen comments that if you get the wrong mm size it could if too much cause you to overheat, if not enough to overheat.

 

Trust me when i tell you, return the card, pool your money, and get a 970M or 980M. You WILL need thermal pads on the bottom VRMs, whether there is metal or not does not matter. Put thermal pads underneath them when you install the MXM card. As for the other ones under the main heatsink, use 1.0MM on all of them

 

@Raidriar
I know what new career I don't want, Laptop Memory Card MXM Thermal Pad replacer. I never had such a time putting on thermal pads like I did with this. I installed the card and will now attempt a driver install without mod the bios at all. Will use the Nvidia drivers you originally had me use (not the drivers of last resort the all inclusive
http://international.download.nvidi...notebook-win8-win7-64bit-international.hf.exe

Will use the 344.11 and then report back on results.

Got the correct x bracket in the mail today, forgot to add that, since I was waiting on that for the correct model.

Primary goal tonight, is to get the card recognized properly. If successful will then go on the hunt for good thermal pads (even those the ones I had were Laird) and good thermal paste. I heated the paste up as instructed online by one of those on NBR. However the paste stuck to the syringe and as I attempted to get it off, it essentially rolled around the tip till the paste was around the tip and not on the GPU. I did get it to finally lay down the middle as youtube videos showed.


OH MY GOSH I AM SORRY ALL CAPS - SCREAMING IN EXCITEMENT, THE 680M INSTALLED AND DOESN'T HAVE ANY ERROR CODE OR LITTLE EXCLAMATION POINT IN YELLOW OVER THE DISPLAY ADAPTER.

woohoo!!!!!!!!!!!!

Now of course I need to settle down and figure out what I am suppose to do next (other than replacing the thermal pads and paste)

Okay I have a working 680m, MSI. Bios is 80.04.33.00.24 (all the boxes this time are filled in on the gpu-z software page for the Nvidia). Only problem is that is says Memory size is 4096mb, for what should be a 2048? I now have sensor reports coming in as well.
GPU Core: 135, mem 162, temp 42c, load levels (gpu, memory, vid eng, bus) 0%, memory ded/dyn 0mb, perfcap reason Util, Vddc: 0.8370v.

Where do I go from here then? How to verify Optimus status? Should I flash the vbios with the modified one you had me download or stay with what I have currently. I don't plan to OC the card (not a believer in OC'ing, just too risky for me to risk). What should I do to test the card but not stress it?

Thanks so much for getting me through this massively bumpy, obstacle in lane, deer in headlights path, now I hope we are in the home stretch (my fingers have been taped crossed).

 

GPU Shark now shows the 680m as well, before I could never get the Nvidia listed.(it also says 4096mb memory. The temp is climbing it is at 51c already in just 2 minutes of running gpu shark and having the detailed view open (52 now) HIghest so far is 59 just moving the mouse around, no games using aero desktop and slideshow after display goes into power saving mode after 1 min no activity.

Nv Control Panel responds and doesn't crash either or refuse to open (as with the previous card we started with).

Intel Extreme Tuning Utility sees the graphics card, its ramdac, driver date along with the HD 4000, something that didn't work under the old card. Also says 4096mb for the 680m

 

So it sounds like your problems are solved, which is a good thing. Enjoy your M17xR4!

 

@Raidriar If I can still bother you on a few items, I am grateful for the help you have given, so I will stick with you on advice.
1) Do I still need to flash the bios or is it fine the way it is?

2) The memory still confuses me, however I assume that the line in the nvflash (I have bolded the only part that confuses me (this also is not the EEP
EEPROM ID (C12,2012) : MXMX25L2005 2.7-3.6V 2048Kx1S
is not the actual memory of the card. So how can I have nvflash tell me how much memory is on the card, if it can?

3) What test/burnin should be done for making sure it won't overheat and most important to validate the memory as not having any bad modules?

4) If I am crazy enough to ever go to Win 10 on the M17x R4, I understand I would need to use UEFI. Will that mean I will need to flash the gpu bios to a uefi version?

5) Is there a better application than Nvidia Control Panel to use to tune the card for each application? Which program can I use to control the fan for the GPU (I assume the card communicates with the mobo to tell it how hot is is and then the fan adjust accordingly - so I don't know if adjusting fan speed can be done with a 3rd party application)

7) Which of the "tuning" applications should I avoid (for example EVGA Precision X shouldn't be used in Win 10 it sounds like).

8) When I update the GPU driver, I assume I need the modified nvdmi.inf for the current driver version which I can pick up over at TechInferno or just see what they modified in the 344.11 and adjust the nvdmi.inf for the driver I am updating to? I know to delete all but those few directories as shown on TI to avoid screwing up the install.

9) Last but not least, how do I tell the Nvidia not to handle the mundane task like the boot into Windows, so it won't draw any power (drain the battery) while I am doing nothing graphic intensive in Windows. Essentially remain idle/0 watts until I actually use an app that I told it via Nv Control Panel to take over the graphic duties.

10) Best thermal paste and best thermal pads that you can tell the difference between 2mm and .5 mm pads. Link or reference?
Once again thank you and yes I will finally be able to start enjoying my laptop again!

 

I've been away for a while, sorry.

1. I would still flash the svl7 vBIOS.

2. Ignore this. Check with GPU-Z, it will tell you exactly how much memory you have.

3. Do not use Furmark or OCCT. Simply run 3Dmark11 a few times, and if it passes every time, you are stable. I use 10 times test as a rule of thumb.

4. You don't need UEFI for Win10, that is only if you want to use 980M or 970M. Legacy is easier to work with.

5. Nvidia Inspector

6. ....(you skipped this number)

7. I would avoid EVGA precision X. I use nvidia inspector exclusively, does the job for me.

8. I just extract the latest nvidia driver, look for j95's post for the latest driver, and overwrite the existing INFs, then run the installer.

9. I do not know the answer to this, never used optimus.

10. I use 1MM pads all around, works fine for me. make sure you cover the VRMs and the mosfets

 

Sorry if I created any confusion that caused you to think I failed to respond to you.

Glad your back, was worried after dealing with me you threw you keyboard out so didn't have to do anymore responses

I don't know how I missed 6, now I need to have my niece to teach me my 1 to 10's again, or I didn't want to spook the 680m (actually only noticed I missed it now)

1. I would assume then the same version you initially had me use. Will report back if there are any issues and won't force a flash until cleared with you.

2. It is a 4GB MSI. The company who sold it to me on Amazon, Compeve, I gave them a complete write up via e-mail (I made sure they knew the credit went to TechInferno and Notebookreview - I didn't use your name as I hadn't got your okay to release it at the time I wrote the message to them)

3. I have tried the Uningine 4.0 Valley Benchmark. Here are the results from one test.
My 680m, running the Uningine Valley Benchmark with:
DX11
High Quality
Tesselation Normal
Stereo/MultiMon - disabled
AA off
Full screen with system resolution (system is 1920x1080 I would think).
Final Score 1011, min/max 7.5/90.9 fps, fps 40.1
Temp 66c was highest recorded.

At idle, it sits around the 44-48c. If I load a game like Star Trek Online (which I don't know how much of it is CPU dependent) full screen with the settings at middle to middle-high, the temps get up to about 56c.

GPU-Z seems capped at 60c. GPU-Shark reports that it gets to 66c, but doesn't report clock speeds or anything (just GPU usage).

4. Since I will be using a Samsung 850 Evo 1Tb as my boot drive, it seems the consensus is to use GPT for reliability and performance, but you can't boot a GPT in Legacy. So I haven't started my rebuild since I don't know if I should go with GPT/MBR and UEFI/No-UEFI. I spent the time since I last responded googling the internet to death over uefi and its effect on hardware (optical drive, card reader, usb, etc.) and whether the 850 Evo is a true 4k, 512, or 512e and what is best for it to be used as. The drive that will be used as the backup of the SSD is a WD 1tb Blue Cav that is Advance Format (wish I had the money for another SSD but was lucky to get the first one at $199 at Frys, special deal with company) and whether GPT or MBR was the best.

5. I used Nvidia Inspector but apparently the 680m doesn't have a fan control chip (forgot the technical name), so I can't force the fan to come on sooner or stay on longer (however during the Uningine test I heard it spin up when it temps were going up and spin down when temps were falling. I assume I would make a "profile" for each game that I want running on the 680m but wasn't sure or if I was missing some sort of different GUI.

6. Forgot last time so have to be consistent.

7. When I click on the unhide overclock it shows performance level 2 - P0. Other options are Level 0-P8, and Level 1-P5.
In reference to a desktop 770 GTX 4GB EVGA FTW would you also advise against the EVGA Prec. tool?

8. Perfect!

9. There is a spike when I move my mouse around (Logitech Wireless) where the clock hits 771 and then drops back to 135 (voltage also spikes as well to max at same time and interval, however GPU usage remains 0%, per Nvidia Inspector little screen tool with status on all things, and max temp is 48c.

10. I used the following as a guide for the pad placement
https://www.google.com/search?q=ima...=I7W7V-jVLoKwjwOFhp24Dg#imgrc=WQoPLjm_8zadrM:

I had issues with getting the pads to stay on and not come off on me (even when using anti-static gloves). On the underside, the x-bracket covered the vram running down the side, I still had the thermal pad from the 7970m that was sitting on the motherboard (didn't have a piece that big to replace it either) to cover the other vram chips. The only reason I question this pad and the others, is that the parts they circled sit highest on the board and didn't know how the thermal pad would stay on the chips since those 3 items would have another 1mm added to the their "height" while the rest like the VRam would still not be any closer in height to those 3 circled items. Lets just say I wasted a great deal of the thermal pads they sent me (It reminded me of playing with Playdooh as a child)


-------------
Beyond these issues card has been working fine. I have avoided pushing it beyond the Uni 4.0 test as I know the thermal paste and pads need time to become effective (burn-in but over months). I looked at replacement thermal pads to redo the card now that I know it works, but a sheet of FujiPoly was nearly $100 (since I don't speak in meters, I don't know if that pad was enough for redoing the card 2 times or let alone to do it one time. The paste seems to be either Gelid GCD Extreme (about $8-10 for a few applications), IC Diamond 7 (saw a 24 carat but figure marketing hype) and MX-4, all for the same approx price as Gelid.

So thanks again for the continued tech support, I need to buy a support contract from you!

 

@Raidriar
Wasn't sure if I was to move ahead with the flash, so attempt to do so.
I renamed the file of 80.04.5b.00.02_oc_rev00.ROM for Dell 680m, to 6805BOC.ROM
When I attempted to flash with:
nvflash ROMFILE - error mismatch boards
nvflash -6 ROMFILE - error mismatch boards
nvflash -5 -6 6805BOC.rom - results shown below

Firmware Image Board ID (e10d) does not match adapter board ID (E10E).
[Then the warnings] and type YES to confirm
Overiding Board ID Mismatch
Current - Version: 80.04.33.00.24 ID:10DE:11A0: 1462:10BC
GK104 Board - 2051b002 (normal board)
Replace with Version:80.04.5B.00.02 ID:10DE:11A0: 1028:0551
GK104 Board - 2051b001 (normal board)

1) The differences other than the bios name I bolded.
The first bolded set shows as far as I know, is MSI:10BC
I can't find 10BC anywhere in but vga bios lookup, link to the only entry with 10bc:
https://www.techpowerup.com/vgabios/131796/msi-gtx680m-4096-120604

2) The second which is bolded and underlined,
Shows a 2015b002 vs 2015b001 (so the replacement bios would change it 001 from 002)

I went to the net and really spun around to no where. Here is what I did find.
B001 seems to refer to Nvidia 680m 4GB which says same bios as clevo, clocks are 719/900 (gpu/mem)
https://www.techpowerup.com/vgabios/127709/nvidia-gtx680m-4096-120507

When I did the b002, seems to refer to MSI 680m 4GB with clocks at 771/900
https://www.techpowerup.com/vgabios/131796/msi-gtx680m-4096-120604

Now I went back into windows and got the subsys of 05511028, which of course is Alienware M17xR4 (0551) Dell (1028), the vendor of course was 10DE and dev 11A0.

So I am not sure how to procede.

I found this on TI by deadsmiley
Well I opened them up and it looks like the boards are slightly different. Just a guess on my part...
Name: GK104 P2051B SKU 001
Board: GK104 Board - 2051b001
Version: 80.04.5B.00.02

Name: GK104 P2051B SKU 002
Board: GK104 Board - 2051b503
Version: 80.04.33.00.32

Then a few post down from Mariussx (only 26 rep and 45 posts)
Dell 680m - 80.04.5B.00.02_'OCedition'_revised_00 - This is the latest version of BIOS, which enables GTX 680m to boot on UEFI systems. Another version is the older vBIOS for the same card as far as I am aware. I tried both on my laptop in the signature. 5b version can boot into Windows 8.1 in full UEFI mode when Intel GPU is disabled. When 680m is flashed with 33 version, card does not boot into Windows in full UEFI - laptop just reboots and re-enables Intel GPU and Optimus and boots into Windows using Intel card. 5b is better is you do not want optimus for some reason as it allows UEFI systems to run fully on nVidia GPU and have Intel GPU disabled. 33 is better for overclocking as some versions have OV, which allows higher stable OC. Unless someone adds the same OV options to the latest vBIOS

Now since I don't know yet since I am booting from a Samsung 850 Evo SSD and whether to use GPT (which would need UEFI but supposedly is more secure, at least some say) or MBR (which has just as many people saying no as yes and supposedly not as secure but there was reference to this meaning for older SSD's so don't know) I don't know would a UEFI video bios boot if I didn't use UEFI system bios? So now I am back to the .5B vs 33 for the vbios?

Help!

 

I'm not sure really, never owned a 680M. All i know for a fact is UEFI/GPT causes more headaches than its worth. I use MBR because it works and is simple to configure.

 

@Raidriar
Okay then I will go for mbr on my SSD. Should I go with the .5B vbios or go with the .33
I see in the kepler package from TI there is:

• Dell 680m - 80.04.33.00.32__'OCedition'_revised_00.zip
• Dell 680m - 80.04.33.00.32__'OCedition'_revised_00 - OV 1000v.zip
• Plus 2 other OV's which I don't think is the direction I want to go (safety first)
• MSI 680m - 80.04.33.00.24_'OCedition'_revised_02.zip
• Plus the same 3 OV's like the dell 1000, 1025 and 1050

I assume the vbios on my card is the stock so while it says 80.04.33.00.32, it is the MSI and not the modded version from svl7.

Would you like your NBR screen name added to that tech note for Compeve (I am going to be contacting them by e-mail about a different receipt to say it is a 4gb not 2gb) as a reference for others to contact if they buy from Compeve or do you prefer not to have your name attached to a company in case they misuse it?

Also sorry made a mistake on (you probably are way ahead of me on this mistake but just for rounding out this thread for others, will correct)
Name: GK104 P2051B SKU 002
Board: GK104 Board - 2051b503
Version: 80.04.33.00.32
Should be b002 not 503

@Raidriar it appears that STO is senstive to optimus or using the Nvidia. Once I switched the game to only use the Intel, no crashing. It happened with the AMD so I know it isn't faulty hardware (unless this was the start of the AMD failing - but it had been doing this for a long time so I am discounting this as an issue).

I still can't seem to find recommended temps for idle and uningine load. I wish there was a database for the Uningine Heaven test, appears they don't publish it.

Following up also on the Thermal pad comments 2 post above and paste issue.

Etiquette question. When we use the @username, is that consider proper or is it consider the same as waving your hands and jumping up and down screaming to get ones attention. I only started to use the @ in the last few months on a few sites but I don't see others doing it as frequently so don't know what is the proper time to use @.

 

I don't recommend using any overvolt modifications. vBIOS version, i could go MSI svl7 vBIOS since you have a MSI card.

Please do not attach my name to any receipts, people can use the internet to look for help if need be (maybe they find this thread)

temp wise: under 60 is normal for idle, under 70 for load at stock voltage.

Etiquette: Just don't be inflammatory, everything else is fine. I personally don't use the @ system, I just quote wherever needed.

 

Okay downloaded and used the non-overvolte MSI bios and no issues except initially forgot the -6, but once added flashed no issues and no issues in windows (in terms of redetecting or windows reactivate). Getting ready to do the fresh Win 7 install on the SSD and go from there. I guess I will leave the thermal pads/paste that is currently on it on for now and if I see problems with the temp then I will come back here for advice.

Just want to say thanks again.

Admin/Mod: This thread can be closed as problem has been solved by @Raidriar
Thank you!

 

Absolutely.