Go Back > Overclocking & Modding > Hardware Modding

Hardware Modding Soldering, deciphering datasheets, tricks and tips.

Thread Tools Search this Thread Display Modes
Old 08-08-2016, 10:00 PM
Khenglish Khenglish is offline
Join Date: Mar 2012
Posts: 28
Default 980 core swap to 980m PCB success

Due to a stupid accident by me, I acquired a 980m with a chunk knocked out of the core. Not wanting to scrap a perfectly good top end PCB for parts, I wanted to replace the core. You can see the gouge in the core to the left of the TFC918.01W writing near the left edge of the die.

First I had to get the dead core off:

With no sellers on ebay selling GM204 cores, my only option was to buy a full card off ebay. With no mobile cards under $500,I had to get a desktop card. And with this much effort involved to do the repair, of course I got a 980 instead of a 970.

Below is the dead 980 I got off ebay:

You can see for some reason someone removed a bunch of components between the core and PCI-E slot. I have no idea why anyone would do this. I tried the card and it was error 43. PCB bend seemed to be too little to kill the card, so those missing components had to be it.

GPUs can be dead because someone removed or installed a heatsink wrong and broke a corner of the core off, so buying cards for cores on ebay is a gamble. This core is not even scratched:

Preheating the card prior to high heat to pull the core:

And core pulled. It survived the pull:

Next is the 980 core on the left cleaned of solder. On the right is the original 980m core:

Here's the 980 core with new balls:

The card is cooling with the new core soldered on:

Here's a quick render test on the old ATITOOL. I ran something light because the card's 3rd phase was out at the time, and blowing out the other phases seemed like a bad idea.

So a summary of the core swap results:

980 core worked with no mods other than the core swap. No vBIOS, hardware strap, or even driver mods required.

The core ran in the full 2048 configuration, despite the 980m only having 1536 natively.

The hardware ID changed to the Desktop 980 device ID. Nvidia does not do laser cuts anymore. They now flash the device ID and configuration onto the core itself like Intel & AMD do with CPUID.

Of course I fixed the 3rd power phase when I saw the card worked:

1.2V 1502/7160 firestrike 16078 GPU score:

1.1875V 1495/7160 3dm11 21748 GPU score (needed to reduce voltage due to exceeding the mobo's AC power limit):

I only ran the CPU at 4350MHz because adding voltage for higher clocks was cutting into GPU voltage & clocks to prevent AC cutout. Normally I would run benchmarks at 4.6GHz.

I'm pretty sure seeing a Desktop 980 device ID with 8GB memory is throwing futuremark and is why it says unrecognized GPU. The MXM 980 cards with 8GB have their own device ID.

I also switched 7 of the 980's memory chips over to the 980(m?) and the results above are with those 7 chips on. I broke the 8th chip. Stable clocks improved from 6652MHz to 6884MHz. Memory voltage is at 1.5V. The stock chips are the same exact part, but only rated for 6GHz at 1.5V instead of the 980's 7GHz.

I could actually bench at 7250, but I got a failure when trying at 7200 and assumed it was the memory's fault, but I think it was really a core crash. As a result I did not push higher when the laptop was on the top of the high speed air filter. Without it I need to drop 10-15MHz on core, which likely offset the 100MHz on memory.

I have added a lot of capacitors to the card as well. Originally the card could only bench up to 1464MHz at 1.2V. Now it does 1502.

What's next:

I am eyeing a dead 970 with Samsung memory on ebay now for 8 more 7GHz memory chips. I could then pull another 7GHz chip that I put on my MSI 980m. There is a seller on ebay for the memory directly & preballed, but for some reason Paypal is failing somewhere between us.

I found on the MSI card that adding wires between the caps near the power phases and the caps directly under the core reduces Vdroop and increases clocks by around 10MHz. I am holding off on that mod until after I finish the memory switching, I still think even with this my voltage is still significantly less stable than on the GALAX board. This chip was probably pretty beast on the original card.

I've seen 3dm11 hit 281W TGP. Kinda funny doing that on the standard 100W rated MXM slot. And people were complaining about the RX480 going a few watts out of spec on the PCI-E slot...
Reply With Quote
Old 08-09-2016, 01:31 AM
TiN's Avatar
TiN TiN is offline
Volt-Nut :)
Join Date: Jan 2010
Location: Taipei
Posts: 1,458
Send a message via ICQ to TiN Send a message via Skype™ to TiN

You are hardcore, man.
Hats off to you. Love those reball pics
Reply With Quote
Old 08-11-2016, 10:35 PM
Khenglish Khenglish is offline
Join Date: Mar 2012
Posts: 28

So Samsung is currently sampling 9GHz rated GDDR5 chips. I asked them if I could sample some, and they had the nerve to ignore my request! Who could have seen that coming! (sarcasm)
Reply With Quote
Old 08-14-2016, 02:07 PM
VPII VPII is offline
Join Date: Oct 2010
Posts: 8

Oh my word.... brilliant khenglish, great job. I am totally amazed at what some people are able to do with hardware
Reply With Quote
Old 11-28-2016, 07:28 AM
kicsipapucs kicsipapucs is offline
Join Date: Mar 2015
Posts: 2

Hats off a couple months later!
Reply With Quote

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

All times are GMT -5. The time now is 04:30 AM.

Contact Us - - Archive - Top
Powered by vBulletin® Version 3.8.3 Copyright ©2000 - 2018, Jelsoft Enterprises Ltd.