Jump to content
HWBOT Community Forums

MSI GTX 560Ti 448cores Facts 'n Fiction

Eeky NoX

Recommended Posts







Hye all,


I start this new thread to regroup all infos we could have on MSI GTX 560Ti 448cores card. Mods, bioses, issues and so on...


Feel free to post all you know or questions you have on this card.


In advance, thank you.







Link to comment
Share on other sites

Here's the VID mod I worked up. I have not tried it (nor do I have the card). I went off of photos and datasheets for similar controllers. Eeky tried it and said they had no luck, but I still have faith!


The VID pins are as follows:


Pin 54 - VID7

Pin 55 - VID6

Pin 56 - VID5

Pin 57 - VID4

Pin 58 - VID3

Pin 59 - VID2

Pin 60 - VID1

Pin 61 - VID0


These pins on the 1981 are by a double bank of resistors which is how VID is designed. When I refer to the first bank of resistors, I mean the bank closest to the controller (the orientation of the two images is different, so it can get confusing).


Here's an image:




(Thanks TPU for the original)


Assuming these pins are the VID pins, here are the values I got from the images of the 560 above:


Pin 54 - VID7 0 1 (Bank 1 Bank 2)

Pin 55 - VID6 0 1

Pin 56 - VID5 0 1

Pin 57 - VID4 1 0

Pin 58 - VID3 1 0

Pin 59 - VID2 0 1

Pin 60 - VID1 1 0

Pin 61 - VID0 1 0


Note that the first bank of numbers (the resistors closest to the chip) are the ones called out in the VID table. When one bank contains a resistor, the other will not. The 1 represents where the resistor goes.


So the two banks above translate into:


7 6 5 4 3 2 1 0 (VIDx)

0 0 0 1 1 0 1 1


Quick binary tutorial:


1 - 0000 0001

2 - 0000 0010

3 - 0000 0011

4 - 0000 0100

5 - 0000 0101

6 - 0000 0110

7 - 0000 0111

8 - 0000 1000


Referring to the uPI datasheet attached, you can see that our number (0001 1011) corresponds to a Voffset of 168.75mV. If you measure the voltage that the controller is providing the GPU, you can use this to determine the reference voltage (Vref) for the controller (the voltage upon which all of the VID adjustments are based).


Vref = vGPU - Voffset


For example, if the vGPU is 1.20V, the reference voltage for the controller is 1.20V - 0.16875V = 1.03125V. Knowing this reference voltage, assume we want a vGPU of 1.50V. To find the Voffset, we would do 1.50V - 1.03125V = 0.46875V = 468.75mV. Looking to our tables, this means we need a VID configuration of 0100 1011.




So you want to perform this mod?


Here's what you need to do:


Note the stock voltage (under load) of your GPU in Windows. If software voltage monitoring is not available, measure the vGPU of your card with a DMM. Look at the VID section (Pins 53-60) on the card and write down the first bank (closest to the controller). Use a 1 for a resistor and a 0 for no resistor. Once you have your 8-digit binary number, compare this to the VID table. Find the Voffset that your binary number corresponds to and subtract this from your vGPU. This will give you the reference voltage for the controller. Once you know how much voltage you want, subtract the Vref from your goal. This will give you the Voffset. Look to the tables to determine what binary you need and make the modification to the card (move the resistors)!





To simplify the experimental process of trying different VID configurations, it may be wise to device a switching configuration. I would recommend using 8 single pole dual throw (SPDT) switches (example). This allows one input and switches between two outputs. One output would connect to a +5V source and the other output would connect to GND.


You would have the circuit below. It allows you to use one resistor and use a SPDT switch to connect it to either +5V or GND. When connected to the +5V source, this corresponds to a logical 1. When connected to GND, it corresponds to a logical 0.




I believe the resistors are 10kΩ, but measure to be sure.



uPI VID Offset Chart


Link to comment
Share on other sites

Hi guys,


I'm the personal assistant of Mister Eeky, a.k.a his s*x slave.


Thanks to Tolsty post, I found the vmod and I share it here.




Sorry for the bad quality of the picture, took it with an iphone. Pin 8 of uP1981A is linked to the capacitor C1067. Fully fonctionnal for me :)

Link to comment
Share on other sites

more info for mods on the IC here-> http://www.overclock.net/t/1009397/msi-6950-tfiii-pe-oc-testing-w-cl3p20/40_40#post_14394941


..its same IC as 6950 PE/OC and GTX570 PE/OC..




**Per Ch. OCP and total OCP brought solid gains to both 6950 and 570.. seems circuits were a bit current limited even at stock clocks. Make sure your PCB does not have current shunts though.. as these will require additional mods to surpass amperage limitations.

Link to comment
Share on other sites



unlock voltage in bios to 1.21 V



I had OCP at 1.2 V 1100 MHz




measuring points
















VID mod



if you use VID6 then the Vcore is 1.4 V

set the voltage down to 0.913 V with software then the Vcore is 1.3 V load

Now you can change the voltage from 1.3 V to +1.5 V


all mods works on my card

all mods are made at your own risk

Link to comment
Share on other sites

all mods are made at your own risk

Yup beware of jumping caps :D


Thank you Zzolio :)


Unfortunatly not tested mine under cascade saturday... I give up this time but if my team mate manage to qualify we'll be back!!


And of course if not I will for hardware too :D

Edited by Eeky NoX
Link to comment
Share on other sites

@ zzolio - what is your +12v input reading at under load - as measured at the PCI-E input?


*nice job on the mods btw! I will be modding the same, except without VID [do not think VID benefits this design, by moving OCP/ILIM as VID increases]. You mentioned you are getting OCP still .. or not?

Link to comment
Share on other sites

  • 3 weeks later...
  • 1 month later...










all mods works on my card

all mods are made at your own risk





did you short both point when you remove 2 resistors? or just one that shown on your pic ?


because datasheet says: to disable OCP short ilim pin to rosc pin .

Link to comment
Share on other sites

Yeah, same here. First time on dice with this card I had only a normal 120mm fan against the VRM area and after 5-10min it smelled of burned rubber from the elastic eraser and I could not go higher than 1055mhz core, with Delta 220cfm fan 1141mhz was possible. So You probably need heatsinks on the VRM stuff. :D

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.

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...