Disable optimus on XPS 15 (L521x)

xpamamadeus · July 23, 2013

Archive.zip

Can u guys help me guide how to disable OPTIMUS in this laptop.

there its no bios option for doing this.

I read many threads about DSDT edits but there its no aio guide how to do it.

I have attached all acpi tables in archive.zip.

I found method OFF in ssdt-6 but cannot compile it becouse of errors.

This dell seems to be too complicated for me.

Specs

Intel® Core™ i7-3632QM

Intel® HM77
NVIDIA® GeForce® GT 640M with 2GB GDDR5 VRAM

DarwinDumper_2.7.8_INSYDE_Corp._ML_dell.zip

FelipeZé · July 30, 2013

You already make it?? If no I can help you, let me know,

xpamamadeus · July 30, 2013

No still nothing.

artur_pt · July 30, 2013

hello

take a look here

http://olarila.com/forum/viewtopic.php?f=7&t=3084

good hack

WaldMeister · August 2, 2013

hello

take a look here

http://olarila.com/forum/viewtopic.php?f=7&t=3084

good hack

Hi,

The images are not working somehow, and it is in Spanish i think?

Adding the information as described in the topic, will produce uncommon errors, i'll see if i manage to create a guide this evening. Will edit this dsdt first.

The card should be disabled now.

I have only fixed the errors to compile the DSDT, so you will have to patch it to function with OS X.

The _T_ patch has been applied, that is it, you will need to do the rest yourself, or request it

I'll start working on a guide for this now.

PS: If you want to know what has been done, search for PINI, and look under PEGP.

DSDT.aml.zip

xpamamadeus · August 6, 2013

Thx for edits.

But nvidia card its still listed.

qwerty12 · August 6, 2013

DSDT.aml.zip

WaldMeister · August 7, 2013

Hi qwerty,

Mind sharing what i forgot?

qwerty12 · August 7, 2013

Hi Lejenk,

Looking at your DSDT, it seems you were calling PINI inside PEGP's _INI. I don't know if _INI there is called on every boot; however, _SB.PCI0._INI definitely is (I forcibly enable my laptop's Wi-Fi LED in _INI and it turns on some milliseconds after I see the Apple logo). (Also according to THe KiNG, PINI apparently stands for PCI0 Wake/Ini, although of course it's arbitrarily named.)

I was lazy compared to you: I didn't move the functions from PEG0 etc. into the DSDT proper (duplicate device definitions are ignored by the ACPI subsystem, which I learnt from reading this) but I declared the functions as External (if you can keep the DSDT size down, go for it as I think you make boot just a tiny bit faster). When dealing with External stuff, iasl always messes up something and I don't know how to forcibly override its flawed autodetection of things. In this case, it thought that the \_SB.PCI0.PEG0.PEGP._PS3 call was an additional argument to the _DSM call. The workaround I used this time was to put _PS3 in a "If (One)" call.

After reading this, I've also taken to turning off Optimus through the _DSM + _PS3 functions which is the "right" way to do it according to Lekensteyn here: http://unix.stackexchange.com/a/49932

EDIT: I looked again at your DSDT and realised that the _PS3 call doesn't do anything. It checks to see if OMPR == 0x03 but OMPR (which is a named variable, not a register, so it's not going to get changed externally) is only changed through calls to _DSM.

Edited August 8, 2013 by qwerty12

FelipeZé · August 7, 2013

Hi Lejenk,

Looking at your DSDT, it seems you were calling PINI inside PEGP's _INI. I don't know if _INI there is called on every boot; however, _SB.PCI0._INI definitely is (I forcibly enable my laptop's Wi-Fi LED in _INI and it turns on some milliseconds after I see the Apple logo). (Also according to THe KiNG, PINI apparently stands for PCI0 Wake/Ini, although of course it's arbitrarily named.)

I was lazier than you: I didn't move the functions from PEG0 etc. into the DSDT proper (duplicate device definitions are ignored by the ACPI subsystem, which I learnt from reading this) but I declared the functions as External (if you can keep the DSDT size down, go for it as I think you make boot just a tiny bit faster). When dealing with External stuff, iasl always messes up something and I don't know how to forcibly override its flawed autodetection of things. In this case, it thought that the \_SB.PCI0.PEG0.PEGP._PS3 call was an additional argument to the _DSM call. The workaround I used this time was to put _PS3 in a "If (One)" call.

After reading this, I've also taken to turning off Optimus through the _DSM + _PS3 functions which is the "right" way to do it according to Lekensteyn here: http://unix.stackexchange.com/a/49932

Can you be a bit more specific on how you disabled it / post the code?? thanks!

qwerty12 · August 7, 2013

Can you be a bit more specific on how you disabled it / post the code?? thanks!

cd ~/Downloads
mkdir Archive && cd Archive
unzip ../Archive.zip #from first post

#Decompile, recompile and decompile the dsdt to ensure we have a DSDT with iasl's optimisations applied
iasl -d dsdt.aml
iasl dsdt.dsl
iasl -d dsdt.aml

mv dsdt.dsl dsdt.orig.dsl
unzip ../DSDT.aml.zip #from my post
iasl -d dsdt.aml
iasl dsdt.dsl
iasl -d dsdt.aml

diff -u dsdt.orig.dsl DSDT.dsl #which produces:

--- dsdt.orig.dsl	2013-08-07 16:23:47.000000000 +0100
+++ DSDT.dsl	2013-08-07 16:26:04.000000000 +0100
@@ -3,23 +3,23 @@
  * AML Disassembler version 20130725-64 [Jul 31 2013]
  * Copyright (c) 2000 - 2013 Intel Corporation
  * 
- * Disassembly of DSDT.aml, Wed Aug  7 16:23:47 2013
+ * Disassembly of DSDT.aml, Wed Aug  7 16:26:04 2013
  *
  * Original Table Header:
  *     Signature        "DSDT"
- *     Length           0x0000A0CC (41164)
+ *     Length           0x0000A11E (41246)
  *     Revision         0x01 **** 32-bit table (V1), no 64-bit math support
- *     Checksum         0x79
+ *     Checksum         0x03
  *     OEM ID           "DELL  "
  *     OEM Table ID     "CL09   "
  *     OEM Revision     0x00000000 (0)
- *     Compiler ID      "ASL "
- *     Compiler Version 0x00040000 (262144)
+ *     Compiler ID      "INTL"
+ *     Compiler Version 0x20130725 (538117925)
  */
 DefinitionBlock ("DSDT.aml", "DSDT", 1, "DELL  ", "CL09   ", 0x00000000)
 {
     /*
-     * iASL Warning: There were 6 external control methods found during
+     * iASL Warning: There were 8 external control methods found during
      * disassembly, but additional ACPI tables to resolve these externals
      * were not specified. This resulting disassembler output file may not
      * compile because the disassembler did not know how many arguments
@@ -30,6 +30,8 @@
      *     iasl -e <dsdt.aml,ssdt2.aml...> -d <ssdt1.aml>
      */
     External (_SB_.PCI0.IEIT.EITV, MethodObj)    // Warning: Unresolved Method, guessing 0 arguments (may be incorrect, see warning above)
+    External (_SB_.PCI0.PEG0.PEGP._DSM, MethodObj)    // Warning: Unresolved Method, guessing 4 arguments (may be incorrect, see warning above)
+    External (_SB_.PCI0.PEG0.PEGP._PS3, MethodObj)    // Warning: Unresolved Method, guessing 0 arguments (may be incorrect, see warning above)
     External (ECST, MethodObj)    // Warning: Unresolved Method, guessing 1 arguments (may be incorrect, see warning above)
     External (HDOS, MethodObj)    // Warning: Unresolved Method, guessing 0 arguments (may be incorrect, see warning above)
     External (HNOT, MethodObj)    // Warning: Unresolved Method, guessing 1 arguments (may be incorrect, see warning above)
@@ -2850,8 +2852,25 @@
         If (LEqual (DBGS, Zero)) {}
     }
 
+    Method (PINI, 0, NotSerialized)
+    {
+        \_SB.PCI0.PEG0.PEGP._DSM (Buffer (0x10)
+            {
+                /* 0000 */   0xF8, 0xD8, 0x86, 0xA4, 0xDA, 0x0B, 0x1B, 0x47,
+                /* 0008 */   0xA7, 0x2B, 0x60, 0x42, 0xA6, 0xB5, 0xBE, 0xE0
+            }, 0x0100, 0x1A, Buffer (0x04)
+            {
+                 0x01, 0x00, 0x00, 0x03
+            })
+        If (One)
+        {
+            \_SB.PCI0.PEG0.PEGP._PS3 ()
+        }
+    }
+
     Method (_WAK, 1, Serialized)  // _WAK: Wake
     {
+        PINI ()
         If (LAnd (LEqual (\_SB.PCI0.LPCB.EC0.AAST, One), LEqual (\_SB.PCI0.LPCB.EC0.AAEN, One)))
         {
             Store (Zero, GP53)
@@ -3545,6 +3564,7 @@
         Method (_INI, 0, NotSerialized)  // _INI: Initialize
         {
             Store (0x07D0, OSYS)
+            PINI ()
             If (CondRefOf (\_OSI, Local0))
             {
                 If (_OSI ("Windows 2001"))

FelipeZé · August 7, 2013

cd ~/Downloads
mkdir Archive && cd Archive
unzip ../Archive.zip #from first post

#Decompile, recompile and decompile the dsdt to ensure we have a DSDT with iasl's optimisations applied
iasl -d dsdt.aml
iasl dsdt.dsl
iasl -d dsdt.aml

mv dsdt.dsl dsdt.orig.dsl
unzip ../DSDT.aml.zip #from my post
iasl -d dsdt.aml
iasl dsdt.dsl
iasl -d dsdt.aml

diff -u dsdt.orig.dsl DSDT.dsl #which produces:

--- dsdt.orig.dsl	2013-08-07 16:23:47.000000000 +0100
+++ DSDT.dsl	2013-08-07 16:26:04.000000000 +0100
@@ -3,23 +3,23 @@
  * AML Disassembler version 20130725-64 [Jul 31 2013]
  * Copyright (c) 2000 - 2013 Intel Corporation
  * 
- * Disassembly of DSDT.aml, Wed Aug  7 16:23:47 2013
+ * Disassembly of DSDT.aml, Wed Aug  7 16:26:04 2013
  *
  * Original Table Header:
  *     Signature        "DSDT"
- *     Length           0x0000A0CC (41164)
+ *     Length           0x0000A11E (41246)
  *     Revision         0x01 **** 32-bit table (V1), no 64-bit math support
- *     Checksum         0x79
+ *     Checksum         0x03
  *     OEM ID           "DELL  "
  *     OEM Table ID     "CL09   "
  *     OEM Revision     0x00000000 (0)
- *     Compiler ID      "ASL "
- *     Compiler Version 0x00040000 (262144)
+ *     Compiler ID      "INTL"
+ *     Compiler Version 0x20130725 (538117925)
  */
 DefinitionBlock ("DSDT.aml", "DSDT", 1, "DELL  ", "CL09   ", 0x00000000)
 {
     /*
-     * iASL Warning: There were 6 external control methods found during
+     * iASL Warning: There were 8 external control methods found during
      * disassembly, but additional ACPI tables to resolve these externals
      * were not specified. This resulting disassembler output file may not
      * compile because the disassembler did not know how many arguments
@@ -30,6 +30,8 @@
      *     iasl -e <dsdt.aml,ssdt2.aml...> -d <ssdt1.aml>
      */
     External (_SB_.PCI0.IEIT.EITV, MethodObj)    // Warning: Unresolved Method, guessing 0 arguments (may be incorrect, see warning above)
+    External (_SB_.PCI0.PEG0.PEGP._DSM, MethodObj)    // Warning: Unresolved Method, guessing 4 arguments (may be incorrect, see warning above)
+    External (_SB_.PCI0.PEG0.PEGP._PS3, MethodObj)    // Warning: Unresolved Method, guessing 0 arguments (may be incorrect, see warning above)
     External (ECST, MethodObj)    // Warning: Unresolved Method, guessing 1 arguments (may be incorrect, see warning above)
     External (HDOS, MethodObj)    // Warning: Unresolved Method, guessing 0 arguments (may be incorrect, see warning above)
     External (HNOT, MethodObj)    // Warning: Unresolved Method, guessing 1 arguments (may be incorrect, see warning above)
@@ -2850,8 +2852,25 @@
         If (LEqual (DBGS, Zero)) {}
     }
 
+    Method (PINI, 0, NotSerialized)
+    {
+        \_SB.PCI0.PEG0.PEGP._DSM (Buffer (0x10)
+            {
+                /* 0000 */   0xF8, 0xD8, 0x86, 0xA4, 0xDA, 0x0B, 0x1B, 0x47,
+                /* 0008 */   0xA7, 0x2B, 0x60, 0x42, 0xA6, 0xB5, 0xBE, 0xE0
+            }, 0x0100, 0x1A, Buffer (0x04)
+            {
+                 0x01, 0x00, 0x00, 0x03
+            })
+        If (One)
+        {
+            \_SB.PCI0.PEG0.PEGP._PS3 ()
+        }
+    }
+
     Method (_WAK, 1, Serialized)  // _WAK: Wake
     {
+        PINI ()
         If (LAnd (LEqual (\_SB.PCI0.LPCB.EC0.AAST, One), LEqual (\_SB.PCI0.LPCB.EC0.AAEN, One)))
         {
             Store (Zero, GP53)
@@ -3545,6 +3564,7 @@
         Method (_INI, 0, NotSerialized)  // _INI: Initialize
         {
             Store (0x07D0, OSYS)
+            PINI ()
             If (CondRefOf (\_OSI, Local0))
             {
                 If (_OSI ("Windows 2001"))

Thanks Pal, got it to work this way too, this really seems the right way to do this... Altought I don't know it's advantages than just calling OFF function =p.

xpamamadeus · August 7, 2013

sry to must report that qwerty12 dsdt still dont disable my card.

What can i do to help to make right edits to disable this card?

qwerty12 · August 8, 2013

After "fixing" your SSDT-6 so that iasl will actually produce an aml file from the dsl, I put a print statement in the _PS3 method telling me if it's actually going to call _OFF, turning off the card, which would only happen if OMPR equals 0x03. Lo and behold, it does (the first instance of the message is due to being called from _INI and the second from me telling acpiexec to call it again):

After looking at your ACPI tables, the *only* way OMPR can be changed is through _DSM (when calling it with the specific arguments that are in the PINI method) and _PS3 itself. OMPR is a named variable; it is not pointing to a register and cannot be changed externally. In addition, it is initialised with 0x02 so _PS3 will not work until the call to _DSM has been made from PINI.

This means that the _DSM way is the right way for your laptop because all other ways would've just made _PS3 not call _OFF. (You could, of course, skip _PS3 and call _OFF directly but you don't really want to do that according to Lekensteyn.) I didn't mess up with the External declarations - they are indeed for the right functions in the right place - because acpiexec is finding the _PS3 function fine (and we know it's found _DSM because of OMPR changing).

The only reason I can think of as to why it's failing is perhaps SSDT tables are being dropped?

SRSR333 · August 8, 2013

Hey, mind if I add in a little statement? For me, despite whether having a DSDT or not, somehow my boot process overrides edits in my DSDT related to graphics. When I boot to OS X from Windows, my Nvidia 525M is enabled, when I restart to OS X again, its gone... So no matter what I do to my DSDT, this has stuck.. Just some info from my experience.

FelipeZé · August 8, 2013

If you Drop your SSDT as I do, you need to copy all the needed methods from your SSDT and patch them into DSDT (In the right section) this is what I do...

xpamamadeus · August 8, 2013

If you Drop your SSDT as I do, you need to copy all the needed methods from your SSDT and patch them into DSDT (In the right section) this is what I do...

thx for info,but i dont know to do that.

I am using clover and it automaticaly drop all ssdt becouse it generate a new one for powermanagment.

If u guys can make dsdt with methods from ssdt-6 that would be perfect.

FelipeZé · August 8, 2013

Oh, good, you have Power management working, give the DSDT I attached a try. I maybe have fixed sleep on lid close too...

DSDT_xpama.aml.zip

Remember to rename dsdt to DSDT.aml and report me back.

WaldMeister · August 9, 2013

Thanks qwerty for the information, will definitly read through it this weekend!

xpamamadeus, try booting with -f, this will skip the kernel caches, just an idea

Disable optimus on XPS 15 (L521x)

Recommended Posts

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites