Overclock.net › Forums › Intel › Intel CPUs › [MOD] LGA775 Support For LGA771 Xeon CPUs
New Posts  All Forums:Forum Nav:

[MOD] LGA775 Support For LGA771 Xeon CPUs - Page 631

post #6301 of 12466
I was wondering the same thing when i put my X5470 on the UD3P. Basically its not necessary that the C-states work but its nice to have and saves 20W of power and heat when the CPU is idle. So I looked around a bit and there is a simple solution to this. Its what some of the better volt mods use (not the pencil and pray ones).

The VRM controller IC that controlls the power delivery to the CPU has a feature called "offset adjustment" it is implemented from factory and described in the datasheet. Basically its a pin on the chip that you can connect to Vcc or ground via a 50k variable resistor trimmer to adjust your Vdrop (BIOS setting to actual idle). The Vdroop (idle to load) will still be the same. Using this offset adjustment you can always leave the bios setting to auto (leaving C-states enabled) and set the actual voltage via the trimmer.

What you need is some SMD soldering skill, a multi turn 50kOhm variable resistor, 5cm of isolates thin wire and maybe a drop of epoxy to secure the wire to the board since the solder pad is really small and cant bear much load. From factory the offset pin is connected to ground via a resistor (setting a negative offset of 0.04V in my case). This resistor has to be unsoldered. The pad connected to the IC can then be connected to the ICs Vcc (I think its 5V) via the 50k trimmer to set a positive offset. The Trimmer should be set to its maximum resistance in the beginning. The solder job is a little delicate wink.gif.

When using this you have to keep in mind that your actual voltage is not what the VID setting in BIOS woud indicate depending on what offset you set.
Edited by rewease - 3/29/15 at 2:26pm
post #6302 of 12466
Quote:
Originally Posted by rewease View Post

I was wondering the same thing when i put my X5470 on the UD3P. Basically its not necessary that the C-states work but its nice to have and saves 20W of power and heat when the CPU is idle. So I looked around a bit and there is a simple solution to this. Its what some of the better volt mods use (not the pencil and pray ones).

The PWM controller IC that controlls the power delivery to the CPU has a feature called "offset adjustment" it is implemented from factory and described in the datasheet. Basically its a pin on the chip that you can connect to Vcc or ground via a 50k variable resistor trimmer to adjust your Vdrop (BIOS setting to actual idle). The Vdroop (idle to load) will still be the same. Using this offset adjustment you can always leave the bios setting to auto (leaving C-states enabled) and set the actual voltage via the trimmer.

What you need is some SMD soldering skill, a multi turn 50kOhm variable resistor, 5cm of isolates thin wire and maybe a drop of epoxy to secure the wire to the board since the solder pad is really small and cant bear much load. From factory the offset pin is connected to ground via a resistor (setting a negative offset of 0.04V in my case). This resistor has to be unsoldered. The pad connected to the IC can then be connected to the ICs Vcc (I think its 5V) via the 50k trimmer to set a positive offset. The Trimmer should be set to its maximum resistance in the beginning. The solder job is a little delicate wink.gif.

When using this you have to keep in mind that your actual voltage is not what the VID setting in BIOS woud indicate depending on what offset you set.

Jeez, that's hardcore for LGA775. So with the solder mod, you would be able to get voltage drops even when overvolting? Haswell is crazy efficient when it comes to voltage idling, which would be ideal for server/htpc usage. This is probably the reason op moved to Haswell after sellin off his LGA775 Xeon system.
post #6303 of 12466
Yes, i can bassically set any voltage and the C-state voltage level would still be switched in when idle (increased by the positive offset).
So if I have a CPU with a VID of 1.25V and the board has a Vdrop of 0.04 it idles at 1.21V. With a Vdroop of 0.05V it runs at 1.16V load.

Now I set the offset to +0.04V eliminating the Vdrop to 0V. The CPU now idles at 1.25V and runs at 1.20V load.
Suppose the C-state VID of the CPU would be 1.00V. The Offset would increase this to 1.04V.

I made some nice pics when I did the mod but I cant find them. So I made a quick and dirty one right now so you have an overview. Due to the epoxy blob (not the nicest I have done, but necessary because even bending the wire would lift the mini solderpad off the board otherwise) you cant see the exact pad where the original resistor to ground was located.



For your reading pleasure biggrin.gif:
http://www.intersil.com/content/dam/Intersil/documents/isl6/isl6336-a.pdf
Edited by rewease - 3/29/15 at 2:45pm
post #6304 of 12466
Quote:
Originally Posted by rewease View Post

Yes, i can bassically set any voltage and the C-state voltage level would still be switched in when idle (increased by the positive offset).
So if I have a CPU with a VID of 1.25V and the board has a Vdrop of 0.04 it idles at 1.21V. With a Vdroop of 0.05V it runs at 1.16V load.

Now I set the offset to +0.04V eliminating the Vdrop to 0V. The CPU now idles at 1.25V and runs at 1.20V load.
Suppose the C-state VID of the CPU would be 1.00V. The Offset would increase this to 1.04V.

I made some nice pics when I did the mod but I cant find them. So I made a quick and dirty one right now so you have an overview. Due to the epoxy blob (not the nicest I have done, but necessary because even bending the wire would lift the mini solderpad off the board otherwise) you cant see the exact pad where the original resistor to ground was located.



For your reading pleasure biggrin.gif:
http://www.intersil.com/content/dam/Intersil/documents/isl6/isl6336-a.pdf

Sounds like what you did is a hard LLC mod. If all you can do is change the offset voltage, this gets added to the C-state VID so the idle voltage increases by the offset voltage. Please elaborate how you can still maintain 1.04v as your c-state idle voltage if you wanted to increase the vcore to like 1.35v for overclocking.
post #6305 of 12466
Its not an LLC since it leaves the VdrOOp (idle to full load transition) alone. All it changes is the VdrOp (VID/BIOS to actual idle). The trick is to always keep the Vcore setting in BIOS at AUTO. This is the only setting at which C-state voltage is switched in. All changes to the actual Vcore are then made via the offset trimmer.
In the example I would have to set an offset of +0.14V to reach 1.35V actual idle. That would translate to 1.14V C-state voltage. The offset is always the same. The controller only gets another VID input when switching to C-state, that detail cant be helped. On the other hand 1.14V is still nicely low and you have to consider that the FSB is not reduced (only the multi as far as I know) in C-state so some extra voltage is not misplaced.
Edited by rewease - 3/29/15 at 3:04pm
post #6306 of 12466
Quote:
Originally Posted by rewease View Post

Its not an LLC since it leaves the VdrOOp (idle to full load transition) alone. All it changes is the VdrOp (VID/BIOS to actual idle). The trick is to always keep the Vcore setting in BIOS at AUTO. This is the only setting at which C-state voltage is switched in. All changes to the actual Vcore are then made via the offset trimmer.

You're right that it's not doing LLC if the VDroop is left active. I suppose this solution is better than running at full load voltage all the time, even though the offset voltage will be added to the c-state idle voltage. It sounds like higher and higher offsets to get more OC will eventually ride out any idle voltage headroom.
post #6307 of 12466
Hello,
can anyone help me with modded BIOS for asus p5qd turbo? I am going to use X5460.

Thank you!
post #6308 of 12466
Quote:
Originally Posted by sleepnot View Post

Hello everybody!

I have a Gigabyte P965-DS3 rev 1.0 mobo. Has anyone had any luck with this certain mobo?

Thanks!

Back with some updates. Still haven't made it to work. I have tested the CPU on another PC and it's ok. My mobo is also ok with my old CPU. Also, I have gone through this guide from delidded to update my BIOS microcode. What's left for me to check? Maybe the socket itself again but other than that I cannot come up with something else. rolleyes.gif
post #6309 of 12466
Quote:
Originally Posted by sleepnot View Post

Back with some updates. Still haven't made it to work. I have tested the CPU on another PC and it's ok. My mobo is also ok with my old CPU. Also, I have gone through this guide from delidded to update my BIOS microcode. What's left for me to check? Maybe the socket itself again but other than that I cannot come up with something else. rolleyes.gif

Hmm, if it isn't a problem with the mod itself than maybe 965P-DS3 (rev. 1.0) has hardware incompatible with 45nm / 1333FSB processors. 1.0 has N/A for them in cpu support list when 3.3 has "Beta support". Not only that, but both of them are using exactly the same BIOS and 3.3 was confirmed working with Xeons. So if they didn't make some blockade in boot sector of the bios that recognizes both - PCB 1.0 and whitelists processors my bet would be a hardware. Maybe old type of some ics that can't handle additional stress or signal noise, who knows.
Alien
(16 items)
 
  
CPUMotherboardGraphicsRAM
Intel Xeon X5450 E0 ASUS P5Q3 Deluxe/WiFi-AP @n Sapphire Radeon HD 6670 Corsair Vengeance CMZ16GX3M4X2133C11R with Sams... 
Hard DriveHard DriveOptical DriveCooling
WD Caviar Green WD10EARS-00MVWB0 WD Caviar Green WDC WD20EARX-00PASB0 HL-DT-ST DVD-ROM Szrot cooler, stock from Acer, hand polished co... 
OSMonitorKeyboardPower
Windows 7 64-bit Ultimate Fujitsu-Siemens 19'' A19-1 Microsoft Sidewinder X4 Corsair AX860 
CaseMouseMouse PadAudio
A-Top Z-Alien Black A4-Tech SWOP-3 Blue Elecom Samsung SMS-5100 
  hide details  
Reply
Alien
(16 items)
 
  
CPUMotherboardGraphicsRAM
Intel Xeon X5450 E0 ASUS P5Q3 Deluxe/WiFi-AP @n Sapphire Radeon HD 6670 Corsair Vengeance CMZ16GX3M4X2133C11R with Sams... 
Hard DriveHard DriveOptical DriveCooling
WD Caviar Green WD10EARS-00MVWB0 WD Caviar Green WDC WD20EARX-00PASB0 HL-DT-ST DVD-ROM Szrot cooler, stock from Acer, hand polished co... 
OSMonitorKeyboardPower
Windows 7 64-bit Ultimate Fujitsu-Siemens 19'' A19-1 Microsoft Sidewinder X4 Corsair AX860 
CaseMouseMouse PadAudio
A-Top Z-Alien Black A4-Tech SWOP-3 Blue Elecom Samsung SMS-5100 
  hide details  
Reply
post #6310 of 12466
Hey
Here is a little story for you all..

In January i started reading about this mod for my brothers computer in February I got a e5430 in the same day it arrived i sadly killed the board while cutting the black plastic and with no choice i rushed out to buy a new board.
while trying to mod the new board i slightly bent a pin which i manged to fix but i decided to stay with the e5400 that was there before.
so of curse 2 weeks later i got an old computer from a friend with the same board i had killed and today at 1 am i had the bright idea to try again with the board i got for free.
so again i tried cutting the plastic with no success.
so here comes the bright idea I said to myself " I have a soldering gun, it can melt stuff.. melt the plastic." after 2 attempts I with my hands shaking booted the PC up with the xenon e5430 inside and it worked!!!
yay tomorrow i will try the same with the new board i prefer to use it because its newer and better.

well the moral of the story is don't give up and never forget to melt stuff!

Thank you for all your help
-Adam Ericson

EDIT: did it on the new board and it worked.



Edited by adamdbz - 3/31/15 at 2:54am
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › [MOD] LGA775 Support For LGA771 Xeon CPUs