Author Topic: [MOVED] Hacking NVidia Cards into their Professional Counterparts  (Read 1305999 times)

0 Members and 7 Guests are viewing this topic.

Offline Black Phoenix

  • Frequent Contributor
  • **
  • Posts: 565
  • Country: hk
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1400 on: September 24, 2019, 07:33:51 am »
Related with this subject, an Youtuber that is registered in this forum just used this topic as a reference for his project:




 

Offline namgorf

  • Contributor
  • Posts: 6
  • Country: us
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1401 on: September 24, 2019, 10:30:02 pm »
I watched these last night, very awesome. New sub from me.
 

Offline NobleX13

  • Newbie
  • Posts: 2
  • Country: us
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1402 on: October 11, 2019, 02:01:28 pm »
I really dont get it.
I flashed ROM from namgorf and confirmed that it has InfoROM
Still having this screen on Dell R720
I updated latest firmware on R720 as well...
It works fine on HP Z420 but not on Dell R720

What am I missing  |O |O |O |O |O |O |O |O |O |O |O |O |O

[attachimg=1]

Did you ever get past this?  You could say I am following in your footsteps from the shadows.  I just converted a Tesla K10 into an NVIDIA GRID K2 yesterday and I get the same result on my PowerEdge R720.

I soldered the strap resistors, reassembled the card with new thermal paste, and flashed the "2014" BIOS to each GPU.  I did notice the "PLX" entry in nvflash but didn't do anything with that.  Was I supposed to? The SHA256 hash of the BIOS file I flashed was BB04DF8552BF60B827E1C963B1D8527386D9448D554BC01799CE7F8605763951

My R720 has dual E5-2650 V2s, 128GB of PC3L-12800R, and dual 750W power supplies.  I saw the other post here about power delivery, but this card has a max power consumption at full load of maybe 200W, and my system power readout is only sitting at 280W with this GPU installed.  I am using a quick and dirty homemade power cable.  Please excuse the WAGO Lever Nuts.  I triple-checked the pinout and continuity with a multimeter, for what it's worth.

Right now I am using the riser connected to CPU2, because it was easier to install the card over there.  I ran out of time for troubleshooting last night.  I did not try the other riser, removing all other PCI-e cards, or enabling "Above 4G decoding" in the BIOS yet.  That is on the agenda for tonight.

I am running a BIOS version from 2016 to avoid the Spectre/Meltdown microcode mitigation.
« Last Edit: October 11, 2019, 03:14:38 pm by NobleX13 »
 

Offline foxxer

  • Newbie
  • Posts: 2
  • Country: ru
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1403 on: October 25, 2019, 04:51:35 am »
Hello. Sorry for my English(using translator). Who knows how to use the HEX editor and understanding of the work of video card BIOS? I ask for help, it will not take much time, if there is knowledge.

I have a Quadro 2000 1Gb DDR5 card
There is a desire to flash it as a GTS450

There is info that for the reverse operation, i.e. for GTS450 firmware ---> Quadro 2000 need to alter the straps with the command:
C: \> nvflash --index = X --strap 0x7FFC2FFF 0x10006000 0x7fffffff 0x00000000

Those. IDs for cards differ, for picking up by drivers:
GTS 450 - 0x0DC4 - 0000 1101 1100 0100
Quadro 2000 - 0x0DD8 - 0000 1101 1101 1000

Tell me, what command should I type in NVFLASH with Q2000 so that the ID from the GTS450 is registered in BIOS?
Goal - so that the system after flashing a QUADRO 2000 card identifies it as a GTS450 and works with Geforce drivers

Thank you for support
 

Offline acidice333

  • Newbie
  • Posts: 1
  • Country: ca
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1404 on: October 31, 2019, 09:18:22 pm »
Could I get the K2 with the inforom dump too? :D
 

Offline foxxer

  • Newbie
  • Posts: 2
  • Country: ru
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1405 on: November 01, 2019, 05:45:00 am »
here is rom for my Quadro 2000
 

Offline n_lona

  • Newbie
  • Posts: 1
  • Country: it
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1406 on: December 07, 2019, 01:30:31 pm »
Hi thanks for let me in.
Is there any previous attempt to turn an EVGA gtx 1050ti and gt 710 into their respectivetly professional couterparty (suppose the first should be a quadro m2000 pascal and ..)
Is there any way to hack cudnn libraries making system believing that on board card is a quadro or better?
« Last Edit: December 07, 2019, 01:42:45 pm by n_lona »
 

Offline martos

  • Contributor
  • Posts: 5
  • Country: fr
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1407 on: December 17, 2019, 11:42:06 am »
Hello all,

I have a real Nvidia K10 , and i want transform it in Nvidia Grid 2
so
tesla k10           none                                                40K                                                  None                                         25k                                                                             
TO
grid k2              none                                                40K                                                  20K                                           none

I have the real gpu bios of Grid K2 ( Craft Computing have post a link in youtube, and i tested with the software KeplerBiosTweaker_1.27 , it seem ok )

I see where is it  ( at time 12:00 )

I see a link to the buy a 40k https://www.digikey.com/products/en?keywords=P40.2KDCCT-ND
But i need a 20K too

Could you validate the filter i put :

Filter : Panasonic ERA-xA Series 0.1 % 0402 1005 25 PPM / C 50 V SMD/SMT Thin Film Resistors Thin Film Resistors - SMD
result : https://www.mouser.fr/Passive-Components/Resistors/Film-Resistors/Thin-Film-Resistors-SMD/ERA-xA-Series/_/N-7gz44?P=1z0wdwsZ1z0z7ymZ1z0vpm5Z1z0x83kZ1yzmoucZ1yzmoubZ1y95jjmZ1y9oo9v


If someone want a dump of the tesla K10 biso , i can send it.

 

Offline martos

  • Contributor
  • Posts: 5
  • Country: fr
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1408 on: December 17, 2019, 07:20:08 pm »
Could someone can confirm this reference :

10K
ERA-2AEB103X
15K
ERA-2AEB153X
20 K
ERA-2AEB203X‎
25K
667-ERA-2AEB2492X
40 k
ERA-2AEB4022X‎
 

Offline DDSD

  • Newbie
  • Posts: 1
  • Country: uz
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1409 on: January 17, 2020, 06:37:07 am »
Hello. If any solution can you help me with PNY GTS450 to Q2000.
 

Offline gotofbi

  • Contributor
  • Posts: 29
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1410 on: January 29, 2020, 11:21:17 pm »
I really dont get it.
I flashed ROM from namgorf and confirmed that it has InfoROM
Still having this screen on Dell R720
I updated latest firmware on R720 as well...
It works fine on HP Z420 but not on Dell R720

What am I missing  |O |O |O |O |O |O |O |O |O |O |O |O |O

[attachimg=1]

Did you ever get past this?  You could say I am following in your footsteps from the shadows.  I just converted a Tesla K10 into an NVIDIA GRID K2 yesterday and I get the same result on my PowerEdge R720.

I soldered the strap resistors, reassembled the card with new thermal paste, and flashed the "2014" BIOS to each GPU.  I did notice the "PLX" entry in nvflash but didn't do anything with that.  Was I supposed to? The SHA256 hash of the BIOS file I flashed was BB04DF8552BF60B827E1C963B1D8527386D9448D554BC01799CE7F8605763951

My R720 has dual E5-2650 V2s, 128GB of PC3L-12800R, and dual 750W power supplies.  I saw the other post here about power delivery, but this card has a max power consumption at full load of maybe 200W, and my system power readout is only sitting at 280W with this GPU installed.  I am using a quick and dirty homemade power cable.  Please excuse the WAGO Lever Nuts.  I triple-checked the pinout and continuity with a multimeter, for what it's worth.

Right now I am using the riser connected to CPU2, because it was easier to install the card over there.  I ran out of time for troubleshooting last night.  I did not try the other riser, removing all other PCI-e cards, or enabling "Above 4G decoding" in the BIOS yet.  That is on the agenda for tonight.

I am running a BIOS version from 2016 to avoid the Spectre/Meltdown microcode mitigation.

I did everything I could do on R720 but it didnt work.
I was thinking about purchasing real K2 for test but got more important projects to do.

I tested on two different R720 but same issue.
One was with V1 cpu and the other was V2

Only common thing that I can see now is 750W power supply.

Have you solved the problem?
« Last Edit: January 29, 2020, 11:43:44 pm by gotofbi »
 

Offline martos

  • Contributor
  • Posts: 5
  • Country: fr
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1411 on: February 02, 2020, 05:28:10 pm »
Hello,
an another solution to try if it's works , take an PC Desktop in windows 10 , put the real power VGA cable ( One 8 pins and One 6 pins ) , of course keep a video card ( so you need 2 * 16X PCI E ) .
If windows detect Grid K2, ( first good point)
  launch "GPU Z" software, choose the card 1 or 2, and make a backup of the Bios.
Compare the bios with K2 bios ( with no error it's the second good point )
Use KeplerBiosTweaker_1.27
and you can optimise the consumation of the card and putin the modfied bios in your 2 cards

I make a image of what i need to change in the K10 to change it in Grid K2, could you confirme its ok please?

 

Offline gotofbi

  • Contributor
  • Posts: 29
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1412 on: February 04, 2020, 08:17:26 am »
Hello,
an another solution to try if it's works , take an PC Desktop in windows 10 , put the real power VGA cable ( One 8 pins and One 6 pins ) , of course keep a video card ( so you need 2 * 16X PCI E ) .
If windows detect Grid K2, ( first good point)
  launch "GPU Z" software, choose the card 1 or 2, and make a backup of the Bios.
Compare the bios with K2 bios ( with no error it's the second good point )
Use KeplerBiosTweaker_1.27
and you can optimise the consumation of the card and putin the modfied bios in your 2 cards

I make a image of what i need to change in the K10 to change it in Grid K2, could you confirme its ok please?

correct
 

Offline martos

  • Contributor
  • Posts: 5
  • Country: fr
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1413 on: February 06, 2020, 12:27:49 am »
Thank's a lot !
another trouble who can have in some server, the pci-e on some server don't give th good power. ( i have the trouble on some proliant G5 )
So i add a pci-x riser with a molex :
https://www.pc-adapter.net/product/732.html
 

Offline martos

  • Contributor
  • Posts: 5
  • Country: fr
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1414 on: February 19, 2020, 08:15:30 am »
some news :
 

Offline Pavlogal

  • Newbie
  • Posts: 1
  • Country: ba
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1415 on: February 29, 2020, 09:01:13 pm »
I got an ASUS P104-100 mining specific GPU which has the same PCB and specs as an ASUS GTX 1080 TURBO (other than maybe CUDA cores) and thus could be flashed into a GTX 1080 or 1070 GDDR5X in case there really are only 1920 CUDA cores on board. The reason anyone would want to turn it into a 1080/1070 is beacuse P104 is SERIOUSLY BIOS limited. PCIe is dropped to 1.1 (PCIe 3.0 hardware), VRAM is limited to 4GB (8GB installed) and video output is disabled (even though this card has video output ports). The device ID of a 1080 is x1B80 and P104's is x1B87, so I just need to find the resistors that determine the 4th symbol and then test what resistances equate to what values.

I don't know if something has changed in the meantime, someone on the LTT forum said they transplanted an entire P106 chip onto a GTX 1060 board and it didn't work unless there was a P106 BIOS flashed, basically it behaved like straps don't exist or are somehow the same for both cards. Maybe that's not correct and strap modding would work, but if it is correct maybe it's baked into the chip or some other component, I simply don't have a way of knowing because I couldn't find any info about that on Pascal.

Here are the docs I found: https://buildmedia.readthedocs.org/media/pdf/envytools/latest/envytools.pdf

I would like to know how device ID is determined on Pascal, if it's same as it has been before. What's the easiest way to identify which resistors are used for the 4th value of device ID? I know there is probably an answer in this thread but it's so massive my eyes fell off reading it...
I would also like to know about any secondary strap functions that could have been changed on the P104 (in case video output is disabled via straps I would need to change that too).

I really hope nothing has changed and that a simple strap mod would unlock it's full potential. Also if anyone has a GTX 1070 with GDDR5X memory please PM me your GPU BIOS and GPU-Z screenshot, so I have everything I need in case it does only have 1920 CUDA cores on board (which it most likely does).
« Last Edit: March 01, 2020, 10:33:51 am by Pavlogal »
 

Offline colokarl

  • Contributor
  • Posts: 23
  • Country: de
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1416 on: March 09, 2020, 06:13:40 pm »
Hi Modders,

first off, thank you very much for sharing this awsome knowledge - :clap:

.. got myself a gt640 wich will undergo some resistor soldering soon.

I am writing this post to ask, if anybody knows if the Tesla K40 is modabble in a similar way ?

I think its the same PCB like a Quatto 6000, is that correct ?

Could it be modded into a K2 or any other Card for that matter ?

I would like to use it in the VMware Host with the "vCPU" Feature.

Looking forward to get some insights from the experts.

Thanks !


« Last Edit: March 09, 2020, 07:11:59 pm by colokarl »
 

Offline gotofbi

  • Contributor
  • Posts: 29
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1417 on: March 18, 2020, 09:51:22 am »
Hi Modders,

first off, thank you very much for sharing this awsome knowledge - :clap:

.. got myself a gt640 wich will undergo some resistor soldering soon.

I am writing this post to ask, if anybody knows if the Tesla K40 is modabble in a similar way ?

I think its the same PCB like a Quatto 6000, is that correct ?

Could it be modded into a K2 or any other Card for that matter ?

I would like to use it in the VMware Host with the "vCPU" Feature.

Looking forward to get some insights from the experts.

Thanks !

Only known GRID converion is from Tesla K10 to GRID K2
Just get K10 from ebay and convert to K2 for vGPU
 

Offline PKTKS

  • Frequent Contributor
  • **
  • Posts: 705
  • Country: br
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1418 on: March 18, 2020, 03:00:21 pm »
(..)
(not duplicated URIs again)

I follow this thread rather curious...

I wonder how that folk with a substantial expensive hardware
like that ... (he's not cheap)  .. 

How he can breathe high density FLUX just 5 cm  with DIRECT INHALE
and DIRECT EYE contact...  may be that glass of alcohol helps...

Paul  :-+
 
The following users thanked this post: colokarl

Offline gotofbi

  • Contributor
  • Posts: 29
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1419 on: March 19, 2020, 07:05:05 am »
I really dont get it.
I flashed ROM from namgorf and confirmed that it has InfoROM
Still having this screen on Dell R720
I updated latest firmware on R720 as well...
It works fine on HP Z420 but not on Dell R720

What am I missing  |O |O |O |O |O |O |O |O |O |O |O |O |O

[attachimg=1]

Did you ever get past this?  You could say I am following in your footsteps from the shadows.  I just converted a Tesla K10 into an NVIDIA GRID K2 yesterday and I get the same result on my PowerEdge R720.

I soldered the strap resistors, reassembled the card with new thermal paste, and flashed the "2014" BIOS to each GPU.  I did notice the "PLX" entry in nvflash but didn't do anything with that.  Was I supposed to? The SHA256 hash of the BIOS file I flashed was BB04DF8552BF60B827E1C963B1D8527386D9448D554BC01799CE7F8605763951

My R720 has dual E5-2650 V2s, 128GB of PC3L-12800R, and dual 750W power supplies.  I saw the other post here about power delivery, but this card has a max power consumption at full load of maybe 200W, and my system power readout is only sitting at 280W with this GPU installed.  I am using a quick and dirty homemade power cable.  Please excuse the WAGO Lever Nuts.  I triple-checked the pinout and continuity with a multimeter, for what it's worth.

Right now I am using the riser connected to CPU2, because it was easier to install the card over there.  I ran out of time for troubleshooting last night.  I did not try the other riser, removing all other PCI-e cards, or enabling "Above 4G decoding" in the BIOS yet.  That is on the agenda for tonight.

I am running a BIOS version from 2016 to avoid the Spectre/Meltdown microcode mitigation.

I am still having same issue.
I tried multiple K2 BIOS but the issue is still there.
I also tried DELL K2 bios updater but didnt fix the problem (https://www.dell.com/support/home/uk/en/ukbsdt1/drivers/driversdetails?driverid=598p8)
I found inforom from CISCO iso but it didnt really help.
I also suspected my RISER cable which was 9H6FV but wiring is correct and it provided 12V/GND in proper pins (tested with multi meter)

I googled alot and no one complained about compatability issue other than you and me.
Also saw few people uses K2 in R720 without any issue.

At this point, there are only three things left.

1. Requires 1100W
2. Proper inforom
3. 40k ohm is wrong.

Since K2 is working fine on other system, I think its gonna be either option 1 or option 2.
 

Offline gotofbi

  • Contributor
  • Posts: 29
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1420 on: March 20, 2020, 05:51:49 am »
I really dont get it.
I flashed ROM from namgorf and confirmed that it has InfoROM
Still having this screen on Dell R720
I updated latest firmware on R720 as well...
It works fine on HP Z420 but not on Dell R720

What am I missing  |O |O |O |O |O |O |O |O |O |O |O |O |O

[attachimg=1]

Did you ever get past this?  You could say I am following in your footsteps from the shadows.  I just converted a Tesla K10 into an NVIDIA GRID K2 yesterday and I get the same result on my PowerEdge R720.

I soldered the strap resistors, reassembled the card with new thermal paste, and flashed the "2014" BIOS to each GPU.  I did notice the "PLX" entry in nvflash but didn't do anything with that.  Was I supposed to? The SHA256 hash of the BIOS file I flashed was BB04DF8552BF60B827E1C963B1D8527386D9448D554BC01799CE7F8605763951

My R720 has dual E5-2650 V2s, 128GB of PC3L-12800R, and dual 750W power supplies.  I saw the other post here about power delivery, but this card has a max power consumption at full load of maybe 200W, and my system power readout is only sitting at 280W with this GPU installed.  I am using a quick and dirty homemade power cable.  Please excuse the WAGO Lever Nuts.  I triple-checked the pinout and continuity with a multimeter, for what it's worth.

Right now I am using the riser connected to CPU2, because it was easier to install the card over there.  I ran out of time for troubleshooting last night.  I did not try the other riser, removing all other PCI-e cards, or enabling "Above 4G decoding" in the BIOS yet.  That is on the agenda for tonight.

I am running a BIOS version from 2016 to avoid the Spectre/Meltdown microcode mitigation.

I am still having same issue.
I tried multiple K2 BIOS but the issue is still there.
I also tried DELL K2 bios updater but didnt fix the problem (https://www.dell.com/support/home/uk/en/ukbsdt1/drivers/driversdetails?driverid=598p8)
I found inforom from CISCO iso but it didnt really help.
I also suspected my RISER cable which was 9H6FV but wiring is correct and it provided 12V/GND in proper pins (tested with multi meter)

I googled alot and no one complained about compatability issue other than you and me.
Also saw few people uses K2 in R720 without any issue.

At this point, there are only three things left.

1. Requires 1100W
2. Proper inforom
3. 40k ohm is wrong.

Since K2 is working fine on other system, I think its gonna be either option 1 or option 2.

I just ordered 2x 1100W power supploy for R720 and will arrive end of next week.
This will eliminate one of possibility.

I will update the thread once I test it.
 

Offline colokarl

  • Contributor
  • Posts: 23
  • Country: de
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1421 on: March 21, 2020, 10:43:19 pm »
did two gtx640 to grid k1 and one gtx670 to grid k2.

That SMD soldering is done best with good soldering equipment.

I realized due to a non working resistor contact to PCB, one gtx640 id'd itself as a grid k340, so i guess you could also mod it to a k340.
« Last Edit: March 23, 2020, 09:44:16 am by colokarl »
 

Offline gotofbi

  • Contributor
  • Posts: 29
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1422 on: April 04, 2020, 12:43:05 am »
I really dont get it.
I flashed ROM from namgorf and confirmed that it has InfoROM
Still having this screen on Dell R720
I updated latest firmware on R720 as well...
It works fine on HP Z420 but not on Dell R720

What am I missing  |O |O |O |O |O |O |O |O |O |O |O |O |O

[attachimg=1]

Did you ever get past this?  You could say I am following in your footsteps from the shadows.  I just converted a Tesla K10 into an NVIDIA GRID K2 yesterday and I get the same result on my PowerEdge R720.

I soldered the strap resistors, reassembled the card with new thermal paste, and flashed the "2014" BIOS to each GPU.  I did notice the "PLX" entry in nvflash but didn't do anything with that.  Was I supposed to? The SHA256 hash of the BIOS file I flashed was BB04DF8552BF60B827E1C963B1D8527386D9448D554BC01799CE7F8605763951

My R720 has dual E5-2650 V2s, 128GB of PC3L-12800R, and dual 750W power supplies.  I saw the other post here about power delivery, but this card has a max power consumption at full load of maybe 200W, and my system power readout is only sitting at 280W with this GPU installed.  I am using a quick and dirty homemade power cable.  Please excuse the WAGO Lever Nuts.  I triple-checked the pinout and continuity with a multimeter, for what it's worth.

Right now I am using the riser connected to CPU2, because it was easier to install the card over there.  I ran out of time for troubleshooting last night.  I did not try the other riser, removing all other PCI-e cards, or enabling "Above 4G decoding" in the BIOS yet.  That is on the agenda for tonight.

I am running a BIOS version from 2016 to avoid the Spectre/Meltdown microcode mitigation.

I am still having same issue.
I tried multiple K2 BIOS but the issue is still there.
I also tried DELL K2 bios updater but didnt fix the problem (https://www.dell.com/support/home/uk/en/ukbsdt1/drivers/driversdetails?driverid=598p8)
I found inforom from CISCO iso but it didnt really help.
I also suspected my RISER cable which was 9H6FV but wiring is correct and it provided 12V/GND in proper pins (tested with multi meter)

I googled alot and no one complained about compatability issue other than you and me.
Also saw few people uses K2 in R720 without any issue.

At this point, there are only three things left.

1. Requires 1100W
2. Proper inforom
3. 40k ohm is wrong.

Since K2 is working fine on other system, I think its gonna be either option 1 or option 2.

I just ordered 2x 1100W power supploy for R720 and will arrive end of next week.
This will eliminate one of possibility.

I will update the thread once I test it.

I got 1100W today....
Same shit happens...

 |O |O |O |O |O |O |O |O |O |O |O
 

Offline colokarl

  • Contributor
  • Posts: 23
  • Country: de
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1423 on: April 12, 2020, 01:28:56 pm »
A short udpate:

I managed to get the gtx 670 => K2 and the gtx 640 (2GB) => K1 to get recognized by esx 5.5 (it was a bumby ride)

Driver used: "NVIDIA-kepler-VMware_ESXi_5.5_Host_Driver_367.106-1OEM.550.0.0.1331820"

After installing vCenter i also can enable 3D Support for the guest and the vm (must be Generation 8 or higher) is able to run Furmark inside the VM.

I think its running in sVGA Mode. As i am fairly new to esx i am not sure how to set it to vgpu in esx, anyone can give me a hint ?

My gtx 640 (1GB) => K1 is not recognized by esx, but i think i made a mistake while modding the bios - i will try to fix that today.

 

Offline mcerveny

  • Contributor
  • Posts: 7
  • Country: cz
Re: [MOVED] Hacking NVidia Cards into their Professional Counterparts
« Reply #1424 on: April 12, 2020, 08:33:21 pm »
Hello.

I checked this thread after many years but nothing changes (still replacing resistors on SPI lines  :D).
A few years ago I challenged myself to enable GTX/RTX/Quadro cards for vGPU paravirtualization and I was successful.

Check this: https://gridforums.nvidia.com/default/topic/8934/

The presented solution is proving that there is no differences between the same chip in GTX-Quadro-Tesla lines and all is about "software".
The solution is useful for virtualization only! It enables the vGPU feature for compatible GTX/RTX/Quadro cards to TESLA/GRID counterparts like M10,M60,P4,T4. It does not modify host installed card (no HW mod, no vBIOS mod, no SW driver modification, probably EULA compliant). It also relaxes all NVIDIA "crippled/throttled" vGPU features to guests (like GPU MEM limits, number of emulated monitors (max 4), resolution of monitors, CUDA...). It does not remove NVIDIA license (to be compliant).

I also made my virtualization stack and universal/lowlatency VDI protocol to replace all overpriced/overlicenced/overcomplicated commercial hypervisors.

Check this: https://gridforums.nvidia.com/default/topic/752/

So happy hacking! :-+
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf