Jump to content
XPEnology Community

DVA3219 loader development thread


Recommended Posts

On 8/30/2022 at 5:19 PM, dimakv2014 said:

The problem with dva3219, dva3221 is that not all of pascal/turing gpus are supported because of an old driver in Nvidia Runtime Library spk. Current driver used for both dva3221 and dva3219 is 440.44 which is old since it was released in late 2019, many pascal and turing gpu revisions (for example with gddr6 memory) were released after this driver. Once synology updates drivers or a way to edit/replace drivers in ssh will be found things will be more interesting )

@dimakv2014 I don't think that is a problem, because as Nvidia described this version include a lot more hardware - take a look here https://www.nvidia.com/Download/driverResults.aspx/156086/en-us/ 

 

Probably Synology have a custom builded driver, based on this one. From last 3 days I personally test vGPU with spoofed PCI ID's, and I can confirm that. I've Proxmox installation of this. Now I'm searching an arg: based solution for this. Currently I'm with Nvidia T600 and every other VM works with spoofed ID's

 

Edited by Agisofttm
  • Thanks 1
Link to comment
Share on other sites

В 01.09.2022 в 10:54, Agisofttm сказал:

@dimakv2014 I don't think that is a problem, because as Nvidia described this version include a lot more hardware - take a look here https://www.nvidia.com/Download/driverResults.aspx/156086/en-us/ 

 

Probably Synology have a custom builded driver, based on this one. From last 3 days I personally test vGPU with spoofed PCI ID's, and I can confirm that. I've Proxmox installation of this. Now I'm searching an arg: based solution for this. Currently I'm with Nvidia T600 and every other VM works with spoofed ID's

 

Its the same linux driver, if you will go to ssh from xpenology box, run command nvidia-smi, you should see that its totally same driver as any linux driver just compiled for synology. You can even download NvidiaRuntimeLib.spk and extract driver from there, download from nvidia link that you provided and compare that its the same exact driver. As official gpu in dva3221 was GTX1650 ggdr5 and GTX1050Ti for DVA3219 both of which are included in this driver version there was no point to modify it by synology. Besides that driver was updated in Nvidia Runtime Lib to 440.44 way after GTX1650 gddr6 was released, if synology modified it why this gpu and many others don't work then )? Also why would they leave others gpu that work instead of cutting the list to GTX1050Ti and GTX1650 during modification? Synology is too lazy company to modify things, they do software updates very slow because hardware is most that gives them profit if something is already done they wouldnt do a thing to modify it.

You said you spoofed id of your gpu somehow, maybe this was a trick since driver 440.44 supports T1000 so maybe this is why you got AI working, but people with GTX1650 gddr6 as well as T400 cannot get AI to work (DVA and Face Rec). Maybe you can share info how you spoofed id and which ids you used? Your gpu's original pci id: 10de:1fb1 so interesting what is it currently. Thanks 

IMG_20220903_004211.jpg

IMG_20220903_004502.jpg

IMG_20220903_004850.jpg

 

IMG_20220903_023409.png

Edited by dimakv2014
Link to comment
Share on other sites

  • 2 months later...

Tested out with SUCCESS baremetal

tcrp v9.0.2.9 great job Pocopico and Peter Suh, many thanks :)

CPU: J4125

GPU: gtx1650 gddr5

NIC: RTL8169

 

DVA3219 with Friend DSM 7.1.1-42962 Update 1

DVA3219 Jot Mod DSM 7.1.1-42962 Update 1

DVA3219 Jun Mod DSM 7.0.1-42218 Update 5 (no other versions are available)

 

Used commands:

 

curl --location "https://github.com/PeterSuh-Q3/tinycore-redpill/raw/main/my.sh.gz" -O;tar -zxvf my.sh.gz;

./my.sh

./my.sh DVA3219F - for Friend

./my.sh DVA3219 - for Jot 

./my.sh DVA3219J - for Jun

sudo reboot

 

1. All Mods Work. Installs from first time no errors during installation.

2. Friend works little slower than others

3. GTX 1650 - AI (Face Rec and DVA) does not work!

It is recognized however gives "Network busy" error on Face Recognition or DVA. Tried older versions of SS and NvidiaRunTimeLibrary .spk still no luck.

Reinstalling SS or reset DSM didn't help either.

Works perfectly with AI on DVA3221, so I guess there is restriction to gpu generation somewhere to Pascal gpus only.

 

 

DVA3219-1.jpg

Edited by dimakv2014
  • Like 1
Link to comment
Share on other sites

В 25.07.2022 в 17:13, Peter Suh сказал:

DVA3219 Jot Mode

Hey Peter, strange thing going on, I purchased another gpu today GTX 1050 Ti and Face Rec as well as DVA doesn't work still, reinstalled SS and NvidiaLibrary didn't help same error Network Busy :( I thought it locked to Pascal gpus, bought pascal and still doesnt work, clearly something is wrong

Screenshot_2022-11-06-00-29-45-574_com.android.chrome.jpg

Edited by dimakv2014
Link to comment
Share on other sites

3 часа назад, dimakv2014 сказал:

Hey Peter, strange thing going on, I purchased another gpu today GTX 1050 Ti and Face Rec as well as DVA doesn't work still, reinstalled SS and NvidiaLibrary didn't help same error Network Busy :( I thought it locked to Pascal gpus, bought pascal and still doesnt work, clearly something is wrong

Screenshot_2022-11-06-00-29-45-574_com.android.chrome.jpg

Update:

I fixed it by full reset DSM and then did automatic download and update from synology .pat. It updated Jun mode dva3219 to 7.1-42661 Update 4 and face rec started to work... The only problem now that with 4 face rec tasks gpu loads about 92-93% and heats up till 80c degrees. 

IMG_20221106_041339.jpg

IMG_20221106_041311.jpg

IMG_20221106_043450.jpg

Edited by dimakv2014
Link to comment
Share on other sites

On 11/5/2022 at 3:25 AM, dimakv2014 said:

Tested out with SUCCESS baremetal

tcrp v9.0.2.9 great job Pocopico and Peter Suh, many thanks :)

CPU: J4125

GPU: gtx1650 gddr5

NIC: RTL8169

 

DVA3219 with Friend DSM 7.1.1-42962 Update 1

DVA3219 Jot Mod DSM 7.1.1-42962 Update 1

DVA3219 Jun Mod DSM 7.0.1-42218 Update 5 (no other versions are available)

 

Used commands:

 

curl --location "https://github.com/PeterSuh-Q3/tinycore-redpill/raw/main/my.sh.gz" -O;tar -zxvf my.sh.gz;

./my.sh

./my.sh DVA3219F - for Friend

./my.sh DVA3219 - for Jot 

./my.sh DVA3219J - for Jun

sudo reboot

 

1. All Mods Work. Installs from first time no errors during installation.

2. Friend works little slower than others

3. GTX 1650 - AI (Face Rec and DVA) does not work!

It is recognized however gives "Network busy" error on Face Recognition or DVA. Tried older versions of SS and NvidiaRunTimeLibrary .spk still no luck.

Reinstalling SS or reset DSM didn't help either.

Works perfectly with AI on DVA3221, so I guess there is restriction to gpu generation somewhere to Pascal gpus only.

 

 

DVA3219-1.jpg

Install tenté sur un N54l pour le fun, je n'arrive pas à pinger sur le réseau le pc....

il y a une config minimum pour ce type de "montage"?

Link to comment
Share on other sites

45 минут назад, krisz67 сказал:

Install tenté sur un N54l pour le fun, je n'arrive pas à pinger sur le réseau le pc....

il y a une config minimum pour ce type de "montage"?

AMD cpus dont work in BAREMETAL DVA, minimum is Intel CPUs Haswell generation. You can try emulate Intel CPU using Proxmox VE or QEMU

This is English forum, please use english next time ;)

Edited by dimakv2014
Link to comment
Share on other sites

10 minutes ago, dimakv2014 said:

AMD cpus dont work in BAREMETAL DVA, minimum is Intel CPUs minimum Haswell generation. You can try emulate Intel CPU using Proxmox VE or QEMU

This is English forum, please use english next time ;)

sorry, I wrote quickly, because I struggle a bit with my old pc, to launch the loader....

Link to comment
Share on other sites

  • 2 weeks later...
В 15.11.2022 в 15:42, yeric79 сказал:

hello i try install DVA3219 on ESXI but dsm install failed at 55%

 

 

with hdd data on sata or scsi

 

an idea ?

 

More info please, which cpu, how many virtual cores and ram, virtual network card used, which loader, version, step by step vm creation you have used?

Edited by dimakv2014
Link to comment
Share on other sites

  • 1 month later...

hi
sorry for long time with no answer

 

i try test install with tinycore-redpill.v0.8.0.0-conv.vmdk on base HDD

i use your script, 

./my.sh DVA3219
or ./myv.sh DVA3219

 

sometimes with differents settings no disk is detected

sometimes install failed at 55%

 

VM is on esxi 7 

4 vCPU

4 GO 

Vmxnet 3

 

Real CPU is CPU i9 9900K

 

other VM running DS918+ , DS3622, 920 are ok

 

Thank's for your work

Link to comment
Share on other sites

7 hours ago, yeric79 said:

hi
sorry for long time with no answer

 

i try test install with tinycore-redpill.v0.8.0.0-conv.vmdk on base HDD

i use your script, 

./my.sh DVA3219
or ./myv.sh DVA3219

 

sometimes with differents settings no disk is detected

sometimes install failed at 55%

 

VM is on esxi 7 

4 vCPU

4 GO 

Vmxnet 3

 

Real CPU is CPU i9 9900K

 

other VM running DS918+ , DS3622, 920 are ok

 

Thank's for your work

 

 

Current TCRP is being discussed centering on TCRP FRIEND. Did you also use TCRP FRIEND?

 

In TCRP FRIEND, the probability of encountering a 55% disk corruption message has been greatly reduced.

Link to comment
Share on other sites

  • 2 weeks later...
Le 26/01/2023 à 03:54, Peter Suh a dit :

 

 

Current TCRP is being discussed centering on TCRP FRIEND. Did you also use TCRP FRIEND?

 

In TCRP FRIEND, the probability of encountering a 55% disk corruption message has been greatly reduced.

hi

it's ok with TCRP 0.9.4

and with no update user_config.json (with update no hdd found)

 

but GPU not detected

 

image.thumb.png.34555b26d9d85e4e05492c4c55d86122.png

 

 

image.thumb.png.9d2ca3ab6755a7a58ec0d62925980d2e.png

 

 

 

image.png

Edited by yeric79
Link to comment
Share on other sites

  • 1 month later...
  • 4 months later...

I have DVA3219 working with GT 1030 with Facial Recognition and Deep Video Analytics. I'd say a good low-power alternative to enable AI tasks.

 

I used arpl-i18n loader. DVA3219 is under beta platform.

 

+-----------------------------------------------------------------------------+
| NVIDIA-SMI 440.44       Driver Version: 440.44       CUDA Version: 10.2     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GeForce GT 1030     On   | 00000000:01:00.0 Off |                  N/A |
| 58%   60C    P0    N/A /  30W |   1516MiB /  2001MiB |     74%      Default |
+-------------------------------+----------------------+----------------------+
                                                                               
+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|    0     26525      C   ...ceStation/target/synoface/bin/synofaced   746MiB |
|    0     26527      C   ...anceStation/target/synodva/bin/synodvad   760MiB |
+-----------------------------------------------------------------------------+

 

Screenshot2023-08-05at22_05_53.thumb.png.ff7740e6cde82cda34bcd2f938abc0ca.pngScreenshot2023-08-05at22_08_28.thumb.png.566b4dedd03ac2c19fc9c3dc9782b2aa.pngScreenshot2023-08-05at22_20_14.thumb.png.089f0a78db52ee81ba422348124bccba.pngScreenshot2023-08-05at22_13_01.thumb.png.78ce436f6e751188e1ab79e0e94f664e.png

Edited by jerico
  • Like 1
Link to comment
Share on other sites

  • 3 weeks later...
On 8/7/2023 at 6:52 AM, jerico said:

I have DVA3219 working with GT 1030 with Facial Recognition and Deep Video Analytics. I'd say a good low-power alternative to enable AI tasks.

 

I used arpl-i18n loader. DVA3219 is under beta platform.

 

+-----------------------------------------------------------------------------+
| NVIDIA-SMI 440.44       Driver Version: 440.44       CUDA Version: 10.2     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GeForce GT 1030     On   | 00000000:01:00.0 Off |                  N/A |
| 58%   60C    P0    N/A /  30W |   1516MiB /  2001MiB |     74%      Default |
+-------------------------------+----------------------+----------------------+
                                                                               
+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|    0     26525      C   ...ceStation/target/synoface/bin/synofaced   746MiB |
|    0     26527      C   ...anceStation/target/synodva/bin/synodvad   760MiB |
+-----------------------------------------------------------------------------+

 

Screenshot2023-08-05at22_05_53.thumb.png.ff7740e6cde82cda34bcd2f938abc0ca.pngScreenshot2023-08-05at22_08_28.thumb.png.566b4dedd03ac2c19fc9c3dc9782b2aa.pngScreenshot2023-08-05at22_20_14.thumb.png.089f0a78db52ee81ba422348124bccba.pngScreenshot2023-08-05at22_13_01.thumb.png.78ce436f6e751188e1ab79e0e94f664e.png

And what did you do to make it work? Which brand, model, revision and memory type of gt 1030 did you use? More info would be great :)

Edited by dimakv2014
Link to comment
Share on other sites

  • 5 months later...

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  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...