DVA3221 loader development thread

[iceman]

So much time has passed, is there really no one who can help with solving my problem?

[phone guy]

@iceman your hba card must support IT mode to be used. IT mode does individual drive pass through to the host/os. Raid card do not, they handle all the drives, and then the host/os sees the card as 1 big storage device. (this explanation is a oversimplified, but it gets the point across)

 

The card needs to pass each and every drive directly through to the os, so the os sees drive1/2/3/4 instead of "megaraid1" drive.  So either see if your card has a jbod mode, or initiate target (IT) mode.  if those are not possible, you need a it mode hba card.

[phone guy]

On 9/19/2022 at 1:40 AM, Kyar said:

Hello someone succeeded to install on Proxmox server ?

 

I tried many method and all times "corrupted files" at 55%

 

 

Many people got it working under proxmox. If you are getting 55% error, you have your proxmox settings wrong, or your loader (usb/sata) is visible dsm install process, and it shouldnt be. Try again. Look in my signature theres a how to redpill proxmox guide I wrote, I prefer to use a virtual usb as loader, because its simple, and it always works. Others like using a vdmk disk, which is still virtual.  Real boxes use a usb type module, so that what I use...plus it works.  Last make sure your cpu can handle whichever platform your testing (3221) must have fma3 or be haskell gen or newer. Very your methods, you missed something. good luck

[iceman]

38 минут назад, phone guy сказал:

@iceman your hba card must support IT mode to be used. IT mode does individual drive pass through to the host/os. Raid card do not, they handle all the drives, and then the host/os sees the card as 1 big storage device. (this explanation is a oversimplified, but it gets the point across)

 

The card needs to pass each and every drive directly through to the os, so the os sees drive1/2/3/4 instead of "megaraid1" drive.  So either see if your card has a jbod mode, or initiate target (IT) mode.  if those are not possible, you need a it mode hba card.

tnx for reply

 

I wrote earlier that my controller has JBOD mode and at the moment the disks installed in the slots are configured as JBOD and I still do not see disks for installation.

Another controller on this server does not have a specific installation slot.

[WiteWulf]

JBOD will still be presenting all your physical disks to the guest as one single disk. As has been pointed out, you must have your HBA in IT mode to pass through individual disks to the guest.

[iceman]

7 часов назад, WiteWulf сказал:

JBOD will still be presenting all your physical disks to the guest as one single disk. As has been pointed out, you must have your HBA in IT mode to pass through individual disks to the guest.

No, in my case, JBOD provides the guest with each disk separately.

[iceman]

@pocopico

 

Friend please can you help me?

Edited September 26, 2022 by iceman

[phone guy]

As an example, if your card does jbod or it mode, if you put that card and drives in a windows pc, it should see all those drives as independent drives automatically like they are connected to the local sata ports, and you should see serial/smart data aswell. if not, get a new card. Or find someone who is using your card in an xpen, truenas, unraid setup. in the end, we all want each drives passed thru to the host os. Some cards do it, while others just dont.

Edited September 29, 2022 by phone guy

[manalishii]

On 5/19/2022 at 10:21 PM, alexsahka said:

This is not a guide, this is my personal experience with the installation of DVA3221 on the ESXI.

 

Software/files used: 
tinycore-redpill.v0.4.6.img.gz
StarWind V2V Converter 
Putty for windows
WinSCP for windows 
ESXI host with 7.0 
Synology x2 pat files
dva3221_42218.pat and dva3221_42661.pat

Extract tinycore-redpill.v0.4.6.img.gz

With StarWind V2V Converter convert tinycore-redpill.v0.4.6.img to vmdk file. (redpill provided .vmdk file not working for some reasons)

Upload tinycore-redpill.v0.4.6.vmdk file to ESXI

 

Create VM in ESXI.

 

BOIS: EFI

CPU 4 cores
8G Ram (2G min)
Network:  VMXNET3 MAC Manual
SATA Controller 0
SATA Controller 1
Video card (Default)
Sound card (HD audio)
ADD LATER PCI device (GPU passthrough)  

Add Existing HDD (tinycore-redpill.v0.4.6.vmdk file converted above from .img file) SATA 0:0
Add New HDD 32G (your choice) SATA 1:0

 

Power On VM.

Find tinycore IP,
With WinSCP or Putty add additional config from first topic to custom_config.json to handle this loader

./rploader.sh update now
./rploader.sh fullupgrade now

./rploader.sh serialgen DVA3221       (Save the MAC for later use.)

 

./rploader.sh ext denverton-7.0.1-42218 add https://raw.githubusercontent.com/OrpheeGT/redpill-ext/master/acpid/rpext-index.json
./rploader.sh ext denverton-7.0.1-42218 add https://raw.githubusercontent.com/OrpheeGT/redpill-load/develop/redpill-virtio/rpext-index.json

 

(After this step, I got an error /home/tc/custom-module/*42218*.pat': No such file or directory 
Solution: manually uploaded dva3221_42218.pat to /home/tc/custom-module/    folder.)

 

Run two times build denverton-7.0.1-42218 (recommended somewhere on the forum)

 

./rploader.sh build denverton-7.0.1-42218
./rploader.sh build denverton-7.0.1-42218
./rploader.sh clean now
./rploader.sh build denverton-7.1.0-42661

 

Power down the VM.

Change the Network adapter with MAC from step above.
Block the future Synology IP for accessing to Internet. (for Installation time and first boot)

 

Powerup VM

 

Select SATA, Verbose.

 

Install DVA3221 with the dva3221_42661.pat file.

 

I experienced more troubles with passthrough GPU to DVA3221,

if someone needed I can post it later, maybe my hardware related. (GTX1650 used)

 

I followed your instructions, but after boot the VM ends up with Kernel Panic. Using HP-DL380 esxi 6.0U3. Any idea what is going wrong here?

DVA3222-7.0.1-42218-serial_output.txt

[dimakv2014]

В 29.09.2022 в 21:15, manalishii сказал:

I followed your instructions, but after boot the VM ends up with Kernel Panic. Using HP-DL380 esxi 6.0U3. Any idea what is going wrong here?

DVA3222-7.0.1-42218-serial_output.txt 40.17 \u041a\u0431 · 1 download

Change virtual network to e1000e

Edited October 3, 2022 by dimakv2014
Typo

[rudymiel]

Thanks for this interesting information!

[rudymiel]

The benefits of custom software are enormous. You can simplify your workforce, gain access to the latest technologies, and make your business more transparent and reliable. Custom software development https://mlsdev.com also reduces your learning curve and requires less maintenance. Moreover, customized software allows you to upgrade your application in a future, should new technology become available.

Edited October 16, 2022 by rudymiel

[jeffestewart]

I have a P1000 I was hoping to use in my build.  It is recognized by DSM but when I attempt to make changes to Facial Recognition or Analytics, I get the error like so many others.  I did attempt to replace the necessary files as suggested in several other posts.  After this it seems I am getting the error message less often but still getting it and neither feature works.  

[c_zs]

Questions about DVA3219/DVA3221 installation certification

I have a very small micro server hpe gen10 plus machine with esxi virtualization system running on it. I want to install dva3221 to use Deep Video Analytics, but after searching google and youtube, I can't find a real and effective paired sn and mac, and I can't solve the problem of AME 2.0. Now there are two options as follows:

a. Install DVA3219 (with real, valid and matching SN and MAC, which can perfectly solve AME 2.0 certification)

b. Install DVA3221 and stay at DSM7.0.1-42218 until the problem of AME 2.0 is solved perfectly and then install the new version

Is there a better option, and can the DVA3219 be used normally if the GTX1650 is installed?

[c_zs]

If someone is willing to share the real and effective sn/mac of DVA3221 to me for certification ame, I will be very grateful, and I am willing to provide reasonable compensation

[dookei]

Hi there,

I been, for some time now, trying to adapt an old desktop server that I used in the past for 3D design.

Currently is installed the denverton-7.1-42661 update 4.

Everything is working fine execept the facial recognition (or related tasks that use the the nvidia card) that dont work at all.

The Card is detected on both nvdia-smi and DSM.

Here is my Output from nvidia-smi and lspci -tvnn.

 

$ nvidia-smi
Sun Oct 23 20:38:10 2022
+-----------------------------------------------------------------------------+
| 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 GTX 770     On   | 00000000:01:00.0 N/A |                  N/A |
| 34%   41C    P8    N/A /  N/A |    175MiB /  4035MiB |     N/A      Default |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|    0                    Not Supported                                       |
+-----------------------------------------------------------------------------+
$ lspci -tvnn
-[0000:00]-+-00.0  Intel Corporation 4th Gen Core Processor DRAM Controller [8086:0c00]
           +-01.0-[01]--+-00.0  NVIDIA Corporation GK104 [GeForce GTX 770] [10de:1184]
           |            \-00.1  NVIDIA Corporation GK104 HDMI Audio Controller [10de:0e0a]
           +-14.0  Intel Corporation 9 Series Chipset Family USB xHCI Controller [8086:8cb1]
           +-16.0  Intel Corporation 9 Series Chipset Family ME Interface #1 [8086:8cba]
           +-19.0  Intel Corporation Ethernet Connection (2) I218-V [8086:15a1]
           +-1a.0  Intel Corporation 9 Series Chipset Family USB EHCI Controller #2 [8086:8cad]
           +-1c.0-[02]--
           +-1c.3-[03-0a]----00.0-[04-0a]--+-01.0-[05]--
           |                               +-03.0-[06]----00.0  ASMedia Technology Inc. ASM1062 Serial ATA Controller [1b21:0612]
           |                               +-04.0-[07]--
           |                               +-05.0-[08]--
           |                               +-06.0-[09]----00.0  Intel Corporation I211 Gigabit Network Connection [8086:1539]
           |                               \-07.0-[0a]--
           +-1c.4-[0b]----00.0  ASMedia Technology Inc. ASM1062 Serial ATA Controller [1b21:0612]
           +-1c.6-[0c]----00.0  ASMedia Technology Inc. ASM1042A USB 3.0 Host Controller [1b21:1142]
           +-1d.0  Intel Corporation 9 Series Chipset Family USB EHCI Controller #1 [8086:8ca6]
           +-1f.0  Intel Corporation Z97 Chipset LPC Controller [8086:8cc4]
           +-1f.2  Intel Corporation 9 Series Chipset Family SATA Controller [AHCI Mode] [8086:8c82]
           \-1f.3  Intel Corporation 9 Series Chipset Family SMBus Controller [8086:8ca2]

 

This is also detected under DSM.

 

Is there anything I could try to make this graphic card work under the surveillance station?

Many thanks

Edited October 23, 2022 by dookei

[c_zs]

I installed a GTX 1650 on esxi and passthrough to the virtual machine DVA3221, the GPU works fine on the DVA3221 as expected, when restarting the DVA3221, the problem starts:

1. If you only restart the DVA3221 virtual machine, open the [Control Panel] of Synology and enter the [Information Center] after startup, and you will find that the GPU-related information field shows that the device cannot be found. and related hardware transcoding, "Deep Video Analytics" will not work

2. If you restart the ESXI host and start the DVA3221 there is no problem,

Reason: I looked up articles online saying this is a known issue when using DirectPath I/O on unsupported desktop hardware. Even after the virtual machine is shut down, PCI devices passed directly to the virtual machine remain "locked" and require a hypervisor restart to function again. This particular problem cannot be solved other than using supported and certified hardware.

Solution: I saw the solution of a netizen on the Enshan forum when the gpu was directly connected to the virtual machine Win10. The GPU was disabled through a script before the virtual machine Win10 was shut down, and the script was used to automatically enable the GPU after restarting. Manually test this way everything works.

Let me ask: How to automatically disable and enable devices under pci in Synology, I hope some experts can solve this problem.

[marcuswrk]

I just tested a P4000 card, it is working with DVA3219 not with DVA3221

I will do more testing!

[yeric79]

hi 

 

My dva3221 can not detect gtx 1650 gigabyte 

 

passthrough enable on ESXI 7.0.3g but not detected on dva3221 GUI

 

on cli : 

 nvidia-debugdump -l -z
nvmlInit succeeded
Listing all GPUs.
Found 1 NVIDIA devices
Error: nvmlDeviceGetHandleByIndex(): Unknown Error
FAILED to get details on GPU (0x0): Unknown Error
 

 

an idea ?

[yeric79]

Il y a 2 heures, yeric79 a dit :

hi 

 

My dva3221 can not detect gtx 1650 gigabyte 

 

passthrough enable on ESXI 7.0.3g but not detected on dva3221 GUI

 

on cli : 

 nvidia-debugdump -l -z
nvmlInit succeeded
Listing all GPUs.
Found 1 NVIDIA devices
Error: nvmlDeviceGetHandleByIndex(): Unknown Error
FAILED to get details on GPU (0x0): Unknown Error
 

 

an idea ?

other info with lspci

 

03:00.0 Class 0300: Device 10de:2188 (rev a1)
        DeviceName: pciPassthru0
        Subsystem: Device 1458:4027
        Flags: bus master, fast devsel, latency 64, IRQ 18
        Memory at fc000000 (32-bit, non-prefetchable) [size=16M]
        Memory at d0000000 (64-bit, prefetchable) [size=256M]
        Memory at e4000000 (64-bit, prefetchable) [size=32M]
        I/O ports at 4000 [size=128]
        Capabilities: [60] Power Management version 3
        Capabilities: [68] MSI: Enable- Count=1/1 Maskable- 64bit+
        Capabilities: [78] Express Endpoint, MSI 00
        Capabilities: [100] Virtual Channel
        Capabilities: [250] Latency Tolerance Reporting
        Capabilities: [258] L1 PM Substates
        Capabilities: [128] Power Budgeting <?>
        Capabilities: [420] Advanced Error Reporting
        Capabilities: [600] Vendor Specific Information: ID=0001 Rev=1 Len=024 <?>
        Capabilities: [900] Secondary PCI Express
        Capabilities: [bb0] Resizable BAR <?>
        Kernel driver in use: nvidia

 

Edited October 27, 2022 by yeric79

[dimakv2014]

7 часов назад, yeric79 сказал:

other info with lspci

 

03:00.0 Class 0300: Device 10de:2188 (rev a1)
        DeviceName: pciPassthru0
        Subsystem: Device 1458:4027
        Flags: bus master, fast devsel, latency 64, IRQ 18
        Memory at fc000000 (32-bit, non-prefetchable) [size=16M]
        Memory at d0000000 (64-bit, prefetchable) [size=256M]
        Memory at e4000000 (64-bit, prefetchable) [size=32M]
        I/O ports at 4000 [size=128]
        Capabilities: [60] Power Management version 3
        Capabilities: [68] MSI: Enable- Count=1/1 Maskable- 64bit+
        Capabilities: [78] Express Endpoint, MSI 00
        Capabilities: [100] Virtual Channel
        Capabilities: [250] Latency Tolerance Reporting
        Capabilities: [258] L1 PM Substates
        Capabilities: [128] Power Budgeting <?>
        Capabilities: [420] Advanced Error Reporting
        Capabilities: [600] Vendor Specific Information: ID=0001 Rev=1 Len=024 <?>
        Capabilities: [900] Secondary PCI Express
        Capabilities: [bb0] Resizable BAR <?>
        Kernel driver in use: nvidia

 

Does your gpu have gddr6 or gddr5 vram?

If this one https://www.techpowerup.com/vgabios/227447/227447

With gddr6 I have bad news for you, revisions of gtx1650 (non super) released after driver 440.44 so its on on driver list and it wont work without driver update.

Edited October 27, 2022 by dimakv2014

[dimakv2014]

В 26.10.2022 в 11:58, c_zs сказал:

I installed a GTX 1650 on esxi and passthrough to the virtual machine DVA3221, the GPU works fine on the DVA3221 as expected, when restarting the DVA3221, the problem starts:

1. If you only restart the DVA3221 virtual machine, open the [Control Panel] of Synology and enter the [Information Center] after startup, and you will find that the GPU-related information field shows that the device cannot be found. and related hardware transcoding, "Deep Video Analytics" will not work

2. If you restart the ESXI host and start the DVA3221 there is no problem,

Reason: I looked up articles online saying this is a known issue when using DirectPath I/O on unsupported desktop hardware. Even after the virtual machine is shut down, PCI devices passed directly to the virtual machine remain "locked" and require a hypervisor restart to function again. This particular problem cannot be solved other than using supported and certified hardware.

Solution: I saw the solution of a netizen on the Enshan forum when the gpu was directly connected to the virtual machine Win10. The GPU was disabled through a script before the virtual machine Win10 was shut down, and the script was used to automatically enable the GPU after restarting. Manually test this way everything works.

Let me ask: How to automatically disable and enable devices under pci in Synology, I hope some experts can solve this problem.

Out of curiousity which version of GTX1650 you have gddr5 or gddr6? In your case its probably has something to do with esxi, which version of esxi do you use? Also do you have other gpu like igpu maybe on in bios? In order to passthrough gpu you need to have one extra for display output of esxi itself, so to pass it to vm you need 2 gpus

Edited October 27, 2022 by dimakv2014

[Cali0028]

Has anyone documented how to get the nvidia card to work with Jellyfin, either using docker or community package? 

[yeric79]

@dimakv2014

 

yes it's DDR6

 

it's KO ! 

 

no way ?

[yeric79]

i have old nvidia gtx 1050 

 

is it functionnal with this ?