Intel Open VINO R4 corei7 8750H iGPU support？

[taka]

Can not OpenVINO run on Corei i7 8750H iGPU

The following message will be displayed

"[ERROR] failed to create engine: Device lookup failed -  unsupported device id:0x3E9B"

clDNN source(gpu_devices.inc) does not mention the device id

Will it be improved in the next version?

Reference URL
https://github.com/intel/clDNN/blob/master/src/caps/public/gpu_devices.inc
https://ark.intel.com/products/134906/Intel-Core-i7-8750H-Processor-9M-Cache-up-to-4-10-GHz-

[nikos1]

Hello Takamitsu-san,

Looks like a typo in the clDNN code. Unofficial wiki https://en.wikipedia.org/wiki/List_of_Intel_graphics_processing_units shows your CPU has an 0x3E9B GPU and this is indeed what is reporting. 0x3E98 looks like a an 8 vs. B typo .

I can see the issue https://github.com/intel/clDNN/issues/47 was created

In the meantime, you may want to build clDNN from source and replace 0x3E98 by 0x3E9B

Cheers,

Nikos

[taka]

Hello Nikos,

0x3E98 exists. Is not not a mistake but a forgotten addition?

https://ark.intel.com/products/134896/Intel-Core-i5-9600K-Processor-9M-Cache-up-to-4-60-GHz-

Replacing openvino 's cldnn64.dll with cldnn' s build will fail with initialization

[nikos1]

Yes indeed, needs to be added! I was not aware of this CPU.

> Replacing openvino 's cldnn64.dll with cldnn' s build will fail with initialization

What compiler are you using? I would also try to build from source and link to my products. Maybe I will hit the same issue. Trying now..

[taka]

cldnn was built with visual studio2015.(Windows10)
openvino sample is also built with vs2015.

[nikos1]

Thanks! Also built with VS2015 64-bit Windows 10.  Just passed the default enabled unit tests - will try a repro

[==========] 659 tests from 98 test cases ran. (177287 ms total)
[  PASSED  ] 659 tests.

  YOU HAVE 17246 DISABLED TESTS
 

[nikos1]

Yes, getting issues here too. Maybe master is not stable. I am going to try earlier version

git checkout prod-2646
Checking out files: 100% (7903/7903), done.
Note: checking out 'prod-2646'.

You are in 'detached HEAD' state. You can look around, make experimental
changes and commit them, and you can discard any commits you make in this
state without impacting any branches by performing another checkout.

If you want to create a new branch to retain commits you create, you may
do so (now or later) by using -b with the checkout command again. Example:

  git checkout -b <new-branch-name>

HEAD is now at 5ef52733 offline data for mxnet

[nikos1]

Getting issues with prod-2646 too

[nikos1]

Anyone knows what version of clDNN was used in OpenVino 2018 R4 ?

Release notes mention Updated clDNN version to 9.1

How can we git checkout clDNN version 9.1 - what is the commit # ? Cannot find the product tag.

Thanks,

Nikos

[nikos1]

Better modify and build clDNN from  https://github.com/opencv/dldt/tree/2018_R4

https://github.com/opencv/dldt/tree/2018_R4/inference-engine/thirdparty/clDNN

R4 seems to be using clDNN Drop 9.1

[taka]

Thank you!

I was able to build clDNN Drop 9.1. Replacing cldnn64.dll, it worked properly with Corei i7 8750H.

[Loza__dayannex]

taka

Could you to build with i7-8750H ?, I' ve tried to build with visual studio 2017 and Sources Code 2019 R1  on Windows 10. I generated the solution inference-engine and dldt but when compiling an null exception is generated