Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Gao__Chao
Beginner
671 Views

Question about Intel Dev cloud job submitting

We are trying to implement some Matrix Multiplication across heterogeneous platform in Dev Cloud, however, our job can't be executed in the jupyter Lab, I attached our screen snipe below.

Gao__Chao_0-1601590740715.png

 

0 Kudos
13 Replies
GouthamK_Intel
Moderator
652 Views

Hi,

Thanks for reaching out to us!

As your issue is related to devcloud, we are moving your thread to dedicated Devcloud forum support for the faster response.


Thanks & Regards

Goutham


JananiC_Intel
Moderator
633 Views

Hi,


We tried from our side and we didn't notice such issues.Can you elaborate more about your issue like steps followed and type of sample used?


Gao__Chao
Beginner
618 Views

Hi, 

I submit two demo to intel dev cloud written in DPC++,

one Matrix Multiplication on GPU;

one Matrix Vector Multiplication on FPGA emulator;

I couldn't get the report for FPGA.

Could you plz check this for me?

Gao__Chao_0-1602104308137.png

 

Gao__Chao
Beginner
611 Views

Hi,

I still wonder why that all my gpu demo will be finished quickly and the vtune will response very qucikly, however,  when it comes to FPGA emulator, it would be very slow and I never see the vtune outcome. Is because FPGA needs more  time or something wrong with the devcloud? 

Gao__Chao_0-1602110242336.png

 

JananiC_Intel
Moderator
596 Views

Hi,


Thanks for the update.


We suggest you to try fpga oneapi samples from devcloud using oneapi-cli command also run each commands individually in devcloud.


Try this and let us know the updates.


Gao__Chao
Beginner
590 Views

Hi,

Thanks for the replying, sir.

Sorry, I am not quite sure what oneapi-cli means here, I did try to run the command from PUTTY, also the code have some issue here, could you please take a look at that?

Gao__Chao_0-1602175572560.png

 

Gao__Chao_2-1602176393923.png

 

 

 

Thanks

Chao

JananiC_Intel
Moderator
557 Views

Hi,


Thanks for the update.


Screenshot shows that you are in gpu node,running fpga sample in gpu node will throw "FPGA device is not found" error.Hence get into fpga node with qsub -I -l nodes=1:fpga_runtime:ppn=2 -d. command and run the sample.Also try to run each commands individually.Try this and let us know the updates.




Gao__Chao
Beginner
551 Views

Hi,

sorry about the confusion sir, I finally figure out how FPGA works in intel dev-cloud, I guess you are using FPGA emulator to simulate the performance of FPGA,

However, I do have another question, did you support the computation for GPU and FPGA interaction?

For example, I run a Matrix Matrix Multiplication in GPU and use the result and multiple with another Vector in FPGA, is this able to be finished within one file?

Thanks

Chao

JananiC_Intel
Moderator
534 Views

Hi,

You can launch two kernels with different device selector, after launching 1st kernel (gpu_selector) add queue.wait() for synchronization and then launch 2nd kernel with intel::fpga_selector. So this is also possible in same single file.

Thanks.

 

JananiC_Intel
Moderator
513 Views

Hi,


Is your issue resolved?Do you have any updates?


Gao__Chao
Beginner
506 Views

Not test, sir
I haven’t figured out what compile instructions I should use, you see that compile instructions are related to your devices usage isn’t it?
Thanks
JananiC_Intel
Moderator
474 Views

Hi,


We suggest you to post FPGA related queries in FPGA forum(https://community.intel.com/t5/Intel-High-Level-Design/bd-p/high-level-design) for a faster response.Also we would like to know about your initial devcloud job submission query.Is it resolved?


Thanks


JananiC_Intel
Moderator
401 Views

Hi,


We are assuming that the solution provided helped and would no longer be monitoring this issue.Please raise a new thread if you have any further issues.


Thanks


Reply