DevCloud - Issues With Stratix10 Nodes

[Christoph9]

Hello,

I hope this is the right sub-forum. The Intel Customer Support told me to issue my problems here in the community forums.

I currently experience problems with some nodes in the DevCloud containing Stratix10 FPGAs. My programs there can not detect any real FPGA hardware. To my knowledge, in the nodes s001-n143 and s001-n144 are affected. I tried sycl-ls to prove that my programs are not at fault:

u153009@login-2:~/thesis/pure_conversion/build$ qsub -I -l nodes=1:fpga_runtime:stratix10:ppn=2 -d .
qsub: waiting for job 1982602.v-qsvr-1.aidevcloud to start
qsub: job 1982602.v-qsvr-1.aidevcloud ready


########################################################################
#      Date:           Sat Sep 10 04:35:53 PDT 2022
#    Job ID:           1982602.v-qsvr-1.aidevcloud
#      User:           u153009
# Resources:           neednodes=1:fpga_runtime:stratix10:ppn=2,nodes=1:fpga_runtime:stratix10:ppn=2,walltime=06:00:00
########################################################################

u153009@s001-n143:~/thesis/pure_conversion/build$ sycl-ls
[opencl:acc:0] Intel(R) FPGA Emulation Platform for OpenCL(TM), Intel(R) FPGA Emulation Device 1.2 [2022.13.3.0.16_160000]
[opencl:cpu:1] Intel(R) OpenCL, Intel(R) Xeon(R) Platinum 8256 CPU @ 3.80GHz 3.0 [2022.13.3.0.16_160000]
[host:host:0] SYCL host platform, SYCL host device 1.2 [1.2]
u153009@s001-n143:~/thesis/pure_conversion/build$ 

Node 144 also shows the same behavior with sycl-ls, so I guess both nodes have some issues with their FPGAs?

Would be great if someone could look into it, the only other Stratix10 node I know that works is 142, which is occupied all the time (at least always when I try) and I would prefer a dynamic session to test my programs.

Best regards,
Christoph

[BoonBengT_Intel · Accepted solution]

Hi @Christoph9,

Thank you for posting in Intel community forum and hope all is well.

My guess is similar and seems like a hardware issues, may I know which example design did you compile and tried executed with? (just to reconfirm that it is hardware issues)

Unfortunately currently there are only the 3 mention nodes with stratix10 devices (i.e., n142, n143 (job), n144), for the time being would recommend to connect to n142 as per mention is working.

At the same time from my end I will be submitting internal request to check on checked and worked on the 2 mention nodes with issues.

Apologies for the inconvenience, please do let us know if there is further clarification/doubts.

Best Wishes

BB

[Christoph9]

Thanks for your reply!

We currently port a GPU benchmark to FPGAs. However our DPC++ programs as stated only find the same devices as sycl-ls on these nodes, so the programs fail already at the device initialization.

If the problems are submitted to the corresponding tech team my issue is fullfilled I think, as the report was the only reason behind it (as there are no other ways of reporting such issues with nodes to the DevCloud team). I hope there will soon be replacements as only 1 node is probably not enough most times.

Best regards,
Christoph

[BoonBengT_Intel]

Hi @Christoph9,

Appreciate your patients, request have been submitted and the platform team will look into those accordingly and implement fixes accordingly.

For future references, is there is potential hardware/platform failure that you notice, below would be an alternative channels to raise to us accordingly.

http://supporttickets.intel.com/supportrequest?lang=en-US&productId=200042:11572

Hope that clarify your doubts, as there is no further clarification on this thread, it will be transitioned to community support for further help on doubts in this thread.

Thank you for the questions and as always pleasure having you here.

Best Wishes

BB