Forum Discussion

JSchr20's avatar
JSchr20
Icon for Occasional Contributor rankOccasional Contributor
5 years ago

OpenCL: What's best practice for replicating a task kernel in-FPGA?

I've written a fairly simple kernel that multiplies multi-limb operands. I've written two versions: an ndrange version and a task version. Both work fine, but the performance of the task version is s...
AnilErinch_A_Intel's avatar
AnilErinch_A_Intel
Icon for Frequent Contributor rankFrequent Contributor
5 years ago

Hi ,

Would recommend to try out the code in v19.4 as mentioned in previous answer .

Please refer to the programming guide for some more clarification regarding recommended practices.

Please adhere to the statements in section

5.4.4. Restrictions in the Implementation of Intel FPGA SDK for OpenCL Channels Extension

like the one below.

"Performance of a kernel that has multiple accesses (reads or writes) to the same channel might be poor."

Also please refer to the following thread for some rules about multiple channel writes.

https://forums.intel.com/s/question/0D50P00003yyMFISA2/compiler-error-multiple-channel-write-sites

For implementing fence also please refer to

https://www.intel.com/content/dam/altera-www/global/en_US/pdfs/literature/hb/opencl-sdk/aocl_programming_guide.pdf

Please let us know , once you have rewritten the code following the guidelines.

Thanks and Regards

Anil