rheine
New ContributorJoined 5 years ago
User Profile
New Contributor
User Widgets
Contributions
Re: OpenCL 18.1 Command Queue does not flush/finish/update.
We have attempted every tool from 17 to 19. None of them work as per the original 15. There seems to be something fundamentally broken in the openCL libraries as the command queues never flush or complete. This does not occur with our original v15 built tools. Due to Intel not providing the chips we need, we no longer have time to deal with a long winded minimal support solution. I will need real support for this, not hope that someone will get back to me on a forum.2.1KViews0likes0CommentsRe: OpenCL 18.1 Command Queue does not flush/finish/update.
As the best as we can tell, the latest tools are fundamentally broken and do not allow us to run opencl on the Cyclone 5 SOC ARM. We are at a critical impasse due to the lack of original parts from Intel. Updating tools has been attempted and do not work. There must be an archive somewhere to maintain full support for Intel products.2.2KViews0likes0CommentsOpenCL 18.1 Command Queue does not flush/finish/update.
We upgraded our tools to from 15.0 to 18.1. Previously working opencl code now does not work. Command Queues never flush or complete. Underneath we can see that the FPGA kernels do start and run one time, but then the queue is blocked in a unflushed state. A simple solution would be to find the quartus 15.0 tools again. Is that possible?2.2KViews0likes7CommentsCyclone 5 devices missing after reprogram.
After reprogramming the fpga on a Cyclone 5 SOC, I cannot seem to find the device anymore with the following error: MMD INFO : [acl0] HPS-to-FPGA read test failed, read 0x2 after 1 attempts ERROR: CL_DEVICE_NOT_FOUND However this was a very minor change to the design and it appears if we put it through the full boot process the driver is fine. It also occurs when I'm trying a new aocx file with opencl. Diagnostic repeats this error. Reprogramming with original .rbf file does not restore it to working condition. Ideas why?? Quartus environment version 18.1.734Views0likes1Commentacl_hal_mmd.c:755:assert failure, Quartus 18.1 OpenCL
Utilizing the 3.10 Linux build and 18.1 OpenCL runtime environment, we're stuck trying to get the aocx file to be accepted due to a version mismatch: FAILED to read auto-discovery string at byte 15: Expected version is 13, found 14 Error: The currently programmed/flashed design is no longer supported in this release. Please recompile the design with the present version of the SDK and re-program/flash the board. acl_hal_mmd.c:755:assert failure: Failed to initialize kernel interfacecyclone: acl_hal_mmd.c:755: l_try_device: Assertion `0' failed. Diagnostics repeat the same error. If everything is upto date on the with the same build (18.1.0.625), how is there a version mismatch?871Views0likes3CommentsError: hps: The pin information for the Hard Processor System could not be determin
Error: system.acl_iface.hps: The pin information for the Hard Processor System could not be determined. Please check whether your edition of Quartus Prime supports the selected device. How is this happenning for the hard processor in a cyclone 5 suddenly? I made no code changes, just went to rebuild and now I cannot get any hard processor design to compile.541Views0likes2CommentsQuartus 18.1, cyclone V hard processor lost pinouts in Qsys
Mature design utilizing Quartus 18.1 and a Cyclone 5 (5cseba6u23i7n). Built many times and suddenly stopped with this error: Error: system.acl_iface.hps: The pin information for the Hard Processor System could not be determined. Please check whether your edition of Quartus Prime supports the selected device. Our licenses are valid and other designs without a hard processor work fine. What is going on? Any clue?1.1KViews0likes2Comments