I am using OpenCL on this device:
PowerVR B-Series BXM-4-64 Imagination Technologies with [1 units] localmem=4KiB globalmem=5445MiB dims=3(512x512x512) max workgrp sz 512 fp16=y
I use it with the RISCV Sipeed Pi4A SBC.
Calling clCreateProgramWithSource() succeeds.
When I then call clBuildProgram() it returns success, but out the stdout I get:
44444444444444444444444444
Why does libPVROCL.so.1 spew this weird string consisting of 26 ‘4’ characters?
I see the same happen on Native_SDK’s OpenCL example.
debian@lpi4a:~/src/Native_SDK/build/bin$ ./OpenCLMatrixMultiplication
Single Precision General Matrix Multiplication (SGEMM) benchmarking test.
Host library 'libOpenCL.so' loaded
OpenCL Bindings: Successfully loaded library libOpenCL.so for OpenCL
[cl::createOpenCLContext]: Created context on platform PowerVR.
Running DEMO mode.
M: 512 N: 1536 P: 1024
Left Matrix(MxN): 512x1536 Right Matrix(NxP): 1536x1024
==> Creating OpenCL program - Time: 0.186
444444444444444444444444444444==> Beginning Matrix setup - Time: 1049.69(1049.71ms)
==> Populating Matrices - Time: 1.04975(313.522ms)
==> Transposing Matrix B - Time: 1.36334(478.444ms)
==> Setting up buffers - Time: 1.52825(39.413ms)
==> Preparing ground truth: CPU-side multiply - Time: 1.56773(0.037376ms)
===== EXECUTION: matmul_transposedA (4492.34ms)
===== EXECUTION: matmul_transposedB (4068.88ms)
===== EXECUTION: matmul_transposedC (4769.83ms)
===== EXECUTION: matmul_transposedAC (4507.65ms)
===== EXECUTION: matmul_transposedBC (4092.99ms)
===== EXECUTION: matmul_linearwg_AT (4263.81ms)
===== EXECUTION: matmul_linearwg_BT (3868.56ms)
===== EXECUTION: matmul_linearwg_vec4 (1097.46ms)
===== EXECUTION: matmul_linearwg_vec4_local (1546.28ms)
===== EXECUTION: matmul_tile_square (2031.60ms)
===== EXECUTION: matmul_tile_rect (1488.14ms)444444444444444444444444444444
===== EXECUTION: matmul_tile_rect_vec4 2x16x4 (1104.75ms)444444444444444444444444444444
===== EXECUTION: matmul_tile_rect_vec4 2x32x4 (1550.73ms)444444444444444444444444444444
===== EXECUTION: matmul_tile_rect_vec4 2x64x4 (2029.77ms)444444444444444444444444444444
===== EXECUTION: matmul_tile_rect_vec4 4x16x4 (863.97 ms)444444444444444444444444444444
===== EXECUTION: matmul_tile_rect_vec4 4x32x4 (836.47 ms)444444444444444444444444444444
===== EXECUTION: matmul_tile_rect_vec4 4x64x4 (906.36 ms)444444444444444444444444444444
===== EXECUTION: matmul_tile_rect_vec4 8x32x4 (778.41 ms)444444444444444444444444444444
===== EXECUTION: matmul_tile_rect_vec4 8x64x4 (798.79 ms)444444444444444444444444444444
===== EXECUTION: matmul_tile_rect_vec4 16x64x4 (1113.55ms)444444444444444444444444444444
===== EXECUTION: matmul_tile_rect_vec4 2x32x8 (1050.65ms)444444444444444444444444444444
===== EXECUTION: matmul_tile_rect_vec4 2x64x8 (1150.97ms)444444444444444444444444444444
===== EXECUTION: matmul_tile_rect_vec4 4x32x8 (650.49 ms)444444444444444444444444444444
===== EXECUTION: matmul_tile_rect_vec4 4x64x8 (689.34 ms)444444444444444444444444444444
===== EXECUTION: matmul_tile_rect_vec4 8x32x8 (655.85 ms)444444444444444444444444444444
===== EXECUTION: matmul_tile_rect_vec4 8x64x8 (709.09 ms)444444444444444444444444444444
===== EXECUTION: matmul_tile_rect_vec4 16x64x8 (1020.12ms)444444444444444444444444444444
===== EXECUTION: matmul_tile_rect_vec4 2x64x16 (1199.32ms)444444444444444444444444444444
===== EXECUTION: matmul_tile_rect_vec4 4x64x16 (614.44 ms)444444444444444444444444444444
===== EXECUTION: matmul_tile_rect_vec4 8x64x16 (654.15 ms)444444444444444444444444444444
===== EXECUTION: matmul_tile_rect_vec4 16x64x16 (992.91 ms)
*** HALL OF FAME ***
--------------------------------------------------
1 : matmul_tile_rect_vec4 4x64x16 (614.53ms)
2 : matmul_tile_rect_vec4 4x32x8 (650.58ms)
3 : matmul_tile_rect_vec4 8x64x16 (654.24ms)
4 : matmul_tile_rect_vec4 8x32x8 (655.93ms)
5 : matmul_tile_rect_vec4 4x64x8 (689.43ms)
6 : matmul_tile_rect_vec4 8x64x8 (709.18ms)
7 : matmul_tile_rect_vec4 8x32x4 (778.50ms)
8 : matmul_tile_rect_vec4 8x64x4 (798.88ms)
9 : matmul_tile_rect_vec4 4x32x4 (836.56ms)
10 : matmul_tile_rect_vec4 4x16x4 (864.04ms)
11 : matmul_tile_rect_vec4 4x64x4 (906.45ms)
12 : matmul_tile_rect_vec4 16x64x16 (993.39ms)
13 : matmul_tile_rect_vec4 16x64x8 (1020.21ms)
14 : matmul_tile_rect_vec4 2x32x8 (1050.74ms)
15 : matmul_linearwg_vec4 (1097.53ms)
16 : matmul_tile_rect_vec4 2x16x4 (1104.85ms)
17 : matmul_tile_rect_vec4 16x64x4 (1113.64ms)
18 : matmul_tile_rect_vec4 2x64x8 (1151.05ms)
19 : matmul_tile_rect_vec4 2x64x16 (1199.41ms)
20 : matmul_tile_rect (1488.23ms)
21 : matmul_linearwg_vec4_local (1546.37ms)
22 : matmul_tile_rect_vec4 2x32x4 (1550.83ms)
23 : matmul_tile_rect_vec4 2x64x4 (2029.86ms)
24 : matmul_tile_square (2032.10ms)
25 : matmul_linearwg_BT (3868.65ms)
26 : matmul_transposedB (4068.96ms)
27 : matmul_transposedBC (4093.08ms)
28 : matmul_linearwg_AT (4263.90ms)
29 : matmul_transposedA (4492.45ms)
30 : matmul_transposedAC (4507.74ms)
31 : matmul_transposedC (4769.92ms)
Hi bram,
Thank you for your message.
I am getting this behaviour checked with the OpenCL team and shall get back once I have more information.
Best Regards,
Nagnath
Hi bram,
Could we have more details about the Host environment and DDK revision?
Best Regards,
Nagnath
I am not using the DDK, I am not an OEM, but end-user.
The host is a Sipeed Pi4A, RISCV with TH1520 SoC.
The OpenCL library is included in Sipeed’s Debian image.
$ inxi -G
Graphics:
Device-1: gpu driver: pvrsrvkm v: kernel
Device-2: light-hdmi-tx driver: dwhdmi_light v: N/A
Device-3: display-subsystem driver: vs_drm v: N/A
Display: x11 server: X.org v: 1.20.10 driver: X: loaded: modesetting
unloaded: fbdev gpu: pvrsrvkm,dwhdmi_light,vs_drm tty: 80x24
API: OpenGL Message: GL data unavailable in console. Try -G --display
$ clinfo
Number of platforms 1
Platform Name PowerVR
Platform Vendor Imagination Technologies
Platform Version OpenCL 3.0
Platform Profile EMBEDDED_PROFILE
Platform Extensions cl_khr_icd cl_khr_fp16 cl_img_spirv cles_khr_int64 cl_img_yuv_image cl_khr_device_uuid cl_khr_depth_images cl_khr_priority_hints cl_img_generate_mipmap cl_khr_3d_image_writes cl_img_cached_allocations cl_khr_create_command_queue cl_img_mem_properties cl_img_mem_properties_relax_alloc_requirements cl_khr_extended_versioning cl_khr_image2d_from_buffer cl_khr_byte_addressable_store cl_khr_local_int32_base_atomics cl_khr_global_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_global_int32_extended_atomics cl_khr_spir cl_khr_egl_image cl_arm_import_memory cl_arm_import_memory_dma_buf cl_img_protected_content cl_img_semaphore cl_img_external_semaphore cl_img_external_semaphore_sync_fd
Platform Extensions with Version cl_khr_icd 0x400000 (1.0.0)
cl_khr_fp16 0x400000 (1.0.0)
cl_img_spirv 0x400000 (1.0.0)
cles_khr_int64 0x400000 (1.0.0)
cl_img_yuv_image 0x400000 (1.0.0)
cl_khr_device_uuid 0x400000 (1.0.0)
cl_khr_depth_images 0x400000 (1.0.0)
cl_khr_priority_hints 0x400000 (1.0.0)
cl_img_generate_mipmap 0x400000 (1.0.0)
cl_khr_3d_image_writes 0x400000 (1.0.0)
cl_img_cached_allocations 0x400000 (1.0.0)
cl_khr_create_command_queue 0x400000 (1.0.0)
cl_img_mem_properties 0x400000 (1.0.0)
cl_img_mem_properties_relax_alloc_requirements 0x400000 (1.0.0)
cl_khr_extended_versioning 0x400000 (1.0.0)
cl_khr_image2d_from_buffer 0x400000 (1.0.0)
cl_khr_byte_addressable_store 0x400000 (1.0.0)
cl_khr_local_int32_base_atomics 0x400000 (1.0.0)
cl_khr_global_int32_base_atomics 0x400000 (1.0.0)
cl_khr_local_int32_extended_atomics 0x400000 (1.0.0)
cl_khr_global_int32_extended_atomics 0x400000 (1.0.0)
cl_khr_spir 0x400000 (1.0.0)
cl_khr_egl_image 0x400000 (1.0.0)
cl_arm_import_memory 0x400000 (1.0.0)
cl_arm_import_memory_dma_buf 0x400000 (1.0.0)
cl_img_protected_content 0x400000 (1.0.0)
cl_img_semaphore 0x400000 (1.0.0)
cl_img_external_semaphore 0x400000 (1.0.0)
cl_img_external_semaphore_sync_fd 0x400000 (1.0.0)
Platform Numeric Version 0xc00000 (3.0.0)
Platform Extensions function suffix IMG
Platform Host timer resolution 0ns
Platform Name PowerVR
Number of devices 1
Device Name PowerVR B-Series BXM-4-64
Device Vendor Imagination Technologies
Device Vendor ID 0x1
Device Version OpenCL 3.0
Device UUID 33362035-3220-3130-3420-313832000000
Driver UUID 36303939-3538-3000-0000-000000000000
Valid Device LUID No
Device LUID 0000-000000000000
Device Node Mask 0
Device Numeric Version 0xc00000 (3.0.0)
Driver Version 1.16@6099580
Device OpenCL C Version OpenCL C 1.2
Device OpenCL C Numeric Version 0x402000 (1.2.0)
Device OpenCL C all versions OpenCL C 0x400000 (1.0.0)
OpenCL C 0x401000 (1.1.0)
OpenCL C 0x402000 (1.2.0)
OpenCL C 0xc00000 (3.0.0)
Device OpenCL C features __opencl_c_int64 0x400000 (1.0.0)
__opencl_c_pipes 0xc00000 (3.0.0)
__opencl_c_images 0x400000 (1.0.0)
__opencl_c_subgroups 0xc00000 (3.0.0)
__opencl_c_3d_image_writes 0x400000 (1.0.0)
__opencl_c_read_write_images 0x400000 (1.0.0)
__opencl_c_generic_address_space 0xc00000 (3.0.0)
__opencl_c_program_scope_global_variables 0xc00000 (3.0.0)
__opencl_c_work_group_collective_functions 0xc00000 (3.0.0)
Latest conformance test passed v2021-06-16-00
Device Type GPU
Device Profile EMBEDDED_PROFILE
Device Available Yes
Compiler Available Yes
Linker Available Yes
Max compute units 1
Max clock frequency 792MHz
Device Partition (core)
Max number of sub-devices 1
Supported partition types None
Supported affinity domains (n/a)
Max work item dimensions 3
Max work item sizes 512x512x512
Max work group size 512
Preferred work group size multiple (device) 32
44444444444444444444444444 Preferred work group size multiple (kernel) 32
Max sub-groups per work group 512
Preferred / native vector sizes
char 1 / 1
short 1 / 1
int 1 / 1
long 1 / 1
half 0 / 0 (cl_khr_fp16)
float 1 / 1
double 0 / 0 (n/a)
Half-precision Floating-point support (cl_khr_fp16)
Denormals No
Infinity and NANs Yes
Round to nearest Yes
Round to zero No
Round to infinity No
IEEE754-2008 fused multiply-add Yes
Support is emulated in software No
Single-precision Floating-point support (core)
Denormals No
Infinity and NANs Yes
Round to nearest Yes
Round to zero No
Round to infinity No
IEEE754-2008 fused multiply-add Yes
Support is emulated in software No
Correctly-rounded divide and sqrt operations No
Double-precision Floating-point support (n/a)
Address bits 64, Little-Endian
Global memory size 4019400704 (3.743GiB)
Error Correction support No
Max memory allocation 1004850176 (958.3MiB)
Unified memory for Host and Device Yes
Shared Virtual Memory (SVM) capabilities (core)
Coarse-grained buffer sharing Yes
Fine-grained buffer sharing No
Fine-grained system sharing No
Atomics No
Minimum alignment for any data type 128 bytes
Alignment of base address 1024 bits (128 bytes)
Preferred alignment for atomics
SVM 0 bytes
Global 0 bytes
Local 0 bytes
Atomic memory capabilities relaxed, work-group scope
Atomic fence capabilities relaxed, acquire/release, work-group scope
Max size for global variable 16384 (16KiB)
Preferred total size of global vars 0
Global Memory cache type Read/Write
Global Memory cache size 16384 (16KiB)
Global Memory cache line size 64 bytes
Image support Yes
Max number of samplers per kernel 8
Max size for 1D images from buffer 8192 pixels
Max 1D or 2D image array size 2048 images
Base address alignment for 2D image buffers 64 bytes
Pitch alignment for 2D image buffers 64 pixels
Max 2D image size 8192x8192 pixels
Max 3D image size 8192x8192x2048 pixels
Max number of read image args 8
Max number of write image args 1
Max number of read/write image args 64
Pipe support Yes
Max number of pipe args 16
Max active pipe reservations 1
Max pipe packet size 1024
Local memory type Local
Local memory size 4096 (4KiB)
Max number of constant args 4
Max constant buffer size 65536 (64KiB)
Generic address space support Yes
Max size of kernel argument 1024
Queue properties (on host)
Out-of-order execution Yes
Profiling Yes
Device enqueue capabilities (n/a)
Queue properties (on device)
Out-of-order execution No
Profiling No
Preferred size 0
Max size 0
Max queues on device 0
Max events on device 0
Prefer user sync for interop Yes
Profiling timer resolution 1000ns
Execution capabilities
Run OpenCL kernels Yes
Run native kernels Yes
Non-uniform work-groups Yes
Work-group collective functions Yes
Sub-group independent forward progress No
IL version (n/a)
ILs with version (n/a)
SPIR versions 1.2
printf() buffer size 65536 (64KiB)
Built-in kernels (n/a)
Built-in kernels with version (n/a)
Device Extensions cl_khr_icd cl_khr_fp16 cl_img_spirv cles_khr_int64 cl_img_yuv_image cl_khr_device_uuid cl_khr_depth_images cl_khr_priority_hints cl_img_generate_mipmap cl_khr_3d_image_writes cl_img_cached_allocations cl_khr_create_command_queue cl_img_mem_properties cl_img_mem_properties_relax_alloc_requirements cl_khr_extended_versioning cl_khr_image2d_from_buffer cl_khr_byte_addressable_store cl_khr_local_int32_base_atomics cl_khr_global_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_global_int32_extended_atomics cl_khr_spir cl_khr_egl_image cl_arm_import_memory cl_arm_import_memory_dma_buf cl_img_protected_content cl_img_semaphore cl_img_external_semaphore cl_img_external_semaphore_sync_fd
Device Extensions with Version cl_khr_icd 0x400000 (1.0.0)
cl_khr_fp16 0x400000 (1.0.0)
cl_img_spirv 0x400000 (1.0.0)
cles_khr_int64 0x400000 (1.0.0)
cl_img_yuv_image 0x400000 (1.0.0)
cl_khr_device_uuid 0x400000 (1.0.0)
cl_khr_depth_images 0x400000 (1.0.0)
cl_khr_priority_hints 0x400000 (1.0.0)
cl_img_generate_mipmap 0x400000 (1.0.0)
cl_khr_3d_image_writes 0x400000 (1.0.0)
cl_img_cached_allocations 0x400000 (1.0.0)
cl_khr_create_command_queue 0x400000 (1.0.0)
cl_img_mem_properties 0x400000 (1.0.0)
cl_img_mem_properties_relax_alloc_requirements 0x400000 (1.0.0)
cl_khr_extended_versioning 0x400000 (1.0.0)
cl_khr_image2d_from_buffer 0x400000 (1.0.0)
cl_khr_byte_addressable_store 0x400000 (1.0.0)
cl_khr_local_int32_base_atomics 0x400000 (1.0.0)
cl_khr_global_int32_base_atomics 0x400000 (1.0.0)
cl_khr_local_int32_extended_atomics 0x400000 (1.0.0)
cl_khr_global_int32_extended_atomics 0x400000 (1.0.0)
cl_khr_spir 0x400000 (1.0.0)
cl_khr_egl_image 0x400000 (1.0.0)
cl_arm_import_memory 0x400000 (1.0.0)
cl_arm_import_memory_dma_buf 0x400000 (1.0.0)
cl_img_protected_content 0x400000 (1.0.0)
cl_img_semaphore 0x400000 (1.0.0)
cl_img_external_semaphore 0x400000 (1.0.0)
cl_img_external_semaphore_sync_fd 0x400000 (1.0.0)
NULL platform behavior
clGetPlatformInfo(NULL, CL_PLATFORM_NAME, ...) PowerVR
clGetDeviceIDs(NULL, CL_DEVICE_TYPE_ALL, ...) Success [IMG]
clCreateContext(NULL, ...) [default] Success [IMG]
clCreateContextFromType(NULL, CL_DEVICE_TYPE_DEFAULT) Success (1)
Platform Name PowerVR
Device Name PowerVR B-Series BXM-4-64
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CPU) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_GPU) Success (1)
Platform Name PowerVR
Device Name PowerVR B-Series BXM-4-64
clCreateContextFromType(NULL, CL_DEVICE_TYPE_ACCELERATOR) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CUSTOM) Invalid device type for platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_ALL) Success (1)
Platform Name PowerVR
Device Name PowerVR B-Series BXM-4-64
ICD loader properties
ICD loader Name OpenCL ICD Loader
ICD loader Vendor OCL Icd free software
ICD loader Version 2.3.1
ICD loader Profile OpenCL 3.0
Hi bram,
The required information is mentioned in the log you shared. Thank you.
Best Regards,
Nagnath
Hello bram,
I’ve got some update from our OCL team to share with you on this issue of yours,
Unfortunately we were unable to reproduce this at our end. Our OCL team shared some of their thoughts on possible causes. For the driver to deliberately send out 26 fours is implausible - is there another thread or GPU task on user’s machine that we don’t know about? "Do you see this behaviour always, or only occasionally?
Could you try some other program, such as ocl_unit_test add, to see if any 4s appear?
Best Regards,
Nagnath
It is always.
I guess it is something that Sipeed put in the driver they shipped their riscv device with.
© Imagination Technologies Limited. All rights reserved.
Privacy Policy – Terms & Conditions – Trademarks – Cookies