math_brute_force / mad PASS

Test metadata

Name mad
Category math_brute_force
Suite Khronos OpenCL CTS
Suite version 80a4a833
Executable path /test_conformance/math_brute_force/bin/test_bruteforce
Arguments mad -m -s -v

Run metadata

Parent job Job 138
Revision f84bac329ba6c9f… / [NFC][Py Reformat] Reformat lit.local.cfg python files in llvm
Driver version NEO 22.34.24023; IGC 1.0.11702.1
Date run 05/19/2023
Duration 11 minutes
Status PASS

Runner output

Stdout from the runner of the corresponding OpenCL CTS test:

/home/bot/Workspace/cts/build/test_conformance/math_brute_force/bin/test_bruteforce mad -m -s -v
Test binary built Sep 8 2022 23:41:18
sizeof( void*) = 8
ARCH: x86_64
system name = Linux
node name = testrunner1
release = 5.19.0-38-generic
version = #39~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Fri Mar 17 21:16:15 UTC 2
machine = x86_64
*** PreventSleep() is not implemented on this platform.
Stopping at first error.
Initializing random seed to 0.
Requesting Default device based on command line for platform index 0 and device index 0
Compute Device Name = Intel(R) UHD Graphics 630 [0x3e98], Compute Device Vendor = Intel(R) Corporation, Compute Device Version = OpenCL 3.0 NEO , CL C Version = OpenCL C 1.2
Device latest conformance version passed: v2022-04-22-00
Supports single precision denormals: YES
sizeof( void*) = 8 (host)
sizeof( void*) = 8 (device)

Compute Device info:
Platform Version: OpenCL 3.0
Device Name: Intel(R) UHD Graphics 630 [0x3e98]
Vendor: Intel(R) Corporation
Device Version: OpenCL 3.0 NEO
CL C Version: OpenCL C 1.2
Driver Version: 22.34.0
Device Frequency: 1200 MHz
Subnormal values supported for floats? YES
Correctly rounded divide and sqrt supported for floats? YES
Testing with correctly rounded float divide and sqrt? YES
Testing with FTZ mode ON for floats? NO
Testing single precision? YES
Testing fast relaxed math? YES
Fast relaxed math has derived implementations? YES
Testing double precision? YES
Is Embedded? NO
Tininess is detected before rounding? NO
ERROR: Running single threaded because thread count < 2.
Worker threads: 1
Testing vector sizes: 1 2 3 4 8 16
Verbose? YES

1: mad fp32 rlx................passed 0.00 @ {0x0p+0, 0x0p+0, 0x0p+0}
2: mad fp32 ................passed 0.00 @ {0x0p+0, 0x0p+0, 0x0p+0}
3: mad fp64 ................passed 0.00 @ {0x0p+0, 0x0p+0, 0x0p+0}
mad passed
PASSED 3 of 3 sub-tests.
PASSED test.
*** ResumeSleep() is not implemented on this platform.

Stderr from the runner of the corresponding OpenCL CTS test:

Runner dumps

The files below were generated by the Intel Graphics Compiler after setting the IGC_ShaderDumpEnable=1 variable. OCL_asmX_before_spirv_backend.spv is the input binary SPIR-V generated by the IGC's OpenCL frontend, from which the SPIRV-LLVM-Translator generates OCL_asmX_before_spirv_backend.ll module which is passed to the SPIR-V backend. The backend outputs OCL_asmX_after_spirv_backend.spv SPIR-V binary which returns to IGC to be translated to LLVM module OCL_asmX_after_spirv_backend.ll for further compilation.