Improve cmake build behaviour with non-host cpu targets (#2246)

1. Supply appropriate values for C/Z GEMM unroll when cross-compiling for CORE2 or ARMV7
2. Add the required xLOCAL_BUFFER_SIZE parameters for cross-compiling CORE2
3. Add -DFORCE_<target> option to getarch when building with -DTARGET=target
for #2245
This commit is contained in:
Martin Kroeker 2019-09-03 22:41:17 +02:00 committed by GitHub
parent ee498525e0
commit fde8a8e6a0
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 17 additions and 5 deletions

View File

@ -105,6 +105,7 @@ if (DEFINED CORE AND CMAKE_CROSSCOMPILING AND NOT (${HOST_OS} STREQUAL "WINDOWSS
# Perhaps this should be inside a different file as it grows larger
file(APPEND ${TARGET_CONF_TEMP}
"#define ${TCORE}\n"
"#define CORE_${TCORE}\n"
"#define CHAR_CORENAME \"${TCORE}\"\n")
if ("${TCORE}" STREQUAL "CORE2")
file(APPEND ${TARGET_CONF_TEMP}
@ -119,15 +120,19 @@ if (DEFINED CORE AND CMAKE_CROSSCOMPILING AND NOT (${HOST_OS} STREQUAL "WINDOWSS
"#define HAVE_SSE\n"
"#define HAVE_SSE2\n"
"#define HAVE_SSE3\n"
"#define HAVE_SSSE3\n")
"#define HAVE_SSSE3\n"
"#define SLOCAL_BUFFER_SIZE\t16384\n"
"#define DLOCAL_BUFFER_SIZE\t16384\n"
"#define CLOCAL_BUFFER_SIZE\t16384\n"
"#define ZLOCAL_BUFFER_SIZE\t16384\n")
set(SGEMM_UNROLL_M 8)
set(SGEMM_UNROLL_N 4)
set(DGEMM_UNROLL_M 4)
set(DGEMM_UNROLL_N 4)
set(CGEMM_DEFAULT_UNROLL_M 4)
set(CGEMM_DEFAULT_UNROLL_N 2)
set(ZGEMM_DEFAULT_UNROLL_M 2)
set(ZGEMM_DEFAULT_UNROLL_N 2)
set(CGEMM_UNROLL_M 4)
set(CGEMM_UNROLL_N 2)
set(ZGEMM_UNROLL_M 2)
set(ZGEMM_UNROLL_N 2)
elseif ("${TCORE}" STREQUAL "ARMV7")
file(APPEND ${TARGET_CONF_TEMP}
"#define L1_DATA_SIZE\t65536\n"
@ -143,6 +148,10 @@ if (DEFINED CORE AND CMAKE_CROSSCOMPILING AND NOT (${HOST_OS} STREQUAL "WINDOWSS
set(SGEMM_UNROLL_N 4)
set(DGEMM_UNROLL_M 4)
set(DGEMM_UNROLL_N 4)
set(CGEMM_UNROLL_M 2)
set(CGEMM_UNROLL_N 2)
set(ZGEMM_UNROLL_M 2)
set(ZGEMM_UNROLL_N 2)
elseif ("${TCORE}" STREQUAL "ARMV8")
file(APPEND ${TARGET_CONF_TEMP}
"#define L1_DATA_SIZE\t32768\n"
@ -331,6 +340,9 @@ else(NOT CMAKE_CROSSCOMPILING)
set(GETARCH_FLAGS ${GETARCH_FLAGS} -DFORCE_GENERIC)
else()
list(APPEND GETARCH_SRC ${PROJECT_SOURCE_DIR}/cpuid.S)
if (DEFINED TARGET_CORE)
set(GETARCH_FLAGS ${GETARCH_FLAGS} -DFORCE_${TARGET_CORE})
endif ()
endif ()
if ("${CMAKE_SYSTEM_NAME}" STREQUAL "WindowsStore")