COMPMID-3152: Initial Bfloat16 support
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Change-Id: Ie6959e37e13731c86b2ee29392a99a293450a1b4
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/2824
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
diff --git a/support/Bfloat16.h b/support/Bfloat16.h
new file mode 100644
index 0000000..d897e42
--- /dev/null
+++ b/support/Bfloat16.h
@@ -0,0 +1,140 @@
+/*
+ * Copyright (c) 2020 ARM Limited.
+ *
+ * SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+ * deal in the Software without restriction, including without limitation the
+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+ * sell copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#ifndef ARM_COMPUTE_BFLOAT16_H
+#define ARM_COMPUTE_BFLOAT16_H
+
+#include <cstdint>
+
+namespace arm_compute
+{
+namespace
+{
+/** Convert float to bfloat16
+ *
+ * @param[in] v Floating-point value to convert to bfloat
+ *
+ * @return Converted value
+ */
+inline uint16_t float_to_bf16(const float v)
+{
+ const uint32_t *fromptr = reinterpret_cast<const uint32_t *>(&v);
+#if defined(__ARM_FEATURE_BF16_VECTOR_ARITHMETIC) || defined(ARM_COMPUTE_FORCE_BF16)
+ uint16_t res;
+
+ __asm __volatile(
+ "ldr s0, [%[fromptr]]\n"
+ ".inst 0x1e634000\n" // BFCVT h0, s0
+ "str h0, [%[toptr]]\n"
+ :
+ : [fromptr] "r"(fromptr), [toptr] "r"(&res)
+ : "v0", "memory");
+#else /* defined(__ARM_FEATURE_BF16_VECTOR_ARITHMETIC) || defined(ARM_COMPUTE_FORCE_BF16) */
+ uint16_t res = (*fromptr >> 16);
+ const uint16_t error = (*fromptr & 0x0000ffff);
+ uint16_t bf_l = res & 0x0001;
+ if((error > 0x8000) || ((error == 0x8000) && (bf_l != 0)))
+ {
+ res += 1;
+ }
+#endif /* defined(__ARM_FEATURE_BF16_VECTOR_ARITHMETIC) || defined(ARM_COMPUTE_FORCE_BF16) */
+ return res;
+}
+
+/** Convert bfloat16 to float
+ *
+ * @param[in] v Bfloat16 value to convert to float
+ *
+ * @return Converted value
+ */
+inline float bf16_to_float(const uint16_t &v)
+{
+ const uint32_t lv = (v << 16);
+ const float *fp = reinterpret_cast<const float *>(&lv);
+
+ return *fp;
+}
+}
+
+/** Brain floating point representation class */
+class bfloat16
+{
+public:
+ /** Default Constructor */
+ bfloat16()
+ : value(0)
+ {
+ }
+ /** Constructor
+ *
+ * @param[in] v Floating-point value
+ */
+ explicit bfloat16(float v)
+ : value(float_to_bf16(v))
+ {
+ }
+ /** Assignment operator
+ *
+ * @param[in] v Floating point value to assign
+ *
+ * @return The updated object
+ */
+ bfloat16 &operator=(float v)
+ {
+ value = float_to_bf16(v);
+ return *this;
+ }
+ /** Floating point conversion operator
+ *
+ * @return Floating point representation of the value
+ */
+ operator float() const
+ {
+ return bf16_to_float(value);
+ }
+ /** Lowest representative value
+ *
+ * @return Returns the lowest finite value representable by bfloat16
+ */
+ static bfloat16 lowest()
+ {
+ bfloat16 val;
+ val.value = 0xFF7F;
+ return val;
+ }
+ /** Largest representative value
+ *
+ * @return Returns the largest finite value representable by bfloat16
+ */
+ static bfloat16 max()
+ {
+ bfloat16 val;
+ val.value = 0x7F7F;
+ return val;
+ }
+
+private:
+ uint16_t value;
+};
+} // namespace arm_compute
+#endif /* ARM_COMPUTE_BFLOAT16_H */
\ No newline at end of file