Q4.11: NEON Intrinsics
4 likes6,969 views
This document discusses NEON intrinsics and how to use them to optimize code for ARM processors that support SIMD instructions. It provides an overview of NEON, describes the data types and some common instructions, and gives examples of using intrinsics for tasks like color space conversion. Performance tests show intrinsics code can be 5-7 times faster than plain C and on par with hand-written assembly. Guidelines are provided for writing efficient NEON intrinsics code.
![Strided load
uint8x8x2_t vld2_u8
(const uint8_t *)
Form of expected instruction(s):
vld2.8 {d0, d1}, [r0]](https://image.slidesharecdn.com/intrinsics-demo-140320094326-phpapp01/85/Q4-11-NEON-Intrinsics-15-320.jpg)
Q4.11: NEON Intrinsics
- 1. Michael Hope, Toolchain bzr branch lp:~michaelh1/+junk/intrinsics-demo NEON Intrinsics
- 2. What's NEON? ●Ch 19 'Introducting NEON' http://infocenter.arm.com/help/topic/com.arm.doc.den0013a/
- 3. SIMD is... Same instruction, many values Anything involving signals is great for SIMD
- 5. ● Easier to read and write ● Easier (better?) register allocation ● Compiler knows how to schedule ● ABI neutral Advantages
- 6. Works across compilers > gcc-mcpu=cortex-a9 -mfpu=neon -O3 -c test.c > armcc --cpu Cortex-A9 --c99 -O3 -c test.c > clang -mcpu=cortex-a9 -mfpu=neon -O3 -c test.c
- 7. Tune for the architecture -mtune=cortex-a9 -mtune=cortex-a8 -mtune=cortex-a5
- 9. Writing
- 10. Environment #include <arm_neon.h> gcc -march=armv7-a -mfpu=neon
- 11. Data types <type>x<lanes>_t (uint8x4_t) <type>x<lanes>x<# registers>_t (int16x2x4_t)
- 13. Add uint16x4_t vadd_u16 ( uint16x4_t left, uint16x4_t right )
- 14. Multiply uint64x2_t vmlal_u32 (uint64x2_t, uint32x2_t, uint32x2_t) int32x4_t vqdmlal_s16 (int32x4_t, int16x4_t, int16x4_t)
- 15. Strided load uint8x8x2_t vld2_u8 (const uint8_t *) Form of expected instruction(s): vld2.8 {d0, d1}, [r0]
- 16. Documentation GCC http://gcc.gnu.org/onlinedocs/gcc/ARM-NEON-Intrinsics.html ARM http://infocenter.arm.com/help/topic/com.arm.doc.den0013a Blog posts Search for “Coding with NEON” on http://blogs.arm.com
- 17. Writing
- 18. Colour space conversion Y = 0.2126 R + 0.7152 G + 0.0722 B HD television (ITU BT.709)
- 19. Versions
- 24. Performance Plain C 48.481 s Assembly 8.727 s (5.55 x faster) Intrinsics 8.728 s (5.55 x faster)
- 25. Bigger Routines “libpixelflinger: Add ARM NEON optimized scanline_t32cb16” http://wiki.linaro.org/RichardSandiford/Sandbox/IntrinsicsPerformance Hand-written 2.831 s Intrinsics 2.637 s (7.4 % faster)