SSE _mm_movemask_epi8 equivalent method for ARM NEON

by

I know this post is quite outdated but I found it useful to give my (validated) solution. It assumes all ones/all zeroes in every lane of the Input argument.

const uint8_t __attribute__ ((aligned (16))) _Powers[16]= 
    { 1, 2, 4, 8, 16, 32, 64, 128, 1, 2, 4, 8, 16, 32, 64, 128 };

// Set the powers of 2 (do it once for all, if applicable)
uint8x16_t Powers= vld1q_u8(_Powers);

// Compute the mask from the input
uint64x2_t Mask= vpaddlq_u32(vpaddlq_u16(vpaddlq_u8(vandq_u8(Input, Powers))));

// Get the resulting bytes
uint16_t Output;
vst1q_lane_u8((uint8_t*)&Output + 0, (uint8x16_t)Mask, 0);
vst1q_lane_u8((uint8_t*)&Output + 1, (uint8x16_t)Mask, 8);

(Mind http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47553, anyway.)

Similarly to Michael, the trick is to form the powers of the indexes of the non-null entries, and to sum them pairwise three times. This must be done with increasing data size to double the stride on every addition. You reduce from 2 x 8 8-bit entries to 2 x 4 16-bit, then 2 x 2 32-bit and 2 x 1 64-bit. The low byte of these two numbers gives the solution. I don’t think there is an easy way to pack them together to form a single short value using NEON.

Takes 6 NEON instructions if the input is in the suitable form and the powers can be preloaded.

More Related Contents:

  1. Arm Neon Intrinsics vs hand assembly
  2. How is load->store reordering possible with in-order commit?
  3. Methods to vectorise histogram in SIMD?
  4. What is the difference between FIQ and IRQ interrupt system?
  5. TrustZone monitor mode and IFAR, IFSR, DFAR, DFSR
  6. Explaination of ARM (especifically mobile) Peripherals Addressing and Bus architecture?
  7. Explicitly accessing banked registers on ARM
  8. what is the right way to update MMU translation table
  9. qemu-system-aarch64 entering el1 when emulating a53 power up
  10. Why unsigned types are more efficient in arm cpu?
  11. How to run code from RAM on ARM architecture
  12. print a __m128i variable
  13. Is it possible to use SSE and SSE2 to make a 128-bit wide integer?
  14. How to efficiently convert an 8-bit bitmap to array of 0/1 integers with x86 SIMD [duplicate]
  15. Fast counting the number of set bits in __m128i register
  16. Linux kernel ARM exception stack init
  17. Is there a way to compile for ARM rather than Thumb in Xcode 4?
  18. In GNU C inline asm, what are the size-override modifiers for xmm/ymm/zmm for a single operand?
  19. How do I find ARM Linux entry point when it fails to uncompress?
  20. What is the interface for ARM system calls and where is it defined in the Linux kernel?
  21. Do I get a performance penalty when mixing SSE integer/float SIMD instructions
  22. cpu dispatcher for visual studio for AVX and SSE
  23. Find the physical address of exception vector table from kernel module
  24. What are the semantics of ADRP and ADRL instructions in ARM assembly?
  25. Convention for displaying vector registers
  26. Fastest way to do horizontal vector sum with AVX instructions [duplicate]
  27. Find the first instance of a character using simd
  28. How to specify an individual register as constraint in ARM GCC inline assembly?
  29. Cross Compile OpenSSH for ARM
  30. Handling ARM TrustZones

Leave a Comment