FastLED 3.9.15
Loading...
Searching...
No Matches

◆ forward() [2/2]

template<typename T>
T && fl::fl::forward ( typename remove_reference< T >::type & t)
constexpr

Definition at line 234 of file s16x16x4.h.

245 {
246
247// Forward declaration for cross-type operations
248struct s0x32x4;
249
252struct s16x16x4 {
253 simd::simd_u32x4 raw; // 4× i32 values in Q16 format
254
255 // ---- Construction ------------------------------------------------------
256
259 result.raw = r;
260 return result;
261 }
262
263 // Load 4 s16x16 values from memory (unaligned access supported)
264 static FASTLED_FORCE_INLINE s16x16x4 load(const s16x16* ptr) {
265 return from_raw(simd::platforms::load_u32_4(reinterpret_cast<const u32*>(ptr))); // ok reinterpret cast
266 }
267
268 // Store 4 s16x16 values to memory (unaligned access supported)
269 FASTLED_FORCE_INLINE void store(s16x16* ptr) const {
270 simd::platforms::store_u32_4(reinterpret_cast<u32*>(ptr), raw); // ok reinterpret cast
271 }
272
273 // Broadcast single s16x16 value to all 4 lanes
274 static FASTLED_FORCE_INLINE s16x16x4 set1(s16x16 value) {
275 return from_raw(simd::platforms::set1_u32_4(static_cast<u32>(value.raw())));
276 }
277
278 // ---- SIMD arithmetic (s16x16x4 OP s16x16x4 → s16x16x4) -----------------
279
280 FASTLED_FORCE_INLINE s16x16x4 operator+(s16x16x4 b) const {
281 return from_raw(simd::add_i32_4(raw, b.raw));
282 }
283
284 FASTLED_FORCE_INLINE s16x16x4 operator-(s16x16x4 b) const {
285 return from_raw(simd::sub_i32_4(raw, b.raw));
286 }
287
288 FASTLED_FORCE_INLINE s16x16x4 operator*(s16x16x4 b) const {
289 // Q16 × Q16 = Q32 → shift right 16 → Q16
290 return from_raw(simd::mulhi_i32_4(raw, b.raw));
291 }
292
293 FASTLED_FORCE_INLINE s16x16x4 operator-() const {
294 // Unary negation: -x = 0 - x
295 auto zero = simd::set1_u32_4(0);
296 return from_raw(simd::sub_i32_4(zero, raw));
297 }
298
299 FASTLED_FORCE_INLINE s16x16x4 operator>>(int shift) const {
300 return from_raw(simd::sra_i32_4(raw, shift));
301 }
302
303 FASTLED_FORCE_INLINE s16x16x4 operator<<(int shift) const {
304 return from_raw(simd::sll_u32_4(raw, shift));
305 }
306
307 // Cross-type multiply: s16x16x4 × s0x32x4 → s16x16x4 (commutative)
308 // Implemented after s0x32x4 is defined
309 FASTLED_FORCE_INLINE s16x16x4 operator*(s0x32x4 b) const;
310
311 // ---- Math functions -------------------------------------------------------
312
314 FASTLED_FORCE_INLINE s16x16x4 abs() const {
315 // mask = -1 if negative (sign extended), 0 if positive
316 auto mask = simd::sra_i32_4(raw, 31);
317 // flip bits if negative, then add 1 (two's complement)
318 auto flipped = simd::xor_u32_4(raw, mask);
319 return from_raw(simd::sub_i32_4(flipped, mask));
320 }
321
323 FASTLED_FORCE_INLINE s16x16x4 min(s16x16x4 b) const {
324 return from_raw(simd::min_i32_4(raw, b.raw));
325 }
326
328 FASTLED_FORCE_INLINE s16x16x4 max(s16x16x4 b) const {
329 return from_raw(simd::max_i32_4(raw, b.raw));
330 }
331
333 FASTLED_FORCE_INLINE s16x16x4 clamp(s16x16x4 lo, s16x16x4 hi) const {
334 return max(lo).min(hi);
335 }
336
339 FASTLED_FORCE_INLINE s16x16x4 lerp(s16x16x4 b, s16x16 t) const {
340 auto t_vec = s16x16x4::set1(t);
341 auto diff = b - (*this);
342 return (*this) + (diff * t_vec);
343 }
344
347 FASTLED_FORCE_INLINE void sincos(s16x16x4& out_sin, s16x16x4& out_cos) const {
348 // Convert radians to 24-bit angle units (same as scalar s16x16)
349 // RAD_TO_24 = 2^24 / (2π) in Q16
350 static constexpr i32 RAD_TO_24 = 2670177; // from s16x16.h
351
352 // Convert 4 angles: mulhi_i32_4 does (i64*i64) >> 16
353 auto angles_u32 = simd::mulhi_su32_4(raw, simd::set1_u32_4(static_cast<u32>(RAD_TO_24)));
354
355 // Call vectorized sincos
356 auto sc = sincos32_simd(angles_u32);
357
358 // Shift results right by 15 to convert from raw sin32 output to Q16.16
359 out_sin = from_raw(simd::sra_i32_4(sc.sin_vals, 15));
360 out_cos = from_raw(simd::sra_i32_4(sc.cos_vals, 15));
361 }
362
364 FASTLED_FORCE_INLINE s16x16x4 sin() const {
365 s16x16x4 sin_out, cos_out;
366 sincos(sin_out, cos_out);
367 return sin_out;
368 }
369
371 FASTLED_FORCE_INLINE s16x16x4 cos() const {
372 s16x16x4 sin_out, cos_out;
373 sincos(sin_out, cos_out);
374 return cos_out;
375 }
376};
377
378// Include simd_ops.h to implement cross-type operations
379// Must come after all types are defined
380#include "fl/math/fixed_point/simd_ops.h" // allow-include-after-namespace
381
382} // namespace fl
FL_DISABLE_WARNING_PUSH U constexpr common_type_t< T, U > min(T a, U b) FL_NOEXCEPT
Memory functions are available in fl:: namespace via fl/stl/cstring.h Using declarations cannot work ...
Definition math.h:71
constexpr common_type_t< T, U > max(T a, U b) FL_NOEXCEPT
Definition math.h:75
constexpr enable_if< is_fixed_point< T >::value, T >::type abs(T x) FL_NOEXCEPT
static uint32_t t
Definition Luminova.h:55
FASTLED_FORCE_INLINE CRGB * operator+(const CRGBSet &pixels, int offset)
Retrieve a pointer to a CRGB array, using a CRGBSet and an LED offset.
Definition pixelset.h:488
platforms::simd_u32x4 simd_u32x4
Definition s16x16x4.h:27
ostream & operator<<(ostream &os, const hex_t &) FL_NOEXCEPT
FASTLED_FORCE_INLINE SinCos32_simd sincos32_simd(simd::simd_u32x4 angles) FL_NOEXCEPT
Process 4 angles simultaneously, returning vectorized sin/cos values SIMD-optimized: vectorized angle...
Definition sin32.h:145
enable_if< is_fixed_point< T >::value, void >::type sincos(T angle, T &out_sin, T &out_cos) FL_NOEXCEPT
FASTLED_FORCE_INLINE CRGB operator*(const CRGB &p1, u8 d) FL_NOEXCEPT
Multiply each of the channels by a constant, saturating each channel at 0xFF.
Definition crgb.hpp:198
FASTLED_FORCE_INLINE CRGB operator-(const CRGB &p1, const CRGB &p2) FL_NOEXCEPT
Subtract one CRGB from another, saturating at 0x00 for each channel.
Definition crgb.hpp:190
expected< T, E > result
Alias for expected (Rust-style naming)
Definition result.h:31
enable_if< is_fixed_point< T >::value, T >::type cos(T angle) FL_NOEXCEPT
enable_if< is_fixed_point< T >::value, T >::type sin(T angle) FL_NOEXCEPT
constexpr enable_if< is_fixed_point< T >::value, T >::type clamp(T x, T lo, T hi) FL_NOEXCEPT
FASTLED_FORCE_INLINE float lerp(float t, float a, float b)
#define FASTLED_FORCE_INLINE
Cross-type SIMD fixed-point operations (implemented after all types are defined)
4-wide s0x32 vector (normalized values [-1, 1]) Backed by 128-bit SIMD register (4× i32 in Q31 format...
Definition s16x16x4.h:17
4-wide s16x16 vector (general fixed-point) Backed by 128-bit SIMD register (4× i32 in Q16 format)
Definition s16x16x4.h:19

Referenced by fl::pair< vec2< u16 >, u8 >::pair(), fl::tuple< Head, Tail... >::tuple(), fl::shared_ptr< filebuf >::allocate_shared, fl::allocator< U >::construct(), fl::allocator_inlined< U, N, typename BaseAllocator::template rebind< U >::other >::construct(), fl::allocator_psram< U >::construct(), fl::allocator_realloc< U >::construct(), fl::allocator_slab< T, SLAB_SIZE >::construct(), fl::deque< int >::emplace(), fl::flat_map< int, FxPtr >::emplace(), fl::flat_set< Key, Less >::emplace(), fl::iterator<, u8 >::emplace(), fl::MapRedBlackTree< Key, T, Compare, fl::allocator_slab< char > >< Key, T, Compare >::emplace(), fl::MultiMapTree< Key, T, Compare, fl::allocator_slab< char > >::emplace(), fl::MultiSetTree< Key, Compare, fl::allocator_slab< char > >::emplace(), fl::Optional< T >::emplace(), fl::priority_queue_stable< T, Compare >::emplace(), fl::PriorityQueue< T, Compare, VectorT >::emplace(), fl::queue< T, Container >::emplace(), fl::RedBlackTree< value_type, PairCompare, Allocator >::emplace(), fl::set< T, fl::allocator_inlined_slab< T, N > >::emplace(), fl::SetRedBlackTree< Key, fl::less< Key >, Allocator >::emplace(), fl::unordered_map_small< Key, Value, Equal >::emplace(), fl::unordered_set< Key, Hash, KeyEqual >::emplace(), fl::VectorSet< Key >::emplace(), fl::VectorSetFixed< Key, N >::emplace(), fl::circular_buffer< T, N >::emplace_back(), fl::circular_buffer_core< T >::emplace_back(), fl::deque< int >::emplace_back(), fl::FixedVector< fl::u32, kMaxBatchSize >::emplace_back(), fl::list< T >::emplace_back(), fl::vector< fl::i16 >::emplace_back(), fl::circular_buffer< T, N >::emplace_front(), fl::circular_buffer_core< T >::emplace_front(), fl::deque< int >::emplace_front(), fl::list< T >::emplace_front(), fl::flat_map< int, FxPtr >::emplace_hint(), fl::flat_set< Key, Less >::emplace_hint(), fl::iterator<, u8 >::emplace_hint(), fl::MapRedBlackTree< Key, T, Compare, fl::allocator_slab< char > >< Key, T, Compare >::emplace_hint(), fl::FL_ALIGN_AS_T(), fl::Optional< T && >::get(), fl::MapRedBlackTree< Key, T, Compare, fl::allocator_slab< char > >< Key, T, Compare >::insert_or_assign(), fl::MapRedBlackTree< Key, T, Compare, fl::allocator_slab< char > >< Key, T, Compare >::insert_or_assign(), fl::MapRedBlackTree< Key, T, Compare, fl::allocator_slab< char > >< Key, T, Compare >::insert_or_assign(), fl::MapRedBlackTree< Key, T, Compare, fl::allocator_slab< char > >< Key, T, Compare >::insert_or_assign(), fl::RedBlackTree< value_type, PairCompare, Allocator >::insertImpl(), fl::invoke(), fl::invoke(), fl::invoke(), fl::make_pair(), fl::make_scope_exit(), fl::shared_ptr< filebuf >::make_shared, fl::make_shared_ptr(), fl::shared_ptr< filebuf >::make_shared_with_deleter, fl::make_tuple(), fl::make_unique(), fl::BindResult< R(Args...)>::operator()(), fl::greater< void >::operator()(), fl::less< void >::operator()(), fl::RpcHandle< R(Args...)>::operator()(), fl::Optional< T && >::operator*(), fl::json_value::operator=(), fl::iterator<, u8 >::try_emplace(), fl::iterator<, u8 >::try_emplace(), fl::MapRedBlackTree< Key, T, Compare, fl::allocator_slab< char > >< Key, T, Compare >::try_emplace(), fl::MapRedBlackTree< Key, T, Compare, fl::allocator_slab< char > >< Key, T, Compare >::try_emplace(), fl::MapRedBlackTree< Key, T, Compare, fl::allocator_slab< char > >< Key, T, Compare >::try_emplace(), fl::MapRedBlackTree< Key, T, Compare, fl::allocator_slab< char > >< Key, T, Compare >::try_emplace(), fl::Optional< T && >::value(), fl::spi::MultiLaneDevice::write(), and fl::Spi::write().