exp16 (reference) in projects: third_party - OpenGrok search results

Repository(s)

Full Search		Search through all text tokens(words,strings,identifiers,numbers) in index.
Definition		Only finds symbol definitions(where e.g a variable(function,...) is defined).
Symbol		Only finds symbol(e.g. methods classes,function,variables).
File Path		Path of the source file(use "/").If you want just exact path,enclose it in "".Source files end with: .jar/.bz2/.a/.h/.java...
History		History log comments.
Type		Type of analyzer used to filter file types include with selected(e.g. just C sources).

Help

Searched refs:exp16 (Results 1 - 4 of 4) sorted by relevance

/third_party/vk-gl-cts/framework/delibs/debase/
H A D	deFloat16.c	220 int exp16; /* exp16: biased exponent for 16-bit floats / in deFloat32To16Round() local 229 exp16 = (int) (exp32) - 127 + 15; / 15/127: exponent bias for 16-bit/32-bit floats / in deFloat32To16Round() 256 * If exp16 is less than 0, we are experiencing underflow for the exponent. To encode this underflowed exponent, in deFloat32To16Round() 258 * The real exponent is exp16 - 15. A denormalized 16-bit float can represent -14 via its exponent. in deFloat32To16Round() 260 * So, we just need to right shift the mantissa -exp16 bits. in deFloat32To16Round() 261 * * If exp16 is 0, mantissa shifting requirement is similar to the above. in deFloat32To16Round() 262 * * If exp16 is greater than 30 (0b11110), we are experiencing overflow for the exponent of 16-bit normalized floats. in deFloat32To16Round() 265 else if (exp16 < -10) in deFloat32To16Round() 272 else if (exp16 < in deFloat32To16Round() 377 int exp16; /* exp16: biased exponent for 16-bit floats */ deFloat64To16Round() local [all...]
/third_party/ffmpeg/tests/
H A D	tiny_psnr.c	`62 static unsigned int exp16(unsigned int a){`
/third_party/skia/third_party/externals/swiftshader/src/Pipeline/
H A D	ShaderCore.cpp	`567 auto exp16 = halfBits & SIMD::UInt(0x7C00); in halfToFloatBits() local 569 auto isDnormOrZero = CmpEQ(exp16, SIMD::UInt(0)); in halfToFloatBits() 570 auto isInfOrNaN = CmpEQ(exp16, SIMD::UInt(0x7C00)); in halfToFloatBits() 574 auto exp32 = (exp16 + SIMD::UInt(0x1C000)) << 13; in halfToFloatBits()`
/third_party/mesa3d/src/broadcom/compiler/
H A D	nir_to_vir.c	1335 struct qreg exp16 = vir_ADD(c, exp32, vir_uniform_ui(c, -127 + 15)); in f2f16_rtz() local 1337 /* if (exp16 > 30) / in f2f16_rtz() 1338 inst = vir_MIN_dest(c, vir_nop_reg(), exp16, vir_uniform_ui(c, 30)); in f2f16_rtz() 1343 / if (exp16 <= 30) / in f2f16_rtz() 1346 vir_SHL(c, exp16, vir_uniform_ui(c, 10))), in f2f16_rtz() 1350 / if (exp16 <= 0) / in f2f16_rtz() 1351 inst = vir_MIN_dest(c, vir_nop_reg(), exp16, vir_uniform_ui(c, 0)); in f2f16_rtz() 1355 tmp = vir_SHR(c, tmp, vir_SUB(c, vir_uniform_ui(c, 14), exp16)); in f2f16_rtz() 1363 exp16 < - 10 \|\| exp32 == 0 \|\| exp32 == 0xff in f2f16_rtz() 1367 * !((exp16 > in f2f16_rtz() [all...]

Completed in 8 milliseconds