Searched refs:calculations (Results 1 – 25 of 142) sorted by relevance
123456
36 the 'input' data. The 'mode' attribute controls exactly which calculations are127 so you should always use the output ports as the range for further calculations.132 operations that have to perform further calculations on them.
43 gains by merging the transformation calculations.
56 and we can get some latency gains by merging the transformation calculations.
32 magnitude. Using squared magnitude can avoid extra calculations.
18 the 'input' data. The 'mode' attribute controls exactly which calculations are
35 by comparing results to calculations done out of the shader code.
41 ; Shadow calculations must happen after the check.
5 ; Testing for failures in divergence calculations when divergent intrinsic is lowered during instru…
8 ; The required pointer calculations for the alloca'd actually requires
8 ; laid out because of the CFG dependency from top.fakephi. The calculations show
25 Skia Perf has the ability to perform calculations over the test data
43 uploads to allow for proper diff calculations.
103 When making exact calculations—such as for buffer lengths—it's often necessary104 to know when those calculations trigger an overflow, undefined behavior, or124 Certain classes of calculations—such as coordinate calculations—require133 // Use clamped arithmetic since inset calculations might overflow.
14 ; Boundaries calculations before this patch:
2 ; intermediate calculations for a comparison performed in a GPR spills the
74 ; Vectors are converted (if necessary) to int16_t for calculations.
8 ;; one set of indexing calculations and a load
39 // calculations of t and x_t below overflow and produce an incorrect interpolant (which then174 // Type-specific calculations: isRadiusIncreasing, focalParams, and the gradient matrix.
8 gemmlowp allows to perform calculations on matrices on uint8 values, but these12 quantization paradigm affects the calculations that gemmlowp itself needs to50 mean that the result of the quantized calculations do no longer readily provide
10 ; "optsize" attribute to disable all interleaving calculations. A cost of 4
2 ; makes our array dimension size calculations inaccurate. (PR32383)
27 results of their calculations. This is to avoid accidentally