code wiki / simd

topic: simd

24 modules sharing the simd name family (derived from the tree's prefix discipline).

The 'simd' topic family in the Nishi sovereign ecosystem focuses on vectorized operations and performance benchmarking for SIMD (Single Instruction, Multiple Data) processing. The nx_simd module provides lane-wise vector operations and a vectorizing transform, enabling efficient data parallelism. The nx_simd_dot_bench module compares x86 integer-SIMD dot products with scalar implementations, highlighting performance gains. Meanwhile, nx_simd_census evaluates the vector backend's competitiveness through internal benchmarks, ensuring the system's reliability and efficiency.

auto-narrated by the local model from this topic's module headers; links verified against the wiki index.

narrated overview -- maintained by the narration lane, module links verified against this wiki.

moduledescriptionlinesfuncs
nx_simd.nxsovereign SIMD: lane-wise vector ops + a VECTORIZING TRANSFORM, the547
nx_simd_avx2_kat_gate.nxSOVEREIGN 256-bit AVX2 vector run-KAT (the VEX rung): proves the1137
nx_simd_bench.nxhonest perf comparison: scalar i64 add vs1163
nx_simd_census.nxSOVEREIGN vector-backend competitive census. The TEAM measures its OWN15610
nx_simd_dot_bench.nxx86 integer-SIMD dot (vpmaddwd via __i16x16_madd) vs scalar: verify + speedup.1254
nx_simd_i16x16_test.nxexercise the i16x16 SIMD intrinsics.1187
nx_simd_i32x8_test.nxexercise the i32x8 SIMD intrinsics.1286
nx_simd_i8x32_test.nxexercise the i8x32 SIMD intrinsics.1066
nx_simd_isa.nxa sovereign VECTOR ISA + SIMD emulator, so vectorized kernels502
nx_simd_isa_test.nxthe vectorized FMA kernel as real vector MACHINE CODE,733
nx_simd_lever_gate.nxPROVE the SIMD lever that WINS the PyTorch h2h, using the REAL supported intrinsic496
nx_simd_madd_probe.nxtest the x86 integer-SIMD primitive __i16x16_madd (vpmaddwd) on nx_cc_known_good.652
nx_simd_minmax_test.nxi16x16 per-lane and horizontal min/max.1468
nx_simd_probe.nxconfirm the __simd i16x16 vdot intrinsics compile + work in a nx_syscalls organ.302
nx_simd_probe2.nxverify the POINTER-FORM __simd_vdot_i16_x16 compiles AND computes on x86.362
nx_simd_sat_test.nxi16x16 signed saturating add/sub.1486
nx_simd_sse_kat_gate.nxSOVEREIGN 128-bit SSE2 vector run-KAT: the1277
nx_simd_sse_rot_gate.nxSOVEREIGN 128-bit SSE2 packed-shift/bitwise run-KAT1167
nx_simd_sse_shuf_gate.nxSOVEREIGN 128-bit pshufd run-KAT (rung R2+R3:1097
nx_simd_test.nxprove the vectorizing transform is 1:1 with the scalar kernel603
nx_simd_usat_shift_test.nxunsigned saturating arith + per-lane shifts (i16x16).1597
nx_simd_vdot_test.nxwidening dot product i16x16 -> i64.1498
nx_simd_x86_test.nxAVX2-subset SIMD smoke (excludes VMUL which491
simd.nxSIMD vector types + intrinsics (spec + soft-fallback). 20313