Merge pull request #752 from pq-code-package/namespace_work

Various namespacing changes
pq-code-package · Feb 6, 2025 · e16cf03 · e16cf03 · oqs-bot · Feb 6, 2025
2 parents b5ef959 + 8f9a74e
commit e16cf03
Show file tree

Hide file tree

Showing 32 changed files with 1,132 additions and 312 deletions.
diff --git a/dev/x86_64/src/align.h b/dev/x86_64/src/align.h
@@ -13,7 +13,7 @@
 #include <immintrin.h>
 #include <stdint.h>
 
-#define ALIGNED_INT16(N)        \
+#define MLK_ALIGNED_INT16(N)    \
   union                         \
   {                             \
     int16_t coeffs[N];          \

diff --git a/dev/x86_64/src/arith_native_x86_64.h b/dev/x86_64/src/arith_native_x86_64.h
@@ -11,9 +11,8 @@
 #include <stdint.h>
 #include "consts.h"
 
-#define REJ_UNIFORM_AVX_NBLOCKS 3 /* See MLKEM_GEN_MATRIX_NBLOCKS */
-#define REJ_UNIFORM_AVX_BUFLEN \
-  (3 * 168) /* REJ_UNIFORM_AVX_BUFLEN * SHAKE128_RATE */
+#define MLK_AVX2_REJ_UNIFORM_BUFLEN \
+  (3 * 168) /* REJ_UNIFORM_NBLOCKS * SHAKE128_RATE */
 
 #define rej_uniform_avx2 MLK_NAMESPACE(rej_uniform_avx2)
 unsigned rej_uniform_avx2(int16_t *r, const uint8_t *buf);

diff --git a/dev/x86_64/src/basemul.S b/dev/x86_64/src/basemul.S
@@ -15,7 +15,7 @@
 
 /* Polynomials to be multiplied are denoted a+bX (rsi arg) and c+dX (rdx arg) */
 .macro schoolbook off
-vmovdqa		AVX2_BACKEND_DATA_OFFSET_16XQINV*2(%rcx),%ymm0
+vmovdqa		MLK_AVX2_BACKEND_DATA_OFFSET_16XQINV*2(%rcx),%ymm0
 vmovdqa		(64*\off+ 0)*2(%rsi),%ymm1		# a0
 vmovdqa		(64*\off+16)*2(%rsi),%ymm2		# b0
 vmovdqa		(64*\off+32)*2(%rsi),%ymm3		# a1
@@ -63,7 +63,7 @@ vpmullw		%ymm7,%ymm12,%ymm7			# b1c1.lo
 vpmullw		%ymm8,%ymm12,%ymm12			# b1d1.lo
 
 /* Compute 2nd high multiplication in Montgomery multiplication */
-vmovdqa		AVX2_BACKEND_DATA_OFFSET_16XQ*2(%rcx),%ymm8
+vmovdqa		MLK_AVX2_BACKEND_DATA_OFFSET_16XQ*2(%rcx),%ymm8
 vpmulhw		%ymm8,%ymm13,%ymm13
 vpmulhw		%ymm8,%ymm9,%ymm9
 vpmulhw		%ymm8,%ymm5,%ymm5
@@ -121,7 +121,7 @@ mov		%rsp,%r8
 and		$-32,%rsp
 sub		$32,%rsp
 
-lea		(AVX2_BACKEND_DATA_OFFSET_ZETAS_EXP+176)*2(%rcx),%r9
+lea		(MLK_AVX2_BACKEND_DATA_OFFSET_ZETAS_EXP+176)*2(%rcx),%r9
 schoolbook	0
 
 add		$32*2,%r9

diff --git a/dev/x86_64/src/compress_avx2.c b/dev/x86_64/src/compress_avx2.c
@@ -24,7 +24,7 @@ void poly_compress_d4_avx2(uint8_t r[MLKEM_POLYCOMPRESSEDBYTES_D4],
   unsigned int i;
   __m256i f0, f1, f2, f3;
   const __m256i v =
-      _mm256_load_si256(&qdata.vec[AVX2_BACKEND_DATA_OFFSET_16XV / 16]);
+      _mm256_load_si256(&qdata.vec[MLK_AVX2_BACKEND_DATA_OFFSET_16XV / 16]);
   const __m256i shift1 = _mm256_set1_epi16(1 << 9);
   const __m256i mask = _mm256_set1_epi16(15);
   const __m256i shift2 = _mm256_set1_epi16((16 << 8) + 1);
@@ -65,7 +65,7 @@ void poly_decompress_d4_avx2(__m256i *MLK_RESTRICT r,
   __m128i t;
   __m256i f;
   const __m256i q =
-      _mm256_load_si256(&qdata.vec[AVX2_BACKEND_DATA_OFFSET_16XQ / 16]);
+      _mm256_load_si256(&qdata.vec[MLK_AVX2_BACKEND_DATA_OFFSET_16XQ / 16]);
   const __m256i shufbidx =
       _mm256_set_epi8(7, 7, 7, 7, 6, 6, 6, 6, 5, 5, 5, 5, 4, 4, 4, 4, 3, 3, 3,
                       3, 2, 2, 2, 2, 1, 1, 1, 1, 0, 0, 0, 0);
@@ -91,7 +91,7 @@ void poly_compress_d10_avx2(uint8_t r[MLKEM_POLYCOMPRESSEDBYTES_D10],
   __m256i f0, f1, f2;
   __m128i t0, t1;
   const __m256i v =
-      _mm256_load_si256(&qdata.vec[AVX2_BACKEND_DATA_OFFSET_16XV / 16]);
+      _mm256_load_si256(&qdata.vec[MLK_AVX2_BACKEND_DATA_OFFSET_16XV / 16]);
   const __m256i v8 = _mm256_slli_epi16(v, 3);
   const __m256i off = _mm256_set1_epi16(15);
   const __m256i shift1 = _mm256_set1_epi16(1 << 12);
@@ -175,7 +175,7 @@ void poly_compress_d5_avx2(uint8_t r[MLKEM_POLYCOMPRESSEDBYTES_D5],
   __m256i f0, f1;
   __m128i t0, t1;
   const __m256i v =
-      _mm256_load_si256(&qdata.vec[AVX2_BACKEND_DATA_OFFSET_16XV / 16]);
+      _mm256_load_si256(&qdata.vec[MLK_AVX2_BACKEND_DATA_OFFSET_16XV / 16]);
   const __m256i shift1 = _mm256_set1_epi16(1 << 10);
   const __m256i mask = _mm256_set1_epi16(31);
   const __m256i shift2 = _mm256_set1_epi16((32 << 8) + 1);
@@ -218,7 +218,7 @@ void poly_decompress_d5_avx2(__m256i *MLK_RESTRICT r,
   __m256i f;
   int16_t ti;
   const __m256i q =
-      _mm256_load_si256(&qdata.vec[AVX2_BACKEND_DATA_OFFSET_16XQ / 16]);
+      _mm256_load_si256(&qdata.vec[MLK_AVX2_BACKEND_DATA_OFFSET_16XQ / 16]);
   const __m256i shufbidx =
       _mm256_set_epi8(9, 9, 9, 8, 8, 8, 8, 7, 7, 6, 6, 6, 6, 5, 5, 5, 4, 4, 4,
                       3, 3, 3, 3, 2, 2, 1, 1, 1, 1, 0, 0, 0);
@@ -248,7 +248,7 @@ void poly_compress_d11_avx2(uint8_t r[MLKEM_POLYCOMPRESSEDBYTES_D11],
   __m256i f0, f1, f2;
   __m128i t0, t1;
   const __m256i v =
-      _mm256_load_si256(&qdata.vec[AVX2_BACKEND_DATA_OFFSET_16XV / 16]);
+      _mm256_load_si256(&qdata.vec[MLK_AVX2_BACKEND_DATA_OFFSET_16XV / 16]);
   const __m256i v8 = _mm256_slli_epi16(v, 3);
   const __m256i off = _mm256_set1_epi16(36);
   const __m256i shift1 = _mm256_set1_epi16(1 << 13);
@@ -320,7 +320,7 @@ void poly_decompress_d11_avx2(__m256i *MLK_RESTRICT r,
   unsigned int i;
   __m256i f;
   const __m256i q =
-      _mm256_load_si256(&qdata.vec[AVX2_BACKEND_DATA_OFFSET_16XQ / 16]);
+      _mm256_load_si256(&qdata.vec[MLK_AVX2_BACKEND_DATA_OFFSET_16XQ / 16]);
   const __m256i shufbidx =
       _mm256_set_epi8(13, 12, 12, 11, 10, 9, 9, 8, 8, 7, 6, 5, 5, 4, 4, 3, 10,
                       9, 9, 8, 7, 6, 6, 5, 5, 4, 3, 2, 2, 1, 1, 0);
Benchmark suite	Current: `e16cf03`	Previous: `b5ef959`	Ratio
`ML-KEM-512 keypair`	`28966` cycles	`28965` cycles	`1.00`
`ML-KEM-512 encaps`	`34252` cycles	`34251` cycles	`1.00`
`ML-KEM-512 decaps`	`44728` cycles	`44731` cycles	`1.00`
`ML-KEM-768 keypair`	`49307` cycles	`49308` cycles	`1.00`
`ML-KEM-768 encaps`	`54569` cycles	`54570` cycles	`1.00`
`ML-KEM-768 decaps`	`69425` cycles	`69426` cycles	`1.00`
`ML-KEM-1024 keypair`	`71914` cycles	`71916` cycles	`1.00`
`ML-KEM-1024 encaps`	`80609` cycles	`80615` cycles	`1.00`
`ML-KEM-1024 decaps`	`100360` cycles	`100364` cycles	`1.00`
Benchmark suite	Current: `e16cf03`	Previous: `b5ef959`	Ratio
`ML-KEM-512 keypair`	`9292` cycles	`9301` cycles	`1.00`
`ML-KEM-512 encaps`	`10772` cycles	`10817` cycles	`1.00`
`ML-KEM-512 decaps`	`14728` cycles	`14751` cycles	`1.00`
`ML-KEM-768 keypair`	`15936` cycles	`15944` cycles	`1.00`
`ML-KEM-768 encaps`	`17258` cycles	`17284` cycles	`1.00`
`ML-KEM-768 decaps`	`22956` cycles	`22973` cycles	`1.00`
`ML-KEM-1024 keypair`	`21323` cycles	`21401` cycles	`1.00`
`ML-KEM-1024 encaps`	`23257` cycles	`23329` cycles	`1.00`
`ML-KEM-1024 decaps`	`30776` cycles	`30797` cycles	`1.00`
Benchmark suite	Current: `e16cf03`	Previous: `b5ef959`	Ratio
`ML-KEM-512 keypair`	`29630` cycles	`29666` cycles	`1.00`
`ML-KEM-512 encaps`	`35307` cycles	`35430` cycles	`1.00`
`ML-KEM-512 decaps`	`45941` cycles	`46054` cycles	`1.00`
`ML-KEM-768 keypair`	`47097` cycles	`47168` cycles	`1.00`
`ML-KEM-768 encaps`	`55511` cycles	`55575` cycles	`1.00`
`ML-KEM-768 decaps`	`67551` cycles	`67579` cycles	`1.00`
`ML-KEM-1024 keypair`	`71741` cycles	`71837` cycles	`1.00`
`ML-KEM-1024 encaps`	`82085` cycles	`82156` cycles	`1.00`
`ML-KEM-1024 decaps`	`99388` cycles	`99400` cycles	`1.00`
Benchmark suite	Current: `e16cf03`	Previous: `b5ef959`	Ratio
`ML-KEM-512 keypair`	`15923` cycles	`15928` cycles	`1.00`
`ML-KEM-512 encaps`	`18002` cycles	`18003` cycles	`1.00`
`ML-KEM-512 decaps`	`24513` cycles	`24545` cycles	`1.00`
`ML-KEM-768 keypair`	`27356` cycles	`27349` cycles	`1.00`
`ML-KEM-768 encaps`	`28916` cycles	`28916` cycles	`1`
`ML-KEM-768 decaps`	`38328` cycles	`38330` cycles	`1.00`
`ML-KEM-1024 keypair`	`36971` cycles	`36988` cycles	`1.00`
`ML-KEM-1024 encaps`	`39900` cycles	`39889` cycles	`1.00`
`ML-KEM-1024 decaps`	`52416` cycles	`52383` cycles	`1.00`
Benchmark suite	Current: `e16cf03`	Previous: `b5ef959`	Ratio
`ML-KEM-512 keypair`	`16965` cycles	`16977` cycles	`1.00`
`ML-KEM-512 encaps`	`18662` cycles	`18681` cycles	`1.00`
`ML-KEM-512 decaps`	`24081` cycles	`24062` cycles	`1.00`
`ML-KEM-768 keypair`	`28717` cycles	`28713` cycles	`1.00`
`ML-KEM-768 encaps`	`29802` cycles	`29802` cycles	`1`
`ML-KEM-768 decaps`	`37586` cycles	`37589` cycles	`1.00`
`ML-KEM-1024 keypair`	`41789` cycles	`41719` cycles	`1.00`
`ML-KEM-1024 encaps`	`44072` cycles	`44091` cycles	`1.00`
`ML-KEM-1024 decaps`	`54297` cycles	`54312` cycles	`1.00`
Benchmark suite	Current: `e16cf03`	Previous: `b5ef959`	Ratio
`ML-KEM-512 keypair`	`17729` cycles	`17721` cycles	`1.00`
`ML-KEM-512 encaps`	`20981` cycles	`20976` cycles	`1.00`
`ML-KEM-512 decaps`	`27658` cycles	`27657` cycles	`1.00`
`ML-KEM-768 keypair`	`30516` cycles	`30520` cycles	`1.00`
`ML-KEM-768 encaps`	`33423` cycles	`33418` cycles	`1.00`
`ML-KEM-768 decaps`	`42976` cycles	`42978` cycles	`1.00`
`ML-KEM-1024 keypair`	`44139` cycles	`44138` cycles	`1.00`
`ML-KEM-1024 encaps`	`49443` cycles	`49438` cycles	`1.00`
`ML-KEM-1024 decaps`	`62362` cycles	`62362` cycles	`1`
Benchmark suite	Current: `e16cf03`	Previous: `b5ef959`	Ratio
`ML-KEM-512 keypair`	`11301` cycles	`11287` cycles	`1.00`
`ML-KEM-512 encaps`	`12847` cycles	`12851` cycles	`1.00`
`ML-KEM-512 decaps`	`17674` cycles	`17703` cycles	`1.00`
`ML-KEM-768 keypair`	`19643` cycles	`19634` cycles	`1.00`
`ML-KEM-768 encaps`	`20592` cycles	`20584` cycles	`1.00`
`ML-KEM-768 decaps`	`27657` cycles	`27647` cycles	`1.00`
`ML-KEM-1024 keypair`	`26289` cycles	`26298` cycles	`1.00`
`ML-KEM-1024 encaps`	`28177` cycles	`28185` cycles	`1.00`
`ML-KEM-1024 decaps`	`37610` cycles	`37654` cycles	`1.00`
Benchmark suite	Current: `e16cf03`	Previous: `b5ef959`	Ratio
`ML-KEM-512 keypair`	`46469` cycles	`46428` cycles	`1.00`
`ML-KEM-512 encaps`	`54663` cycles	`54593` cycles	`1.00`
`ML-KEM-512 decaps`	`70389` cycles	`70319` cycles	`1.00`
`ML-KEM-768 keypair`	`76685` cycles	`76752` cycles	`1.00`
`ML-KEM-768 encaps`	`87366` cycles	`87358` cycles	`1.00`
`ML-KEM-768 decaps`	`107873` cycles	`107739` cycles	`1.00`
`ML-KEM-1024 keypair`	`112167` cycles	`112122` cycles	`1.00`
`ML-KEM-1024 encaps`	`126095` cycles	`126277` cycles	`1.00`
`ML-KEM-1024 decaps`	`152295` cycles	`152314` cycles	`1.00`
Benchmark suite	Current: `e16cf03`	Previous: `b5ef959`	Ratio
`ML-KEM-512 keypair`	`39488` cycles	`39495` cycles	`1.00`
`ML-KEM-512 encaps`	`47517` cycles	`47534` cycles	`1.00`
`ML-KEM-512 decaps`	`61796` cycles	`61829` cycles	`1.00`
`ML-KEM-768 keypair`	`63918` cycles	`63896` cycles	`1.00`
`ML-KEM-768 encaps`	`75275` cycles	`75263` cycles	`1.00`
`ML-KEM-768 decaps`	`94145` cycles	`93820` cycles	`1.00`
`ML-KEM-1024 keypair`	`95481` cycles	`95500` cycles	`1.00`
`ML-KEM-1024 encaps`	`108989` cycles	`109013` cycles	`1.00`
`ML-KEM-1024 decaps`	`132485` cycles	`132492` cycles	`1.00`
Benchmark suite	Current: `e16cf03`	Previous: `b5ef959`	Ratio
`ML-KEM-512 keypair`	`35508` cycles	`35500` cycles	`1.00`
`ML-KEM-512 encaps`	`40646` cycles	`40643` cycles	`1.00`
`ML-KEM-512 decaps`	`51642` cycles	`51642` cycles	`1`
`ML-KEM-768 keypair`	`58486` cycles	`58475` cycles	`1.00`
`ML-KEM-768 encaps`	`65245` cycles	`65242` cycles	`1.00`
`ML-KEM-768 decaps`	`80460` cycles	`80455` cycles	`1.00`
`ML-KEM-1024 keypair`	`88191` cycles	`88171` cycles	`1.00`
`ML-KEM-1024 encaps`	`96961` cycles	`96951` cycles	`1.00`
`ML-KEM-1024 decaps`	`116638` cycles	`116622` cycles	`1.00`