Skip to content

Commit

Permalink
Add corrected range jets.
Browse files Browse the repository at this point in the history
  • Loading branch information
@sigilante
sigilante committed Aug 13, 2024
1 parent 840760d commit 3d462d3
Show file tree
Hide file tree
Showing 2 changed files with 45 additions and 42 deletions.
2 changes: 1 addition & 1 deletion WORKSPACE.bazel
View file
Original file line number Diff line number Diff line change
Expand Up @@ -297,7 +297,7 @@ versioned_http_archive(
strip_prefix = "SoftBLAS-{version}",
# sha256 = "",
url = "https://github.com/urbit/SoftBLAS/archive/{version}.tar.gz",
version = "29daa2f2fd0ad5070e405ad287f3623804f8fc67",
version = "cbffb33f19ea02f9ffbd184d445123c57929ec53",
)

versioned_http_archive(
Expand Down
85 changes: 44 additions & 41 deletions pkg/noun/jets/i/lagoon.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -10,6 +10,11 @@
#include <math.h> // for pow()
#include <stdio.h>

#define f16_ceil(a) f16_roundToInt( a, softfloat_round_max, false )
#define f32_ceil(a) f32_roundToInt( a, softfloat_round_max, false )
#define f64_ceil(a) f64_roundToInt( a, softfloat_round_max, false )
#define f128M_ceil(a, b) f128M_roundToInt( a, softfloat_round_max, false, b )

union half {
float16_t h;
c3_w c;
Expand Down Expand Up @@ -1964,15 +1969,14 @@
u3r_bytes(0, 2, (c3_y*)&(a16.v), a);
u3r_bytes(0, 2, (c3_y*)&(b16.v), b);
u3r_bytes(0, 2, (c3_y*)&(interval16.v), d);
c3_d n16 = f16_to_i64(f16_div(f16_sub(b16, a16), interval16), softfloat_round_minMag, false);
c3_y* x_bytes16 = (c3_y*)u3a_malloc(((n16+1)*2+1)*sizeof(c3_y));
for (c3_d i = 1; i <= n16; i++) {
((float16_t*)x_bytes16)[n16-i] = f16_add(a16, f16_mul(i32_to_f16(i), interval16));
}
((float16_t*)x_bytes16)[n16] = a16;
// ((float16_t*)x_bytes16)[0] = b16;
x_bytes16[(n16+1)*2] = 0x1; // pin head
r_data = u3i_bytes(((n16+1)*2+1)*sizeof(c3_y), x_bytes16);
c3_d n16 = f16_to_i64(f16_ceil(f16_div(f16_sub(b16, a16), interval16)), softfloat_round_minMag, false);
c3_y* x_bytes16 = (c3_y*)u3a_malloc(((n16+1)*2)*sizeof(c3_y));
((float16_t*)x_bytes16)[0] = a16;
for (c3_d i = 1; i < n16; i++) {
((float16_t*)x_bytes16)[i] = f16_add(a16, f16_mul(i32_to_f16(i), interval16));
}
((float16_t*)x_bytes16)[n16].v = 0x1; // pin head
r_data = u3i_bytes(((n16+1)*2)*sizeof(c3_y), x_bytes16);
u3a_free(x_bytes16);
break;}

Expand All @@ -1981,15 +1985,14 @@
u3r_bytes(0, 4, (c3_y*)&(a32.v), a);
u3r_bytes(0, 4, (c3_y*)&(b32.v), b);
u3r_bytes(0, 4, (c3_y*)&(interval32.v), d);
c3_d n32 = f32_to_i64(f32_div(f32_sub(b32, a32), interval32), softfloat_round_minMag, false);
c3_y* x_bytes32 = (c3_y*)u3a_malloc(((n32+1)*4+1)*sizeof(c3_y));
for (c3_d i = 1; i <= n32; i++) {
((float32_t*)x_bytes32)[n32-i] = f32_add(a32, f32_mul(i32_to_f32(i), interval32));
}
((float32_t*)x_bytes32)[n32] = a32;
// ((float32_t*)x_bytes32)[0] = b32;
x_bytes32[(n32+1)*4] = 0x1; // pin head
r_data = u3i_bytes(((n32+1)*4+1)*sizeof(c3_y), x_bytes32);
c3_d n32 = f32_to_i64(f32_ceil(f32_div(f32_sub(b32, a32), interval32)), softfloat_round_minMag, false);
c3_y* x_bytes32 = (c3_y*)u3a_malloc(((n32+1)*4)*sizeof(c3_y));
((float32_t*)x_bytes32)[0] = a32;
for (c3_d i = 1; i < n32; i++) {
((float32_t*)x_bytes32)[i] = f32_add(a32, f32_mul(i32_to_f32(i), interval32));
}
((float32_t*)x_bytes32)[n32].v = 0x1; // pin head
r_data = u3i_bytes(((n32+1)*4)*sizeof(c3_y), x_bytes32);
u3a_free(x_bytes32);
break;}

Expand All @@ -1998,38 +2001,38 @@
u3r_bytes(0, 8, (c3_y*)&(a64.v), a);
u3r_bytes(0, 8, (c3_y*)&(b64.v), b);
u3r_bytes(0, 8, (c3_y*)&(interval64.v), d);
c3_d n64 = f64_to_i64(f64_div(f64_sub(b64, a64), interval64), softfloat_round_minMag, false);
c3_y* x_bytes64 = (c3_y*)u3a_malloc(((n64+1)*8+1)*sizeof(c3_y));
c3_d n64 = f64_to_i64(f64_ceil(f64_div(f64_sub(b64, a64), interval64)), softfloat_round_minMag, false);
c3_y* x_bytes64 = (c3_y*)u3a_malloc(((n64+1)*8)*sizeof(c3_y));
((float64_t*)x_bytes64)[0] = a64;
for (c3_d i = 1; i < n64; i++) {
((float64_t*)x_bytes64)[n64-i] = f64_add(a64, f64_mul(i32_to_f64(i), interval64));
((float64_t*)x_bytes64)[i] = f64_add(a64, f64_mul(i32_to_f64(i), interval64));
}
((float64_t*)x_bytes64)[n64] = a64;
// ((float64_t*)x_bytes64)[0] = b64;
x_bytes64[(n64+1)*8] = 0x1; // pin head
r_data = u3i_bytes(((n64+1)*8+1)*sizeof(c3_y), x_bytes64);
((float64_t*)x_bytes64)[n64].v = 0x1; // pin head
r_data = u3i_bytes(((n64+1)*8)*sizeof(c3_y), x_bytes64);
u3a_free(x_bytes64);
break;}

case 7: {
float128_t a128, b128, interval128;
u3r_bytes(0, 16, (c3_y*)&(a128.v[0]), a);
u3r_bytes(0, 16, (c3_y*)&(b128.v[0]), b);
u3r_bytes(0, 16, (c3_y*)&(interval128.v), d);
u3r_bytes(0, 16, (c3_y*)&(interval128.v[0]), d);
float128_t tmp;
f128M_sub(&b128, &a128, &tmp);
f128M_div(&tmp, &interval128, &interval128);
f128M_div(&tmp, &interval128, &tmp);
f128M_ceil(&tmp, &tmp);
c3_d n128 = f128M_to_i64(&tmp, softfloat_round_minMag, false);
c3_y* x_bytes128 = (c3_y*)u3a_malloc(((n128+1)*16+1)*sizeof(c3_y));
c3_y* x_bytes128 = (c3_y*)u3a_malloc(((n128+1)*16)*sizeof(c3_y));
float128_t i128;
((float128_t*)x_bytes128)[0] = a128;
for (c3_d i = 1; i < n128; i++) {
i32_to_f128M(i, &i128);
f128M_mul(&i128, &interval128, &((float128_t*)x_bytes128)[n128-i]);
f128M_add(&a128, &((float128_t*)x_bytes128)[n128-i], &((float128_t*)x_bytes128)[n128-i]);
f128M_mul(&i128, &interval128, &((float128_t*)x_bytes128)[i]);
f128M_add(&a128, &((float128_t*)x_bytes128)[i], &((float128_t*)x_bytes128)[i]);
}
((float128_t*)x_bytes128)[n128] = a128;
// ((float128_t*)x_bytes128)[0] = b128;
x_bytes128[(n128+1)*16] = 0x1; // pin head
r_data = u3i_bytes(((n128+1)*16+1)*sizeof(c3_y), x_bytes128);
((float128_t*)x_bytes128)[n128].v[0] = 0x1; // pin head
((float128_t*)x_bytes128)[n128].v[1] = 0x0; // pin head
r_data = u3i_bytes(((n128+1)*16)*sizeof(c3_y), x_bytes128);
u3a_free(x_bytes128);
break;}
}
Expand Down Expand Up @@ -3154,19 +3157,19 @@
u3r_bytes(0, 2, (c3_y*)&a_, a);
u3r_bytes(0, 2, (c3_y*)&b_, b);
u3r_bytes(0, 2, (c3_y*)&d_, d);
n_ = f16_to_i64(f16_div(f16_sub((float16_t){b_}, (float16_t){a_}), (float16_t){d_}), softfloat_round_minMag, false);
n_ = f16_to_i64(f16_ceil(f16_div(f16_sub((float16_t){b_}, (float16_t){a_}), (float16_t){d_})), softfloat_round_minMag, false) - 1;
break;
case 5:
u3r_bytes(0, 4, (c3_y*)&a_, a);
u3r_bytes(0, 4, (c3_y*)&b_, b);
u3r_bytes(0, 4, (c3_y*)&d_, d);
n_ = f32_to_i64(f32_div(f32_sub((float32_t){b_}, (float32_t){a_}), (float32_t){d_}), softfloat_round_minMag, false);
n_ = f32_to_i64(f32_ceil(f32_div(f32_sub((float32_t){b_}, (float32_t){a_}), (float32_t){d_})), softfloat_round_minMag, false) - 1;
break;
case 6:
u3r_bytes(0, 8, (c3_y*)&a_, a);
u3r_bytes(0, 8, (c3_y*)&b_, b);
u3r_bytes(0, 8, (c3_y*)&d_, d);
n_ = f64_to_i64(f64_div(f64_sub((float64_t){b_}, (float64_t){a_}), (float64_t){d_}), softfloat_round_minMag, false);
n_ = f64_to_i64(f64_ceil(f64_div(f64_sub((float64_t){b_}, (float64_t){a_}), (float64_t){d_})), softfloat_round_minMag, false) - 1;
break;
case 7: {
c3_d a__[2], b__[2], d__[2];
Expand All @@ -3176,11 +3179,12 @@
float128_t tmp;
f128M_sub((float128_t*)&b__, (float128_t*)&a__, &tmp);
f128M_div(&tmp, (float128_t*)&d__, &tmp);
n_ = f128M_to_i64(&tmp, softfloat_round_minMag, false);
f128M_ceil(&tmp, &tmp);
n_ = f128M_to_i64(&tmp, softfloat_round_minMag, false) - 1;
break;}
}
u3_noun n = u3i_chub(n_+1);
x_shape = u3nt(u3k(n), 0x1, u3_nul);
x_shape = u3nc(u3k(n), u3_nul);
return u3nc(u3nq(u3k(x_shape), u3k(x_bloq), u3k(x_kind), u3k(x_tail)), r_data);

default:
Expand Down Expand Up @@ -3268,7 +3272,7 @@
// Each argument is a ray, [=meta data=@ux]
u3_noun x_meta, x_data,
y_meta, y_data;
fprintf(stderr, "mmul 1\n");

if ( c3n == u3r_mean(cor,
u3x_sam_4, &x_meta,
u3x_sam_5, &x_data,
Expand Down Expand Up @@ -3297,7 +3301,6 @@
} else {
switch (x_kind) {
case c3__i754:
fprintf(stderr, "mmul 2\n");
_set_rounding(rnd);
u3_noun r_data = u3qi_la_mmul_i754(x_data, y_data, x_shape, y_shape, x_bloq);
// result is already [meta data]
Expand Down

0 comments on commit 3d462d3

Please sign in to comment.