vllm.models.deepseek_v41.nvidia.specialized_decode.ampere.mhc
¶
The mHC layer boundary of DeepSeek-V4.1 on SM80, in CuTe DSL.
DeepSeek-V4.1 carries hc = 4 residual streams per token (manifold
hyper-connections, mHC). Every sublayer boundary does two things with them:
- Post and collapse (
reference.mhc_post_collapse). The sublayer outputxenters the streams through the previous boundary's coefficients,r'[j] = post[j] x + sum_k comb[k, j] r[k](rounded to BF16, the stored residual), and the next sublayer's input is the RMSNorm of the streams collapsed with the previous pre-mix,norm(sum_j pre[j] r'[j]). The collapse uses the carried pre-mix, so this step does not wait for this boundary's own coefficients. - Coefficients (
reference.mhc_coefficients). A projection of the flattened streamsr'(4 x hiddenwide) onto 24 mixes and their RMS, then sigmoids for the pre- and post-mixes and a 20-step Sinkhorn normalization for the 4 x 4combmatrix. They are used only at the next boundary.
The stock path runs both in two TileLang kernels with one CTA per token for the coefficient math and the collapse; at a handful of decode rows they are latency-bound (about 14 us per boundary at six rows on A100). Here each step is its own kernel:
PostCollapse-- one CTA per row, one 16-byte vector of every stream per thread: the post-mix, the stored residual, the collapse and its RMSNorm (one block reduction), and optionally the stream mean that draft models read.Coefficients--hidden / 128CTAs, each owning 128 hidden columns of all four streams for every row, taken in groups of 16 rows. A CTA keeps the slice of the projection weights in registers, recomputesr'for its slice (so it does not wait forPostCollapse) and writes per-row partial dot products. The last CTA to finish a row group (an atomic ticket per group) sums the group's partials in CTA order and runs the Sinkhorn normalization, one 16-lane group per row; the CTAs start at different groups, so that different CTAs finish them. The ticket counters reset themselves, so the kernel needs no initialization per launch and is CUDA-graph safe.CoefficientsMma-- the same work on tensor cores, taken on SM80 for the four streams when the hidden size is a multiple of 320 (SM90, whose small decode batches run these kernels too, keepsCoefficients, which is faster there at six rows):hidden / 320CTAs, each stagingr'of its 320 columns for a row group in shared memory as BF16, then multiplying it by the projection withmma.syncm16n8k16 (each of eight warps one stream's half of the slice, all 24 mixes). The FP32 projection is split into two BF16 terms,hi = bf16(w)andlo = bf16(w - hi), multiplied separately and summed in FP32. The sums of squares stay on CUDA cores; the partials, tickets and Sinkhorn finish are those ofCoefficients.
Because the coefficients are needed only at the next boundary, the model
runs them on a side stream inside FULL CUDA graphs (see mhc_post_pre):
the critical path keeps only PostCollapse. At DeepSeek-V4.1's hidden
size of 5120 the side kernel holds 16 SMs (CoefficientsMma) or 40
(Coefficients) at any batch. With a CTA per slice and row group it took
120 CTAs at 48 rows, more than A100's 108 SMs, and PostCollapse, whose
640-thread CTAs do not fit on an SM beside one of them, waited about 14 us
for SMs at every boundary. Fewer side CTAs leave more SMs to the sublayers
that run beside them.
Numerics: the residual and the collapse round exactly as the stock kernels
do; the projection reads the BF16 residual (as the PyTorch reference and
the stock large-batch path do; the stock fused small-batch kernel projects
the unrounded FP32 values instead, a difference within BF16 rounding of the
projection's input), and the sums run in a different order, so results
differ from the stock kernels by FP32 rounding of the reductions. The
coefficients stay within about 1e-6 of an FP64 reference. The two BF16
terms of CoefficientsMma keep about 16 significant bits of each
projection weight, more than the TF32 projection of the stock SM90 kernel;
their effect on the coefficients is bounded in the tests.
The ticket counters are shared per device, so launches of Coefficients
must be ordered on one stream at a time, which the model's boundaries are.
Classes:
-
Coefficients–The next pre-, post- and comb mixes; see the module docstring and
-
CoefficientsMma–Coefficientsof the four residual streams with the projection on -
PostCollapse–Post-mix, collapse and RMSNorm of one row per CTA; see the module
Functions:
-
bf16_high–The FP32 value of the high BF16 of a packed pair.
-
bf16_low–The FP32 value of the low BF16 of a packed pair.
-
column_sum–Sum over lanes
e ^ 4ande ^ 8: a comb column of a 16-lane -
fence_gpu–Order this thread's memory accesses at GPU scope (acquire-release).
-
fma–a * b + cwith one rounding. -
load_f32–One FP32 global load.
-
load_f32_shared_by_ctas–One FP32 load from L2 of a value another CTA of this grid wrote; it
-
load_f32x4–Four FP32 values (16-byte aligned) in one vector load.
-
load_weights2–Two FP32 weights (read-only for the kernel) in one vector load.
-
load_weights4–Four FP32 weights (read-only for the kernel) in one vector load.
-
load_words–count(2 or 4) consecutive 32-bit words in one vector load. -
mhc_coefficients–reference.mhc_coefficientson the GPU: the post-, comb- and -
mhc_post_collapse–reference.mhc_post_collapseon the GPU. -
mhc_post_pre–A boundary between sublayers:
mhc_post_collapseand -
mhc_pre–A boundary without a sublayer output to add: the coefficients of
-
mma_bf16–c + a bfor one m16n8k16 BF16 tile with FP32 accumulation, given -
pack–FP32 values rounded to BF16 and packed in pairs.
-
pack_bf16x2–Two FP32 values rounded to nearest-even BF16, packed low-first.
-
prepare–Allocate the ticket counters and the stream-zero pre-mix, and compile
-
rounded–The packed BF16 words of
valuesand their FP32 values. -
sigmoid–1 / (1 + exp(-x)). -
store_f32–One FP32 global store.
-
store_words–Store 2 or 4 consecutive 32-bit words in one vector store.
-
supports–Whether these kernels take a boundary of
rowsrows. -
ticket–Atomically add one to the counter at
address; the old value. -
transpose_sum–Warp sums of the first
countvalues of every lane, taken as 16 -
unpack–BF16 values of packed words as FP32, in memory order.
-
warp_sum–Butterfly sum over aligned groups of
widthlanes.
Coefficients
¶
The next pre-, post- and comb mixes; see the module docstring and
reference.mhc_coefficients.
streams is 4 for the residual streams, or 1 when the projection reads
one hidden-wide row (the first layer's broadcast embedding). With
has_post the streams are the post-mix of x and residual
(PostCollapse's output, recomputed); otherwise residual itself.
Methods:
-
accumulate–This lane's share of row
local's dot products (and, in warp -
accumulate_squares–Row
local's sum of squares over this lane's columns. -
finish–Mixes of two rows per warp, a 16-lane group per row: lane
e -
finish_group–In the row group's last CTA: reset its counter, sum every row's
-
gather–Statistic
pair % STATSof rowpair // STATS, summed over -
publish–Warp sums of the group's statistics into its partials at
-
stage–One row's four columns per lane of every stream into shared
-
store_pre_post–Pre- and post-mix
eof a row. -
stream_words–Four columns of every stream of a row as two words of BF16 pairs
-
sum_partials–The sum, in CTA order, of one statistic's partials.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 | |
accumulate(sValues, weights, acc, squares, local, lane, warp)
¶
This lane's share of row local's dot products (and, in warp
0, its sum of squares).
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
accumulate_squares(squares, v, local)
¶
Row local's sum of squares over this lane's columns.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
finish(sStats, lane, warp, first, rows, scale, base, post_out, comb_out, pre_out, rms_eps, pre_eps, sinkhorn_eps, post_mult, iterations)
¶
Mixes of two rows per warp, a 16-lane group per row: lane e
owns comb entry (e // 4, e % 4) and, below 4, the pre- and
post-mix e.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 | |
finish_group(sStats, partials, counter, tid, group, first, rows, scale, base, post_out, comb_out, pre_out, rms_eps, pre_eps, sinkhorn_eps, post_mult, iterations)
¶
In the row group's last CTA: reset its counter, sum every row's partials in CTA order and compute its mixes.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
gather(sStats, partials, pair, group, first, rows)
¶
Statistic pair % STATS of row pair // STATS, summed over
the CTAs in order (zero for a row past the batch).
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
publish(acc, squares, address, valid, lane, warp, mix0)
¶
Warp sums of the group's statistics into its partials at
address, for its first valid rows.
A transposed butterfly: at each of the offsets 1, 2, 4 and 8 a lane
keeps one half of its values (by its lane bit) and adds its
partner's copy of that half, so the 48 sums (16 for the squares)
take 48 shuffles (16) instead of five each. The additions are those
of warp_sum, so the sums are bitwise the same. Lane l and
l ^ 16 end with the row whose index is the low four bits of
l reversed.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
stage(sValues, x, residual, post, comb, local, row, lane, column)
¶
One row's four columns per lane of every stream into shared
memory as BF16 pairs, post-mixed and rounded with has_post.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
store_pre_post(sStats, stats, e, row, inverse_rms, scale, base, post_out, pre_out, pre_eps, post_mult)
¶
Pre- and post-mix e of a row.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
stream_words(x, residual, post, comb, row, column)
¶
Four columns of every stream of a row as two words of BF16 pairs
each, post-mixed and rounded with has_post.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
sum_partials(address)
¶
The sum, in CTA order, of one statistic's partials.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
CoefficientsMma
¶
Bases: Coefficients
Coefficients of the four residual streams with the projection on
tensor cores, in hidden / 320 CTAs; see the module docstring.
A CTA owns 320 hidden columns of the four streams, a projection width of
1280, which its eight warps split into a stream's half each. A warp
holds its 24 x 160 slice of the FP32 projection in registers as two BF16
terms, hi = bf16(w) and lo = bf16(w - hi), in mma.sync
fragment order. For each group of 16 rows the CTA stages the post-mixed
streams in shared memory, each warp multiplies them by both terms
(m16n8k16, FP32 accumulation) and sums the squares of its fragments, and
the warps' partials are added in warp order into the CTA's partials, which
the group's last CTA finishes as Coefficients does.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 | |
PostCollapse
¶
Post-mix, collapse and RMSNorm of one row per CTA; see the module
docstring and reference.mhc_post_collapse.
Methods:
-
finish_row–The post-mix (with the sublayer output's values
xs, eight per -
load_row–Everything but the sublayer output that thread
tidof row
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 | |
finish_row(xs, loaded, residual_out, layer_input, aux, eps, sSums, row, tid)
¶
The post-mix (with the sublayer output's values xs, eight per
thread), stored streams, stream mean, collapse and RMSNorm of a row,
from load_row's values; every thread of the CTA takes part.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
load_row(residual, post, comb, pre, pre_stride, weight, row, tid)
¶
Everything but the sublayer output that thread tid of row
row reads: its vector of the norm weight and of every stream, and
the row's mixes. A caller can issue these loads early.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
_address(tensor)
¶
A tensor's device address as a kernel argument (0 for None).
_asm(result, operands, text, constraints, side_effects, loc, ip)
¶
One inline PTX statement.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
_coefficients_kernel(device_index, hidden, streams, has_post)
cached
¶
Coefficients compiled for a device and variant.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
_coefficients_mma_kernel(device_index, hidden, has_post)
cached
¶
CoefficientsMma compiled for a device and variant.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
_counters(device_index)
cached
¶
The per-device ticket counters of Coefficients, zeroed once.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
_fake_stream()
¶
_post_collapse_kernel(device_index, hidden, has_post, write_aux)
cached
¶
PostCollapse compiled for a device and variant.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
_select_first_stream(device_index)
cached
¶
The pre-mix (1, 0, 0, 0) that selects stream zero, for every row.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
_uses_mma(streams, hidden)
¶
Whether CoefficientsMma projects these streams: the four residual
streams in 320-column slices, on SM80. One hidden-wide row (the first
layer's broadcast embedding), other widths and SM90 project on CUDA
cores: on H100 at TP8 with five draft tokens the tensor-core kernel made
the decode step 1.7% slower at 6 rows and 0.5-0.7% faster at 24 and 48.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
bf16_high(word)
¶
bf16_low(word)
¶
column_sum(value)
¶
Sum over lanes e ^ 4 and e ^ 8: a comb column of a 16-lane
group.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
fence_gpu(*, loc=None, ip=None)
¶
Order this thread's memory accesses at GPU scope (acquire-release).
fma(a, b, c, *, loc=None, ip=None)
¶
a * b + c with one rounding.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
load_f32(address, *, loc=None, ip=None)
¶
One FP32 global load.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
load_f32_shared_by_ctas(address, *, loc=None, ip=None)
¶
One FP32 load from L2 of a value another CTA of this grid wrote; it stays after the ticket that orders it.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
load_f32x4(address, *, loc=None, ip=None)
¶
Four FP32 values (16-byte aligned) in one vector load.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
load_weights2(address, *, loc=None, ip=None)
¶
Two FP32 weights (read-only for the kernel) in one vector load.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
load_weights4(address, *, loc=None, ip=None)
¶
Four FP32 weights (read-only for the kernel) in one vector load.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
load_words(address, count, *, loc=None, ip=None)
¶
count (2 or 4) consecutive 32-bit words in one vector load.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
mhc_coefficients(x, residual, post_mix, comb_mix, fn, hc_scale, hc_base, rms_eps, hc_pre_eps, hc_sinkhorn_eps, hc_post_mult_value, sinkhorn_repeat, *, stream=None)
¶
reference.mhc_coefficients on the GPU: the post-, comb- and
pre-mix [rows, 4], [rows, 4, 4], [rows, 4] (FP32).
residual is [rows, 4, hidden] or, for a projection of one
hidden-wide row, [rows, hidden]; with x the streams are first
post-mixed as in mhc_post_collapse. With stream the kernel runs
there after the work queued so far on the current stream; the outputs
are allocated on the current stream, which must wait for stream
before reading them.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 | |
mhc_post_collapse(x, residual, post_mix, comb_mix, pre_mix, norm_weight, norm_eps, write_aux=False)
¶
reference.mhc_post_collapse on the GPU.
With x the streams are first post-mixed (post_mix [rows, 4],
comb_mix [rows, 4, 4]); pre_mix None selects stream zero.
Returns the streams (new, or residual itself without x), the
normalized layer input and the stream mean (empty without
write_aux).
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
mhc_post_pre(x, residual, post_mix, comb_mix, fn, hc_scale, hc_base, rms_eps, hc_pre_eps, hc_sinkhorn_eps, hc_post_mult_value, sinkhorn_repeat, pre_mix, norm_weight, norm_eps, capture_aux=False, *, stream=None)
¶
A boundary between sublayers: mhc_post_collapse and
mhc_coefficients of the sublayer output x.
Returns, as the stock fused post + pre op does, the new streams, the
post-, comb-mix [rows, 4, 1], [rows, 4, 4], the layer input, the
next pre-mix and the stream mean (empty without capture_aux). With
stream the coefficients run there, concurrently with the collapse
and whatever follows it; join stream before reading them.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
mhc_pre(residual, fn, hc_scale, hc_base, rms_eps, hc_pre_eps, hc_sinkhorn_eps, hc_post_mult_value, sinkhorn_repeat, pre_mix, x, norm_weight, norm_eps, *, stream=None)
¶
A boundary without a sublayer output to add: the coefficients of
residual (or of the one-row projection input x) and its collapse
with pre_mix (stream zero when None), as the stock delayed mHC pre
op returns them: post-, comb-mix, layer input and next pre-mix.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
mma_bf16(c, a, b, *, loc=None, ip=None)
¶
c + a b for one m16n8k16 BF16 tile with FP32 accumulation, given
this thread's fragments: four words of a, two of b and four values
of c (the PTX mma.sync layouts).
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
pack(values)
¶
FP32 values rounded to BF16 and packed in pairs.
pack_bf16x2(low, high, *, loc=None, ip=None)
¶
Two FP32 values rounded to nearest-even BF16, packed low-first.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
prepare(device_index, hidden)
¶
Allocate the ticket counters and the stream-zero pre-mix, and compile every kernel variant the model uses, before any CUDA-graph capture.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
rounded(values)
¶
The packed BF16 words of values and their FP32 values.
sigmoid(x)
¶
store_f32(address, value, *, loc=None, ip=None)
¶
One FP32 global store.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
store_words(address, words, *, loc=None, ip=None)
¶
Store 2 or 4 consecutive 32-bit words in one vector store.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
supports(rows, hidden, hc_mult)
¶
Whether these kernels take a boundary of rows rows.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
ticket(address, *, loc=None, ip=None)
¶
Atomically add one to the counter at address; the old value.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
transpose_sum(values, count, lane)
¶
Warp sums of the first count values of every lane, taken as 16
groups of count / 16 consecutive values (group g is
values[g * count / 16 :]): on return, the first count / 16 values
of lane l hold the sums of group g, whose 4-bit index is
l % 16 reversed. The additions are those of warp_sum(value, 32).
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
unpack(words)
¶
BF16 values of packed words as FP32, in memory order.
Source code in vllm/models/deepseek_v41/nvidia/specialized_decode/ampere/mhc.py
warp_sum(value, width)
¶
Butterfly sum over aligned groups of width lanes.