Compiling ipu-codegen v0.1.0 (/srv/home/gc-sdk/ipu-stack/crates/ipu-codegen) warning: unused import: `region_program` --> crates/ipu-codegen/src/estimate/mod.rs:22:23 | 22 | operation_cycles, region_program, | ^^^^^^^^^^^^^^ | = note: `#[warn(unused_imports)]` (part of `#[warn(unused)]`) on by default warning: unused import: `program::program_cycles` --> crates/ipu-codegen/src/estimate/mod.rs:25:16 | 25 | pub(crate) use program::program_cycles; | ^^^^^^^^^^^^^^^^^^^^^^^ warning: unused import: `memory::*` --> crates/ipu-codegen/src/estimate/mod.rs:39:16 | 39 | pub(crate) use memory::*; | ^^^^^^^^^ warning: unused import: `contiguous_axis_blocks` --> crates/ipu-codegen/src/storage/mod.rs:12:49 | 12 | ByteTraversal, StridedSpan, byte_traversal, contiguous_axis_blocks, contiguous_axis_stride, panel_byte_traversal, | ^^^^^^^^^^^^^^^^^^^^^^ warning: unused imports: `axis_tiling` and `broadcast_operand_tiling` --> crates/ipu-codegen/src/tensor/mod.rs:99:5 | 99 | axis_tiling, broadcast_operand_tiling, project_tiling, same_distribution, | ^^^^^^^^^^^ ^^^^^^^^^^^^^^^^^^^^^^^^ warning: methods `objectives` and `total_with_exchange` are never used --> crates/ipu-codegen/src/estimate/memory.rs:57:19 | 46 | impl MemoryPeaks { | ---------------- methods in this implementation ... 57 | pub(crate) fn objectives(self) -> [u64; 5] { | ^^^^^^^^^^ ... 67 | pub(crate) fn total_with_exchange(self) -> u64 { | ^^^^^^^^^^^^^^^^^^^ | = note: `#[warn(dead_code)]` (part of `#[warn(unused)]`) on by default warning: function `tensor_memory` is never used --> crates/ipu-codegen/src/estimate/memory.rs:106:15 | 106 | pub(crate) fn tensor_memory(tensor: &TensorType) -> MemoryUsage { | ^^^^^^^^^^^^^ warning: function `operation_cycles` is never used --> crates/ipu-codegen/src/estimate/mid.rs:26:15 | 26 | pub(crate) fn operation_cycles<'a>( | ^^^^^^^^^^^^^^^^ warning: function `maximum_axis_shard_extent` is never used --> crates/ipu-codegen/src/estimate/tensor.rs:25:15 | 25 | pub(crate) fn maximum_axis_shard_extent(tensor: &TensorType, axis: usize) -> u64 { | ^^^^^^^^^^^^^^^^^^^^^^^^^ warning: constant `WORKER_CALL_CYCLES` is never used --> crates/ipu-codegen/src/kernel/copy.rs:18:18 | 18 | pub(crate) const WORKER_CALL_CYCLES: u64 = ipu_target::ipu21::costs::COSTS.local_copy_call_cycles; | ^^^^^^^^^^^^^^^^^^ warning: field `complete_rows` is never read --> crates/ipu-codegen/src/kernel/binding.rs:326:9 | 323 | pub(crate) struct OutputCapability { | ---------------- field in this struct ... 326 | pub complete_rows: bool, | ^^^^^^^^^^^^^ warning: function `supports_row_major_population` is never used --> crates/ipu-codegen/src/kernel/rearrange.rs:247:15 | 247 | pub(crate) fn supports_row_major_population(order: ElementOrder) -> bool { | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ warning: function `is_single_use` is never used --> crates/ipu-codegen/src/mid/rewrite.rs:44:15 | 44 | pub(crate) fn is_single_use( | ^^^^^^^^^^^^^ warning: function `producer_through_copies` is never used --> crates/ipu-codegen/src/mid/rewrite.rs:141:15 | 141 | pub(crate) fn producer_through_copies( | ^^^^^^^^^^^^^^^^^^^^^^^ warning: function `fusion_pays` is never used --> crates/ipu-codegen/src/mid/rewrite.rs:171:15 | 171 | pub(crate) fn fusion_pays<'a>( | ^^^^^^^^^^^ warning: method `input_axis` is never used --> crates/ipu-codegen/src/tensor/mod.rs:54:19 | 38 | impl<'a> Broadcast<'a> { | ---------------------- method in this implementation ... 54 | pub(crate) fn input_axis(self, output_axis: usize) -> Option { | ^^^^^^^^^^ warning: method `minimum_extent` is never used --> crates/ipu-codegen/src/tensor/resolved.rs:50:19 | 15 | impl ResolvedAxis { | ----------------- method in this implementation ... 50 | pub(crate) fn minimum_extent(&self) -> u32 { | ^^^^^^^^^^^^^^ warning: function `broadcast_operand_tiling` is never used --> crates/ipu-codegen/src/tensor/resolved.rs:503:15 | 503 | pub(crate) fn broadcast_operand_tiling( | ^^^^^^^^^^^^^^^^^^^^^^^^ warning: function `axis_tiling` is never used --> crates/ipu-codegen/src/tensor/resolved.rs:514:15 | 514 | pub(crate) fn axis_tiling(tensor: &TensorType, axis: usize) -> Option<&AxisTiling> { | ^^^^^^^^^^^ warning: `ipu-codegen` (lib test) generated 19 warnings (run `cargo fix --lib -p ipu-codegen --tests` to apply 5 suggestions) Finished `release` profile [optimized] target(s) in 33.12s Running unittests src/lib.rs (target/release/deps/ipu_codegen-8a4acda8960e2f33) running 3 tests COST elementwise 1 1 0 582 582 COST elementwise 1 1 1 582 582 COST elementwise 1 3 0 972 1098 COST elementwise 1 3 1 1212 1362 COST elementwise 2 1 0 558 546 COST elementwise 2 1 1 558 546 COST elementwise 2 1 2 1440 1428 COST elementwise 2 1 3 1566 1488 COST elementwise 2 1 4 1254 1110 COST elementwise 2 1 5 996 1188 COST elementwise 2 3 0 558 558 COST elementwise 2 3 1 1068 1074 COST elementwise 2 3 2 4056 4020 COST elementwise 2 3 3 4434 4200 COST elementwise 2 3 4 3510 3078 COST elementwise 2 3 5 1872 2100 COST elementwise 3 1 0 972 1098 COST elementwise 3 1 1 972 1098 COST elementwise 3 3 0 972 954 COST elementwise 3 3 1 1092 1074 COST elementwise 4 1 0 522 522 COST elementwise 4 1 1 522 522 COST elementwise 4 1 2 1518 1470 COST elementwise 4 1 3 1644 1536 COST elementwise 4 1 4 1254 1110 COST elementwise 4 1 5 1302 1260 COST elementwise 4 3 0 522 522 COST elementwise 4 3 1 1026 1026 COST elementwise 4 3 2 4290 4146 COST elementwise 4 3 3 4668 4344 COST elementwise 4 3 4 3510 3078 COST elementwise 4 3 5 2358 2316 COST elementwise 6 1 0 558 558 COST elementwise 6 1 1 558 558 COST elementwise 6 1 2 1440 1470 COST elementwise 6 1 3 1566 1530 COST elementwise 6 1 4 1254 1152 COST elementwise 6 1 5 996 1188 COST elementwise 6 3 0 594 588 COST elementwise 6 3 1 1068 1086 COST elementwise 6 3 2 4056 4146 COST elementwise 6 3 3 4434 4332 COST elementwise 6 3 4 3510 3204 COST elementwise 6 3 5 1872 2100 COST elementwise 7 1 0 972 954 COST elementwise 7 1 1 972 954 COST elementwise 7 3 0 1164 1218 COST elementwise 7 3 1 1284 1314 COST elementwise 12 1 0 522 522 COST elementwise 12 1 1 522 522 COST elementwise 12 1 2 1518 1512 COST elementwise 12 1 3 1644 1578 COST elementwise 12 1 4 1254 1152 COST elementwise 12 1 5 1302 1260 COST elementwise 12 3 0 540 540 COST elementwise 12 3 1 1026 1026 COST elementwise 12 3 2 4290 4278 COST elementwise 12 3 3 4668 4482 SOFTMAX softmax 1 1 732 726 SOFTMAX softmax 2 1 732 726 SOFTMAX softmax 5 1 732 726 SOFTMAX softmax 6 1 732 726 COST elementwise 12 3 4 3510 3204 COST elementwise 12 3 5 2358 2316 COST elementwise 14 1 0 594 576 COST elementwise 14 1 1 594 576 SOFTMAX softmax 7 1 1230 1242 SOFTMAX softmax 8 1 1230 1242 SOFTMAX softmax 12 1 1230 1242 COST elementwise 14 1 2 1548 1572 COST elementwise 14 1 3 1710 1668 SOFTMAX softmax 17 1 1728 1758 SOFTMAX softmax 1 2 744 732 SOFTMAX softmax 2 2 744 738 SOFTMAX softmax 5 2 744 738 SOFTMAX softmax 6 2 744 738 COST elementwise 14 1 4 1302 1194 COST elementwise 14 1 5 1056 1248 SOFTMAX softmax 7 2 1254 1254 SOFTMAX softmax 8 2 1254 1266 SOFTMAX softmax 12 2 1254 1266 SOFTMAX softmax 17 2 1764 1794 COST elementwise 14 3 0 666 648 COST elementwise 14 3 1 1176 1164 SOFTMAX softmax 1 15 1158 1152 SOFTMAX softmax 2 15 1158 1200 SOFTMAX softmax 5 15 1158 1200 SOFTMAX softmax 6 15 1158 1200 COST elementwise 14 3 2 4380 4452 COST elementwise 14 3 3 4866 4740 COST elementwise 14 3 4 3654 3330 SOFTMAX softmax 7 15 2082 2094 SOFTMAX softmax 8 15 2082 2190 SOFTMAX softmax 12 15 2082 2190 SOFTMAX softmax 17 15 3006 3180 COST elementwise 14 3 5 2052 2280 COST elementwise 16 1 0 522 522 SOFTMAX softmax 1 16 690 684 COST elementwise 16 1 1 522 522 COST elementwise 16 1 2 1566 1518 SOFTMAX softmax 2 16 690 750 SOFTMAX softmax 5 16 690 750 SOFTMAX softmax 6 16 690 750 SOFTMAX softmax 7 16 1158 1164 COST elementwise 16 1 3 1716 1614 COST elementwise 16 1 4 1302 1158 COST elementwise 16 1 5 1302 1260 SOFTMAX softmax 8 16 1158 1296 SOFTMAX softmax 12 16 1158 1296 SOFTMAX softmax 17 16 1626 1842 SOFTMAX softmax 1 17 1038 1032 COST elementwise 16 3 0 540 540 COST elementwise 16 3 1 1026 1026 SOFTMAX softmax 2 17 1038 1092 SOFTMAX softmax 5 17 1038 1092 SOFTMAX softmax 6 17 1038 1092 SOFTMAX softmax 7 17 1842 1854 COST elementwise 16 3 2 4434 4290 COST elementwise 16 3 3 4884 4578 SOFTMAX softmax 8 17 1842 1974 COST elementwise 16 3 4 3654 3222 COST elementwise 16 3 5 2358 2316 SOFTMAX softmax 12 17 1842 1974 SOFTMAX softmax 17 17 2646 2856 SOFTMAX softmax 1 31 1464 1458 COST elementwise 24 1 0 522 522 COST elementwise 24 1 1 522 522 SOFTMAX softmax 2 31 1464 1560 COST elementwise 24 1 2 1566 1560 COST elementwise 24 1 3 1716 1656 SOFTMAX softmax 5 31 1464 1560 SOFTMAX softmax 6 31 1464 1560 SOFTMAX softmax 7 31 2694 2706 SOFTMAX softmax 8 31 2694 2910 COST elementwise 24 1 4 1302 1200 COST elementwise 24 1 5 1302 1260 SOFTMAX softmax 12 31 2694 2910 SOFTMAX softmax 17 31 3924 4260 SOFTMAX softmax 1 64 1428 1422 SOFTMAX softmax 2 64 1428 1686 COST elementwise 24 3 0 558 558 COST elementwise 24 3 1 1026 1026 SOFTMAX softmax 5 64 1428 1686 COST elementwise 24 3 2 4434 4422 COST elementwise 24 3 3 4884 4704 SOFTMAX softmax 6 64 1428 1686 COST elementwise 24 3 4 3654 3348 COST elementwise 24 3 5 2358 2316 SOFTMAX softmax 7 64 2634 2640 SOFTMAX softmax 8 64 2634 3168 SOFTMAX softmax 12 64 2634 3168 SOFTMAX softmax 17 64 3840 4650 SOFTMAX softmax 1 65 1776 1770 COST elementwise 32 1 0 540 540 COST elementwise 32 1 1 540 540 SOFTMAX softmax 2 65 1776 1992 SOFTMAX softmax 5 65 1776 1992 SOFTMAX softmax 6 65 1776 1992 SOFTMAX softmax 7 65 3318 3330 COST elementwise 32 1 2 1686 1644 COST elementwise 32 1 3 1872 1770 SOFTMAX softmax 8 65 3318 3774 SOFTMAX softmax 12 65 3318 3774 SOFTMAX softmax 17 65 4860 5556 SOFTMAX softmax 1 729 12240 12234 COST elementwise 32 1 4 1350 1212 COST elementwise 32 1 5 1374 1332 SOFTMAX softmax 2 729 12240 14694 SOFTMAX softmax 5 729 12240 14694 SOFTMAX softmax 6 729 12240 14694 COST elementwise 32 3 0 576 576 COST elementwise 32 3 1 1080 1080 SOFTMAX softmax 7 729 24246 24258 SOFTMAX softmax 8 729 24246 29178 SOFTMAX softmax 12 729 24246 29178 SOFTMAX softmax 17 729 36252 43662 COST elementwise 32 3 2 4794 4668 COST elementwise 32 3 3 5352 5058 SOFTMAX softmax 1 768 12252 12246 SOFTMAX softmax 2 768 12252 15414 SOFTMAX softmax 5 768 12252 15414 SOFTMAX softmax 6 768 12252 15414 COST elementwise 32 3 4 3798 3384 COST elementwise 32 3 5 2574 2532 SOFTMAX softmax 7 768 24282 24288 SOFTMAX softmax 8 768 24282 30624 SOFTMAX softmax 12 768 24282 30624 SOFTMAX softmax 17 768 36312 45834 COST elementwise 48 1 0 540 540 COST elementwise 48 1 1 540 540 COST elementwise 48 1 2 1734 1728 COST elementwise 48 1 3 1944 1884 COST elementwise 48 1 4 1398 1296 COST elementwise 48 1 5 1374 1332 COST elementwise 48 3 0 612 612 COST elementwise 48 3 1 1080 1080 COST elementwise 48 3 2 4938 4926 COST elementwise 48 3 3 5568 5388 COST elementwise 48 3 4 3942 3636 COST elementwise 48 3 5 2574 2532 COST elementwise 72 1 0 558 558 COST elementwise 72 1 1 558 558 COST elementwise 72 1 2 1902 1896 COST elementwise 72 1 3 2172 2112 COST elementwise 72 1 4 1494 1392 COST elementwise 72 1 5 1446 1404 COST elementwise 72 3 0 666 666 COST elementwise 72 3 1 1134 1134 COST elementwise 72 3 2 5442 5430 COST elementwise 72 3 3 6252 6072 COST elementwise 72 3 4 4230 3924 COST elementwise 72 3 5 2790 2748 COST elementwise 96 1 0 576 576 COST elementwise 96 1 1 576 576 COST elementwise 96 1 2 1983 1980 COST elementwise 96 1 3 2172 2172 COST elementwise 96 1 4 1374 1392 COST elementwise 96 1 5 1518 1476 COST elementwise 96 3 0 720 720 COST elementwise 96 3 1 1188 1188 COST elementwise 96 3 2 5685 5682 COST elementwise 96 3 3 6252 6252 COST elementwise 96 3 4 3870 3924 COST elementwise 96 3 5 3006 2964 COST elementwise 104 1 0 594 594 COST elementwise 104 1 1 594 594 COST elementwise 104 1 2 2097 2088 COST elementwise 104 1 3 2328 2316 COST elementwise 104 1 4 1416 1428 COST elementwise 104 1 5 1590 1548 COST elementwise 104 3 0 738 738 COST elementwise 104 3 1 1242 1242 COST elementwise 104 3 2 6027 6000 COST elementwise 104 3 3 6720 6696 COST elementwise 104 3 4 3996 4032 COST elementwise 104 3 5 3222 3180 COST elementwise 120 1 0 594 594 COST elementwise 120 1 1 594 594 COST elementwise 120 1 2 2097 2094 COST elementwise 120 1 3 2328 2328 COST elementwise 120 1 4 1416 1434 COST elementwise 120 1 5 1590 1548 COST elementwise 120 3 0 774 774 COST elementwise 120 3 1 1242 1242 COST elementwise 120 3 2 6027 6024 COST elementwise 120 3 3 6720 6720 COST elementwise 120 3 4 3996 4050 SOFTMAX softmax-fp8 1 1 930 924 COST elementwise 120 3 5 3222 3180 SOFTMAX softmax-fp8 2 1 930 924 COST elementwise 144 1 0 612 612 SOFTMAX softmax-fp8 5 1 930 924 COST elementwise 144 1 1 612 612 SOFTMAX softmax-fp8 6 1 930 924 COST elementwise 144 1 2 2199 2196 SOFTMAX softmax-fp8 7 1 1626 1608 COST elementwise 144 1 3 2454 2454 SOFTMAX softmax-fp8 8 1 1626 1608 COST elementwise 144 1 4 1446 1464 SOFTMAX softmax-fp8 12 1 1626 1608 COST elementwise 144 1 5 1662 1620 SOFTMAX softmax-fp8 17 1 2322 2292 COST elementwise 144 3 0 828 828 SOFTMAX softmax-fp8 1 2 942 930 COST elementwise 144 3 1 1296 1296 SOFTMAX softmax-fp8 2 2 942 930 COST elementwise 144 3 2 6333 6330 SOFTMAX softmax-fp8 5 2 942 930 COST elementwise 144 3 3 7098 7098 SOFTMAX softmax-fp8 6 2 942 930 COST elementwise 144 3 4 4086 4140 SOFTMAX softmax-fp8 7 2 1650 1620 COST elementwise 144 3 5 3438 3396 SOFTMAX softmax-fp8 8 2 1650 1626 COST elementwise 288 1 0 720 720 SOFTMAX softmax-fp8 12 2 1650 1626 COST elementwise 288 1 1 720 720 SOFTMAX softmax-fp8 17 2 2358 2322 COST elementwise 288 1 2 2847 2844 SOFTMAX softmax-fp8 1 15 1356 1350 COST elementwise 288 1 3 3300 3300 SOFTMAX softmax-fp8 2 15 1356 1392 COST elementwise 288 1 4 1662 1680 SOFTMAX softmax-fp8 5 15 1356 1392 COST elementwise 288 1 5 2094 2052 SOFTMAX softmax-fp8 6 15 1356 1392 COST elementwise 288 3 0 1152 1152 SOFTMAX softmax-fp8 7 15 2478 2460 COST elementwise 288 3 1 1620 1620 SOFTMAX softmax-fp8 8 15 2478 2550 COST elementwise 288 3 2 8277 8274 SOFTMAX softmax-fp8 12 15 2478 2550 COST elementwise 288 3 3 9636 9636 SOFTMAX softmax-fp8 17 15 3600 3708 COST elementwise 288 3 4 4734 4788 SOFTMAX softmax-fp8 1 16 1110 1140 COST elementwise 288 3 5 4734 4692 SOFTMAX softmax-fp8 2 16 1110 1188 COST elementwise 576 1 0 936 936 SOFTMAX softmax-fp8 5 16 1110 1188 COST elementwise 576 1 1 936 936 SOFTMAX softmax-fp8 6 16 1110 1188 COST elementwise 576 1 2 4143 4140 SOFTMAX softmax-fp8 7 16 1986 2040 COST elementwise 576 1 3 4992 4992 SOFTMAX softmax-fp8 8 16 1986 2142 COST elementwise 576 1 4 2094 2112 SOFTMAX softmax-fp8 12 16 1986 2142 SOFTMAX softmax-fp8 17 16 2862 3096 COST elementwise 576 1 5 2958 2916 SOFTMAX softmax-fp8 1 17 1170 1206 COST elementwise 576 3 0 1800 1800 SOFTMAX softmax-fp8 2 17 1170 1260 COST elementwise 576 3 1 2268 2268 SOFTMAX softmax-fp8 5 17 1170 1260 COST elementwise 576 3 2 12165 12162 SOFTMAX softmax-fp8 6 17 1170 1260 COST elementwise 576 3 3 14712 14712 SOFTMAX softmax-fp8 7 17 2106 2172 COST elementwise 576 3 4 6030 6084 SOFTMAX softmax-fp8 8 17 2106 2286 COST elementwise 576 3 5 7326 7284 SOFTMAX softmax-fp8 12 17 2106 2286 COST elementwise 1152 1 0 1368 1368 SOFTMAX softmax-fp8 17 17 3042 3312 COST elementwise 1152 1 1 1368 1368 SOFTMAX softmax-fp8 1 31 1596 1632 COST elementwise 1152 1 2 6735 6732 SOFTMAX softmax-fp8 2 31 1596 1728 COST elementwise 1152 1 3 8376 8376 SOFTMAX softmax-fp8 5 31 1596 1728 COST elementwise 1152 1 4 2958 2976 SOFTMAX softmax-fp8 6 31 1596 1728 COST elementwise 1152 1 5 4686 4644 SOFTMAX softmax-fp8 7 31 2958 3024 COST elementwise 1152 3 0 3096 3096 SOFTMAX softmax-fp8 8 31 2958 3222 COST elementwise 1152 3 1 3564 3564 SOFTMAX softmax-fp8 12 31 2958 3222 COST elementwise 1152 3 2 19941 19938 SOFTMAX softmax-fp8 17 31 4320 4716 SOFTMAX softmax-fp8 1 64 1644 1602 SOFTMAX softmax-fp8 2 64 1644 1794 SOFTMAX softmax-fp8 5 64 1644 1794 SOFTMAX softmax-fp8 6 64 1644 1794 COST elementwise 1152 3 3 24864 24864 COST elementwise 1152 3 4 8622 8676 COST elementwise 1152 3 5 12510 12468 SOFTMAX softmax-fp8 7 64 3066 2976 SOFTMAX softmax-fp8 8 64 3066 3360 SOFTMAX softmax-fp8 12 64 3066 3360 SOFTMAX softmax-fp8 17 64 4488 4926 SOFTMAX softmax-fp8 1 65 2142 2124 SOFTMAX softmax-fp8 2 65 2142 2340 SOFTMAX softmax-fp8 5 65 2142 2340 COST elementwise 1728 1 0 1800 1800 COST elementwise 1728 1 1 1800 1800 SOFTMAX softmax-fp8 6 65 2142 2340 SOFTMAX softmax-fp8 7 65 4050 4008 SOFTMAX softmax-fp8 8 65 4050 4446 SOFTMAX softmax-fp8 12 65 4050 4446 COST elementwise 1728 1 2 9327 9324 COST elementwise 1728 1 3 11760 11760 SOFTMAX softmax-fp8 17 65 5958 6552 SOFTMAX softmax-fp8 1 729 14304 14118 SOFTMAX softmax-fp8 2 729 14304 16572 SOFTMAX softmax-fp8 5 729 14304 16572 COST elementwise 1728 1 4 3822 3840 COST elementwise 1728 1 5 6414 6372 SOFTMAX softmax-fp8 6 729 14304 16572 SOFTMAX softmax-fp8 7 729 28374 27996 SOFTMAX softmax-fp8 8 729 28374 32910 SOFTMAX softmax-fp8 12 729 28374 32910 COST elementwise 1728 3 0 4392 4392 COST elementwise 1728 3 1 4860 4860 SOFTMAX softmax-fp8 17 729 42444 49248 SOFTMAX softmax-fp8 1 768 14316 14142 SOFTMAX softmax-fp8 2 768 14316 16446 SOFTMAX softmax-fp8 5 768 14316 16446 COST elementwise 1728 3 2 27717 27714 COST elementwise 1728 3 3 35016 35016 SOFTMAX softmax-fp8 6 768 14316 16446 SOFTMAX softmax-fp8 7 768 28410 28056 SOFTMAX softmax-fp8 8 768 28410 32664 SOFTMAX softmax-fp8 12 768 28410 32664 COST elementwise 1728 3 4 11214 11268 COST elementwise 1728 3 5 17694 17652 SOFTMAX softmax-fp8 17 768 42504 48882 COST elementwise 2152 1 0 2124 2124 COST elementwise 2152 1 1 2124 2124 COST elementwise 2152 1 2 11271 11268 COST elementwise 2152 1 3 14298 14298 COST elementwise 2152 1 4 4470 4488 COST elementwise 2152 1 5 7710 7668 COST elementwise 2152 3 0 5346 5346 COST elementwise 2152 3 1 5832 5832 COST elementwise 2152 3 2 33549 33546 COST elementwise 2152 3 3 42630 42630 COST elementwise 2152 3 4 13158 13212 COST elementwise 2152 3 5 21582 21540 COST residual 1 1 0 582 582 COST residual 1 1 1 582 582 COST residual 1 3 0 972 1098 SOFTMAX softmax-split 1 729 5208 5190 COST residual 1 3 1 1212 1362 SOFTMAX softmax-split 2 729 5208 6270 COST residual 2 1 0 558 546 SOFTMAX softmax-split 5 729 14556 16662 COST residual 2 1 1 558 546 SOFTMAX softmax-split 6 729 14556 16686 COST residual 2 1 2 1440 1428 SOFTMAX softmax-split 7 729 19356 21456 COST residual 2 1 3 1566 1488 SOFTMAX softmax-split 8 729 19356 22536 COST residual 2 1 4 1254 1110 SOFTMAX softmax-split 12 729 28704 32952 COST residual 2 1 5 996 1188 SOFTMAX softmax-split 17 729 42852 49194 COST residual 2 3 0 558 558 SOFTMAX softmax-split 1 768 5208 5190 COST residual 2 3 1 1068 1074 SOFTMAX softmax-split 2 768 5208 6270 COST residual 2 3 2 4056 4020 SOFTMAX softmax-split 5 768 14556 16662 COST residual 2 3 3 4434 4200 SOFTMAX softmax-split 6 768 14556 16686 COST residual 2 3 4 3510 3078 SOFTMAX softmax-split 7 768 19356 21456 COST residual 2 3 5 1872 2100 SOFTMAX softmax-split 8 768 19356 22536 COST residual 3 1 0 972 1098 SOFTMAX softmax-split 12 768 28704 32952 COST residual 3 1 1 972 1098 SOFTMAX softmax-split 17 768 42852 49194 COST residual 3 3 0 972 954 COST residual 3 3 1 1092 1074 COST residual 4 1 0 522 522 COST residual 4 1 1 522 522 COST residual 4 1 2 1518 1470 COST residual 4 1 3 1644 1536 COST residual 4 1 4 1254 1110 COST residual 4 1 5 1302 1260 COST residual 4 1 10 1386 1182 COST residual 4 1 11 1386 1182 COST residual 4 3 0 522 522 COST residual 4 3 1 1026 1026 COST residual 4 3 2 4290 4146 COST residual 4 3 3 4668 4344 COST residual 4 3 4 3510 3078 COST residual 4 3 5 2358 2316 COST residual 4 3 10 3906 3282 COST residual 4 3 11 3906 3282 COST residual 6 1 0 558 558 COST residual 6 1 1 558 558 COST residual 6 1 2 1440 1470 COST residual 6 1 3 1566 1530 COST residual 6 1 4 1254 1152 COST residual 6 1 5 996 1188 COST residual 6 3 0 594 588 COST residual 6 3 1 1068 1086 COST residual 6 3 2 4056 4146 COST residual 6 3 3 4434 4332 COST residual 6 3 4 3510 3204 COST residual 6 3 5 1872 2100 COST residual 7 1 0 972 954 COST residual 7 1 1 972 954 COST residual 7 3 0 1164 1218 COST residual 7 3 1 1284 1314 COST residual 12 1 0 522 522 COST residual 12 1 1 522 522 COST residual 12 1 2 1518 1512 COST residual 12 1 3 1644 1578 COST residual 12 1 4 1254 1152 COST residual 12 1 5 1302 1260 COST residual 12 1 10 1386 1350 COST residual 12 1 11 1386 1350 COST residual 12 3 0 540 540 COST residual 12 3 1 1026 1026 COST residual 12 3 2 4290 4278 COST residual 12 3 3 4668 4482 COST residual 12 3 4 3510 3204 COST residual 12 3 5 2358 2316 test kernel::attention::tests::report_softmax_costs ... COST residual 12 3 10 3906 3786 COST residual 12 3 11 3906 3786 COST residual 14 1 0 594 576 COST residual 14 1 1 594 576 COST residual 14 1 2 1548 1572 okCOST residual 14 1 3 1710 1668 COST residual 14 1 4 1302 1194 COST residual 14 1 5 1056 1248 COST residual 14 3 0 666 648 COST residual 14 3 1 1176 1164 COST residual 14 3 2 4380 4452 COST residual 14 3 3 4866 4740 COST residual 14 3 4 3654 3330 COST residual 14 3 5 2052 2280 COST residual 16 1 0 522 522 COST residual 16 1 1 522 522 COST residual 16 1 2 1566 1518 COST residual 16 1 3 1716 1614 COST residual 16 1 4 1302 1158 COST residual 16 1 5 1302 1260 COST residual 16 1 10 1434 1338 COST residual 16 1 11 1434 1338 COST residual 16 3 0 540 540 COST residual 16 3 1 1026 1026 COST residual 16 3 2 4434 4290 COST residual 16 3 3 4884 4578 COST residual 16 3 4 3654 3222 COST residual 16 3 5 2358 2316 COST residual 16 3 10 4050 3750 COST residual 16 3 11 4050 3750 COST residual 24 1 0 522 522 COST residual 24 1 1 522 522 COST residual 24 1 2 1566 1560 COST residual 24 1 3 1716 1656 COST residual 24 1 4 1302 1200 COST residual 24 1 5 1302 1260 COST residual 24 1 10 1434 1380 COST residual 24 1 11 1434 1380 COST residual 24 3 0 558 558 COST residual 24 3 1 1026 1026 COST residual 24 3 2 4434 4422 COST residual 24 3 3 4884 4704 COST residual 24 3 4 3654 3348 COST residual 24 3 5 2358 2316 COST residual 24 3 10 4050 3876 COST residual 24 3 11 4050 3876 COST residual 32 1 0 540 540 COST residual 32 1 1 540 540 COST residual 32 1 2 1686 1644 COST residual 32 1 3 1872 1770 COST residual 32 1 4 1350 1212 COST residual 32 1 5 1374 1332 COST residual 32 1 10 1482 1374 COST residual 32 1 11 1482 1374 COST residual 32 3 0 576 576 COST residual 32 3 1 1080 1080 COST residual 32 3 2 4794 4668 COST residual 32 3 3 5352 5058 COST residual 32 3 4 3798 3384 COST residual 32 3 5 2574 2532 COST residual 32 3 10 4194 3858 COST residual 32 3 11 4194 3858 COST residual 48 1 0 540 540 COST residual 48 1 1 540 540 COST residual 48 1 2 1734 1728 COST residual 48 1 3 1944 1884 COST residual 48 1 4 1398 1296 COST residual 48 1 5 1374 1332 COST residual 48 1 10 1530 1440 COST residual 48 1 11 1530 1440 COST residual 48 3 0 612 612 COST residual 48 3 1 1080 1080 COST residual 48 3 2 4938 4926 COST residual 48 3 3 5568 5388 COST residual 48 3 4 3942 3636 COST residual 48 3 5 2574 2532 COST residual 48 3 10 4338 4056 COST residual 48 3 11 4338 4056 COST residual 72 1 0 558 558 COST residual 72 1 1 558 558 COST residual 72 1 2 1902 1896 COST residual 72 1 3 2172 2112 COST residual 72 1 4 1494 1392 COST residual 72 1 5 1446 1404 COST residual 72 1 10 1650 1536 COST residual 72 1 11 1650 1536 COST residual 72 3 0 666 666 COST residual 72 3 1 1134 1134 COST residual 72 3 2 5442 5430 COST residual 72 3 3 6252 6072 COST residual 72 3 4 4230 3924 COST residual 72 3 5 2790 2748 COST residual 72 3 10 4698 4344 COST residual 72 3 11 4698 4344 COST residual 96 1 0 576 576 COST residual 96 1 1 576 576 COST residual 96 1 2 1983 1980 COST residual 96 1 3 2172 2172 COST residual 96 1 4 1374 1392 COST residual 96 1 5 1518 1476 COST residual 96 1 10 1530 1536 COST residual 96 1 11 1530 1536 COST residual 96 3 0 720 720 COST residual 96 3 1 1188 1188 COST residual 96 3 2 5685 5682 COST residual 96 3 3 6252 6252 COST residual 96 3 4 3870 3924 COST residual 96 3 5 3006 2964 COST residual 96 3 10 4338 4344 COST residual 96 3 11 4338 4344 COST residual 104 1 0 594 594 COST residual 104 1 1 594 594 COST residual 104 1 2 2097 2088 COST residual 104 1 3 2328 2316 COST residual 104 1 4 1416 1428 COST residual 104 1 5 1590 1548 COST residual 104 1 10 1596 1596 COST residual 104 1 11 1596 1596 COST residual 104 3 0 738 738 COST residual 104 3 1 1242 1242 COST residual 104 3 2 6027 6000 COST residual 104 3 3 6720 6696 COST residual 104 3 4 3996 4032 COST residual 104 3 5 3222 3180 COST residual 104 3 10 4536 4524 COST residual 104 3 11 4536 4524 COST residual 120 1 0 594 594 COST residual 120 1 1 594 594 COST residual 120 1 2 2097 2094 COST residual 120 1 3 2328 2328 COST residual 120 1 4 1416 1434 COST residual 120 1 5 1590 1548 COST residual 120 1 10 1596 1602 COST residual 120 1 11 1596 1602 COST residual 120 3 0 774 774 COST residual 120 3 1 1242 1242 COST residual 120 3 2 6027 6024 COST residual 120 3 3 6720 6720 COST residual 120 3 4 3996 4050 COST residual 120 3 5 3222 3180 COST residual 120 3 10 4536 4542 COST residual 120 3 11 4536 4542 COST residual 144 1 0 612 612 COST residual 144 1 1 612 612 COST residual 144 1 2 2199 2196 COST residual 144 1 3 2454 2454 COST residual 144 1 4 1446 1464 COST residual 144 1 5 1662 1620 COST residual 144 1 10 1626 1632 COST residual 144 1 11 1626 1632 COST residual 144 3 0 828 828 COST residual 144 3 1 1296 1296 COST residual 144 3 2 6333 6330 COST residual 144 3 3 7098 7098 COST residual 144 3 4 4086 4140 COST residual 144 3 5 3438 3396 COST residual 144 3 10 4626 4632 COST residual 144 3 11 4626 4632 COST residual 288 1 0 720 720 COST residual 288 1 1 720 720 COST residual 288 1 2 2847 2844 COST residual 288 1 3 3300 3300 COST residual 288 1 4 1662 1680 COST residual 288 1 5 2094 2052 COST residual 288 1 10 1914 1920 COST residual 288 1 11 1914 1920 COST residual 288 3 0 1152 1152 COST residual 288 3 1 1620 1620 COST residual 288 3 2 8277 8274 COST residual 288 3 3 9636 9636 COST residual 288 3 4 4734 4788 COST residual 288 3 5 4734 4692 COST residual 288 3 10 5490 5496 COST residual 288 3 11 5490 5496 COST residual 576 1 0 936 936 COST residual 576 1 1 936 936 COST residual 576 1 2 4143 4140 COST residual 576 1 3 4992 4992 COST residual 576 1 4 2094 2112 COST residual 576 1 5 2958 2916 COST residual 576 1 10 2490 2496 COST residual 576 1 11 2490 2496 COST residual 576 3 0 1800 1800 COST residual 576 3 1 2268 2268 COST residual 576 3 2 12165 12162 COST residual 576 3 3 14712 14712 COST residual 576 3 4 6030 6084 COST residual 576 3 5 7326 7284 COST residual 576 3 10 7218 7224 COST residual 576 3 11 7218 7224 COST residual 1152 1 0 1368 1368 COST residual 1152 1 1 1368 1368 COST residual 1152 1 2 6735 6732 COST residual 1152 1 3 8376 8376 COST residual 1152 1 4 2958 2976 COST residual 1152 1 5 4686 4644 COST residual 1152 1 10 3642 3648 COST residual 1152 1 11 3642 3648 COST residual 1152 3 0 3096 3096 COST residual 1152 3 1 3564 3564 COST residual 1152 3 2 19941 19938 COST residual 1152 3 3 24864 24864 COST residual 1152 3 4 8622 8676 COST residual 1152 3 5 12510 12468 COST residual 1152 3 10 10674 10680 COST residual 1152 3 11 10674 10680 COST residual 1728 1 0 1800 1800 COST residual 1728 1 1 1800 1800 COST residual 1728 1 2 9327 9324 COST residual 1728 1 3 11760 11760 COST residual 1728 1 4 3822 3840 COST residual 1728 1 5 6414 6372 COST residual 1728 1 10 4794 4800 COST residual 1728 1 11 4794 4800 COST residual 1728 3 0 4392 4392 COST residual 1728 3 1 4860 4860 COST residual 1728 3 2 27717 27714 COST residual 1728 3 3 35016 35016 COST residual 1728 3 4 11214 11268 COST residual 1728 3 5 17694 17652 COST residual 1728 3 10 14130 14136 COST residual 1728 3 11 14130 14136 COST residual 2152 1 0 2124 2124 COST residual 2152 1 1 2124 2124 COST residual 2152 1 2 11271 11268 COST residual 2152 1 3 14298 14298 COST residual 2152 1 4 4470 4488 COST residual 2152 1 5 7710 7668 COST residual 2152 1 10 5658 5664 COST residual 2152 1 11 5658 5664 COST residual 2152 3 0 5346 5346 COST residual 2152 3 1 5832 5832 COST residual 2152 3 2 33549 33546 COST residual 2152 3 3 42630 42630 COST residual 2152 3 4 13158 13212 COST residual 2152 3 5 21582 21540 COST residual 2152 3 10 16722 16728 COST residual 2152 3 11 16722 16728 COST elementwise-fp8 1 1 0 582 582 COST elementwise-fp8 1 1 1 582 582 COST elementwise-fp8 1 3 0 972 1098 COST elementwise-fp8 1 3 1 1212 1362 COST elementwise-fp8 2 1 0 558 546 COST elementwise-fp8 2 1 1 558 546 COST elementwise-fp8 2 1 2 1440 1428 COST elementwise-fp8 2 1 3 1566 1488 COST elementwise-fp8 2 1 4 1254 1110 COST elementwise-fp8 2 1 5 996 1188 COST elementwise-fp8 2 3 0 558 558 COST elementwise-fp8 2 3 1 1068 1074 COST elementwise-fp8 2 3 2 4056 4020 COST elementwise-fp8 2 3 3 4434 4200 COST elementwise-fp8 2 3 4 3510 3078 COST elementwise-fp8 2 3 5 1872 2100 COST elementwise-fp8 3 1 0 972 1098 COST elementwise-fp8 3 1 1 972 1098 COST elementwise-fp8 3 3 0 972 954 COST elementwise-fp8 3 3 1 1092 1074 COST elementwise-fp8 4 1 0 522 522 COST elementwise-fp8 4 1 1 522 522 COST elementwise-fp8 4 1 2 1518 1470 COST elementwise-fp8 4 1 3 1644 1536 COST elementwise-fp8 4 1 4 1254 1110 COST elementwise-fp8 4 1 5 1302 1260 COST elementwise-fp8 4 1 6 1716 1686 COST elementwise-fp8 4 1 7 1716 2028 COST elementwise-fp8 4 1 8 642 642 COST elementwise-fp8 4 1 9 972 972 COST elementwise-fp8 4 3 0 522 522 COST elementwise-fp8 4 3 1 1026 1026 COST elementwise-fp8 4 3 2 4290 4146 COST elementwise-fp8 4 3 3 4668 4344 COST elementwise-fp8 4 3 4 3510 3078 COST elementwise-fp8 4 3 5 2358 2316 COST elementwise-fp8 4 3 6 4884 4776 COST elementwise-fp8 4 3 7 5352 5748 COST elementwise-fp8 4 3 8 1326 1326 COST elementwise-fp8 4 3 9 2316 2316 COST elementwise-fp8 6 1 0 558 558 COST elementwise-fp8 6 1 1 558 558 COST elementwise-fp8 6 1 2 1440 1470 COST elementwise-fp8 6 1 3 1566 1530 COST elementwise-fp8 6 1 4 1254 1152 COST elementwise-fp8 6 1 5 996 1188 COST elementwise-fp8 6 3 0 594 588 COST elementwise-fp8 6 3 1 1068 1086 COST elementwise-fp8 6 3 2 4056 4146 COST elementwise-fp8 6 3 3 4434 4332 COST elementwise-fp8 6 3 4 3510 3204 COST elementwise-fp8 6 3 5 1872 2100 COST elementwise-fp8 7 1 0 972 954 COST elementwise-fp8 7 1 1 972 954 COST elementwise-fp8 7 3 0 1164 1218 COST elementwise-fp8 7 3 1 1284 1314 COST elementwise-fp8 12 1 0 522 522 COST elementwise-fp8 12 1 1 522 522 COST elementwise-fp8 12 1 2 1518 1512 COST elementwise-fp8 12 1 3 1644 1578 COST elementwise-fp8 12 1 4 1254 1152 COST elementwise-fp8 12 1 5 1302 1260 COST elementwise-fp8 12 1 6 1716 1728 COST elementwise-fp8 12 1 7 1716 2016 COST elementwise-fp8 12 1 8 954 954 COST elementwise-fp8 12 1 9 1200 1200 COST elementwise-fp8 12 3 0 540 540 COST elementwise-fp8 12 3 1 1026 1026 COST elementwise-fp8 12 3 2 4290 4278 COST elementwise-fp8 12 3 3 4668 4482 COST elementwise-fp8 12 3 4 3510 3204 COST elementwise-fp8 12 3 5 2358 2316 COST elementwise-fp8 12 3 6 4884 4908 COST elementwise-fp8 12 3 7 5352 5718 COST elementwise-fp8 12 3 8 2262 2262 COST elementwise-fp8 12 3 9 3000 3000 COST elementwise-fp8 14 1 0 594 576 COST elementwise-fp8 14 1 1 594 576 COST elementwise-fp8 14 1 2 1548 1572 COST elementwise-fp8 14 1 3 1710 1668 COST elementwise-fp8 14 1 4 1302 1194 COST elementwise-fp8 14 1 5 1056 1248 COST elementwise-fp8 14 3 0 666 648 COST elementwise-fp8 14 3 1 1176 1164 COST elementwise-fp8 14 3 2 4380 4452 COST elementwise-fp8 14 3 3 4866 4740 COST elementwise-fp8 14 3 4 3654 3330 COST elementwise-fp8 14 3 5 2052 2280 COST elementwise-fp8 16 1 0 522 522 COST elementwise-fp8 16 1 1 522 522 COST elementwise-fp8 16 1 2 1566 1518 COST elementwise-fp8 16 1 3 1716 1614 COST elementwise-fp8 16 1 4 1302 1158 COST elementwise-fp8 16 1 5 1302 1260 COST elementwise-fp8 16 1 6 1824 1734 COST elementwise-fp8 16 1 7 1824 2022 COST elementwise-fp8 16 1 8 1110 1110 COST elementwise-fp8 16 1 9 1314 1314 COST elementwise-fp8 16 3 0 540 540 COST elementwise-fp8 16 3 1 1026 1026 COST elementwise-fp8 16 3 2 4434 4290 COST elementwise-fp8 16 3 3 4884 4578 COST elementwise-fp8 16 3 4 3654 3222 COST elementwise-fp8 16 3 5 2358 2316 COST elementwise-fp8 16 3 6 5208 4920 COST elementwise-fp8 16 3 7 5676 5730 COST elementwise-fp8 16 3 8 2730 2730 COST elementwise-fp8 16 3 9 3342 3342 COST elementwise-fp8 24 1 0 522 522 COST elementwise-fp8 24 1 1 522 522 COST elementwise-fp8 24 1 2 1566 1560 COST elementwise-fp8 24 1 3 1716 1656 COST elementwise-fp8 24 1 4 1302 1200 COST elementwise-fp8 24 1 5 1302 1260 COST elementwise-fp8 24 1 6 1824 1776 COST elementwise-fp8 24 1 7 1824 2064 COST elementwise-fp8 24 1 8 1422 1422 COST elementwise-fp8 24 1 9 1542 1542 COST elementwise-fp8 24 3 0 558 558 COST elementwise-fp8 24 3 1 1026 1026 COST elementwise-fp8 24 3 2 4434 4422 COST elementwise-fp8 24 3 3 4884 4704 COST elementwise-fp8 24 3 4 3654 3348 COST elementwise-fp8 24 3 5 2358 2316 COST elementwise-fp8 24 3 6 5208 5058 COST elementwise-fp8 24 3 7 5676 5856 COST elementwise-fp8 24 3 8 3666 3666 COST elementwise-fp8 24 3 9 4026 4026 COST elementwise-fp8 32 1 0 540 540 COST elementwise-fp8 32 1 1 540 540 COST elementwise-fp8 32 1 2 1686 1644 COST elementwise-fp8 32 1 3 1872 1770 COST elementwise-fp8 32 1 4 1350 1212 COST elementwise-fp8 32 1 5 1374 1332 COST elementwise-fp8 32 1 6 1938 1914 COST elementwise-fp8 32 1 7 1938 1956 COST elementwise-fp8 32 1 8 1488 1488 COST elementwise-fp8 32 1 9 1500 1500 COST elementwise-fp8 32 3 0 576 576 COST elementwise-fp8 32 3 1 1080 1080 COST elementwise-fp8 32 3 2 4794 4668 COST elementwise-fp8 32 3 3 5352 5058 COST elementwise-fp8 32 3 4 3798 3384 COST elementwise-fp8 32 3 5 2574 2532 COST elementwise-fp8 32 3 6 5550 5460 COST elementwise-fp8 32 3 7 6486 6000 COST elementwise-fp8 32 3 8 3864 3864 COST elementwise-fp8 32 3 9 3900 3900 COST elementwise-fp8 48 1 0 540 540 COST elementwise-fp8 48 1 1 540 540 COST elementwise-fp8 48 1 2 1734 1728 COST elementwise-fp8 48 1 3 1944 1884 COST elementwise-fp8 48 1 4 1398 1296 COST elementwise-fp8 48 1 5 1374 1332 COST elementwise-fp8 48 1 6 2046 1998 COST elementwise-fp8 48 1 7 2046 2286 COST elementwise-fp8 48 1 8 1488 1488 COST elementwise-fp8 48 1 9 1500 1500 COST elementwise-fp8 48 3 0 612 612 COST elementwise-fp8 48 3 1 1080 1080 COST elementwise-fp8 48 3 2 4938 4926 COST elementwise-fp8 48 3 3 5568 5388 COST elementwise-fp8 48 3 4 3942 3636 COST elementwise-fp8 48 3 5 2574 2532 COST elementwise-fp8 48 3 6 5874 5724 COST elementwise-fp8 48 3 7 6810 6996 COST elementwise-fp8 48 3 8 3864 3864 COST elementwise-fp8 48 3 9 3900 3900 COST elementwise-fp8 72 1 0 558 558 COST elementwise-fp8 72 1 1 558 558 COST elementwise-fp8 72 1 2 1902 1896 COST elementwise-fp8 72 1 3 2172 2112 COST elementwise-fp8 72 1 4 1494 1392 COST elementwise-fp8 72 1 5 1446 1404 COST elementwise-fp8 72 1 6 2268 2220 COST elementwise-fp8 72 1 7 2268 2508 COST elementwise-fp8 72 1 8 1488 1488 COST elementwise-fp8 72 1 9 1500 1500 COST elementwise-fp8 72 3 0 666 666 COST elementwise-fp8 72 3 1 1134 1134 COST elementwise-fp8 72 3 2 5442 5430 COST elementwise-fp8 72 3 3 6252 6072 COST elementwise-fp8 72 3 4 4230 3924 COST elementwise-fp8 72 3 5 2790 2748 COST elementwise-fp8 72 3 6 6540 6390 COST elementwise-fp8 72 3 7 7944 8136 COST elementwise-fp8 72 3 8 3864 3864 COST elementwise-fp8 72 3 9 3900 3900 COST elementwise-fp8 96 1 0 576 576 COST elementwise-fp8 96 1 1 576 576 COST elementwise-fp8 96 1 2 1983 1980 COST elementwise-fp8 96 1 3 2172 2172 COST elementwise-fp8 96 1 4 1374 1392 COST elementwise-fp8 96 1 5 1518 1476 COST elementwise-fp8 96 1 6 2403 2358 COST elementwise-fp8 96 1 7 2403 2400 COST elementwise-fp8 96 1 8 1488 1488 COST elementwise-fp8 96 1 9 1500 1500 COST elementwise-fp8 96 3 0 720 720 COST elementwise-fp8 96 3 1 1188 1188 COST elementwise-fp8 96 3 2 5685 5682 COST elementwise-fp8 96 3 3 6252 6252 COST elementwise-fp8 96 3 4 3870 3924 COST elementwise-fp8 96 3 5 3006 2964 COST elementwise-fp8 96 3 6 6945 6804 COST elementwise-fp8 96 3 7 8817 8280 COST elementwise-fp8 96 3 8 3864 3864 COST elementwise-fp8 96 3 9 3900 3900 COST elementwise-fp8 104 1 0 594 594 COST elementwise-fp8 104 1 1 594 594 COST elementwise-fp8 104 1 2 2097 2088 COST elementwise-fp8 104 1 3 2328 2316 COST elementwise-fp8 104 1 4 1416 1428 COST elementwise-fp8 104 1 5 1590 1548 COST elementwise-fp8 104 1 6 2511 2520 COST elementwise-fp8 104 1 7 2511 2808 COST elementwise-fp8 104 1 8 1488 1488 COST elementwise-fp8 104 1 9 1500 1500 COST elementwise-fp8 104 3 0 738 738 COST elementwise-fp8 104 3 1 1242 1242 COST elementwise-fp8 104 3 2 6027 6000 COST elementwise-fp8 104 3 3 6720 6696 COST elementwise-fp8 104 3 4 3996 4032 COST elementwise-fp8 104 3 5 3222 3180 COST elementwise-fp8 104 3 6 7269 7278 COST elementwise-fp8 104 3 7 9609 9960 COST elementwise-fp8 104 3 8 3864 3864 COST elementwise-fp8 104 3 9 3900 3900 COST elementwise-fp8 120 1 0 594 594 COST elementwise-fp8 120 1 1 594 594 COST elementwise-fp8 120 1 2 2097 2094 COST elementwise-fp8 120 1 3 2328 2328 COST elementwise-fp8 120 1 4 1416 1434 COST elementwise-fp8 120 1 5 1590 1548 COST elementwise-fp8 120 1 6 2571 2526 COST elementwise-fp8 120 1 7 2571 2814 COST elementwise-fp8 120 1 8 1488 1488 COST elementwise-fp8 120 1 9 1542 1536 COST elementwise-fp8 120 3 0 774 774 COST elementwise-fp8 120 3 1 1242 1242 COST elementwise-fp8 120 3 2 6027 6024 COST elementwise-fp8 120 3 3 6720 6720 COST elementwise-fp8 120 3 4 3996 4050 COST elementwise-fp8 120 3 5 3222 3180 COST elementwise-fp8 120 3 6 7449 7302 COST elementwise-fp8 120 3 7 9789 9978 COST elementwise-fp8 120 3 8 3864 3864 COST elementwise-fp8 120 3 9 4026 4020 COST elementwise-fp8 144 1 0 612 612 COST elementwise-fp8 144 1 1 612 612 COST elementwise-fp8 144 1 2 2199 2196 COST elementwise-fp8 144 1 3 2454 2454 COST elementwise-fp8 144 1 4 1446 1464 COST elementwise-fp8 144 1 5 1662 1620 COST elementwise-fp8 144 1 6 2727 2682 COST elementwise-fp8 144 1 7 2727 2970 COST elementwise-fp8 144 1 8 1488 1488 COST elementwise-fp8 144 1 9 1860 1860 COST elementwise-fp8 144 3 0 828 828 COST elementwise-fp8 144 3 1 1296 1296 COST elementwise-fp8 144 3 2 6333 6330 COST elementwise-fp8 144 3 3 7098 7098 COST elementwise-fp8 144 3 4 4086 4140 COST elementwise-fp8 144 3 5 3438 3396 COST elementwise-fp8 144 3 6 7917 7776 COST elementwise-fp8 144 3 7 10725 10920 COST elementwise-fp8 144 3 8 3864 3864 COST elementwise-fp8 144 3 9 4980 4980 COST elementwise-fp8 288 1 0 720 720 COST elementwise-fp8 288 1 1 720 720 COST elementwise-fp8 288 1 2 2847 2844 COST elementwise-fp8 288 1 3 3300 3300 COST elementwise-fp8 288 1 4 1662 1680 COST elementwise-fp8 288 1 5 2094 2052 COST elementwise-fp8 288 1 6 3699 3654 COST elementwise-fp8 288 1 7 3699 3696 COST elementwise-fp8 288 1 8 2490 2490 COST elementwise-fp8 288 1 9 2502 2502 COST elementwise-fp8 288 3 0 1152 1152 COST elementwise-fp8 288 3 1 1620 1620 COST elementwise-fp8 288 3 2 8277 8274 COST elementwise-fp8 288 3 3 9636 9636 COST elementwise-fp8 288 3 4 4734 4788 COST elementwise-fp8 288 3 5 4734 4692 COST elementwise-fp8 288 3 6 10833 10692 COST elementwise-fp8 288 3 7 16449 15912 COST elementwise-fp8 288 3 8 6870 6870 COST elementwise-fp8 288 3 9 6906 6906 COST elementwise-fp8 576 1 0 936 936 COST elementwise-fp8 576 1 1 936 936 COST elementwise-fp8 576 1 2 4143 4140 COST elementwise-fp8 576 1 3 4992 4992 COST elementwise-fp8 576 1 4 2094 2112 COST elementwise-fp8 576 1 5 2958 2916 COST elementwise-fp8 576 1 6 5643 5598 COST elementwise-fp8 576 1 7 5643 5640 COST elementwise-fp8 576 1 8 3492 3492 COST elementwise-fp8 576 1 9 3864 3864 COST elementwise-fp8 576 3 0 1800 1800 COST elementwise-fp8 576 3 1 2268 2268 COST elementwise-fp8 576 3 2 12165 12162 COST elementwise-fp8 576 3 3 14712 14712 COST elementwise-fp8 576 3 4 6030 6084 COST elementwise-fp8 576 3 5 7326 7284 COST elementwise-fp8 576 3 6 16665 16524 COST elementwise-fp8 576 3 7 27897 27360 COST elementwise-fp8 576 3 8 9876 9876 COST elementwise-fp8 576 3 9 10992 10992 COST elementwise-fp8 1152 1 0 1368 1368 COST elementwise-fp8 1152 1 1 1368 1368 COST elementwise-fp8 1152 1 2 6735 6732 COST elementwise-fp8 1152 1 3 8376 8376 COST elementwise-fp8 1152 1 4 2958 2976 COST elementwise-fp8 1152 1 5 4686 4644 COST elementwise-fp8 1152 1 6 9531 9486 COST elementwise-fp8 1152 1 7 9531 9528 COST elementwise-fp8 1152 1 8 6498 6498 COST elementwise-fp8 1152 1 9 6870 6870 COST elementwise-fp8 1152 3 0 3096 3096 COST elementwise-fp8 1152 3 1 3564 3564 COST elementwise-fp8 1152 3 2 19941 19938 COST elementwise-fp8 1152 3 3 24864 24864 COST elementwise-fp8 1152 3 4 8622 8676 COST elementwise-fp8 1152 3 5 12510 12468 COST elementwise-fp8 1152 3 6 28329 28188 COST elementwise-fp8 1152 3 7 50793 50256 COST elementwise-fp8 1152 3 8 18894 18894 COST elementwise-fp8 1152 3 9 20010 20010 COST elementwise-fp8 1728 1 0 1800 1800 COST elementwise-fp8 1728 1 1 1800 1800 COST elementwise-fp8 1728 1 2 9327 9324 COST elementwise-fp8 1728 1 3 11760 11760 COST elementwise-fp8 1728 1 4 3822 3840 COST elementwise-fp8 1728 1 5 6414 6372 COST elementwise-fp8 1728 1 6 13419 13374 COST elementwise-fp8 1728 1 7 13419 13416 COST elementwise-fp8 1728 1 8 9504 9504 COST elementwise-fp8 1728 1 9 9876 9876 COST elementwise-fp8 1728 3 0 4392 4392 COST elementwise-fp8 1728 3 1 4860 4860 COST elementwise-fp8 1728 3 2 27717 27714 COST elementwise-fp8 1728 3 3 35016 35016 COST elementwise-fp8 1728 3 4 11214 11268 COST elementwise-fp8 1728 3 5 17694 17652 COST elementwise-fp8 1728 3 6 39993 39852 COST elementwise-fp8 1728 3 7 73689 73152 COST elementwise-fp8 1728 3 8 27912 27912 COST elementwise-fp8 1728 3 9 29028 29028 COST elementwise-fp8 2152 1 0 2124 2124 COST elementwise-fp8 2152 1 1 2124 2124 COST elementwise-fp8 2152 1 2 11271 11268 COST elementwise-fp8 2152 1 3 14298 14298 COST elementwise-fp8 2152 1 4 4470 4488 COST elementwise-fp8 2152 1 5 7710 7668 COST elementwise-fp8 2152 1 6 16335 16290 COST elementwise-fp8 2152 1 7 16335 16578 COST elementwise-fp8 2152 1 8 12510 12510 COST elementwise-fp8 2152 1 9 12522 12522 COST elementwise-fp8 2152 3 0 5346 5346 COST elementwise-fp8 2152 3 1 5832 5832 COST elementwise-fp8 2152 3 2 33549 33546 COST elementwise-fp8 2152 3 3 42630 42630 COST elementwise-fp8 2152 3 4 13158 13212 COST elementwise-fp8 2152 3 5 21582 21540 COST elementwise-fp8 2152 3 6 48741 48594 COST elementwise-fp8 2152 3 7 90861 91056 COST elementwise-fp8 2152 3 8 36930 36930 COST elementwise-fp8 2152 3 9 36966 36966 COST bias-gelu-fp8 1 1 0 582 582 COST bias-gelu-fp8 1 1 1 582 582 COST bias-gelu-fp8 1 3 0 972 1098 COST bias-gelu-fp8 1 3 1 1212 1362 COST bias-gelu-fp8 2 1 0 558 546 COST bias-gelu-fp8 2 1 1 558 546 COST bias-gelu-fp8 2 1 2 1440 1428 COST bias-gelu-fp8 2 1 3 1566 1488 COST bias-gelu-fp8 2 1 4 1254 1110 COST bias-gelu-fp8 2 1 5 996 1188 COST bias-gelu-fp8 2 3 0 558 558 COST bias-gelu-fp8 2 3 1 1068 1074 COST bias-gelu-fp8 2 3 2 4056 4020 COST bias-gelu-fp8 2 3 3 4434 4200 COST bias-gelu-fp8 2 3 4 3510 3078 COST bias-gelu-fp8 2 3 5 1872 2100 COST bias-gelu-fp8 3 1 0 972 1098 COST bias-gelu-fp8 3 1 1 972 1098 COST bias-gelu-fp8 3 3 0 972 954 COST bias-gelu-fp8 3 3 1 1092 1074 COST bias-gelu-fp8 4 1 0 522 522 COST bias-gelu-fp8 4 1 1 522 522 COST bias-gelu-fp8 4 1 2 1518 1470 COST bias-gelu-fp8 4 1 3 1644 1536 COST bias-gelu-fp8 4 1 4 1254 1110 COST bias-gelu-fp8 4 1 5 1302 1260 COST bias-gelu-fp8 4 1 6 1716 1686 COST bias-gelu-fp8 4 1 7 1716 2028 COST bias-gelu-fp8 4 1 8 702 702 COST bias-gelu-fp8 4 1 9 1032 1032 COST bias-gelu-fp8 4 3 0 522 522 COST bias-gelu-fp8 4 3 1 1026 1026 COST bias-gelu-fp8 4 3 2 4290 4146 COST bias-gelu-fp8 4 3 3 4668 4344 COST bias-gelu-fp8 4 3 4 3510 3078 COST bias-gelu-fp8 4 3 5 2358 2316 COST bias-gelu-fp8 4 3 6 4884 4776 COST bias-gelu-fp8 4 3 7 5352 5748 COST bias-gelu-fp8 4 3 8 1446 1446 COST bias-gelu-fp8 4 3 9 2436 2436 COST bias-gelu-fp8 6 1 0 558 558 COST bias-gelu-fp8 6 1 1 558 558 COST bias-gelu-fp8 6 1 2 1440 1470 COST bias-gelu-fp8 6 1 3 1566 1530 COST bias-gelu-fp8 6 1 4 1254 1152 COST bias-gelu-fp8 6 1 5 996 1188 COST bias-gelu-fp8 6 3 0 594 588 COST bias-gelu-fp8 6 3 1 1068 1086 COST bias-gelu-fp8 6 3 2 4056 4146 COST bias-gelu-fp8 6 3 3 4434 4332 COST bias-gelu-fp8 6 3 4 3510 3204 COST bias-gelu-fp8 6 3 5 1872 2100 COST bias-gelu-fp8 7 1 0 972 954 COST bias-gelu-fp8 7 1 1 972 954 COST bias-gelu-fp8 7 3 0 1164 1218 COST bias-gelu-fp8 7 3 1 1284 1314 COST bias-gelu-fp8 12 1 0 522 522 COST bias-gelu-fp8 12 1 1 522 522 COST bias-gelu-fp8 12 1 2 1518 1512 COST bias-gelu-fp8 12 1 3 1644 1578 COST bias-gelu-fp8 12 1 4 1254 1152 COST bias-gelu-fp8 12 1 5 1302 1260 COST bias-gelu-fp8 12 1 6 1716 1728 COST bias-gelu-fp8 12 1 7 1716 2016 COST bias-gelu-fp8 12 1 8 1050 1050 COST bias-gelu-fp8 12 1 9 1296 1296 COST bias-gelu-fp8 12 3 0 540 540 COST bias-gelu-fp8 12 3 1 1026 1026 COST bias-gelu-fp8 12 3 2 4290 4278 COST bias-gelu-fp8 12 3 3 4668 4482 COST bias-gelu-fp8 12 3 4 3510 3204 COST bias-gelu-fp8 12 3 5 2358 2316 COST bias-gelu-fp8 12 3 6 4884 4908 COST bias-gelu-fp8 12 3 7 5352 5718 COST bias-gelu-fp8 12 3 8 2490 2490 COST bias-gelu-fp8 12 3 9 3228 3228 COST bias-gelu-fp8 14 1 0 594 576 COST bias-gelu-fp8 14 1 1 594 576 COST bias-gelu-fp8 14 1 2 1548 1572 COST bias-gelu-fp8 14 1 3 1710 1668 COST bias-gelu-fp8 14 1 4 1302 1194 COST bias-gelu-fp8 14 1 5 1056 1248 COST bias-gelu-fp8 14 3 0 666 648 COST bias-gelu-fp8 14 3 1 1176 1164 COST bias-gelu-fp8 14 3 2 4380 4452 COST bias-gelu-fp8 14 3 3 4866 4740 COST bias-gelu-fp8 14 3 4 3654 3330 COST bias-gelu-fp8 14 3 5 2052 2280 COST bias-gelu-fp8 16 1 0 522 522 COST bias-gelu-fp8 16 1 1 522 522 COST bias-gelu-fp8 16 1 2 1566 1518 COST bias-gelu-fp8 16 1 3 1716 1614 COST bias-gelu-fp8 16 1 4 1302 1158 COST bias-gelu-fp8 16 1 5 1302 1260 COST bias-gelu-fp8 16 1 6 1824 1734 COST bias-gelu-fp8 16 1 7 1824 2022 COST bias-gelu-fp8 16 1 8 1224 1224 COST bias-gelu-fp8 16 1 9 1428 1428 COST bias-gelu-fp8 16 3 0 540 540 COST bias-gelu-fp8 16 3 1 1026 1026 COST bias-gelu-fp8 16 3 2 4434 4290 COST bias-gelu-fp8 16 3 3 4884 4578 COST bias-gelu-fp8 16 3 4 3654 3222 COST bias-gelu-fp8 16 3 5 2358 2316 COST bias-gelu-fp8 16 3 6 5208 4920 COST bias-gelu-fp8 16 3 7 5676 5730 COST bias-gelu-fp8 16 3 8 3012 3012 COST bias-gelu-fp8 16 3 9 3624 3624 COST bias-gelu-fp8 24 1 0 522 522 COST bias-gelu-fp8 24 1 1 522 522 COST bias-gelu-fp8 24 1 2 1566 1560 COST bias-gelu-fp8 24 1 3 1716 1656 COST bias-gelu-fp8 24 1 4 1302 1200 COST bias-gelu-fp8 24 1 5 1302 1260 COST bias-gelu-fp8 24 1 6 1824 1776 COST bias-gelu-fp8 24 1 7 1824 2064 COST bias-gelu-fp8 24 1 8 1572 1572 COST bias-gelu-fp8 24 1 9 1692 1692 COST bias-gelu-fp8 24 3 0 558 558 COST bias-gelu-fp8 24 3 1 1026 1026 COST bias-gelu-fp8 24 3 2 4434 4422 COST bias-gelu-fp8 24 3 3 4884 4704 COST bias-gelu-fp8 24 3 4 3654 3348 COST bias-gelu-fp8 24 3 5 2358 2316 COST bias-gelu-fp8 24 3 6 5208 5058 COST bias-gelu-fp8 24 3 7 5676 5856 COST bias-gelu-fp8 24 3 8 4056 4056 COST bias-gelu-fp8 24 3 9 4416 4416 COST bias-gelu-fp8 32 1 0 540 540 COST bias-gelu-fp8 32 1 1 540 540 COST bias-gelu-fp8 32 1 2 1686 1644 COST bias-gelu-fp8 32 1 3 1872 1770 COST bias-gelu-fp8 32 1 4 1350 1212 COST bias-gelu-fp8 32 1 5 1374 1332 COST bias-gelu-fp8 32 1 6 1938 1914 COST bias-gelu-fp8 32 1 7 1938 1956 COST bias-gelu-fp8 32 1 8 1632 1632 COST bias-gelu-fp8 32 1 9 1644 1644 COST bias-gelu-fp8 32 3 0 576 576 COST bias-gelu-fp8 32 3 1 1080 1080 COST bias-gelu-fp8 32 3 2 4794 4668 COST bias-gelu-fp8 32 3 3 5352 5058 COST bias-gelu-fp8 32 3 4 3798 3384 COST bias-gelu-fp8 32 3 5 2574 2532 COST bias-gelu-fp8 32 3 6 5550 5460 COST bias-gelu-fp8 32 3 7 6486 6000 COST bias-gelu-fp8 32 3 8 4236 4236 COST bias-gelu-fp8 32 3 9 4272 4272 COST bias-gelu-fp8 48 1 0 540 540 COST bias-gelu-fp8 48 1 1 540 540 COST bias-gelu-fp8 48 1 2 1734 1728 COST bias-gelu-fp8 48 1 3 1944 1884 COST bias-gelu-fp8 48 1 4 1398 1296 COST bias-gelu-fp8 48 1 5 1374 1332 COST bias-gelu-fp8 48 1 6 2046 1998 COST bias-gelu-fp8 48 1 7 2046 2286 COST bias-gelu-fp8 48 1 8 1632 1632 COST bias-gelu-fp8 48 1 9 1644 1644 COST bias-gelu-fp8 48 3 0 612 612 COST bias-gelu-fp8 48 3 1 1080 1080 COST bias-gelu-fp8 48 3 2 4938 4926 COST bias-gelu-fp8 48 3 3 5568 5388 COST bias-gelu-fp8 48 3 4 3942 3636 COST bias-gelu-fp8 48 3 5 2574 2532 COST bias-gelu-fp8 48 3 6 5874 5724 COST bias-gelu-fp8 48 3 7 6810 6996 COST bias-gelu-fp8 48 3 8 4236 4236 COST bias-gelu-fp8 48 3 9 4272 4272 COST bias-gelu-fp8 72 1 0 558 558 COST bias-gelu-fp8 72 1 1 558 558 COST bias-gelu-fp8 72 1 2 1902 1896 COST bias-gelu-fp8 72 1 3 2172 2112 COST bias-gelu-fp8 72 1 4 1494 1392 COST bias-gelu-fp8 72 1 5 1446 1404 COST bias-gelu-fp8 72 1 6 2268 2220 COST bias-gelu-fp8 72 1 7 2268 2508 COST bias-gelu-fp8 72 1 8 1632 1632 COST bias-gelu-fp8 72 1 9 1644 1644 COST bias-gelu-fp8 72 3 0 666 666 COST bias-gelu-fp8 72 3 1 1134 1134 COST bias-gelu-fp8 72 3 2 5442 5430 COST bias-gelu-fp8 72 3 3 6252 6072 COST bias-gelu-fp8 72 3 4 4230 3924 COST bias-gelu-fp8 72 3 5 2790 2748 COST bias-gelu-fp8 72 3 6 6540 6390 COST bias-gelu-fp8 72 3 7 7944 8136 COST bias-gelu-fp8 72 3 8 4236 4236 COST bias-gelu-fp8 72 3 9 4272 4272 COST bias-gelu-fp8 96 1 0 576 576 COST bias-gelu-fp8 96 1 1 576 576 COST bias-gelu-fp8 96 1 2 1983 1980 COST bias-gelu-fp8 96 1 3 2172 2172 COST bias-gelu-fp8 96 1 4 1374 1392 COST bias-gelu-fp8 96 1 5 1518 1476 COST bias-gelu-fp8 96 1 6 2403 2358 COST bias-gelu-fp8 96 1 7 2403 2400 COST bias-gelu-fp8 96 1 8 1632 1632 COST bias-gelu-fp8 96 1 9 1644 1644 COST bias-gelu-fp8 96 3 0 720 720 COST bias-gelu-fp8 96 3 1 1188 1188 COST bias-gelu-fp8 96 3 2 5685 5682 COST bias-gelu-fp8 96 3 3 6252 6252 COST bias-gelu-fp8 96 3 4 3870 3924 COST bias-gelu-fp8 96 3 5 3006 2964 COST bias-gelu-fp8 96 3 6 6945 6804 COST bias-gelu-fp8 96 3 7 8817 8280 COST bias-gelu-fp8 96 3 8 4236 4236 COST bias-gelu-fp8 96 3 9 4272 4272 COST bias-gelu-fp8 104 1 0 594 594 COST bias-gelu-fp8 104 1 1 594 594 COST bias-gelu-fp8 104 1 2 2097 2088 COST bias-gelu-fp8 104 1 3 2328 2316 COST bias-gelu-fp8 104 1 4 1416 1428 COST bias-gelu-fp8 104 1 5 1590 1548 COST bias-gelu-fp8 104 1 6 2511 2520 COST bias-gelu-fp8 104 1 7 2511 2808 COST bias-gelu-fp8 104 1 8 1632 1632 COST bias-gelu-fp8 104 1 9 1644 1644 COST bias-gelu-fp8 104 3 0 738 738 COST bias-gelu-fp8 104 3 1 1242 1242 COST bias-gelu-fp8 104 3 2 6027 6000 COST bias-gelu-fp8 104 3 3 6720 6696 COST bias-gelu-fp8 104 3 4 3996 4032 COST bias-gelu-fp8 104 3 5 3222 3180 COST bias-gelu-fp8 104 3 6 7269 7278 COST bias-gelu-fp8 104 3 7 9609 9960 COST bias-gelu-fp8 104 3 8 4236 4236 COST bias-gelu-fp8 104 3 9 4272 4272 COST bias-gelu-fp8 120 1 0 594 594 COST bias-gelu-fp8 120 1 1 594 594 COST bias-gelu-fp8 120 1 2 2097 2094 COST bias-gelu-fp8 120 1 3 2328 2328 COST bias-gelu-fp8 120 1 4 1416 1434 COST bias-gelu-fp8 120 1 5 1590 1548 COST bias-gelu-fp8 120 1 6 2571 2526 COST bias-gelu-fp8 120 1 7 2571 2814 COST bias-gelu-fp8 120 1 8 1632 1632 COST bias-gelu-fp8 120 1 9 1692 1686 COST bias-gelu-fp8 120 3 0 774 774 COST bias-gelu-fp8 120 3 1 1242 1242 COST bias-gelu-fp8 120 3 2 6027 6024 COST bias-gelu-fp8 120 3 3 6720 6720 COST bias-gelu-fp8 120 3 4 3996 4050 COST bias-gelu-fp8 120 3 5 3222 3180 COST bias-gelu-fp8 120 3 6 7449 7302 COST bias-gelu-fp8 120 3 7 9789 9978 COST bias-gelu-fp8 120 3 8 4236 4236 COST bias-gelu-fp8 120 3 9 4416 4410 COST bias-gelu-fp8 144 1 0 612 612 COST bias-gelu-fp8 144 1 1 612 612 COST bias-gelu-fp8 144 1 2 2199 2196 COST bias-gelu-fp8 144 1 3 2454 2454 COST bias-gelu-fp8 144 1 4 1446 1464 COST bias-gelu-fp8 144 1 5 1662 1620 COST bias-gelu-fp8 144 1 6 2727 2682 COST bias-gelu-fp8 144 1 7 2727 2970 COST bias-gelu-fp8 144 1 8 1632 1632 COST bias-gelu-fp8 144 1 9 2004 2004 COST bias-gelu-fp8 144 3 0 828 828 COST bias-gelu-fp8 144 3 1 1296 1296 COST bias-gelu-fp8 144 3 2 6333 6330 COST bias-gelu-fp8 144 3 3 7098 7098 COST bias-gelu-fp8 144 3 4 4086 4140 COST bias-gelu-fp8 144 3 5 3438 3396 COST bias-gelu-fp8 144 3 6 7917 7776 COST bias-gelu-fp8 144 3 7 10725 10920 COST bias-gelu-fp8 144 3 8 4236 4236 COST bias-gelu-fp8 144 3 9 5352 5352 COST bias-gelu-fp8 288 1 0 720 720 COST bias-gelu-fp8 288 1 1 720 720 COST bias-gelu-fp8 288 1 2 2847 2844 COST bias-gelu-fp8 288 1 3 3300 3300 COST bias-gelu-fp8 288 1 4 1662 1680 COST bias-gelu-fp8 288 1 5 2094 2052 COST bias-gelu-fp8 288 1 6 3699 3654 COST bias-gelu-fp8 288 1 7 3699 3696 COST bias-gelu-fp8 288 1 8 2736 2736 COST bias-gelu-fp8 288 1 9 2748 2748 COST bias-gelu-fp8 288 3 0 1152 1152 COST bias-gelu-fp8 288 3 1 1620 1620 COST bias-gelu-fp8 288 3 2 8277 8274 COST bias-gelu-fp8 288 3 3 9636 9636 COST bias-gelu-fp8 288 3 4 4734 4788 COST bias-gelu-fp8 288 3 5 4734 4692 COST bias-gelu-fp8 288 3 6 10833 10692 COST bias-gelu-fp8 288 3 7 16449 15912 COST bias-gelu-fp8 288 3 8 7548 7548 COST bias-gelu-fp8 288 3 9 7584 7584 COST bias-gelu-fp8 576 1 0 936 936 COST bias-gelu-fp8 576 1 1 936 936 COST bias-gelu-fp8 576 1 2 4143 4140 COST bias-gelu-fp8 576 1 3 4992 4992 COST bias-gelu-fp8 576 1 4 2094 2112 COST bias-gelu-fp8 576 1 5 2958 2916 COST bias-gelu-fp8 576 1 6 5643 5598 COST bias-gelu-fp8 576 1 7 5643 5640 COST bias-gelu-fp8 576 1 8 3840 3840 COST bias-gelu-fp8 576 1 9 4212 4212 COST bias-gelu-fp8 576 3 0 1800 1800 COST bias-gelu-fp8 576 3 1 2268 2268 COST bias-gelu-fp8 576 3 2 12165 12162 COST bias-gelu-fp8 576 3 3 14712 14712 COST bias-gelu-fp8 576 3 4 6030 6084 COST bias-gelu-fp8 576 3 5 7326 7284 COST bias-gelu-fp8 576 3 6 16665 16524 COST bias-gelu-fp8 576 3 7 27897 27360 COST bias-gelu-fp8 576 3 8 10860 10860 COST bias-gelu-fp8 576 3 9 11976 11976 COST bias-gelu-fp8 1152 1 0 1368 1368 COST bias-gelu-fp8 1152 1 1 1368 1368 COST bias-gelu-fp8 1152 1 2 6735 6732 COST bias-gelu-fp8 1152 1 3 8376 8376 COST bias-gelu-fp8 1152 1 4 2958 2976 COST bias-gelu-fp8 1152 1 5 4686 4644 COST bias-gelu-fp8 1152 1 6 9531 9486 COST bias-gelu-fp8 1152 1 7 9531 9528 COST bias-gelu-fp8 1152 1 8 7152 7152 COST bias-gelu-fp8 1152 1 9 7524 7524 COST bias-gelu-fp8 1152 3 0 3096 3096 COST bias-gelu-fp8 1152 3 1 3564 3564 COST bias-gelu-fp8 1152 3 2 19941 19938 COST bias-gelu-fp8 1152 3 3 24864 24864 COST bias-gelu-fp8 1152 3 4 8622 8676 COST bias-gelu-fp8 1152 3 5 12510 12468 COST bias-gelu-fp8 1152 3 6 28329 28188 COST bias-gelu-fp8 1152 3 7 50793 50256 COST bias-gelu-fp8 1152 3 8 20796 20796 COST bias-gelu-fp8 1152 3 9 21912 21912 COST bias-gelu-fp8 1728 1 0 1800 1800 COST bias-gelu-fp8 1728 1 1 1800 1800 COST bias-gelu-fp8 1728 1 2 9327 9324 COST bias-gelu-fp8 1728 1 3 11760 11760 COST bias-gelu-fp8 1728 1 4 3822 3840 COST bias-gelu-fp8 1728 1 5 6414 6372 COST bias-gelu-fp8 1728 1 6 13419 13374 COST bias-gelu-fp8 1728 1 7 13419 13416 COST bias-gelu-fp8 1728 1 8 10464 10464 COST bias-gelu-fp8 1728 1 9 10836 10836 COST bias-gelu-fp8 1728 3 0 4392 4392 COST bias-gelu-fp8 1728 3 1 4860 4860 COST bias-gelu-fp8 1728 3 2 27717 27714 COST bias-gelu-fp8 1728 3 3 35016 35016 COST bias-gelu-fp8 1728 3 4 11214 11268 COST bias-gelu-fp8 1728 3 5 17694 17652 COST bias-gelu-fp8 1728 3 6 39993 39852 COST bias-gelu-fp8 1728 3 7 73689 73152 COST bias-gelu-fp8 1728 3 8 30732 30732 COST bias-gelu-fp8 1728 3 9 31848 31848 COST bias-gelu-fp8 2152 1 0 2124 2124 COST bias-gelu-fp8 2152 1 1 2124 2124 COST bias-gelu-fp8 2152 1 2 11271 11268 COST bias-gelu-fp8 2152 1 3 14298 14298 COST bias-gelu-fp8 2152 1 4 4470 4488 COST bias-gelu-fp8 2152 1 5 7710 7668 COST bias-gelu-fp8 2152 1 6 16335 16290 COST bias-gelu-fp8 2152 1 7 16335 16578 COST bias-gelu-fp8 2152 1 8 13776 13776 COST bias-gelu-fp8 2152 1 9 13788 13788 COST bias-gelu-fp8 2152 3 0 5346 5346 COST bias-gelu-fp8 2152 3 1 5832 5832 COST bias-gelu-fp8 2152 3 2 33549 33546 COST bias-gelu-fp8 2152 3 3 42630 42630 COST bias-gelu-fp8 2152 3 4 13158 13212 COST bias-gelu-fp8 2152 3 5 21582 21540 COST bias-gelu-fp8 2152 3 6 48741 48594 COST bias-gelu-fp8 2152 3 7 90861 91056 COST bias-gelu-fp8 2152 3 8 40668 40668 COST bias-gelu-fp8 2152 3 9 40704 40704 test kernel::tests::report_measured_costs ... ok MLP 729x1152x4304: 665542 assignments, 665542 distinct boundaries MLP 729x4304x1152: 696256 assignments, 696256 distinct boundaries test planner::gemm::tests::report_mlp_boundary_diversity ... ok test result: ok. 3 passed; 0 failed; 0 ignored; 0 measured; 299 filtered out; finished in 7.54s