Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[DO NOT MERGE] Tracking Devel #1208

Open
wants to merge 1,311 commits into
base: master
Choose a base branch
from
Open
Show file tree
Hide file tree
Changes from 250 commits
Commits
Show all changes
1311 commits
Select commit Hold shift + click to select a range
cfd8e5d
Merge remote-tracking branch 'upstream/viable/strict' into HEAD
jjsjann123 Nov 9, 2022
f588435
Refactor lower_alias_memory.cpp (#2170)
naoyam Nov 11, 2022
aeecec0
Merge pull request #2172 from csarofeen/upstream_bump_1109
csarofeen Nov 13, 2022
df88344
Misc mutator cleanups/fixes (#2180)
zasdfgbnm Nov 14, 2022
40128c1
fixing my python flake8 issue from merging warning (#2182)
jjsjann123 Nov 14, 2022
c9f8c1d
Fix invalid aliasing (#2178)
naoyam Nov 14, 2022
ca1387a
Add support for `select` op (#2179)
zasdfgbnm Nov 15, 2022
d1bf8c3
More trivial reduction cleanup (#2181)
zasdfgbnm Nov 15, 2022
11c459e
Print debug info about expr transform on GpuLower (#2185)
azazhu Nov 16, 2022
8d3c95a
Reduce the work to add a new expr: remove ExprType (#2186)
zasdfgbnm Nov 16, 2022
bf1596c
Add computeWith to interleave gmem accesses and computations (#2156)
naoyam Nov 17, 2022
52713b8
bug-3607940: fix wave quantization and round bdimx in outer reduction…
liqiangxl Nov 17, 2022
f7f8c3c
Add support for normal distribution RNG (#2171)
xwang233 Nov 18, 2022
3a6197e
Add Float IR node class (#2197)
naoyam Nov 18, 2022
50861ec
Reduce the work to add a new expr: unify the structure of exprs (#2190)
zasdfgbnm Nov 18, 2022
2ab1408
Support int32 Int (#2198)
naoyam Nov 18, 2022
2b754f3
Don't set DataType of IterDomain as Int (#2200)
naoyam Nov 19, 2022
59e497d
delete unused function (#2201)
naoyam Nov 19, 2022
b02b95c
Reduce the work to add a new expr: mutator cleanup (#2199)
zasdfgbnm Nov 19, 2022
1557d69
Refactor scalar IR nodes (#2203)
naoyam Nov 19, 2022
d165f29
Reduce the work to add a new expr: graphviz cleanup (#2205)
zasdfgbnm Nov 19, 2022
d24b8ae
Suppress nvcc warnings (#2206)
naoyam Nov 20, 2022
75c8f0e
use reference, some compilers may complain or treat it as error (#2211)
liqiangxl Nov 21, 2022
b030e41
Reduce the work to add a new expr: rewrite SubstituteInExpr with OptO…
zasdfgbnm Nov 21, 2022
2057f37
Vectorized welford (#2204)
naoyam Nov 21, 2022
9764e05
Add missing clone (#2213)
naoyam Nov 22, 2022
8614772
WAR to make nvcc compile generated code (#2214)
naoyam Nov 22, 2022
a0145c7
Rework vectorization analysis for View (#2124)
csarofeen Nov 23, 2022
52d8943
Fix expr sorting (#2217)
naoyam Nov 23, 2022
3edc643
Add support for index_select OP (backend implementation) (#2183)
azazhu Nov 24, 2022
b482dda
Move tests around. (#2208)
csarofeen Nov 25, 2022
595d39d
Remove dummy outputs before inlining (#2219)
naoyam Nov 26, 2022
18c788f
Add support for index_select OP (frontend implementation) (#2184)
azazhu Nov 28, 2022
3146e89
Minor cleanup, mostly of Transform Replay (#2220)
csarofeen Nov 28, 2022
7dfb553
Buffer overrun detected with asan (#2218)
naoyam Nov 28, 2022
99fc2f2
Don't exact map across data swizzles. (#2222)
csarofeen Nov 28, 2022
5f66884
Add a unit test using an external source file (#2226)
naoyam Nov 29, 2022
e27be91
Make sure predicate elimination info propagated through unarySetInser…
naoyam Nov 29, 2022
47cc7e1
Cleans up the option handling code for PYTORCH_NVFUSER_DUMP/DISABLE/E…
naoyam Nov 29, 2022
829586a
Fix comments due to PR #2208 (#2223)
csarofeen Nov 29, 2022
8c33bec
Cleanup some nonsymmetric BestEffortReplay usage. (#2221)
csarofeen Nov 29, 2022
ffff014
Add performance benchmark for indexselect OP (#2224)
chang-l Nov 29, 2022
95a28c3
Add support for a symbolic output_shape for broadcast_in_dim (#2195)
kevinstephano Nov 30, 2022
ed2c040
Allow IrBuilder to have Int and Int32 inputs in binary expressions (#…
naoyam Nov 30, 2022
acb42b9
Fix #2230 (#2231)
naoyam Dec 1, 2022
7f1bb3f
A new index hoisting and CSE approach based on subexpression equivale…
zasdfgbnm Dec 6, 2022
d258558
Change TensorIndex to store a single Val* (#2234)
zasdfgbnm Dec 6, 2022
70ef3a4
disable index_select fusion for 0-dim input (#2247)
jjsjann123 Dec 6, 2022
673d40c
empty gpu memory cache between different benchmark cases (#2242) (#2243)
liqiangxl Dec 6, 2022
ac09bae
Change indexing of tensor factories (#2248)
zasdfgbnm Dec 7, 2022
f2dc7e9
Hoist entire predicates (#2249)
zasdfgbnm Dec 8, 2022
5573b21
Use non-broadcasting positions in `multiReductionInliner` (#2250)
zasdfgbnm Dec 9, 2022
6d6918c
Grouped grid welford optimized for outer reduction (#2229)
naoyam Dec 10, 2022
962ddd3
Hoist entire index (#2255)
zasdfgbnm Dec 12, 2022
c9ffa81
Scalar hoist self-reusing (#2254)
zasdfgbnm Dec 12, 2022
cbf141b
Expr simplifier step 1: mathematical theory and associative and commu…
zasdfgbnm Dec 13, 2022
3fc6e28
Make expr sort put scalar exprs in the front to enable more hoisting …
zasdfgbnm Dec 13, 2022
a80d7b5
Move IR printing definition to the class, step 1: Val (#2256)
zasdfgbnm Dec 13, 2022
3b4eaa0
Minor refactoring of scheduler (#2253)
naoyam Dec 14, 2022
38408ae
Update and fix comments (#2261)
naoyam Dec 14, 2022
ac17876
Simplify loop extent in codegen (#2257)
zasdfgbnm Dec 14, 2022
95ae957
bug-3607824: add maxregcount in innerPersistentHeuristic to increase …
liqiangxl Dec 14, 2022
bc6312c
Fix arg eval order (#2265)
zasdfgbnm Dec 14, 2022
797e396
Move IR printing definition to the class, step 2: Exprs (#2263)
zasdfgbnm Dec 15, 2022
4f62c5e
ExpressionEvaluator move op impl to Expr::evaluate (#2262)
zasdfgbnm Dec 15, 2022
123a5bf
Add debug printing for expr simplifier (#2267)
zasdfgbnm Dec 15, 2022
1f5a04a
Allow debug dump, enable, and disable options to have arguments (#2268)
zasdfgbnm Dec 15, 2022
5eaf881
expr simplifier: eliminateTrivialComputation (#2266)
zasdfgbnm Dec 15, 2022
0f7df7d
ir_utils::varName (#2271)
zasdfgbnm Dec 15, 2022
e18569d
Expr simplifier: prove::isNonZero (#2270)
zasdfgbnm Dec 15, 2022
74b773e
Allow users to specify which pass to trigger PYTORCH_NVFUSER_DUMP="ex…
zasdfgbnm Dec 15, 2022
f9ece4b
Add auto-grad support for index_select op (#2259)
chang-l Dec 16, 2022
ac96761
Fix CI (#2274)
zasdfgbnm Dec 16, 2022
8582cf2
FlattenedAssocCommOp changes (#2272)
zasdfgbnm Dec 16, 2022
8549baa
Add support for torch.gather OP (backend implementation) (#2232)
ftxj Dec 19, 2022
6799dd3
Fix indexing when swizzle exist (#2283)
zasdfgbnm Dec 19, 2022
32939dc
Fix lower_verbose printing (#2281)
zasdfgbnm Dec 20, 2022
e5f67b8
Merge commit 'd19988093d1b7942d15dce6ef61b62dccfc3b8a3' into HEAD
jjsjann123 Dec 27, 2022
125726e
Merge pull request #2289 from csarofeen/upstream_merge_1221
csarofeen Dec 28, 2022
9216064
Nvfuser build refactor cherry pick (#2292)
jjsjann123 Dec 30, 2022
07ac457
Add outer normalization tests with perf measurement switch (#2236)
naoyam Jan 4, 2023
5e41232
Add a missing override (#2296)
naoyam Jan 5, 2023
40d6d68
Make PairwiseRootDomainMap require the same extent by default (#2298)
naoyam Jan 5, 2023
cf7958b
Allow -maxrregcount value to be overwritten by (#2299)
naoyam Jan 5, 2023
e18ee0f
Code cleaning (#2293)
jjsjann123 Jan 5, 2023
12160a6
Fixes nvfuser installation and examples (#2303)
jjsjann123 Jan 6, 2023
d75bb49
nvFuser benchamrk moving to third_party/nvfuser (#2294)
csarofeen Jan 6, 2023
1cc2e1f
Kernel Database (redo of PR #2193) (#2297)
kevinstephano Jan 6, 2023
4322bc3
Changing FusionCache names in the Python Frontend to reflect that the…
kevinstephano Jan 6, 2023
59fff35
Build fix (#2305)
naoyam Jan 7, 2023
3dc4d94
clang-format (#2306)
naoyam Jan 7, 2023
b46ebd6
Lintrunner compiler warning patch (#2312)
jjsjann123 Jan 9, 2023
98029e4
Revert CMakeLists.txt as build fails with clang due to unused variabl…
naoyam Jan 10, 2023
63240cc
Modify `FusionDefinition.add_output` API to an include an `alias_inpu…
kevinstephano Jan 11, 2023
2c4d40d
add enable option to warn register spill (#2313) (#2317)
liqiangxl Jan 11, 2023
9b19867
replace welford in layer_norm forward with two-pass (#2300) (#2302)
liqiangxl Jan 11, 2023
cedd359
fixing segfault on CI. The order of argument elimination was wrong (#…
jjsjann123 Jan 12, 2023
1b6393c
Code cleaning (#2320)
jjsjann123 Jan 12, 2023
ff9e5fd
removing wrong check in kerneldb test (#2324)
jjsjann123 Jan 12, 2023
bd6d828
check welford input before trial translate to two-pass (#2325)
liqiangxl Jan 12, 2023
a842a31
patching failing segfault in test_jit (#2328)
jjsjann123 Jan 14, 2023
d91ddf5
adding a -Wall in case compile options passed from dependency changed…
jjsjann123 Jan 14, 2023
b84ae9c
Add FusionExternalSrcMatmul_CUDA to test_gpu_external_src.cpp (#2287)
zasdfgbnm Jan 16, 2023
01d7545
Expr simplifier: divisibility analysis (#2258)
zasdfgbnm Jan 16, 2023
3793d8a
Do not warn unused functions (#2333)
naoyam Jan 17, 2023
c5c6e06
Allow disabling nvfuser_zero (#2277)
zasdfgbnm Jan 17, 2023
2ca39e3
Add comment about swizzle at getGlobalProducerStridedIndices (#2334)
zasdfgbnm Jan 17, 2023
b0febfe
Expr simplifier: implement prove::isPositive, prove::isNonNegative, p…
zasdfgbnm Jan 17, 2023
08aa32a
Enable view on TorchScript (#1966)
zasdfgbnm Jan 18, 2023
7b1ab9a
Reduce tensor sizes to avoid OOM on 3080 (#2335)
naoyam Jan 18, 2023
cead7ad
Expr simplifier: simplification passes for matmul (#2275)
zasdfgbnm Jan 18, 2023
7e8c035
Indexing shared memory with `unsigned` and hoist cvta (#2286)
zasdfgbnm Jan 18, 2023
e08f2c1
Revert PR #2328 as nothing seems to fail anymore (#2339)
naoyam Jan 18, 2023
0b7e314
Add support for torch.gather (frontend) (#2280)
ftxj Jan 19, 2023
11c3909
simplify TensorDomain::noReductions and TensorDomain::noBroadcasts us…
liqiangxl Jan 19, 2023
daffb59
Expr simplifier: predicate register usage optimization (#2276)
zasdfgbnm Jan 19, 2023
df39a1f
Fix position of toSmem for scalar hoisting (#2340)
zasdfgbnm Jan 19, 2023
63044e6
Remove obselete `#if defined(USE_CUDA)` (#2341)
zasdfgbnm Jan 19, 2023
3ea7139
Disable nvfuser zero for matmul (#2342)
zasdfgbnm Jan 19, 2023
e9ca115
patching reshape with scalar tensor input (#2343)
jjsjann123 Jan 20, 2023
c68883b
Revert "Revert PR #2328 as nothing seems to fail anymore (#2339)" (#2…
jjsjann123 Jan 20, 2023
93d3290
Remove `FusionInterface` class (c++) and `Fusion` class (python) from…
kevinstephano Jan 20, 2023
4c6465d
Fix gather test to include removal of Fusion object usage. (#2349)
kevinstephano Jan 20, 2023
0ae093f
Disable reuse of x_mean_sub in translateSingleWelford (#2322)
liqiangxl Jan 24, 2023
4bd949f
Fix ExpandOp::toString printing leading comma (#2338)
jacobhinkle Jan 24, 2023
dc92051
Index select patch (#2353)
jjsjann123 Jan 24, 2023
a9d4037
Validate the corretness of results when running matmul benchmarks (#2…
zasdfgbnm Jan 25, 2023
a9c6f92
Reinitialize segments after welford translation (#2350)
naoyam Jan 25, 2023
15e8a69
Insert floating-point type casting at every segmentation step (#2344)
naoyam Jan 25, 2023
2b43368
Do not segment UnaryOps from fusion inputs (#2354)
naoyam Jan 25, 2023
054dbc4
Add PYTORCH_NVFUSER_DISABLE="expr_simplify(pass,names)" (#2363)
zasdfgbnm Jan 25, 2023
61919f4
Add debug print for expr sorting (#2365)
zasdfgbnm Jan 26, 2023
a7fa96f
Improve printting of LoadStoreOp and MMAOp (#2364)
zasdfgbnm Jan 26, 2023
f92d167
Fix resetting expr list of SegmentedGroup (#2368)
naoyam Jan 26, 2023
f139de1
Do not insert fp16 cast if used by select-like ops (#2357)
naoyam Jan 26, 2023
1d94e31
Add randn, randn_like (#2355)
xwang233 Jan 27, 2023
4a28563
this test is NOT failing! (#2371)
zasdfgbnm Jan 30, 2023
7387fdf
fix (#2385)
naoyam Jan 30, 2023
5ecd8d0
python tests moving under nvfuser directory (#2373)
jjsjann123 Jan 31, 2023
d21da75
Fix projected size calculation (#2382)
naoyam Jan 31, 2023
2187b7f
Reduce tensor size in FusionExprSortMatmulLikeSchedule_CUDA (#2395)
zasdfgbnm Jan 31, 2023
5d2d32b
Added python API `nvfuser.version()` (#2393)
jjsjann123 Jan 31, 2023
4121ffc
Add `from_pytorch` function to `FusionDefinition` API and remove numb…
kevinstephano Feb 1, 2023
e4c339b
Rename view to reshape. (#2397)
kevinstephano Feb 1, 2023
9f9c480
Support 0-dimensional shapes as inputs/outputs to view() (#2398)
jacobhinkle Feb 1, 2023
d04de34
FusionExprSortMatmulLikeSchedule_CUDA use double (#2401)
zasdfgbnm Feb 1, 2023
84a0c36
Add `FusionExecutor::disassembledBinary` (#2402)
zasdfgbnm Feb 2, 2023
03069a7
updating scripts for ci test (#2389)
jjsjann123 Feb 2, 2023
d2d36d7
Copy variant cleanup (#2404)
jjsjann123 Feb 3, 2023
ca989b1
Nvfuser version patch (#2407)
jjsjann123 Feb 3, 2023
e83e73b
Allow cacheAfter for inputs that are outputs (#2374)
jacobhinkle Feb 3, 2023
40fe049
Misc changes/cleanups (#2408)
naoyam Feb 3, 2023
14bcaf5
allow load external cuda code from python (#2411)
liqiangxl Feb 3, 2023
d263f93
Remove automatic aliasOutputToInput in addOutput (#2414)
jacobhinkle Feb 3, 2023
0f3df65
Some more about SASS printing/checking (#2415)
zasdfgbnm Feb 4, 2023
c6bb631
Refactor vectorization in the schedulers (#2412)
naoyam Feb 6, 2023
6a384e6
Some C++17 cleanup (#2420)
zasdfgbnm Feb 7, 2023
1ccf33a
DisableOption::IndexHoist should not disable index simplification (#2…
zasdfgbnm Feb 7, 2023
6e8430d
[Scatter 2] Add ScatterOp IR data structure (#2391)
ftxj Feb 8, 2023
6659183
Print float constant values as "...f" (#2427)
naoyam Feb 8, 2023
3533ac8
Refactor squeeze prim to accept a sequence of dimensions. (#2428)
rdspring1 Feb 8, 2023
29136be
Add new op: iota (#2433)
zasdfgbnm Feb 8, 2023
5a05da3
cp.async.cg (#2429)
zasdfgbnm Feb 8, 2023
c411f97
Add dtype argument to define_constant (#2423)
jacobhinkle Feb 9, 2023
aaa3b9c
Simplify expr with the fundamental division-with-remainder property (…
zasdfgbnm Feb 9, 2023
3c1f9b3
Assert LDSM register usage (#2431)
zasdfgbnm Feb 9, 2023
b14abd5
Predicate simplification: `b && b`, `b || b` (#2436)
zasdfgbnm Feb 9, 2023
1dee78d
avoid promote float16 and bfloat16 to float32 in op-where (#2409)
liqiangxl Feb 9, 2023
d1586ec
Add index_select shape expression (#2418)
ftxj Feb 9, 2023
4a95b24
Fix cpAsyncCg (#2439)
zasdfgbnm Feb 9, 2023
3da9822
Just to make nvcc happy with generated code (#2444)
naoyam Feb 9, 2023
1539b6f
Fix input type lowering in segmentation (#2416)
naoyam Feb 9, 2023
da06d15
switching `packaging.version.Version` to `torch.torch_version.Version…
jjsjann123 Feb 10, 2023
6369d0f
Pull some tests from Shiming's PR (#2446)
zasdfgbnm Feb 10, 2023
a8b5062
Add complex rsqrt function to runtime/helpers.cu (#2448)
jacobhinkle Feb 13, 2023
e6f758a
Scalar type patch (#2452)
jjsjann123 Feb 13, 2023
c19ed8f
Add python binding for iota (#2447)
zasdfgbnm Feb 14, 2023
9a60b9c
Convert aligned block sync (#2445)
zasdfgbnm Feb 14, 2023
19ee57e
Temporarily disable cp.async.cg for now (#2457)
zasdfgbnm Feb 15, 2023
fb17875
Fix SASS test, manually set PATH (#2462)
zasdfgbnm Feb 15, 2023
084e340
Clear thread predicates when RAW sync is used (#2460)
naoyam Feb 15, 2023
7b37a83
Enable grid outer persistent scheduling (#2435)
naoyam Feb 15, 2023
3e4ec2f
Generate where with C++ `? ... : ...` rather than using device functi…
naoyam Feb 16, 2023
e310f03
hasToString util template function to print using toString (#2471)
naoyam Feb 16, 2023
65eb427
Fix cp.async barrier (#2470)
zasdfgbnm Feb 16, 2023
d57da94
Improve error message for SASS dumping (#2474)
zasdfgbnm Feb 16, 2023
5007cd0
Move arith to ops directory (#2475)
naoyam Feb 16, 2023
fe49742
Build fix (#2476)
naoyam Feb 16, 2023
785b853
Move our codebase to namespace nvfuser (#2477)
zasdfgbnm Feb 16, 2023
e3e86c8
Skip FusionCpAsyncCommitWait_CUDA on pre Ampere (#2481)
zasdfgbnm Feb 17, 2023
766ee48
Swap order of stdout and stderror read from nvdisasm (#2483)
zasdfgbnm Feb 17, 2023
a0638c4
Moving predicate in cp.async (#2484)
mmigdal-nv Feb 17, 2023
980f6eb
Thread predication of reduction init expr (#2489)
naoyam Feb 17, 2023
08dc16d
Don't initialize if using cpAsync (#2485)
zasdfgbnm Feb 18, 2023
3ed70b2
Fix buffer reusing (#2490)
zasdfgbnm Feb 18, 2023
7eaf4cf
set heuristics to persistent after welford translatte (#2491)
liqiangxl Feb 21, 2023
3a9cb4a
Cleanup debugging print (#2497)
zasdfgbnm Feb 21, 2023
4ea9c16
Return Bool* for logical expr in IrBuilder (#2494)
zasdfgbnm Feb 21, 2023
5f5aefe
Make IrCloner more convenient to use (#2495)
zasdfgbnm Feb 21, 2023
167718b
Add Minimal User Scheduling to Python API (#2438)
kevinstephano Feb 21, 2023
68ce62e
Lower unroll cleanup, make it support IfThenElse (#2496)
zasdfgbnm Feb 22, 2023
559ba79
Debugging helper: assert when accessing global tensor out of bound (#…
zasdfgbnm Feb 22, 2023
b6258d4
Fix errors resulting from fusions with scalar only inputs (#2503)
kevinstephano Feb 22, 2023
369e36e
Fix canOmitStopPredicate (#2504)
zasdfgbnm Feb 22, 2023
ed6b6bd
patching dynamo test (#2498)
jjsjann123 Feb 22, 2023
b7c866b
Refactor DataType to better support complicated dtypes like array and…
zasdfgbnm Feb 22, 2023
e0ee4ec
disable welford translate if the original scheduler before translate …
liqiangxl Feb 23, 2023
3b25576
Fix mismatched signature in addcmul macro (#2512)
jacobhinkle Feb 23, 2023
9340f80
[Scatter 3] Add IndexLowering for ScatterOp (#2437)
ftxj Feb 23, 2023
eb04799
index_select/gather cache patch (#2501)
jjsjann123 Feb 24, 2023
a8b9ceb
Move zero shape tests (#2514)
ftxj Feb 24, 2023
934523c
Refactor `FusionDefinition` to support deserialization. (#2507)
rdspring1 Feb 24, 2023
88946da
Bug fix (#2518)
naoyam Feb 25, 2023
14d918f
cleanup (#2519)
naoyam Feb 25, 2023
820d749
cleanup (#2520)
naoyam Feb 26, 2023
4d990dc
Adding current stream to event record which (#2521)
mmigdal-nv Feb 27, 2023
354a6e9
Add frontend arithmetic ops and shape property (#2479)
jacobhinkle Feb 27, 2023
774bda7
Add prod op (#2511)
jacobhinkle Feb 28, 2023
28990ac
disable a failing test (#2528)
naoyam Feb 28, 2023
244136e
Random op python api (#2430)
jjsjann123 Feb 28, 2023
657080b
reduction on complex numbers, added volatile copy and assignment (#2453)
liqiangxl Feb 28, 2023
cf94a48
disable index_select/gather tests (#2530)
jjsjann123 Feb 28, 2023
dc04f87
Fix #2531 (#2532)
naoyam Mar 1, 2023
591b8ae
New lowering pass: loop rotation (#2500)
zasdfgbnm Mar 1, 2023
0200e7a
Add optional bool `preserve_error` for expr simplifier (#2534)
zasdfgbnm Mar 1, 2023
5913acc
Improve matmul instruction scheduling with loop rotation (#2488)
zasdfgbnm Mar 2, 2023
a40ef69
Fix stop predicate (#2537)
zasdfgbnm Mar 2, 2023
b928665
bump ubuntu version (#2546)
jjsjann123 Mar 6, 2023
16a26a1
cp.async access global tensor via pointer (#2282)
zasdfgbnm Mar 6, 2023
5a69c1b
Rewrite `reducePredicateRegisterUsage` (#2533)
zasdfgbnm Mar 6, 2023
12c1765
Fix downcast_ptr != nullptr in RandomOp (#2547)
rdspring1 Mar 7, 2023
14930f4
Update fusion_record.h (#2536)
zasdfgbnm Mar 7, 2023
4ad1055
Make contiguity ignore broadcasts (#2517)
zasdfgbnm Mar 7, 2023
1a72ba1
Multidevice setup merged with devel (#2318)
csarofeen Mar 7, 2023
d5f10e1
allow prediction on block dim (#2544)
liqiangxl Mar 7, 2023
8ed9540
Output stride order (#2548)
jjsjann123 Mar 7, 2023
2ee05db
skipping ampere tests on pre-ampere devices (#2553)
jjsjann123 Mar 7, 2023
a86f9b0
Add debugging utility RAII guard for printting scopes (#2555)
zasdfgbnm Mar 8, 2023
3b85308
Fixing and improving indexing type handling (#2522)
mmigdal-nv Mar 8, 2023
4f82275
Avoid adding "f" suffix to std::{real,imag} (#2565)
jacobhinkle Mar 9, 2023
e0c1786
Revert "Fixing and improving indexing type handling (#2522)" (#2568)
naoyam Mar 9, 2023
4bc286a
Bcast python api patch (#2561)
jjsjann123 Mar 10, 2023
3c4b3da
Clean up index type handling (#2570)
naoyam Mar 10, 2023
9eb4c20
Change contiguity into `std::vector<c10::optional<bool>>` (#2569)
zasdfgbnm Mar 10, 2023
29bb8c0
Make expr simplifier tests more readable (#2571)
zasdfgbnm Mar 10, 2023
81adad8
Index select empty tensor scalar tensor (#2513)
jjsjann123 Mar 10, 2023
9c62d94
Fix indexing failure with non-view rfactor (#2562)
naoyam Mar 11, 2023
2809661
Fix multidevice tests (#2574)
samnordmann Mar 13, 2023
1e30fee
IterDomain resize for pad, cat, slice (#2480)
naoyam Mar 14, 2023
e8557ec
persistent_use_of_buffer is accumulated over all the resolution point…
naoyam Mar 14, 2023
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
136 changes: 1 addition & 135 deletions .github/generated-ciflow-ruleset.json

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

54 changes: 27 additions & 27 deletions .github/scripts/generate_ci_workflows.py
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@

from dataclasses import asdict, dataclass, field
from pathlib import Path
from typing import Dict, Set, List, Iterable
from typing import Dict, Set, List

import jinja2
import json
Expand Down Expand Up @@ -259,19 +259,19 @@ def generate_workflow_file(self, workflow_template: jinja2.Template) -> None:
labels={LABEL_CIFLOW_DEFAULT, LABEL_CIFLOW_CPU, LABEL_CIFLOW_WIN}
),
),
CIWorkflow(
arch="windows",
build_environment="win-vs2019-cuda11.3-py3",
cuda_version="11.3",
test_runner_type=WINDOWS_CUDA_TEST_RUNNER,
num_test_shards=2,
only_run_smoke_tests_on_pull_request=True,
enable_force_on_cpu_test=1,
ciflow_config=CIFlowConfig(
run_on_canary=True,
labels={LABEL_CIFLOW_DEFAULT, LABEL_CIFLOW_CUDA, LABEL_CIFLOW_WIN}
),
),
# CIWorkflow(
# arch="windows",
# build_environment="win-vs2019-cuda11.3-py3",
# cuda_version="11.3",
# test_runner_type=WINDOWS_CUDA_TEST_RUNNER,
# num_test_shards=2,
# only_run_smoke_tests_on_pull_request=True,
# enable_force_on_cpu_test=1,
# ciflow_config=CIFlowConfig(
# run_on_canary=True,
# labels={LABEL_CIFLOW_DEFAULT, LABEL_CIFLOW_CUDA, LABEL_CIFLOW_WIN}
# ),
# ),
CIWorkflow(
arch="windows",
build_environment="periodic-win-vs2019-cuda11.1-py3",
Expand Down Expand Up @@ -559,12 +559,12 @@ def main() -> None:
loader=jinja2.FileSystemLoader(str(GITHUB_DIR.joinpath("templates"))),
undefined=jinja2.StrictUndefined,
)
template_and_workflows = [
(jinja_env.get_template("linux_ci_workflow.yml.j2"), LINUX_WORKFLOWS),
(jinja_env.get_template("windows_ci_workflow.yml.j2"), WINDOWS_WORKFLOWS),
(jinja_env.get_template("bazel_ci_workflow.yml.j2"), BAZEL_WORKFLOWS),
(jinja_env.get_template("docker_builds_ci_workflow.yml.j2"), DOCKER_WORKFLOWS),
]
# template_and_workflows = [
# (jinja_env.get_template("linux_ci_workflow.yml.j2"), LINUX_WORKFLOWS),
# (jinja_env.get_template("windows_ci_workflow.yml.j2"), WINDOWS_WORKFLOWS),
# (jinja_env.get_template("bazel_ci_workflow.yml.j2"), BAZEL_WORKFLOWS),
# (jinja_env.get_template("docker_builds_ci_workflow.yml.j2"), DOCKER_WORKFLOWS),
# ]
# Delete the existing generated files first, this should align with .gitattributes file description.
existing_workflows = GITHUB_DIR.glob("workflows/generated-*")
for w in existing_workflows:
Expand All @@ -574,13 +574,13 @@ def main() -> None:
print(f"Error occurred when deleting file {w}: {e}")

ciflow_ruleset = CIFlowRuleset()
for template, workflows in template_and_workflows:
# added Iterable check to appease the mypy gods
if not isinstance(workflows, Iterable):
raise Exception(f"How is workflows not iterable? {workflows}")
for workflow in workflows:
workflow.generate_workflow_file(workflow_template=template)
ciflow_ruleset.add_label_rule(workflow.ciflow_config.labels, workflow.build_environment)
# for template, workflows in template_and_workflows:
# # added Iterable check to appease the mypy gods
# if not isinstance(workflows, Iterable):
# raise Exception(f"How is workflows not iterable? {workflows}")
# for workflow in workflows:
# workflow.generate_workflow_file(workflow_template=template)
# ciflow_ruleset.add_label_rule(workflow.ciflow_config.labels, workflow.build_environment)
ciflow_ruleset.generate_json()


Expand Down
23 changes: 23 additions & 0 deletions .github/workflows/cancel_redundant_workflows.yml
Original file line number Diff line number Diff line change
@@ -0,0 +1,23 @@
name: Cancel redundant workflows
on:
workflow_run:
types:
- requested
# NOTE: Make sure to add to this list as you add more workflows running on 'pull_request'
workflows:
- Lint
- Test tools
- TorchBench CI (pytorch-linux-py3.7-cu102)
- clang-format
jobs:
cancel:
# We do not want to cancel reruns on master
if: github.event.workflow_run.head_branch != 'master'
runs-on: ubuntu-18.04
steps:
- name: Cancel duplicate workflow runs
uses: potiuk/cancel-workflow-runs@a81b3c4d59c61e27484cfacdc13897dd908419c9
with:
cancelMode: duplicates
token: ${{ secrets.GITHUB_TOKEN }}
sourceRunId: ${{ github.event.workflow_run.id }}
Loading