Skip to content

Compiling models with Reactant fails (cublasLtCreate) #1601

@mytolisk

Description

@mytolisk

I'm having trouble with compiling models with the latest Lux/Reactant versions. I'm on Rocky linux 8.10

The LSTM tutorial fails, and also this simple script:

using Lux, Reactant, Random

rng = Random.default_rng()
dev = reactant_device()

m=Chain(Conv((3,3), 2=>4, leakyrelu))
ps, st = Lux.setup(rng,m) |> dev
a = rand(Float32,10,10,2,1) |> dev
out, st = @jit Lux.apply(m,a,ps,st)

Julia version is 1.11.4

Pkg versions:
julia> Pkg.status()
Status /path/Project.toml
[7da242da] Enzyme v0.13.109
[b2108857] Lux v1.27.1
[3bd65402] Optimisers v0.4.7
[3c362404] Reactant v0.2.185

julia> Reactant_jll.is_available()
true

julia> Reactant_jll.host_platform
Linux x86_64 {cxxstring_abi=cxx11, gpu=cuda, gpu_version=12.9, julia_version=1.11.4, libc=glibc, libgfortran_version=5.0.0, libstdcxx_version=3.4.30, mode=opt}

It finds hardware and CUDA
julia> ps, st = Lux.setup(rng,m) |> dev
WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
I0000 00:00:1766045551.603328 598355 service.cc:153] XLA service 0x63843a0 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
I0000 00:00:1766045551.603366 598355 service.cc:161] StreamExecutor device (0): NVIDIA A40, Compute Capability 8.6
I0000 00:00:1766045551.604502 598355 se_gpu_pjrt_client.cc:1440] Using BFC allocator.
I0000 00:00:1766045551.604537 598355 gpu_helpers.cc:141] XLA backend allocating 35781869568 bytes on device 0 for BFCAllocator.
I0000 00:00:1766045551.604570 598355 gpu_helpers.cc:182] XLA backend will use up to 11927289856 bytes on device 0 for CollectiveBFCAllocator.
I0000 00:00:1766045551.634117 598355 cuda_dnn.cc:461] Loaded cuDNN version 91400
((layer_1 = (weight = ConcretePJRTArray{Float32, 4, 1}(Float32[0.2168731 -0.21829635 0.021970421; -0.20019113 -0.0011756442 0.07267339; -0.037982095 -0.18488145 -0.025027305;;; -0.08308116 -0.083196364 0.23397905; -0.13060631 -0.09887839 0.15277651; -0.22280036 0.009649015 0.047057267;;;; 0.05517559 -0.01687113 -0.012363862; 0.07931492 0.17914091 -0.1315057; 0.16668728 0.024479002 -0.011142334;;; 0.1636007 0.2022534 -0.118163295; -0.117578246 -0.22701325 -0.15792598; 0.17941107 -0.037423227 0.14115952;;;; 0.1687574 0.21836714 0.17667723; -0.23390144 -0.14125527 0.12690254; 0.09573904 -0.0916368 0.18706667;;; 0.17201677 0.0023923956 0.12816548; 0.008272274 -0.06775288 0.041589584; -0.1558286 -0.062512204 0.18791631;;;; 0.23415983 0.14652681 -0.15826698; -0.01860803 0.19823596 -0.15953152; -0.2299741 -0.1296339 -0.003415974;;; -0.16505618 -0.05167108 0.1538629; 0.19457056 -0.18084836 0.025250599; -0.045280356 -0.071394034 -0.176011]), bias = ConcretePJRTArray{Float32, 1, 1}(Float32[0.061203968, 0.032079708, -0.06641705, 0.028946258])),), (layer_1 = NamedTuple(),))

But then fails critically:

Error:
Invalid handle. Cannot load symbol cublasLtCreate

[598355] signal 6 (-6): Aborted
in expression starting at REPL[12]:1
gsignal at /lib64/libc.so.6 (unknown line)
abort at /lib64/libc.so.6 (unknown line)
unknown function (ip: 0x7f241d8f2cf4)
unknown function (ip: 0x7f241e454c19)
unknown function (ip: 0x7f241e453391)
unknown function (ip: 0x7f241dcfc03d)
unknown function (ip: 0x7f241da56538)
_ZN5cudnn7backend6Engine17finalize_internalEv at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libcudnn_graph.so.9 (unknown line)
_ZN5cudnn7backend10Descriptor8finalizeEv at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libcudnn_graph.so.9 (unknown line)
_ZN5cudnn7backend12EngineConfig17finalize_internalEv at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libcudnn_graph.so.9 (unknown line)
_ZN5cudnn7backend12EngcfgTmpVar7Results23emplace_back_on_successEONS0_12EngineConfigERNS1_10ErrorStackE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libcudnn_graph.so.9 (unknown line)
_ZNK5cudnn7backend16EngineHeuristics12get_internalE27cudnnBackendAttributeName_t27cudnnBackendAttributeType_tlPlPv at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libcudnn_graph.so.9 (unknown line)
cudnnBackendGetAttribute at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libcudnn_graph.so.9 (unknown line)
_ZN14cudnn_frontend19EngineHeuristics_v815getEngineConfigEl at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN14cudnn_frontendL24get_heuristics_list_implE22cudnnBackendHeurMode_tRNS_17OperationGraph_v8ESt8functionIFbPvEEiRSt6vectorISt10shared_ptrINS_20OpaqueBackendPointerEESaISA_EES8_IKNS_16DevicePropertiesEE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN14cudnn_frontendL19get_heuristics_listERKSt6vectorINSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEESaIS6_EERNS_17OperationGraph_v8ESt8functionIFbPvEERS0_ISt10shared_ptrINS_20OpaqueBackendPointerEESaISJ_EEbiSH_IKNS_16DevicePropertiesEE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN14cudnn_frontend19get_heuristics_listILm1EEESt6vectorI13cudnnStatus_tSaIS2_EESt5arrayINSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEEXT_EERNS_17OperationGraph_v8ESt8functionIFbPvEERS1_ISt10shared_ptrINS_20OpaqueBackendPointerEESaISL_EEb at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN15stream_executor3gpu12CudnnSupport18GetConvolveRunnersENS_3dnn15ConvolutionKindENS2_8DataTypeES4_PNS_6StreamERKNS2_15BatchDescriptorENS_17DeviceAddressBaseERKNS2_16FilterDescriptorESA_S9_SA_RKNS2_21ConvolutionDescriptorEbPNS_16ScratchAllocatorERKNS_13EngineOptionsEPSt6vectorISt10unique_ptrIKNS2_8OpRunnerIFvSA_SA_SA_EEESt14default_deleteISR_EESaISU_EE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla3gpu12_GLOBAL__N_113GetAlgorithmsEPN15stream_executor3dnn10DnnSupportENS3_15ConvolutionKindENS3_8DataTypeES7_PNS2_6StreamERKNS0_13GpuConvConfigERKNS2_13EngineOptionsEb at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla3gpu12CudnnBackend19GetSupportedConfigsERKNS_14HloInstructionE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla9Autotuner19GetSupportedConfigsEPNS_14HloInstructionE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla9Autotuner14TuneBestConfigEPNS_14HloInstructionE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla9Autotuner9GetConfigEPNS_14HloInstructionE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla9Autotuner8AutotuneEPNS_9HloModuleERKN4absl12lts_2025081411FunctionRefIFbRKNS_14HloInstructionEEEE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla3gpu13AutotunerPass7RunImplEPNS_9HloModuleERKN4absl12lts_2025081413flat_hash_setISt17basic_string_viewIcSt11char_traitsIcEENS5_18container_internal10StringHashENSB_8StringEqESaISA_EEE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla16HloPassInterface3RunEPNS_9HloModuleERKN4absl12lts_2025081413flat_hash_setISt17basic_string_viewIcSt11char_traitsIcEENS4_18container_internal10StringHashENSA_8StringEqESaIS9_EEE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla15HloPassPipeline9RunHelperIPNS_9HloModuleEEEN4absl12lts_202508148StatusOrIbEEPNS_16HloPassInterfaceET_RKNS5_13flat_hash_setISt17basic_string_viewIcSt11char_traitsIcEENS5_18container_internal10StringHashENSG_8StringEqESaISF_EEE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla15HloPassPipeline17RunPassesInternalIPNS_9HloModuleEEEN4absl12lts_202508148StatusOrIbEET_RKNS_12DebugOptionsERKNS5_13flat_hash_setISt17basic_string_viewIcSt11char_traitsIcEENS5_18container_internal10StringHashENSH_8StringEqESaISG_EEE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla15HloPassPipeline7RunImplEPNS_9HloModuleERKN4absl12lts_2025081413flat_hash_setISt17basic_string_viewIcSt11char_traitsIcEENS4_18container_internal10StringHashENSA_8StringEqESaIS9_EEE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla16HloPassInterface3RunEPNS_9HloModuleERKN4absl12lts_2025081413flat_hash_setISt17basic_string_viewIcSt11char_traitsIcEENS4_18container_internal10StringHashENSA_8StringEqESaIS9_EEE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla3gpu11GpuCompiler31OptimizeHloPostLayoutAssignmentEPNS_9HloModuleEPN15stream_executor14StreamExecutorERKNS_8Compiler14CompileOptionsERKNS7_15GpuTargetConfigEPKNS0_12GpuAliasInfoEPN3tsl6thread10ThreadPoolE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla3gpu13NVPTXCompiler31OptimizeHloPostLayoutAssignmentEPNS_9HloModuleEPN15stream_executor14StreamExecutorERKNS_8Compiler14CompileOptionsERKNS7_15GpuTargetConfigEPKNS0_12GpuAliasInfoEPN3tsl6thread10ThreadPoolE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla3gpu11GpuCompiler17OptimizeHloModuleEPNS_9HloModuleEPN15stream_executor14StreamExecutorERKNS_8Compiler14CompileOptionsERKNS7_15GpuTargetConfigEPKNS0_12GpuAliasInfoE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla3gpu11GpuCompiler12RunHloPassesESt10unique_ptrINS_9HloModuleESt14default_deleteIS3_EEPN15stream_executor14StreamExecutorERKNS_8Compiler14CompileOptionsE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla7Service15BuildExecutableERKNS_14HloModuleProtoESt10unique_ptrINS_15HloModuleConfigESt14default_deleteIS5_EEPNS_7BackendEPN15stream_executor14StreamExecutorERKNS_8Compiler14CompileOptionsEb at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla12LocalService18CompileExecutablesERKNS_14XlaComputationEN4absl12lts_202508144SpanIKPKNS_5ShapeEEERKNS_22ExecutableBuildOptionsE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla11LocalClient7CompileERKNS_14XlaComputationEN4absl12lts_202508144SpanIKPKNS_5ShapeEEERKNS_22ExecutableBuildOptionsE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla24PjRtStreamExecutorClient15CompileInternalERKNS_14XlaComputationERKSt6vectorIPKNS_5ShapeESaIS7_EESt8functionIFN4absl12lts_202508148StatusOrISt4pairIS4_IS5_SaIS5_EES5_EEERKNS_9HloModuleEEENS_14CompileOptionsEb at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla24PjRtStreamExecutorClient7CompileEN4mlir8ModuleOpENS_14CompileOptionsEb at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla24PjRtStreamExecutorClient14CompileAndLoadEN4mlir8ModuleOpENS_14CompileOptionsE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
_ZN3xla23StreamExecutorGpuClient14CompileAndLoadEN4mlir8ModuleOpENS_14CompileOptionsE at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
ClientCompile at /hiddenpath/artifacts/377531aafc691817f8096dca88c7e7b26571dadb/lib/libReactantExtra.so (unknown line)
#25 at /hiddenpath/packages/Reactant/EDxwi/src/xla/PJRT/LoadedExecutable.jl:83
try_compile_dump_mlir at /hiddenpath/packages/Reactant/EDxwi/src/mlir/IR/Pass.jl:134
try_compile_dump_mlir at /hiddenpath/packages/Reactant/EDxwi/src/mlir/IR/Pass.jl:129 [inlined]
#compile#24 at /hiddenpath/packages/Reactant/EDxwi/src/xla/PJRT/LoadedExecutable.jl:82 [inlined]
compile at /hiddenpath/packages/Reactant/EDxwi/src/xla/PJRT/LoadedExecutable.jl:68
jfptr_compile_37042.1 at /hiddenpath/compiled/v1.11/Reactant/p9PzF_5JzdJ.so (unknown line)
#compile_xla#58 at /hiddenpath/packages/Reactant/EDxwi/src/Compiler.jl:3632
compile_xla at /hiddenpath/packages/Reactant/EDxwi/src/Compiler.jl:3560 [inlined]
#compile#59 at /hiddenpath/packages/Reactant/EDxwi/src/Compiler.jl:3664
compile at /hiddenpath/packages/Reactant/EDxwi/src/Compiler.jl:3661
unknown function (ip: 0x7f255d21237d)
jl_apply at /cache/build/builder-amdci5-5/julialang/julia-release-1-dot-11/src/julia.h:2157 [inlined]
do_call at /cache/build/builder-amdci5-5/julialang/julia-release-1-dot-11/src/interpreter.c:126
eval_value at /cache/build/builder-amdci5-5/julialang/julia-release-1-dot-11/src/interpreter.c:223
eval_stmt_value at /cache/build/builder-amdci5-5/julialang/julia-release-1-dot-11/src/interpreter.c:174 [inlined]
eval_body at /cache/build/builder-amdci5-5/julialang/julia-release-1-dot-11/src/interpreter.c:666
jl_interpret_toplevel_thunk at /cache/build/builder-amdci5-5/julialang/julia-release-1-dot-11/src/interpreter.c:824
jl_toplevel_eval_flex at /cache/build/builder-amdci5-5/julialang/julia-release-1-dot-11/src/toplevel.c:943
jl_toplevel_eval_flex at /cache/build/builder-amdci5-5/julialang/julia-release-1-dot-11/src/toplevel.c:886
ijl_toplevel_eval_in at /cache/build/builder-amdci5-5/julialang/julia-release-1-dot-11/src/toplevel.c:994
eval at ./boot.jl:430 [inlined]
eval_user_input at /cache/build/builder-amdci5-5/julialang/julia-release-1-dot-11/usr/share/julia/stdlib/v1.11/REPL/src/REPL.jl:245
repl_backend_loop at /cache/build/builder-amdci5-5/julialang/julia-release-1-dot-11/usr/share/julia/stdlib/v1.11/REPL/src/REPL.jl:342
#start_repl_backend#59 at /cache/build/builder-amdci5-5/julialang/julia-release-1-dot-11/usr/share/julia/stdlib/v1.11/REPL/src/REPL.jl:327
start_repl_backend at /cache/build/builder-amdci5-5/julialang/julia-release-1-dot-11/usr/share/julia/stdlib/v1.11/REPL/src/REPL.jl:324
#run_repl#72 at /cache/build/builder-amdci5-5/julialang/julia-release-1-dot-11/usr/share/julia/stdlib/v1.11/REPL/src/REPL.jl:483
run_repl at /cache/build/builder-amdci5-5/julialang/julia-release-1-dot-11/usr/share/julia/stdlib/v1.11/REPL/src/REPL.jl:469
jfptr_run_repl_10102.1 at /hiddenpath2/julia/julia-1.11.4/share/julia/compiled/v1.11/REPL/u0gqU_PBQaY.so (unknown line)
#1150 at ./client.jl:446
jfptr_YY.1150_14761.1 at /hiddenpath2/julia/julia-1.11.4/share/julia/compiled/v1.11/REPL/u0gqU_PBQaY.so (unknown line)
jl_apply at /cache/build/builder-amdci5-5/julialang/julia-release-1-dot-11/src/julia.h:2157 [inlined]
jl_f__call_latest at /cache/build/builder-amdci5-5/julialang/julia-release-1-dot-11/src/builtins.c:875
#invokelatest#2 at ./essentials.jl:1055 [inlined]
invokelatest at ./essentials.jl:1052 [inlined]
run_main_repl at ./client.jl:430
repl_main at ./client.jl:567 [inlined]
_start at ./client.jl:541
jfptr__start_73560.1 at /hiddenpath2/julia/julia-1.11.4/lib/julia/sys.so (unknown line)
jl_apply at /cache/build/builder-amdci5-5/julialang/julia-release-1-dot-11/src/julia.h:2157 [inlined]
true_main at /cache/build/builder-amdci5-5/julialang/julia-release-1-dot-11/src/jlapi.c:900
jl_repl_entrypoint at /cache/build/builder-amdci5-5/julialang/julia-release-1-dot-11/src/jlapi.c:1059
main at julia (unknown line)
__libc_start_main at /lib64/libc.so.6 (unknown line)
unknown function (ip: 0x4010b8)
Allocations: 73429715 (Pool: 73423826; Big: 5889); GC: 398
Aborted (core dumped)

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions