Knet 1.4.7: libknet8 library not found
jonathan-laurent opened this issue · 4 comments
jonathan-laurent commented
I just tried to install the latest Knet release (1.4.7) in a fresh Julia 1.6.1 installation on Ubuntu 20.04 and I am getting the following error:
julia> using Pkg
julia> Pkg.update()
julia> Pkg.add("Knet")
julia> Pkg.build("Knet")
julia> using Knet; include(Knet.dir("test/gpu.jl"))
Downloaded artifact: CUDA
┌ Warning: libknet8 library not found, some GPU functionality may not be available, try reinstalling Knet.
└ @ Knet ~/.julia/packages/Knet/JXldi/src/Knet.jl:33
CUDA toolkit 11.3.1, artifact installation
CUDA driver 11.3.0
NVIDIA driver 465.19.1
Libraries:
- CUBLAS: 11.5.1
- CURAND: 10.2.4
- CUFFT: 10.4.2
- CUSOLVER: 11.1.2
- CUSPARSE: 11.6.0
- CUPTI: 14.0.0
- NVML: 11.0.0+465.19.1
Downloaded artifact: CUDNN
- CUDNN: 8.20.0 (for CUDA 11.3.0)
Downloaded artifact: CUTENSOR
- CUTENSOR: 1.3.0 (for CUDA 11.2.0)
Toolchain:
- Julia: 1.6.1
- LLVM: 11.0.1
- PTX ISA support: 3.2, 4.0, 4.1, 4.2, 4.3, 5.0, 6.0, 6.1, 6.3, 6.4, 6.5, 7.0
- Device capability support: sm_35, sm_37, sm_50, sm_52, sm_53, sm_60, sm_61, sm_62, sm_70, sm_72, sm_75, sm_80
1 device:
0: NVIDIA GeForce RTX 2070 (sm_75, 7.658 GiB / 7.793 GiB available)
CuDevice(0): NVIDIA GeForce RTX 2070
length(CUDA.devices()) = 1
CUDA.capability(CUDA.device()) = v"7.5.0"
CUDA.warpsize(CUDA.device()) = 32
CUDA.toolkit() = CUDA.Deps.ArtifactToolkit(v"11.3.1", "/home/jonathan/.julia/artifacts/0ed779f9e1d7b0c9905eb5be2c4349d32fbeb050")
CUDA.version() = v"11.3.0"
Mem.info() = (8222932992, 8367570944)
CUDA.synchronize() = nothing
NVML.driver_version() = v"465.19.1"
NVML.version() = v"11.0.0+465.19.1"
NVML.cuda_driver_version() = v"11.3.0"
NVML.memory_info(nvmldev) = (total = 8367570944, free = 8222932992, used = 144637952)
CUBLAS.handle() = Ptr{Nothing} @0x000000000bbfa390
CUBLAS.version() = v"11.5.1"
CUDNN.handle() = Ptr{Nothing} @0x0000000004442370
CUDNN.version() = v"8.20.0"
Knet.LibKnet8.libknet8 = ""
gpu: Test Failed at /home/jonathan/.julia/packages/Knet/JXldi/test/gpu.jl:39
Expression: !(isempty(Knet.LibKnet8.libknet8))
Evaluated: !(isempty(""))
Stacktrace:
[1] macro expansion
@ ~/.julia/packages/Knet/JXldi/test/gpu.jl:39 [inlined]
[2] macro expansion
@ /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.6/Test/src/Test.jl:1151 [inlined]
[3] top-level scope
@ ~/.julia/packages/Knet/JXldi/test/gpu.jl:4
readdir(artifact"libknet8") = ["libknet8.so"]
Test Summary: | Pass Fail Total
gpu | 13 1 14
ERROR: LoadError: Some tests did not pass: 13 passed, 1 failed, 0 errored, 0 broken.
in expression starting at /home/jonathan/.julia/packages/Knet/JXldi/test/gpu.jl:3
denizyuret commented
The libknet8.so library seems to be present in your output, maybe it is a
library version mismatch? Can you run an ldd on the libknet8.so under your
artifacts?
…On Sun, Jul 25, 2021 at 1:59 PM Jonathan Laurent ***@***.***> wrote:
I just tried to install the latest Knet release (1.4.7) in a fresh Julia
1.6.1 installation on Ubuntu 20.04 and I am getting the following error:
julia> using Pkg
julia> Pkg.update()
julia> Pkg.add("Knet")
julia> Pkg.build("Knet")
julia> using Knet; include(Knet.dir("test/gpu.jl"))
Downloaded artifact: CUDA
┌ Warning: libknet8 library not found, some GPU functionality may not be available, try reinstalling Knet.
└ @ Knet ~/.julia/packages/Knet/JXldi/src/Knet.jl:33
CUDA toolkit 11.3.1, artifact installation
CUDA driver 11.3.0
NVIDIA driver 465.19.1
Libraries:
- CUBLAS: 11.5.1
- CURAND: 10.2.4
- CUFFT: 10.4.2
- CUSOLVER: 11.1.2
- CUSPARSE: 11.6.0
- CUPTI: 14.0.0
- NVML: 11.0.0+465.19.1
Downloaded artifact: CUDNN
- CUDNN: 8.20.0 (for CUDA 11.3.0)
Downloaded artifact: CUTENSOR
- CUTENSOR: 1.3.0 (for CUDA 11.2.0)
Toolchain:
- Julia: 1.6.1
- LLVM: 11.0.1
- PTX ISA support: 3.2, 4.0, 4.1, 4.2, 4.3, 5.0, 6.0, 6.1, 6.3, 6.4, 6.5, 7.0
- Device capability support: sm_35, sm_37, sm_50, sm_52, sm_53, sm_60, sm_61, sm_62, sm_70, sm_72, sm_75, sm_80
1 device:
0: NVIDIA GeForce RTX 2070 (sm_75, 7.658 GiB / 7.793 GiB available)
CuDevice(0): NVIDIA GeForce RTX 2070
length(CUDA.devices()) = 1
CUDA.capability(CUDA.device()) = v"7.5.0"
CUDA.warpsize(CUDA.device()) = 32
CUDA.toolkit() = CUDA.Deps.ArtifactToolkit(v"11.3.1", "/home/jonathan/.julia/artifacts/0ed779f9e1d7b0c9905eb5be2c4349d32fbeb050")
CUDA.version() = v"11.3.0"
Mem.info() = (8222932992, 8367570944)
CUDA.synchronize() = nothing
NVML.driver_version() = v"465.19.1"
NVML.version() = v"11.0.0+465.19.1"
NVML.cuda_driver_version() = v"11.3.0"
NVML.memory_info(nvmldev) = (total = 8367570944, free = 8222932992, used = 144637952)
CUBLAS.handle() = Ptr{Nothing} @0x000000000bbfa390
CUBLAS.version() = v"11.5.1"
CUDNN.handle() = Ptr{Nothing} @0x0000000004442370
CUDNN.version() = v"8.20.0"
Knet.LibKnet8.libknet8 = ""
gpu: Test Failed at /home/jonathan/.julia/packages/Knet/JXldi/test/gpu.jl:39
Expression: !(isempty(Knet.LibKnet8.libknet8))
Evaluated: !(isempty(""))
Stacktrace:
[1] macro expansion
@ ~/.julia/packages/Knet/JXldi/test/gpu.jl:39 [inlined]
[2] macro expansion
@ /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.6/Test/src/Test.jl:1151 [inlined]
[3] top-level scope
@ ~/.julia/packages/Knet/JXldi/test/gpu.jl:4
readdir(artifact"libknet8") = ["libknet8.so"]
Test Summary: | Pass Fail Total
gpu | 13 1 14
ERROR: LoadError: Some tests did not pass: 13 passed, 1 failed, 0 errored, 0 broken.
in expression starting at /home/jonathan/.julia/packages/Knet/JXldi/test/gpu.jl:3
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#664>, or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAN43JXEVTXYSKGS4PCOH4DTZPVCFANCNFSM5A6NUAMQ>
.
jonathan-laurent commented
Here it is:
jonathan@aurora:~$ ldd /home/jonathan/.julia/artifacts/672d091cec9838612ebc6ffdba49c8b4add7aff1/libknet8.so
linux-vdso.so.1 (0x00007ffc6d5fc000)
librt.so.1 => /lib/x86_64-linux-gnu/librt.so.1 (0x00007f25d09f1000)
libpthread.so.0 => /lib/x86_64-linux-gnu/libpthread.so.0 (0x00007f25d09ce000)
libdl.so.2 => /lib/x86_64-linux-gnu/libdl.so.2 (0x00007f25d09c8000)
libstdc++.so.6 => /lib/x86_64-linux-gnu/libstdc++.so.6 (0x00007f25d07e7000)
libm.so.6 => /lib/x86_64-linux-gnu/libm.so.6 (0x00007f25d0698000)
libgcc_s.so.1 => /lib/x86_64-linux-gnu/libgcc_s.so.1 (0x00007f25d067d000)
libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x00007f25d0489000)
/lib64/ld-linux-x86-64.so.2 (0x00007f25d1008000)
denizyuret commented
v1.4.8 should fix this, to be released later today...
jonathan-laurent commented
It works now. Thanks!