Build for Android NDK with Vulkan fails #14984
Description
🐛 Describe the bug
Hello,
First of all thanks a lot for this package that really helps me !
I face a problem when I try to build ExecuTorch (branch viable/strict) with Vulkan for Android NDK version 27.3.13750724 (at make step, after running cmake) :
ninja: error: '_deps/executorch/backends/vulkan/runtime/graph/ops/glsl/*', needed by 'vulkan_compute_shaders/spv.cpp', missing and no known rule to make it
Without set(EXECUTORCH_BUILD_VULKAN ON CACHE BOOL "") it builds perfectly. Here is my CMakeLists.txt section to build Executorch :
FetchContent_Declare(
executorch
GIT_REPOSITORY https://github.com/pytorch/executorch.git
GIT_TAG "viable/strict"
GIT_PROGRESS TRUE
GIT_SUBMODULES_RECURSE TRUE
SOURCE_DIR "${CMAKE_CURRENT_BINARY_DIR}/_deps/executorch"
)
set(ET_VENV "${CMAKE_CURRENT_BINARY_DIR}/.venv_et")
execute_process(
COMMAND bash -lc
"/usr/bin/python3.12 -m venv \"${ET_VENV}\" \
&& ${ET_VENV}/bin/pip install --upgrade pip pyyaml \
&& ${ET_VENV}/bin/pip install torch --index-url https://download.pytorch.org/whl/cpu"
COMMAND_ECHO STDOUT
RESULT_VARIABLE _venv_rc
)
if(NOT _venv_rc EQUAL 0)
message(FATAL_ERROR "Error when creating ExecuTorch python venv")
endif()
set(PYTHON_EXECUTABLE "${ET_VENV}/bin/python" CACHE FILEPATH "")
set(EXECUTORCH_ENABLE_LOGGING ON CACHE BOOL "")
set(EXECUTORCH_BUILD_EXTENSION_DATA_LOADER ON CACHE BOOL "")
set(EXECUTORCH_BUILD_EXTENSION_FLAT_TENSOR ON CACHE BOOL "")
set(EXECUTORCH_BUILD_TESTS OFF CACHE BOOL "")
set(EXECUTORCH_BUILD_EXTENSION_MODULE ON CACHE BOOL "")
set(EXECUTORCH_BUILD_EXTENSION_NAMED_DATA_MAP ON CACHE BOOL "")
set(EXECUTORCH_BUILD_PYBIND OFF CACHE BOOL "")
set(EXECUTORCH_BUILD_PYTHON OFF CACHE BOOL "")
set(EXECUTORCH_ENABLE_PYTHON_API OFF CACHE BOOL "")
set(EXECUTORCH_BUILD_EXTENSION_TENSOR ON CACHE BOOL "")
set(EXECUTORCH_BUILD_KERNELS_OPTIMIZED ON CACHE BOOL "")
set(EXECUTORCH_BUILD_XNNPACK ON CACHE BOOL "")
set(EXECUTORCH_BUILD_VULKAN ON CACHE BOOL "")
FetchContent_MakeAvailable(executorch)
I've tried to execute https://github.com/pytorch/executorch/blob/main/backends/vulkan/runtime/gen_vulkan_spv.py with GLSLC from NDK and I have the same issue that is mentioned in : #14507
With the GLSLC from my computer, the gen_vulkan_spv.py is running well.
So like suggested in the issue 14507, I have downloaded a recent version of Vulkan like this :
execute_process(
COMMAND bash -lc
"mkdir -p \"${CMAKE_CURRENT_BINARY_DIR}/vulkan\" \
&& cd \"${CMAKE_CURRENT_BINARY_DIR}/vulkan\" \
&& curl -O https://sdk.lunarg.com/sdk/download/1.4.321.1/linux/vulkansdk-linux-x86_64-1.4.321.1.tar.xz \
&& tar xf vulkansdk-linux-x86_64-1.4.321.1.tar.xz"
COMMAND_ECHO STDOUT
RESULT_VARIABLE _dl_vulkan_rt
)
if(NOT _dl_vulkan_rt EQUAL 0)
message(FATAL_ERROR "Failed to download and extract Vulkan")
endif ()
set(ENV{PATH} "${CMAKE_CURRENT_BINARY_DIR}/vulkan/1.4.321.1/x86_64/bin:$ENV{PATH}")
# tried also with set(GLSLC_PATH ""${CMAKE_CURRENT_BINARY_DIR}/vulkan/1.4.321.1/x86_64/bin/glslc")
but I face the same issue when pointing to this glslc executable.
I'm sure that I missed something to build it correctly with Vulkan. Can you help me or give some advices to debug it ? I will continue to investigate it on my side but I'm quite lost to understand this issue. As I understand, the ShaderLibrary.cmake isn't producing the built shaders.
If needed, my code is avalaible at https://github.com/Ipsedo/PhyVR/tree/develop
Thank you in advance for your time.
Versions
$ python collect_env.py # My computer, not the android device
Collecting environment information...
PyTorch version: N/A
Is debug build: N/A
CUDA used to build PyTorch: N/A
ROCM used to build PyTorch: N/A
OS: Arch Linux (x86_64)
GCC version: (GCC) 15.2.1 20250813
Clang version: 20.1.8
CMake version: version 4.1.2
Libc version: glibc-2.42
Python version: 3.13.7 (main, Aug 15 2025, 12:34:02) [GCC 15.2.1 20250813] (64-bit runtime)
Python platform: Linux-6.17.1-arch1-1-x86_64-with-glibc2.42
Is CUDA available: N/A
CUDA runtime version: 13.0.88
CUDA_MODULE_LOADING set to: N/A
GPU models and configuration: GPU 0: NVIDIA GeForce RTX 3060 Laptop GPU
Nvidia driver version: 580.95.05
cuDNN version: Probably one of the following:
/usr/lib/libcudnn.so.9.12.0
/usr/lib/libcudnn_adv.so.9.12.0
/usr/lib/libcudnn_cnn.so.9.12.0
/usr/lib/libcudnn_engines_precompiled.so.9.12.0
/usr/lib/libcudnn_engines_runtime_compiled.so.9.12.0
/usr/lib/libcudnn_graph.so.9.12.0
/usr/lib/libcudnn_heuristic.so.9.12.0
/usr/lib/libcudnn_ops.so.9.12.0
Is XPU available: N/A
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: N/A
CPU:
Architecture : x86_64
Mode(s) opératoire(s) des processeurs : 32-bit, 64-bit
Tailles des adresses: 48 bits physical, 48 bits virtual
Boutisme : Little Endian
Processeur(s) : 16
Liste de processeur(s) en ligne : 0-15
Identifiant constructeur : AuthenticAMD
Nom de modèle : AMD Ryzen 7 5800H with Radeon Graphics
Famille de processeur : 25
Modèle : 80
Thread(s) par cœur : 2
Cœur(s) par socket : 8
Socket(s) : 1
Révision : 0
Accroissement de fréquence : activé
multiplication des MHz du/des CPU(s) : 42%
Vitesse maximale du processeur en MHz : 4465,2612
Vitesse minimale du processeur en MHz : 403,4880
BogoMIPS : 6387,76
Drapeaux : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl xtopology nonstop_tsc cpuid extd_apicid aperfmperf rapl pni pclmulqdq monitor ssse3 fma cx16 sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_llc mwaitx cpb cat_l3 cdp_l3 hw_pstate ssbd mba ibrs ibpb stibp vmmcall fsgsbase bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local user_shstk clzero irperf xsaveerptr rdpru wbnoinvd cppc arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic v_vmsave_vmload vgif v_spec_ctrl umip pku ospke vaes vpclmulqdq rdpid overflow_recov succor smca fsrm debug_swap
Virtualisation : AMD-V
Cache L1d : 256 KiB (8 instances)
Cache L1i : 256 KiB (8 instances)
Cache L2 : 4 MiB (8 instances)
Cache L3 : 16 MiB (1 instance)
Nœud(s) NUMA : 1
Nœud NUMA 0 de processeur(s) : 0-15
Vulnérabilité Gather data sampling : Not affected
Vulnérabilité Ghostwrite : Not affected
Vulnérabilité Indirect target selection : Not affected
Vulnérabilité Itlb multihit : Not affected
Vulnérabilité L1tf : Not affected
Vulnérabilité Mds : Not affected
Vulnérabilité Meltdown : Not affected
Vulnérabilité Mmio stale data : Not affected
Vulnérabilité Old microcode : Not affected
Vulnérabilité Reg file data sampling : Not affected
Vulnérabilité Retbleed : Not affected
Vulnérabilité Spec rstack overflow : Vulnerable: Safe RET, no microcode
Vulnérabilité Spec store bypass : Mitigation; Speculative Store Bypass disabled via prctl
Vulnérabilité Spectre v1 : Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnérabilité Spectre v2 : Mitigation; Retpolines; IBPB conditional; IBRS_FW; STIBP always-on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected
Vulnérabilité Srbds : Not affected
Vulnérabilité Tsa : Vulnerable: No microcode
Vulnérabilité Tsx async abort : Not affected
Vulnérabilité Vmscape : Mitigation; IBPB before exit to userspace
Versions of relevant libraries:
[pip3] numpy==2.3.3
[conda] Could not collect