[BUG] Internal error in package accelerate and LLVM.PTX backend: CUDA Exception - misaligned address

[sergiodguezc]

Description
I encountered an internal error in the accelerate package while running my code. The error message was:

Internal error in package accelerate
Please submit a bug report at https://github.com/AccelerateHS/accelerate/issues

CUDA Exception: misaligned address

CallStack (from HasCallStack):
  internalError: Data.Array.Accelerate.LLVM.PTX.State:53:9

Steps to reproduce
Currently, I don't have an example to reproduce the error, but I am trying my best to get one. I will attach it to this issue as soon as I have it.

Expected behaviour
I expected the code to run without any errors.

Environment

• Accelerate : 1.3.0.0
• Accelerate backend(s): accelerate-llvm-ptx
  • Device Query

 CUDA Device Query (Runtime API) version (CUDART static linking)

Detected 1 CUDA Capable device(s)

Device 0: "NVIDIA GeForce GTX 760 (192-bit)"
  CUDA Driver Version / Runtime Version          11.4 / 10.2
  CUDA Capability Major/Minor version number:    3.0
  Total amount of global memory:                 1482 MBytes (1554251776 bytes)
  ( 6) Multiprocessors, (192) CUDA Cores/MP:     1152 CUDA Cores
  GPU Max Clock rate:                            889 MHz (0.89 GHz)
  Memory Clock rate:                             2800 Mhz
  Memory Bus Width:                              192-bit
  L2 Cache Size:                                 393216 bytes
  Maximum Texture Dimension Size (x,y,z)         1D=(65536), 2D=(65536, 65536), 3D=(4096, 4096, 4096)
  Maximum Layered 1D Texture Size, (num) layers  1D=(16384), 2048 layers
  Maximum Layered 2D Texture Size, (num) layers  2D=(16384, 16384), 2048 layers
  Total amount of constant memory:               65536 bytes
  Total amount of shared memory per block:       49152 bytes
  Total number of registers available per block: 65536
  Warp size:                                     32
  Maximum number of threads per multiprocessor:  2048
  Maximum number of threads per block:           1024
  Max dimension size of a thread block (x,y,z): (1024, 1024, 64)
  Max dimension size of a grid size    (x,y,z): (2147483647, 65535, 65535)
  Maximum memory pitch:                          2147483647 bytes
  Texture alignment:                             512 bytes
  Concurrent copy and kernel execution:          Yes with 1 copy engine(s)
  Run time limit on kernels:                     Yes
  Integrated GPU sharing Host Memory:            No
  Support host page-locked memory mapping:       Yes
  Alignment requirement for Surfaces:            Yes
  Device has ECC support:                        Disabled
  Device supports Unified Addressing (UVA):      Yes
  Device supports Compute Preemption:            No
  Supports Cooperative Kernel Launch:            No
  Supports MultiDevice Co-op Kernel Launch:      No
  Device PCI Domain ID / Bus ID / location ID:   0 / 1 / 0
  Compute Mode:
     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >

deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 11.4, CUDA Runtime Version = 10.2, NumDevs = 1
Result = PASS

• GHC: 8.10.7
• OS: Arch Linux 6.1.22-1

Additional context
Note that when using the LLVM.Native backend, the code runs without any errors. Please let me know if you need any additional information or if there's anything else I can do to help diagnose and fix this issue. Thank you.

[ivogabe]

We've recently also seen this in a larger project, but we couldn't create a small reproduction from that (yet). It would be really useful if you could create a small reproduction! That would make it a lot easier to debug.