CUDA Fatal Error While Trying to Run Under Windows 8.1

From: Justin Sanders (
Date: Thu Nov 06 2014 - 12:01:58 CST


I am trying to get the latest binaries of NAMD (2014-08-22) for Windows-64
with CUDA to run on my laptop running Windows 8.1. When I started to run a
simulation under VEGA ZZ, it gave me the following error:

FATAL ERROR: CUDA error on Pe 2 (Kaori device 0): All CUDA devices are in
prohibited mode, of compute capability 1.0, unable to map host memory, too
small, or otherwise unusable.

I checked the drivers and they are up to date. I ran samples from the most
recent CUDA toolkit (v 6.5) that I installed. The GPU was available and
communicating with the computer. But when tried to run the sample program,
it failed. I discovered that in my laptop, I had to set for programs to
have access to the GPU. After adjusting the setting, the sample program
worked. But NAMD still gives me the same error.

I was reading in the manual about ld_library_path variable. Is this part of
the problem? And if so, how would I go about setting it in Windows 8.1? I
have set environment variables in WinXP and IRIX before but not in 8.1.

Thank you.


P.S. Attached below is the output from the deviceQuery program distributed
in the CUDA Toolkit with technical information about the driver and GPU.

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

Detected 1 CUDA Capable device(s)

Device 0: "GeForce GT 740M"

  CUDA Driver Version / Runtime Version 6.5 / 6.5

  CUDA Capability Major/Minor version number: 3.5

  Total amount of global memory: 2048 MBytes (2147483648

  ( 2) Multiprocessors, (192) CUDA Cores/MP: 384 CUDA Cores

  GPU Clock rate: 1033 MHz (1.03 GHz)

  Memory Clock rate: 800 Mhz

  Memory Bus Width: 64-bit

  L2 Cache Size: 524288 bytes

  Maximum Texture Dimension Size (x,y,z) 1D=(65536), 2D=(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

  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

  CUDA Device Driver Mode (TCC or WDDM): WDDM (Windows Display
Driver Mo


  Device supports Unified Addressing (UVA): Yes

  Device PCI Bus ID / PCI location ID: 4 / 0

  Compute Mode:

     < Default (multiple host threads can use ::cudaSetDevice() with device

ltaneously) >

deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 6.5, CUDA Runtime

on = 6.5, NumDevs = 1, Device0 = GeForce GT 740M

Result = PASS

This archive was generated by hypermail 2.1.6 : Thu Dec 31 2015 - 23:21:20 CST