facebookresearch / silk

SiLK (Simple Learned Keypoint) is a self-supervised deep learning keypoint model.
GNU General Public License v3.0
644 stars 58 forks source link

silk-torch-script.py, nvrtc error #39

Closed BADAL244 closed 1 year ago

BADAL244 commented 1 year ago

/home/badalkumar/facebook/silk/silk/backbones/superpoint/utils.py:36: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs! if channels == 1: Traceback (most recent call last): File "silk-torch-script.py", line 73, in main() File "silk-torch-script.py", line 66, in main test_on_image_pair(model, script_model, downsampled_images) File "silk-torch-script.py", line 20, in test_on_image_pair positions_1, descriptors1, = script_model(images) File "/home/badalkumar/miniconda3/envs/silk/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl return forward_call(*input, **kwargs) RuntimeError: nvrtc: error: invalid value for --gpu-architecture (-arch)

nvrtc compilation failed:

define NAN __int_as_float(0x7fffffff)

define POS_INFINITY __int_as_float(0x7f800000)

define NEG_INFINITY __int_as_float(0xff800000)

template device T maximum(T a, T b) { return isnan(a) ? a : (a > b ? a : b); }

template device T minimum(T a, T b) { return isnan(a) ? a : (a < b ? a : b); }

extern "C" global void fused_div_mul_batch_norm_sigmoid(float tinput26_1, float tinput31_1, float tdenom_1, float aten_sigmoid, float aten_mul, float constv, float const_v__2, float const_v1, float const_v_0, float const_v_1) { { float v = ldg(const_v_1 + 0ll); float v_1 = ldg(tinput31_1 + (long long)(threadIdx.x) + 512ll * (long long)(blockIdx.x)); float v_2 = ldg(tdenom_1 + ((long long)(threadIdx.x) + 512ll (long long)(blockIdx.x)) % 251424ll + 251424ll (((long long)(threadIdx.x) + 512ll (long long)(blockIdx.x)) / 32182272ll)); aten_mul[(long long)(threadIdx.x) + 512ll (long long)(blockIdx.x)] = v (v_1 / v_2); if (blockIdx.x<983ll ? 1 : 0) { if ((long long)(threadIdx.x) + 512ll (long long)(blockIdx.x)<502848ll ? 1 : 0) { if (blockIdx.x<983ll ? 1 : 0) { float const_v1_1 = __ldg(constv + 0ll); float const_v_0_1 = ldg(const_v_0 + 0ll); float v_3 = ldg(tinput26_1 + (long long)(threadIdx.x) + 512ll * (long long)(blockIdx.x)); float v_4 = ldg(const_v1 + 0ll); float v_5 = __ldg(const_v2 + 0ll); aten_sigmoid[(long long)(threadIdx.x) + 512ll (long long)(blockIdx.x)] = 1.f / (1.f + (expf(0.f - (v_3 ((rsqrtf(const_v__1_1 + 9.999999747378752e-06f)) const_v_0_1) + (v_4 - v_5 ((rsqrtf(const_v__1_1 + 9.999999747378752e-06f)) * const_v_0_1)))))); } } }} }

gleize commented 1 year ago

Hi @BADAL244,

This could be caused by nvidia drivers that are too old (c.f. here). Try updating your drivers and see if that solves your issue.

gleize commented 1 year ago

No activity. Closing now.