Can we run a trame application using Docker and in a Windows machine and still get access to the GPU? #829

amine0110 · 2025-11-16T11:16:46Z

amine0110
Nov 16, 2025

Hello,
I have a Trame application that contains several features, including volume rendering and 3D meshes (STLs) visualization. The app works great when we run it natively, but when we build a Docker image using this base kitware/trame:py3.10-glvnd, the app runs very slowly.

I did a lot of searching, but most of the time, I get to a point where it says, We have the GPU experience of a Trame app when dockerized on a Windows machine.

For that, I wanted to confirm this information, or if there is an alternative to be able to Dockerize the app and run it in a Windows machine, and still have the app run smoothly with the GPU as it runs without Docker.

Thank you in advance for your response.

Answered by amine0110

Nov 17, 2025

Thanks @jourdain , I already did and in fact the missing env var was this: GALLIUM_DRIVER=d3d12, it was the missing part in my Dockerfile. I hope it helps the others :)

View full answer

jourdain · 2025-11-16T17:05:37Z

jourdain
Nov 16, 2025
Maintainer

Docker with GPU can run on Windows but there is some limitation in term of GPU memory available. I don't think you hit that limit.

Did you install the nvidia runtime to get access to the GPU? Do you pass the --gpus all argument?

One way to validate which rendering backend get used is to run that app inside docker until the EGL rendering from NVidia get selected.

HTH

0 replies

amine0110 · 2025-11-16T17:27:10Z

amine0110
Nov 16, 2025
Author

Hi @jourdain

Thank you for your response.

I have a question, is there a way to know the limit or when we hit it?

Because I tried several Dockerfile configs, with OSMESA (which uses CPU) and I also tried using EGL but I didn't see a difference.

I tried also using the argument --gpu all but nothing changed.

I run the command that you sent inside docker, here is the output:

2025-11-16 17:17:38.684 (   0.386s) [    7A75BB66B1C0]vtkXOpenGLRenderWindow.:1458  WARN| bad X server connection. DISPLAY=
EGL version string:  1.5
EGL vendor string:  Mesa Project
EGL client APIs:  OpenGL OpenGL_ES
EGL extensions:  EGL_EXT_create_context_robustness EGL_KHR_cl_event2 EGL_KHR_config_attribs EGL_KHR_context_flush_control EGL_KHR_create_context EGL_KHR_create_context_no_error EGL_KHR_fence_sync EGL_KHR_get_all_proc_addresses EGL_KHR_gl_colorspace EGL_KHR_gl_renderbuffer_image EGL_KHR_gl_texture_2D_image EGL_KHR_gl_texture_3D_image EGL_KHR_gl_texture_cubemap_image EGL_KHR_image_base EGL_KHR_no_config_context EGL_KHR_reusable_sync EGL_KHR_surfaceless_context EGL_EXT_pixel_format_float EGL_KHR_wait_sync EGL_MESA_configless_context EGL_MESA_drm_image EGL_MESA_query_driver
OpenGL vendor string:  Mesa
OpenGL renderer string:  llvmpipe (LLVM 15.0.7, 256 bits)
OpenGL version string:  4.5 (Compatibility Profile) Mesa 23.2.1-1ubuntu3.1~22.04.3
OpenGL extensions:
  GL_3DFX_texture_compression_FXT1
  GL_AMD_conservative_depth
  GL_AMD_draw_buffers_blend
  GL_AMD_multi_draw_indirect
  GL_AMD_pinned_memory
  GL_AMD_query_buffer_object
  GL_AMD_seamless_cubemap_per_texture
  GL_AMD_shader_stencil_export
  GL_AMD_shader_trinary_minmax
  GL_AMD_texture_texture4
  GL_AMD_vertex_shader_layer
  GL_AMD_vertex_shader_viewport_index
  GL_ANGLE_texture_compression_dxt3
  GL_ANGLE_texture_compression_dxt5
  GL_APPLE_packed_pixels
  GL_ARB_ES2_compatibility
  GL_ARB_ES3_1_compatibility
  GL_ARB_ES3_2_compatibility
  GL_ARB_ES3_compatibility
  GL_ARB_arrays_of_arrays
  GL_ARB_base_instance
  GL_ARB_blend_func_extended
  GL_ARB_buffer_storage
  GL_ARB_clear_buffer_object
  GL_ARB_clear_texture
  GL_ARB_clip_control
  GL_ARB_color_buffer_float
  GL_ARB_compatibility
  GL_ARB_compressed_texture_pixel_storage
  GL_ARB_compute_shader
  GL_ARB_conditional_render_inverted
  GL_ARB_conservative_depth
  GL_ARB_copy_buffer
  GL_ARB_copy_image
  GL_ARB_cull_distance
  GL_ARB_debug_output
  GL_ARB_depth_buffer_float
  GL_ARB_depth_clamp
  GL_ARB_depth_texture
  GL_ARB_derivative_control
  GL_ARB_direct_state_access
  GL_ARB_draw_buffers
  GL_ARB_draw_buffers_blend
  GL_ARB_draw_elements_base_vertex
  GL_ARB_draw_indirect
  GL_ARB_draw_instanced
  GL_ARB_enhanced_layouts
  GL_ARB_explicit_attrib_location
  GL_ARB_explicit_uniform_location
  GL_ARB_fragment_coord_conventions
  GL_ARB_fragment_layer_viewport
  GL_ARB_fragment_program
  GL_ARB_fragment_program_shadow
  GL_ARB_fragment_shader
  GL_ARB_framebuffer_no_attachments
  GL_ARB_framebuffer_object
  GL_ARB_framebuffer_sRGB
  GL_ARB_get_program_binary
  GL_ARB_get_texture_sub_image
  GL_ARB_gl_spirv
  GL_ARB_gpu_shader5
  GL_ARB_gpu_shader_fp64
  GL_ARB_gpu_shader_int64
  GL_ARB_half_float_pixel
  GL_ARB_half_float_vertex
  GL_ARB_indirect_parameters
  GL_ARB_instanced_arrays
  GL_ARB_internalformat_query
  GL_ARB_internalformat_query2
  GL_ARB_invalidate_subdata
  GL_ARB_map_buffer_alignment
  GL_ARB_map_buffer_range
  GL_ARB_multi_bind
  GL_ARB_multi_draw_indirect
  GL_ARB_multisample
  GL_ARB_multitexture
  GL_ARB_occlusion_query
  GL_ARB_occlusion_query2
  GL_ARB_parallel_shader_compile
  GL_ARB_pipeline_statistics_query
  GL_ARB_pixel_buffer_object
  GL_ARB_point_parameters
  GL_ARB_point_sprite
  GL_ARB_polygon_offset_clamp
  GL_ARB_post_depth_coverage
  GL_ARB_program_interface_query
  GL_ARB_provoking_vertex
  GL_ARB_query_buffer_object
  GL_ARB_robust_buffer_access_behavior
  GL_ARB_robustness
  GL_ARB_sample_shading
  GL_ARB_sampler_objects
  GL_ARB_seamless_cube_map
  GL_ARB_seamless_cubemap_per_texture
  GL_ARB_separate_shader_objects
  GL_ARB_shader_atomic_counter_ops
  GL_ARB_shader_atomic_counters
  GL_ARB_shader_ballot
  GL_ARB_shader_bit_encoding
  GL_ARB_shader_clock
  GL_ARB_shader_draw_parameters
  GL_ARB_shader_group_vote
  GL_ARB_shader_image_load_store
  GL_ARB_shader_image_size
  GL_ARB_shader_objects
  GL_ARB_shader_precision
  GL_ARB_shader_stencil_export
  GL_ARB_shader_storage_buffer_object
  GL_ARB_shader_subroutine
  GL_ARB_shader_texture_image_samples
  GL_ARB_shader_texture_lod
  GL_ARB_shader_viewport_layer_array
  GL_ARB_shading_language_100
  GL_ARB_shading_language_420pack
  GL_ARB_shading_language_include
  GL_ARB_shading_language_packing
  GL_ARB_shadow
  GL_ARB_spirv_extensions
  GL_ARB_stencil_texturing
  GL_ARB_sync
  GL_ARB_tessellation_shader
  GL_ARB_texture_barrier
  GL_ARB_texture_border_clamp
  GL_ARB_texture_buffer_object
  GL_ARB_texture_buffer_object_rgb32
  GL_ARB_texture_buffer_range
  GL_ARB_texture_compression
  GL_ARB_texture_compression_bptc
  GL_ARB_texture_compression_rgtc
  GL_ARB_texture_cube_map
  GL_ARB_texture_cube_map_array
  GL_ARB_texture_env_add
  GL_ARB_texture_env_combine
  GL_ARB_texture_env_crossbar
  GL_ARB_texture_env_dot3
  GL_ARB_texture_filter_anisotropic
  GL_ARB_texture_filter_minmax
  GL_ARB_texture_float
  GL_ARB_texture_gather
  GL_ARB_texture_mirror_clamp_to_edge
  GL_ARB_texture_mirrored_repeat
  GL_ARB_texture_multisample
  GL_ARB_texture_non_power_of_two
  GL_ARB_texture_query_levels
  GL_ARB_texture_query_lod
  GL_ARB_texture_rectangle
  GL_ARB_texture_rg
  GL_ARB_texture_rgb10_a2ui
  GL_ARB_texture_stencil8
  GL_ARB_texture_storage
  GL_ARB_texture_storage_multisample
  GL_ARB_texture_swizzle
  GL_ARB_texture_view
  GL_ARB_timer_query
  GL_ARB_transform_feedback2
  GL_ARB_transform_feedback3
  GL_ARB_transform_feedback_instanced
  GL_ARB_transform_feedback_overflow_query
  GL_ARB_transpose_matrix
  GL_ARB_uniform_buffer_object
  GL_ARB_vertex_array_bgra
  GL_ARB_vertex_array_object
  GL_ARB_vertex_attrib_64bit
  GL_ARB_vertex_attrib_binding
  GL_ARB_vertex_buffer_object
  GL_ARB_vertex_program
  GL_ARB_vertex_shader
  GL_ARB_vertex_type_10f_11f_11f_rev
  GL_ARB_vertex_type_2_10_10_10_rev
  GL_ARB_viewport_array
  GL_ARB_window_pos
  GL_ARM_shader_framebuffer_fetch_depth_stencil
  GL_ATI_blend_equation_separate
  GL_ATI_draw_buffers
  GL_ATI_fragment_shader
  GL_ATI_meminfo
  GL_ATI_separate_stencil
  GL_ATI_texture_compression_3dc
  GL_ATI_texture_env_combine3
  GL_ATI_texture_float
  GL_ATI_texture_mirror_once
  GL_EXT_EGL_image_storage
  GL_EXT_EGL_sync
  GL_EXT_abgr
  GL_EXT_bgra
  GL_EXT_blend_color
  GL_EXT_blend_equation_separate
  GL_EXT_blend_func_separate
  GL_EXT_blend_minmax
  GL_EXT_blend_subtract
  GL_EXT_compiled_vertex_array
  GL_EXT_copy_texture
  GL_EXT_debug_label
  GL_EXT_direct_state_access
  GL_EXT_draw_buffers2
  GL_EXT_draw_instanced
  GL_EXT_draw_range_elements
  GL_EXT_fog_coord
  GL_EXT_framebuffer_blit
  GL_EXT_framebuffer_multisample
  GL_EXT_framebuffer_multisample_blit_scaled
  GL_EXT_framebuffer_object
  GL_EXT_framebuffer_sRGB
  GL_EXT_gpu_program_parameters
  GL_EXT_gpu_shader4
  GL_EXT_memory_object
  GL_EXT_memory_object_fd
  GL_EXT_multi_draw_arrays
  GL_EXT_packed_depth_stencil
  GL_EXT_packed_float
  GL_EXT_packed_pixels
  GL_EXT_pixel_buffer_object
  GL_EXT_point_parameters
  GL_EXT_polygon_offset_clamp
  GL_EXT_provoking_vertex
  GL_EXT_rescale_normal
  GL_EXT_secondary_color
  GL_EXT_separate_specular_color
  GL_EXT_shader_framebuffer_fetch
  GL_EXT_shader_framebuffer_fetch_non_coherent
  GL_EXT_shader_integer_mix
  GL_EXT_shadow_funcs
  GL_EXT_stencil_two_side
  GL_EXT_stencil_wrap
  GL_EXT_subtexture
  GL_EXT_texture
  GL_EXT_texture3D
  GL_EXT_texture_array
  GL_EXT_texture_buffer_object
  GL_EXT_texture_compression_dxt1
  GL_EXT_texture_compression_latc
  GL_EXT_texture_compression_rgtc
  GL_EXT_texture_compression_s3tc
  GL_EXT_texture_cube_map
  GL_EXT_texture_edge_clamp
  GL_EXT_texture_env_add
  GL_EXT_texture_env_combine
  GL_EXT_texture_env_dot3
  GL_EXT_texture_filter_anisotropic
  GL_EXT_texture_filter_minmax
  GL_EXT_texture_integer
  GL_EXT_texture_lod_bias
  GL_EXT_texture_mirror_clamp
  GL_EXT_texture_object
  GL_EXT_texture_rectangle
  GL_EXT_texture_sRGB
  GL_EXT_texture_sRGB_R8
  GL_EXT_texture_sRGB_RG8
  GL_EXT_texture_sRGB_decode
  GL_EXT_texture_shadow_lod
  GL_EXT_texture_shared_exponent
  GL_EXT_texture_snorm
  GL_EXT_texture_swizzle
  GL_EXT_timer_query
  GL_EXT_transform_feedback
  GL_EXT_vertex_array
  GL_EXT_vertex_array_bgra
  GL_EXT_vertex_attrib_64bit
  GL_IBM_multimode_draw_arrays
  GL_IBM_rasterpos_clip
  GL_IBM_texture_mirrored_repeat
  GL_INGR_blend_func_separate
  GL_INTEL_shader_atomic_float_minmax
  GL_KHR_blend_equation_advanced
  GL_KHR_blend_equation_advanced_coherent
  GL_KHR_context_flush_control
  GL_KHR_debug
  GL_KHR_no_error
  GL_KHR_parallel_shader_compile
  GL_KHR_robust_buffer_access_behavior
  GL_KHR_robustness
  GL_KHR_texture_compression_astc_ldr
  GL_KHR_texture_compression_astc_sliced_3d
  GL_MESA_framebuffer_flip_y
  GL_MESA_pack_invert
  GL_MESA_shader_integer_functions
  GL_MESA_texture_signed_rgba
  GL_MESA_window_pos
  GL_MESA_ycbcr_texture
  GL_NVX_gpu_memory_info
  GL_NV_ES1_1_compatibility
  GL_NV_blend_square
  GL_NV_conditional_render
  GL_NV_copy_depth_to_color
  GL_NV_copy_image
  GL_NV_depth_clamp
  GL_NV_fog_distance
  GL_NV_half_float
  GL_NV_light_max_exponent
  GL_NV_packed_depth_stencil
  GL_NV_primitive_restart
  GL_NV_shader_atomic_float
  GL_NV_texgen_reflection
  GL_NV_texture_barrier
  GL_NV_texture_env_combine4
  GL_NV_texture_rectangle
  GL_OES_EGL_image
  GL_OES_read_format
  GL_S3_s3tc
  GL_SGIS_generate_mipmap
  GL_SGIS_texture_border_clamp
  GL_SGIS_texture_edge_clamp
  GL_SGIS_texture_lod
  GL_SUN_multi_draw_arrays

Here is the Dockerfile, I am using:

FROM kitware/trame:py3.10-glvnd

# extra libs your image already installs
RUN apt-get update \
    && apt-get install -y libxrender1 \
    && rm -rf /var/lib/apt/lists/*

# -------------------------------------
# 1. Application code
# -------------------------------------
COPY --chown=trame-user:trame-user . /deploy

# -------------------------------------
# 2. Custom Apache vhost  (<<< NEW)
# -------------------------------------
COPY apache/001-trame.conf /etc/apache2/sites-available/001-trame.conf
RUN a2dissite 000-default.conf \
 && a2ensite 001-trame.conf

# -------------------------------------
# 3. Build the Trame server directory
# -------------------------------------
ENV TRAME_CLIENT_TYPE=vue2
RUN /opt/trame/entrypoint.sh build

And this is my requirements.txt:

--extra-index-url https://wheels.vtk.org
# trame
# trame-vtk
# trame-vuetify


# vtk==9.3.1
vtk==9.5.2
trame==3.12.0
trame-vtk==2.9.1
trame-vuetify==3.0.2
acvl_utils==0.2.5
tqdm==4.67.1
# pandas==2.3.2
# dynamic-network-architectures==0.4.2
nibabel==5.3.2
matplotlib==3.10.5
seaborn==0.13.2
graphviz==0.21
SimpleITK==2.5.2
dicom2nifti==2.6.2
scipy==1.15.3
imagecodecs==2025.3.30
yacs==0.1.8
python-dotenv==1.1.1
# vedo==2025.5.4
tomli>=2.0.1; python_version<'3.11'
wslink==2.4.0
pydantic==2.11.9

Am I missing something in the configs?

0 replies

jourdain · 2025-11-16T21:58:04Z

jourdain
Nov 16, 2025
Maintainer

EGL vendor string: Mesa Project => CPU

You want NVidia if you want fast rendering. Please read carefully this guide.

6 replies

jourdain Nov 17, 2025
Maintainer

Did you install the nvidia-runtime for docker? If you did not, you will never have access to your GPU inside docker.
A first step would be to run nvidia-smi inside docker to ensure that docker can see your GPU. After that, it could be related to some driver issue specific to EGL implementation selection (nvidia vs mesa).

Also using vtk>=9.5 is all you need as it include X, EGL and OSMesa.

Side note, OSMesa is not the same thing as EGL with Mesa implementation.

amine0110 Nov 17, 2025
Author

Yes I verified all these, and it seems everything is fine, but what I noticed is that when I use this docker run --gpus all image_name it is just showing this:

 * Restarting Apache httpd web server apache2
AH00558: apache2: Could not reliably determine the server's fully qualified domain name, using 172.17.0.2. Set the 'ServerName' directive globally to suppress this message
   ...done.
Starting server...
Starting the wslink launcher at

But the app is not starting, I should use docker compose up in order for the app to run!

jourdain Nov 17, 2025
Maintainer

If you want professional support we can help you get to the bottom of that fast.

amine0110 Nov 17, 2025
Author

Thanks @jourdain , I already did and in fact the missing env var was this: GALLIUM_DRIVER=d3d12, it was the missing part in my Dockerfile. I hope it helps the others :)

Answer selected by amine0110

jourdain Nov 17, 2025
Maintainer

I see that is a specific Windows setup. Thanks for sharing your finding.

jourdain · 2025-11-16T21:59:47Z

jourdain
Nov 16, 2025
Maintainer

Also --gpu all is not the same thing as --gpus all

1 reply

amine0110 Nov 16, 2025
Author

Sorry, it was a typp

Uh oh!

Can we run a trame application using Docker and in a Windows machine and still get access to the GPU? #829

Uh oh!

amine0110 Nov 16, 2025

Replies: 4 comments · 7 replies

Uh oh!

jourdain Nov 16, 2025 Maintainer

Uh oh!

amine0110 Nov 16, 2025 Author

Uh oh!

jourdain Nov 16, 2025 Maintainer

Uh oh!

jourdain Nov 17, 2025 Maintainer

Uh oh!

amine0110 Nov 17, 2025 Author

Uh oh!

jourdain Nov 17, 2025 Maintainer

Uh oh!

Uh oh!

amine0110 Nov 17, 2025 Author

Uh oh!

jourdain Nov 17, 2025 Maintainer

Uh oh!

jourdain Nov 16, 2025 Maintainer

Uh oh!

amine0110 Nov 16, 2025 Author

amine0110
Nov 16, 2025

Replies: 4 comments 7 replies

jourdain
Nov 16, 2025
Maintainer

amine0110
Nov 16, 2025
Author

jourdain
Nov 16, 2025
Maintainer

jourdain Nov 17, 2025
Maintainer

amine0110 Nov 17, 2025
Author

jourdain Nov 17, 2025
Maintainer

amine0110 Nov 17, 2025
Author

jourdain Nov 17, 2025
Maintainer

jourdain
Nov 16, 2025
Maintainer

amine0110 Nov 16, 2025
Author