P: Slow performance on Xeon CPUs

[Dimizu]

I noticed that LR clone and brush tool on my XEON E5-1650 0 3.20GHz (Attention: E5-1650 0 and not E5-1650 v4) can not stress my CPU and after x minutes of working LR slow down, until I have to restart it.

Please see the full diskussion with the problem here: https://forums.adobe.com/thread/2210245 (test with different Lightroom versions, confrontation with a weak laptop, that works fine, tests wit other graphic grafic card, test with other bios settings etc. No results. Only restart LR or minor display resolution helps.)

Can anyone with an XEON E5-1650 0 3.2Ghz confirm this?

[Lufthansen]

I second that. I had better performance for a couple of days, then trouble again. There is a difference: Before throttling the CPU threads, the Lightroom made my whole computer sluggish, i.e. typical memory leak. Now, with the config.lua or a similar effect done with a bat-file (start /affinity F cmd.exe /c "c:\Program Files\Adobe\Adobe Lightroom\lightroom.exe" ), the computer remains stable, but Lightroom crashes frequently, especially when switching between library and develop module. Not yet able to point to what exactly makes it crash. I've noticed earlier that the metadata dialog (which reads from the sql-lite) tends to slow everything dowm when batch editing +100 images. My catalog has some 200.000 images, majority are dng-files. For what it's worth, the throttling of the cpu definitely seems to have isolated the problem to be with the Lightroom.

[FeedbackCommunityMember]

googleapps@getsatisfaction.com

[scottwild]

Agreed Mihael, this is a joke. 

[excorp11]

10.1 was just released.  Can anyone on the Adobe side tell us if it addresses anything on this thread?  Should be still be using the config.lua patch with the new update?

[David Kkozi]

I have the same question 🙂

[simonsaith]

Lr 6.10.1 is a minor release for additional camera and lens support and some critical bug fixes introduced in the last release. The team is still investigating the issue and the config.lua workaround would continue to work in Lr 6.10.1. Depending on the customer feedbacks and internal validation, we might make the config.lua workaround a permanent fix in the coming release.

How does people feel about the current config.lua workaround so far? Thumbs up, thumbs down, makes no difference? It may not solve all the performance problems that people reported in this thread. There are other leads that the team is following up.

[Assaf Frank]

Hi Simon,
for me the main issue is the degradation in performance after working on images, this affects the in develop module, creating panoramas, doing spot removal and file export. So I have to restart every 5 minutes and to be honest many times I just carry on working slow. This is painful on 80MP phase one raw files. however after restart things are much faster for few minutes which shows you can have it fast already.

when lightroom  become slower (after 5 minutes or so) the CPU usage do not reach as high as it would after restart. It as if it has a cap of how fast the software allows it to use the CPU resources.

if there is a switch we can try which will force the allocation 100% priority to the current task (e.g. working in development module or doing export) let me know and I will test it.

it is very frustrating as I have reported the performance degradation in lightroom 8 mouth ago
https://feedback.photoshop.com/photoshop_family/topics/lightroom-image-export-is-much-faster-after-r...

on the positive side, at least you are talking to us now.

Lightroom version: CC 2015.10.1 [ 1117303 ]
License: Creative Cloud
Operating system: Windows 10
Version: 10.0
Application architecture: x64
System architecture: x64
Logical processor count: 16
Processor speed: 3.0 GHz
Built-in memory: 65446.8 MB
Real memory available to Lightroom: 65446.8 MB
Real memory used by Lightroom: 1162.4 MB (1.7%)
Virtual memory used by Lightroom: 1438.3 MB
GDI objects count: 638
USER objects count: 2395
Process handles count: 1370
Memory cache size: 73.3MB / 16105.7MB (0.5%)
Maximum thread count used by Camera Raw: 8
Camera Raw SIMD optimization: SSE2,AVX,AVX2
Camera Raw virtual memory: 226MB / 32723MB (0%)
System DPI setting: 144 DPI (high DPI mode)
Desktop composition enabled: Yes
Displays: 1) 3840x2160
Input types: Multitouch: No, Integrated touch: No, Integrated pen: Yes, External touch: No, External pen: Yes, Keyboard: No

Graphics Processor Info: 
GeForce GTX 980 Ti/PCIe/SSE2

Check OpenGL support: Passed
Vendor: NVIDIA Corporation
Version: 3.3.0 NVIDIA 382.05
Renderer: GeForce GTX 980 Ti/PCIe/SSE2
LanguageVersion: 3.30 NVIDIA via Cg compiler


Application folder: C:\Program Files\Adobe\Adobe Lightroom
Library Path: L:\LR Catalog\LR 6 Catalog Full\LR 6 Catalog Full.lrcat
Settings Folder: C:\Users\User\AppData\Roaming\Adobe\Lightroom

Installed Plugins: 
1) ColorChecker Passport
2) Export to Photomatix Pro
3) jb Search Replace Transfer
4) jf Metadata Wrangler
5) jf Snapshot on Export
6) LR/Transporter
7) Stockboxphoto

Config.lua flags: 
Develop.AdjustMaximumThreadCount = 0.5

Adapter #1: Vendor : 10de
Device : 17c8
Subsystem : 32331462
Revision : a1
Video Memory : 6103
Adapter #2: Vendor : 1414
Device : 8c
Subsystem : 0
Revision : 0
Video Memory : 0
AudioDeviceIOBlockSize: 1024
AudioDeviceName: Speakers (Realtek High Definition Audio)
AudioDeviceNumberOfChannels: 2
AudioDeviceSampleRate: 48000
Build: Uninitialized
Direct2DEnabled: false
GL_ACCUM_ALPHA_BITS: 16
GL_ACCUM_BLUE_BITS: 16
GL_ACCUM_GREEN_BITS: 16
GL_ACCUM_RED_BITS: 16
GL_ALPHA_BITS: 0
GL_BLUE_BITS: 8
GL_DEPTH_BITS: 24
GL_GREEN_BITS: 8
GL_MAX_3D_TEXTURE_SIZE: 2048
GL_MAX_TEXTURE_SIZE: 16384
GL_MAX_TEXTURE_UNITS: 4
GL_MAX_VIEWPORT_DIMS: 16384,16384
GL_RED_BITS: 8
GL_RENDERER: GeForce GTX 980 Ti/PCIe/SSE2
GL_SHADING_LANGUAGE_VERSION: 4.50 NVIDIA
GL_STENCIL_BITS: 8
GL_VENDOR: NVIDIA Corporation
GL_VERSION: 4.5.0 NVIDIA 382.05
GPUDeviceEnabled: false
OGLEnabled: true
GL_EXTENSIONS: GL_AMD_multi_draw_indirect GL_AMD_seamless_cubemap_per_texture GL_AMD_vertex_shader_viewport_index GL_AMD_vertex_shader_layer GL_ARB_arrays_of_arrays GL_ARB_base_instance GL_ARB_bindless_texture 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_conservative_depth GL_ARB_compute_shader GL_ARB_compute_variable_group_size GL_ARB_conditional_render_inverted 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_indirect GL_ARB_draw_elements_base_vertex GL_ARB_draw_instanced GL_ARB_enhanced_layouts GL_ARB_ES2_compatibility GL_ARB_ES3_compatibility GL_ARB_ES3_1_compatibility GL_ARB_ES3_2_compatibility 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_fragment_shader_interlock GL_ARB_framebuffer_no_attachments GL_ARB_framebuffer_object GL_ARB_framebuffer_sRGB GL_ARB_geometry_shader4 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_imaging 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_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_locations 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_storage_buffer_object GL_ARB_shader_subroutine GL_ARB_shader_texture_image_samples GL_ARB_shader_texture_lod GL_ARB_shading_language_100 GL_ARB_shader_viewport_layer_array GL_ARB_shading_language_420pack GL_ARB_shading_language_include GL_ARB_shading_language_packing GL_ARB_shadow GL_ARB_sparse_buffer GL_ARB_sparse_texture GL_ARB_sparse_texture2 GL_ARB_sparse_texture_clamp 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_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_ATI_draw_buffers GL_ATI_texture_float GL_ATI_texture_mirror_once GL_S3_s3tc GL_EXT_texture_env_add GL_EXT_abgr GL_EXT_bgra GL_EXT_bindable_uniform 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_Cg_shader GL_EXT_depth_bounds_test 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_EXTX_framebuffer_mixed_formats GL_EXT_framebuffer_multisample_blit_scaled GL_EXT_framebuffer_object GL_EXT_framebuffer_sRGB GL_EXT_geometry_shader4 GL_EXT_gpu_program_parameters GL_EXT_gpu_shader4 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_post_depth_coverage GL_EXT_provoking_vertex GL_EXT_raster_multisample GL_EXT_rescale_normal GL_EXT_secondary_color GL_EXT_separate_shader_objects GL_EXT_separate_specular_color GL_EXT_shader_image_load_formatted GL_EXT_shader_image_load_store GL_EXT_shader_integer_mix GL_EXT_shadow_funcs GL_EXT_sparse_texture2 GL_EXT_stencil_two_side GL_EXT_stencil_wrap 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_combine GL_EXT_texture_env_dot3 GL_EXT_texture_filter_anisotropic GL_EXT_texture_filter_minmax GL_EXT_texture_integer GL_EXT_texture_lod GL_EXT_texture_lod_bias GL_EXT_texture_mirror_clamp GL_EXT_texture_object GL_EXT_texture_shared_exponent GL_EXT_texture_sRGB GL_EXT_texture_sRGB_decode GL_EXT_texture_storage GL_EXT_texture_swizzle GL_EXT_timer_query GL_EXT_transform_feedback2 GL_EXT_vertex_array GL_EXT_vertex_array_bgra GL_EXT_vertex_attrib_64bit GL_EXT_window_rectangles GL_EXT_import_sync_object GL_IBM_rasterpos_clip GL_IBM_texture_mirrored_repeat GL_KHR_context_flush_control GL_KHR_debug GL_KHR_no_error GL_KHR_robust_buffer_access_behavior GL_KHR_robustness GL_KTX_buffer_region GL_NV_alpha_to_coverage_dither_control GL_NV_bindless_multi_draw_indirect GL_NV_bindless_multi_draw_indirect_count GL_NV_bindless_texture GL_NV_blend_equation_advanced GL_NV_blend_equation_advanced_coherent GL_NV_blend_square GL_NV_command_list GL_NV_compute_program5 GL_NV_conditional_render GL_NV_conservative_raster GL_NV_conservative_raster_dilate GL_NV_copy_depth_to_color GL_NV_copy_image GL_NV_depth_buffer_float GL_NV_depth_clamp GL_NV_draw_texture GL_NV_draw_vulkan_image GL_NV_ES1_1_compatibility GL_NV_ES3_1_compatibility GL_NV_explicit_multisample GL_NV_fence GL_NV_fill_rectangle GL_NV_float_buffer GL_NV_fog_distance GL_NV_fragment_coverage_to_color GL_NV_fragment_program GL_NV_fragment_program_option GL_NV_fragment_program2 GL_NV_fragment_shader_interlock GL_NV_framebuffer_mixed_samples GL_NV_framebuffer_multisample_coverage GL_NV_geometry_shader4 GL_NV_geometry_shader_passthrough GL_NV_gpu_program4 GL_NV_internalformat_sample_query GL_NV_gpu_program4_1 GL_NV_gpu_program5 GL_NV_gpu_program5_mem_extended GL_NV_gpu_program_fp64 GL_NV_gpu_shader5 GL_NV_half_float GL_NV_light_max_exponent GL_NV_multisample_coverage GL_NV_multisample_filter_hint GL_NV_occlusion_query GL_NV_packed_depth_stencil GL_NV_parameter_buffer_object GL_NV_parameter_buffer_object2 GL_NV_path_rendering GL_NV_path_rendering_shared_edge GL_NV_pixel_data_range GL_NV_point_sprite GL_NV_primitive_restart GL_NV_register_combiners GL_NV_register_combiners2 GL_NV_sample_locations GL_NV_sample_mask_override_coverage GL_NV_shader_atomic_counters GL_NV_shader_atomic_float GL_NV_shader_atomic_fp16_vector GL_NV_shader_atomic_int64 GL_NV_shader_buffer_load GL_NV_shader_storage_buffer_object GL_NV_texgen_reflection GL_NV_texture_barrier GL_NV_texture_compression_vtc GL_NV_texture_env_combine4 GL_NV_texture_multisample GL_NV_texture_rectangle GL_NV_texture_shader GL_NV_texture_shader2 GL_NV_texture_shader3 GL_NV_transform_feedback GL_NV_transform_feedback2 GL_NV_uniform_buffer_unified_memory GL_NV_vertex_array_range GL_NV_vertex_array_range2 GL_NV_vertex_attrib_integer_64bit GL_NV_vertex_buffer_unified_memory GL_NV_vertex_program GL_NV_vertex_program1_1 GL_NV_vertex_program2 GL_NV_vertex_program2_option GL_NV_vertex_program3 GL_NV_viewport_array2 GL_NV_viewport_swizzle GL_NVX_conditional_render GL_NVX_gpu_memory_info GL_NVX_multigpu_info GL_NVX_nvenc_interop GL_NV_shader_thread_group GL_NV_shader_thread_shuffle GL_KHR_blend_equation_advanced GL_KHR_blend_equation_advanced_coherent GL_SGIS_generate_mipmap GL_SGIS_texture_lod GL_SGIX_depth_texture GL_SGIX_shadow GL_SUN_slice_accum GL_WIN_swap_hint WGL_EXT_swap_control

[FeedbackCommunityMember]

OK FOLKS: 
I made extensive testing, to answer my curiosity why some guys have no problem and some have and why my 5 year old laptop performs better than 6-core monster workstation. Here are my conclusion: 

Adobe sucks... ok, joke aside. 

Lr normal uses up to 35% of my CPU and cpu cooler does not even wake. For regular slider tweaking it is semi-smooth. For using more intensive slider like barrel and  NR, it starts to "build up" and I nned to leave it to settle. My other processor always pump 100% out of CPU, so it does never "build up" in  the queu line of commands. Lightroom does need to "digest" if I slide to much and go back and forth with images too quickly. So for culing Lr sucks as it is. 

Using perspective correction sliders it stals. CPU is 15% max and just one CPU is peaking at 80-90% while load balancing does jump that single proces to various cores. CPU cooler goes like mad since ONE core doing something. Of course overal CPU temperature is not even warm, but that proofs just one core is doing something. The tool is useless! Almost the same is with local brushes. CPU is going from 10-20% and only evry second thread is used (6 out of my 12). CPU coller spins fast, brushes are jittering and are not easy to work with. 

BUT Lr pulls 100% CPU while importing images, rendering 1:1 previews and EXPORTING. It really blows through 100's of images on my monster machine. :))

NOW famous "config.lua" :))))

Editing is now smooth like butter. YOu can not overload CPU/buffer no matter how quickly and long term you slide like madman everything from A-Z. Even perspective and NR... Now we are talking editing! CPU jumps up to 85%... CPU cooler spins slightly more than nothing. Lr seem to work normal now. But...

There is no real difference in local brush performance. The perspective correction is much improved now and is usable again. 

But not so fast. The "config.lua" have it's price! Sicne we are effectively disconnect half of virtual cores that reflects on importing, rendering and exporting performance, which now tops at 60%. 

And the nVidia GPU problem is there. If enabled, the brush effect/overlay is not visible until you stop brushing. If disabled, the brush effect/overlay can be seen, but movement is not smoot either. It can just jump somewhere... or stick to the mouse when button is already released... 

I will not use the hack, sicne it is not solution, and I mostly use Lr just for DAM. There is other processor that have no probles at all. They just significantly raise the subscription recently which speaks for itself. 

And the answer Why low end systems does not suffer:

The problem as it seems effects only i7 processors. More cores greater the problem. While i3 processors have HyperThreading, it is unlikely that many does use it, and if they, they do not expect super performance out of it, or are probably some begginers that does not hit the roof of Lr capabilities so easily. But most users does have some sort of i5 processors which do not have HyperThreading and so their system does not have such a problem when Lr want to squeeze more threads than it should at the same time, which obviously clogs the pipeline of high end machines with great number of cores and HT. 

[naestyoh]

So why aren't you using Smart Preview edit mode with 80MP phase one files? The whole point is to give you a proxy file to edit. Similarly, you can batch convert your 80MP files into 20MP proxy DNG and swap them out after you edit. 

[ChrisDPage]

I am losing the will to live......adobe needs to get their finger out, my i7 can happily churn using all cores on Lightroom processing, it needs them but when I am working through adding keywords I gradually get slower and slower performance until I get a non responding lightroom. If I kill lightroom before it becomes terminal my PC continues with no issues. Basic software problems on professional, and not cheap, software really isn't acceptable in this day and age.

[scottwild]

Looks like Adobe has finally decided to listen to it's customers. Please take the time and fill out this official survey: http://www.canonrumors.com/do-you-wish-lightroom-was-quicker-adobe-does-too/

[matthewg51914714]

Here's a direct link to the survey. 
https://www.surveymonkey.com/r/LrDesktop_performance

Thanks for calling it out Scott.

[FeedbackCommunityMember]

It saems noticably amount of users stopped using it. Capture One can afford to raise the prices. Or it is just time buying (again). 

[Assaf Frank]

The fear of Adobe to acknowledge to their community that lightroom performance is below standard for a very long time has resulted in a lot of anger hate and suffering to say the least. Use the Force Adobe, Rewrite Lightroom and get us all off the path to the dark side.
THE LIGHTROOM MANIFESTO: ADOBE HAS LOST ITS WAY
ADOBE WANTS TO FIX LIGHTROOM & WANTS YOUR INPUT | BUT DO WE NEED A FIX OR A NEW BUILD?

[FeedbackCommunityMember]

We need new build from scratch. There is no fix possible for Lr anymoore since it is 10+ year design. It is odd where oll billions of dollars of income goes?? It seems development gets only fraction of it. At best.

[Diko.bg]

OK. Lovely explanation. 🙂 Thanks. Now this rises the second logical question: Why 0.51? 😉 Have you simulated it to be the best? Isn't this ratio number optimization of the traffic processor model and even vendor dependent? 😉

[Diko.bg]

Yes.... This is my biggest issue mr. Chen! What is the issue with the local adjustments? Both healing spot and the brush?

[simonsaith]

If you are an existing Lr Classic CC customer, and would like to test out a developer preview build that we think fixes this performance issue. Please send me a note and I'll make sure that you are invited to the prerelease program. Thanks.

[rrrad]

I would definitely be interested in testing. I am a real estate photographer and I just got a Nikon D850. The raw files are huge and I can really feel it in Lightroom.

reed@TripleRPhotography.com

[simonsaith]

Hi Reed, you should see an invitation email to join the program.

[rrrad]

Super, Thanks!

[naestyoh]

How we looking for upcoming 18-core Xeon iMac pro?

[hagenm16374940]

I also would like testing. I have a catalog with 150.000 RAW photos. Speed is always an issue.
info@hagens-world-photography.de

[FeedbackCommunityMember]

Hi Simon, any chance you could invite me to the prerelease program too? This issue is causing me a great deal of frustration (and expense in terms of trying to solve it by throwing more hardware at the problem).

[ChrisDPage]

I hope Adobe are not just solving performance problems on xeons!