ggml: Add POOL2D OP for GPU acceleration to the Vulkan backend in the MobileVLM model. #9763

cyzero-kim · 2024-10-06T14:35:01Z

The MobileVLM model now supports inference acceleration through GPU by utilizing the Vulkan backend.
A GGML_OP_POOL_2D shader has been added. (Pooling)
The encoding performance of the CLIP model improved from 2.8s on the CPU to 0.7s on the GPU.
I have read the contributing guidelines
Self-reported review complexity:
- Low
- Medium
- High
test model : MobileVLM V2 1.7B (https://huggingface.co/ZiangWu/MobileVLM_V2-1.7B-GGUF)
Test image : https://raw.githubusercontent.com/neuralmagic/deepsparse/main/tests/deepsparse/pipelines/sample_images/buddy.jpeg

master (cpu) :

clip_model_load: CLIP using CPU backend

encode_image_with_clip: image encoded in  2887.00 ms by CLIP (   20.05 ms per image patch)

This image captures a delightful scene featuring a tan Labrador Retriever puppy. The puppy, with its yellow fur and black nose, is seen panting with its tongue sticking out, a common sign of excitement and happiness. The puppy is wearing a blue collar around its neck, adding a pop of color to the scene. The backdrop of the image is a tree-lined street, providing a serene and peaceful setting for this adorable pet. The puppy's position and the surrounding environment create a sense of harmony between the viewer and the natural world. The image is a beautiful representation of the joy and innocence that can be found in simple moments in our lives.
llama_perf_context_print:        load time =    5034.15 ms
llama_perf_context_print: prompt eval time =    1544.06 ms /   189 tokens (    8.17 ms per token,   122.40 tokens per second)
llama_perf_context_print:        eval time =    7537.75 ms /   255 runs   (   29.56 ms per token,    33.83 tokens per second)
llama_perf_context_print:       total time =   12628.64 ms /   444 tokens

master (vulkan) :

clip_model_load: CLIP using Vulkan backend
clip_model_load: text_encoder:   0
clip_model_load: vision_encoder: 1
clip_model_load: llava_projector:  1
clip_model_load: minicpmv_projector:  0
clip_model_load: model size:     567.51 MB
clip_model_load: metadata size:  0.13 MB
clip_model_load: params backend buffer size =  567.51 MB (379 tensors)
key clip.vision.image_grid_pinpoints not found in file
key clip.vision.mm_patch_merge_type not found in file
key clip.vision.image_crop_resolution not found in file
clip_model_load: compute allocated memory: 32.89 MB
llama_new_context_with_model: n_ctx      = 2048
llama_new_context_with_model: n_batch    = 2048
llama_new_context_with_model: n_ubatch   = 512
llama_new_context_with_model: flash_attn = 0
llama_new_context_with_model: freq_base  = 10000.0
llama_new_context_with_model: freq_scale = 1
llama_kv_cache_init: Intel(R) Iris(R) Xe Graphics KV buffer size =   320.00 MiB
llama_kv_cache_init: Vulkan_Host KV buffer size =    64.00 MiB
llama_new_context_with_model: KV self size  =  384.00 MiB, K (f16):  192.00 MiB, V (f16):  192.00 MiB
llama_new_context_with_model: Vulkan_Host  output buffer size =     0.12 MiB
llama_new_context_with_model: Intel(R) Iris(R) Xe Graphics compute buffer size =   117.77 MiB
llama_new_context_with_model: Vulkan_Host compute buffer size =    12.01 MiB
llama_new_context_with_model: graph nodes  = 774
llama_new_context_with_model: graph splits = 48
ggml_vulkan: Error: Missing op: POOL_2D
C:\work\llm\cyzero\llama.cpp\ggml\src\ggml-vulkan.cpp:5735: fatal error

PR :

clip_model_load: CLIP using Vulkan backend

encode_image_with_clip: image encoded in   799.99 ms by CLIP (    5.56 ms per image patch)

 In the image, a golden-colored Labrador Retriever is the main subject, happily posing for the camera. The dog's tongue, sticking out, is a playful sign of its contentment. The dog is wearing a blue collar with a red ribbon, adding a pop of color to its golden fur. The dog is standing on a sidewalk, with a sidewalk visible in the background, suggesting an urban setting.

The dog's face is turned towards the camera, giving a direct and engaging gaze. Its eyes are gently closed, as if it is enjoying a moment of tranquility. The dog's expression is one of happiness and contentment, suggesting it's at home in the city. The overall image paints a picture of a joyful and content dog enjoying a walk in the city.
llama_perf_context_print:        load time =    4446.97 ms
llama_perf_context_print: prompt eval time =    2376.80 ms /   189 tokens (   12.58 ms per token,    79.52 tokens per second)
llama_perf_context_print:        eval time =    6710.63 ms /   182 runs   (   36.87 ms per token,    27.12 tokens per second)
llama_perf_context_print:       total time =   11221.29 ms /   371 tokens

test-backend-ops

PS C:\work\llm\cyzero\llama.cpp.latest> .\build\bin\Release\test-backend-ops.exe -o POOL2D
ggml_vulkan: Found 1 Vulkan devices:
Vulkan0: Intel(R) Iris(R) Xe Graphics (Intel Corporation) | uma: 1 | fp16: 1 | warp size: 32
Testing 2 backends

Backend 1/2 (CPU)
  Skipping CPU backend
Backend 2/2 (Vulkan0)
  Backend name: Vulkan0
  1605/1605 tests passed
  Backend Vulkan0: OK

2/2 backends passed
OK

Full logs:

PS C:\work\llm\cyzero\llama.cpp\build\bin\Release> .\llama-llava-cli.exe -m 'C:\work\llm\MobileVLM_V2-1.7B-GGUF\ggml-model-q4_k.gguf' --mmproj C:\work\llm\MobileVLM_V2-1.7B-GGUF\mmproj-model-f16.gguf --image C:\work\llm\buddy.jpeg -ngl 20 -p "describe the image in detail."
ggml_vulkan: Found 1 Vulkan devices:
Vulkan0: Intel(R) Iris(R) Xe Graphics (Intel Corporation) | uma: 1 | fp16: 1 | warp size: 32
build: 3985 (7b59859b) with MSVC 19.35.32217.1 for x64
llama_load_model_from_file: using device Vulkan0 (Intel(R) Iris(R) Xe Graphics) - 16163 MiB free
llama_model_loader: loaded meta data with 23 key-value pairs and 219 tensors from C:\work\llm\MobileVLM_V2-1.7B-GGUF\ggml-model-q4_k.gguf (version GGUF V3 (latest))
llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
llama_model_loader: - kv   0:                       general.architecture str              = llama
llama_model_loader: - kv   1:                               general.name str              = Work
llama_model_loader: - kv   2:                           llama.vocab_size u32              = 32000
llama_model_loader: - kv   3:                       llama.context_length u32              = 2048
llama_model_loader: - kv   4:                     llama.embedding_length u32              = 2048
llama_model_loader: - kv   5:                          llama.block_count u32              = 24
llama_model_loader: - kv   6:                  llama.feed_forward_length u32              = 5632
llama_model_loader: - kv   7:                 llama.rope.dimension_count u32              = 128
llama_model_loader: - kv   8:                 llama.attention.head_count u32              = 16
llama_model_loader: - kv   9:              llama.attention.head_count_kv u32              = 16
llama_model_loader: - kv  10:     llama.attention.layer_norm_rms_epsilon f32              = 0.000001
llama_model_loader: - kv  11:                       llama.rope.freq_base f32              = 10000.000000
llama_model_loader: - kv  12:                          general.file_type u32              = 14
llama_model_loader: - kv  13:                       tokenizer.ggml.model str              = llama
llama_model_loader: - kv  14:                      tokenizer.ggml.tokens arr[str,32000]   = ["<unk>", "<s>", "</s>", "<0x00>", "<...
llama_model_loader: - kv  15:                      tokenizer.ggml.scores arr[f32,32000]   = [0.000000, 0.000000, 0.000000, 0.0000...
llama_model_loader: - kv  16:                  tokenizer.ggml.token_type arr[i32,32000]   = [2, 3, 3, 6, 6, 6, 6, 6, 6, 6, 6, 6, ...
llama_model_loader: - kv  17:                tokenizer.ggml.bos_token_id u32              = 1
llama_model_loader: - kv  18:                tokenizer.ggml.eos_token_id u32              = 2
llama_model_loader: - kv  19:            tokenizer.ggml.padding_token_id u32              = 0
llama_model_loader: - kv  20:               tokenizer.ggml.add_bos_token bool             = true
llama_model_loader: - kv  21:               tokenizer.ggml.add_eos_token bool             = false
llama_model_loader: - kv  22:               general.quantization_version u32              = 2
llama_model_loader: - type  f32:   49 tensors
llama_model_loader: - type q4_K:  162 tensors
llama_model_loader: - type q5_K:    7 tensors
llama_model_loader: - type q6_K:    1 tensors
llm_load_vocab: special_eos_id is not in special_eog_ids - the tokenizer config may be incorrect
llm_load_vocab: special tokens cache size = 3
llm_load_vocab: token to piece cache size = 0.1684 MB
llm_load_print_meta: format           = GGUF V3 (latest)
llm_load_print_meta: arch             = llama
llm_load_print_meta: vocab type       = SPM
llm_load_print_meta: n_vocab          = 32000
llm_load_print_meta: n_merges         = 0
llm_load_print_meta: vocab_only       = 0
llm_load_print_meta: n_ctx_train      = 2048
llm_load_print_meta: n_embd           = 2048
llm_load_print_meta: n_layer          = 24
llm_load_print_meta: n_head           = 16
llm_load_print_meta: n_head_kv        = 16
llm_load_print_meta: n_rot            = 128
llm_load_print_meta: n_swa            = 0
llm_load_print_meta: n_embd_head_k    = 128
llm_load_print_meta: n_embd_head_v    = 128
llm_load_print_meta: n_gqa            = 1
llm_load_print_meta: n_embd_k_gqa     = 2048
llm_load_print_meta: n_embd_v_gqa     = 2048
llm_load_print_meta: f_norm_eps       = 0.0e+00
llm_load_print_meta: f_norm_rms_eps   = 1.0e-06
llm_load_print_meta: f_clamp_kqv      = 0.0e+00
llm_load_print_meta: f_max_alibi_bias = 0.0e+00
llm_load_print_meta: f_logit_scale    = 0.0e+00
llm_load_print_meta: n_ff             = 5632
llm_load_print_meta: n_expert         = 0
llm_load_print_meta: n_expert_used    = 0
llm_load_print_meta: causal attn      = 1
llm_load_print_meta: pooling type     = 0
llm_load_print_meta: rope type        = 0
llm_load_print_meta: rope scaling     = linear
llm_load_print_meta: freq_base_train  = 10000.0
llm_load_print_meta: freq_scale_train = 1
llm_load_print_meta: n_ctx_orig_yarn  = 2048
llm_load_print_meta: rope_finetuned   = unknown
llm_load_print_meta: ssm_d_conv       = 0
llm_load_print_meta: ssm_d_inner      = 0
llm_load_print_meta: ssm_d_state      = 0
llm_load_print_meta: ssm_dt_rank      = 0
llm_load_print_meta: ssm_dt_b_c_rms   = 0
llm_load_print_meta: model type       = ?B
llm_load_print_meta: model ftype      = Q4_K - Small
llm_load_print_meta: model params     = 1.36 B
llm_load_print_meta: model size       = 754.43 MiB (4.64 BPW)
llm_load_print_meta: general.name     = Work
llm_load_print_meta: BOS token        = 1 '<s>'
llm_load_print_meta: EOS token        = 2 '</s>'
llm_load_print_meta: UNK token        = 0 '<unk>'
llm_load_print_meta: PAD token        = 0 '<unk>'
llm_load_print_meta: LF token         = 13 '<0x0A>'
llm_load_print_meta: EOG token        = 2 '</s>'
llm_load_print_meta: max token length = 48
llm_load_tensors: ggml ctx size =    0.20 MiB
llm_load_tensors: offloading 20 repeating layers to GPU
llm_load_tensors: offloaded 20/25 layers to GPU
llm_load_tensors:    Vulkan0 buffer size =   551.56 MiB
llm_load_tensors:        CPU buffer size =   754.43 MiB
...........................................................................................
clip_model_load: model name:   openai/clip-vit-large-patch14-336
clip_model_load: description:  image encoder for LLaVA
clip_model_load: GGUF version: 3
clip_model_load: alignment:    32
clip_model_load: n_tensors:    379
clip_model_load: n_kv:         19
clip_model_load: ftype:        f16

clip_model_load: loaded meta data with 19 key-value pairs and 379 tensors from C:\work\llm\MobileVLM_V2-1.7B-GGUF\mmproj-model-f16.gguf
clip_model_load: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
clip_model_load: - kv   0:                       general.architecture str              = clip
clip_model_load: - kv   1:                      clip.has_text_encoder bool             = false
clip_model_load: - kv   2:                    clip.has_vision_encoder bool             = true
clip_model_load: - kv   3:                   clip.has_llava_projector bool             = true
clip_model_load: - kv   4:                          general.file_type u32              = 1
clip_model_load: - kv   5:                               general.name str              = openai/clip-vit-large-patch14-336
clip_model_load: - kv   6:                        general.description str              = image encoder for LLaVA
clip_model_load: - kv   7:                        clip.projector_type str              = ldpv2
clip_model_load: - kv   8:                     clip.vision.image_size u32              = 336
clip_model_load: - kv   9:                     clip.vision.patch_size u32              = 14
clip_model_load: - kv  10:               clip.vision.embedding_length u32              = 1024
clip_model_load: - kv  11:            clip.vision.feed_forward_length u32              = 4096
clip_model_load: - kv  12:                 clip.vision.projection_dim u32              = 768
clip_model_load: - kv  13:           clip.vision.attention.head_count u32              = 16
clip_model_load: - kv  14:   clip.vision.attention.layer_norm_epsilon f32              = 0.000010
clip_model_load: - kv  15:                    clip.vision.block_count u32              = 23
clip_model_load: - kv  16:                     clip.vision.image_mean arr[f32,3]       = [0.481455, 0.457828, 0.408211]
clip_model_load: - kv  17:                      clip.vision.image_std arr[f32,3]       = [0.268630, 0.261303, 0.275777]
clip_model_load: - kv  18:                              clip.use_gelu bool             = false
clip_model_load: - type  f32:  236 tensors
clip_model_load: - type  f16:  143 tensors
clip_model_load: CLIP using Vulkan backend
clip_model_load: text_encoder:   0
clip_model_load: vision_encoder: 1
clip_model_load: llava_projector:  1
clip_model_load: minicpmv_projector:  0
clip_model_load: model size:     567.51 MB
clip_model_load: metadata size:  0.13 MB
clip_model_load: params backend buffer size =  567.51 MB (379 tensors)
key clip.vision.image_grid_pinpoints not found in file
key clip.vision.mm_patch_merge_type not found in file
key clip.vision.image_crop_resolution not found in file
clip_model_load: compute allocated memory: 32.89 MB
llama_new_context_with_model: n_ctx      = 2048
llama_new_context_with_model: n_batch    = 2048
llama_new_context_with_model: n_ubatch   = 512
llama_new_context_with_model: flash_attn = 0
llama_new_context_with_model: freq_base  = 10000.0
llama_new_context_with_model: freq_scale = 1
llama_kv_cache_init:    Vulkan0 KV buffer size =   320.00 MiB
llama_kv_cache_init: Vulkan_Host KV buffer size =    64.00 MiB
llama_new_context_with_model: KV self size  =  384.00 MiB, K (f16):  192.00 MiB, V (f16):  192.00 MiB
llama_new_context_with_model: Vulkan_Host  output buffer size =     0.12 MiB
llama_new_context_with_model:    Vulkan0 compute buffer size =   117.77 MiB
llama_new_context_with_model: Vulkan_Host compute buffer size =    12.01 MiB
llama_new_context_with_model: graph nodes  = 774
llama_new_context_with_model: graph splits = 48
encode_image_with_clip: image embedding created: 144 tokens

encode_image_with_clip: image encoded in   799.99 ms by CLIP (    5.56 ms per image patch)

 In the image, a golden-colored Labrador Retriever is the main subject, happily posing for the camera. The dog's tongue, sticking out, is a playful sign of its contentment. The dog is wearing a blue collar with a red ribbon, adding a pop of color to its golden fur. The dog is standing on a sidewalk, with a sidewalk visible in the background, suggesting an urban setting.

The dog's face is turned towards the camera, giving a direct and engaging gaze. Its eyes are gently closed, as if it is enjoying a moment of tranquility. The dog's expression is one of happiness and contentment, suggesting it's at home in the city. The overall image paints a picture of a joyful and content dog enjoying a walk in the city.
llama_perf_context_print:        load time =    4446.97 ms
llama_perf_context_print: prompt eval time =    2376.80 ms /   189 tokens (   12.58 ms per token,    79.52 tokens per second)
llama_perf_context_print:        eval time =    6710.63 ms /   182 runs   (   36.87 ms per token,    27.12 tokens per second)
llama_perf_context_print:       total time =   11221.29 ms /   371 tokens
PS C:\work\llm\cyzero\llama.cpp\build\bin\Release>

- The MobileVLM model now supports inference acceleration through GPU by utilizing the Vulkan backend. - A GGML_OP_POOL_2D shader has been added. (Pooling) - The encoding performance of the CLIP model improved from 2.8s on the CPU to 0.7s on the GPU. Signed-off-by: Changyeon Kim <[email protected]>

0cc4m · 2024-10-16T09:26:56Z

Apologies for the delay.

I don't think this is correct yet, I get this when running the unit tests:

» build_vk/bin/test-backend-ops -o POOL_2D                                                                                                                        130 ↵
ggml_vulkan: Found 1 Vulkan devices:
Vulkan0: Intel(R) UHD Graphics (CML GT2) (Intel open-source Mesa driver) | uma: 1 | fp16: 1 | warp size: 32
Testing 2 devices

Backend 1/2: Vulkan0
  Device description: Intel(R) UHD Graphics (CML GT2)
  Device memory: 7843 MB (7843 MB free)

  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=0,p1=0): OK
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=0,p1=1): [POOL_2D] NMSE = 2.030467935 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=1,p1=0): [POOL_2D] NMSE = 2.268705026 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=1,p1=1): OK
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=0,p1=0): [POOL_2D] NMSE = 3.050673744 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=0,p1=1): [POOL_2D] NMSE = 2.416454361 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=1,p1=0): [POOL_2D] NMSE = 2.688794475 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=1,p1=1): [POOL_2D] NMSE = 3.050729022 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=0,p1=0): [POOL_2D] NMSE = 3.232077375 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=0,p1=1): [POOL_2D] NMSE = 3.399366295 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=1,p1=0): [POOL_2D] NMSE = 2.552616749 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=1,p1=1): [POOL_2D] NMSE = 2.562324992 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=0,p1=0): OK
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=0,p1=1): [POOL_2D] NMSE = 1.935281866 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=1,p1=0): [POOL_2D] NMSE = 2.504371237 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=1,p1=1): OK
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=0,p1=0): [POOL_2D] NMSE = 1.336785832 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=0,p1=1): [POOL_2D] NMSE = 1.277472324 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=1,p1=0): [POOL_2D] NMSE = 2.375843840 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=1,p1=1): [POOL_2D] NMSE = 1.445155213 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=0,p1=0): [POOL_2D] NMSE = 3.140820330 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=0,p1=1): [POOL_2D] NMSE = 2.981867067 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=1,p1=0): [POOL_2D] NMSE = 4.346719475 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=1,p1=1): [POOL_2D] NMSE = 2.721390211 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=0,p1=0): [POOL_2D] NMSE = 2.495996146 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=0,p1=1): [POOL_2D] NMSE = 3.342858313 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=1,p1=0): [POOL_2D] NMSE = 2.575888212 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=1,p1=1): [POOL_2D] NMSE = 2.308845151 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=0,p1=0): [POOL_2D] NMSE = 1.030314006 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=0,p1=1): [POOL_2D] NMSE = 1.689569663 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=1,p1=0): [POOL_2D] NMSE = 1.862128496 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=1,p1=1): [POOL_2D] NMSE = 1.421018250 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=0,p1=0): [POOL_2D] NMSE = 1.248297401 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=0,p1=1): [POOL_2D] NMSE = 2.134703374 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=1,p1=0): [POOL_2D] NMSE = 1.371575790 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=1,p1=1): [POOL_2D] NMSE = 1.685003904 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=0,p1=0): [POOL_2D] NMSE = 2.183317800 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=0,p1=1): [POOL_2D] NMSE = 2.172274261 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=1,p1=0): [POOL_2D] NMSE = 2.871110642 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=1,p1=1): [POOL_2D] NMSE = 3.002149438 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=0,p1=0): [POOL_2D] NMSE = 2.754006227 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=0,p1=1): [POOL_2D] NMSE = 3.897215086 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=1,p1=0): [POOL_2D] NMSE = 2.602448242 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=1,p1=1): [POOL_2D] NMSE = 2.994456405 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=0,p1=0): [POOL_2D] NMSE = 1.570852520 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=0,p1=1): [POOL_2D] NMSE = 2.875054193 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=1,p1=0): [POOL_2D] NMSE = 1.251711103 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=1,p1=1): [POOL_2D] NMSE = 1.622370651 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=0,p1=0): OK
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=0,p1=1): [POOL_2D] NMSE = 1.280032333 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=1,p1=0): [POOL_2D] NMSE = 0.985044499 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=1,p1=1): [POOL_2D] NMSE = 0.042384638 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=0,p1=0): [POOL_2D] NMSE = 1.841743576 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=0,p1=1): [POOL_2D] NMSE = 1.988331853 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=1,p1=0): [POOL_2D] NMSE = 2.954893907 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=1,p1=1): [POOL_2D] NMSE = 3.374220742 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=0,p1=0): [POOL_2D] NMSE = 2.344994872 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=0,p1=1): [POOL_2D] NMSE = 3.964606402 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=1,p1=0): [POOL_2D] NMSE = 2.211845678 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=1,p1=1): [POOL_2D] NMSE = 2.354940776 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=0,p1=0): OK
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=0,p1=1): [POOL_2D] NMSE = 1.641584943 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=1,p1=0): [POOL_2D] NMSE = 1.152639983 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=1,p1=1): [POOL_2D] NMSE = 0.250634787 > 0.000000100 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=0,p1=0): OK
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=0.405328 CPU=-340282346638528859811704183484516925440.000000 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.840913 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=1,p1=1): OK
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=0,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.915980 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=0.177821 CPU=-340282346638528859811704183484516925440.000000 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=-0.180461 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.671645 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=0,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.869008 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-0.463808 CPU=-340282346638528859811704183484516925440.000000 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=-0.365907 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=-0.771264 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=0,p1=0): OK
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-0.835242 CPU=-340282346638528859811704183484516925440.000000 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.385887 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=1,p1=1): OK
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=0,p1=0): [POOL_2D] NMSE = 0.677671239 > 0.000000100 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.804175 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.614433 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.484188 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=0,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.664404 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.638225 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.246217 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.968062 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=0,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.403607 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=-0.531504 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.705453 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.768638 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=0,p1=0): [POOL_2D] NMSE = 0.636545302 > 0.000000100 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.792187 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.891921 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.668178 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=0,p1=0): [POOL_2D] NMSE = 0.643276464 > 0.000000100 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=0.294561 CPU=-340282346638528859811704183484516925440.000000 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.860586 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.777860 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=0,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.946818 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=0.843795 CPU=-340282346638528859811704183484516925440.000000 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.918683 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.807116 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=0,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.661981 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=0.901174 CPU=-340282346638528859811704183484516925440.000000 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.583113 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.947309 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=0,p1=0): [POOL_2D] NMSE = 0.720412981 > 0.000000100 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-0.031811 CPU=-340282346638528859811704183484516925440.000000 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.705607 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.930897 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=0,p1=0): OK
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.951253 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.404126 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.777326 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=0,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.954501 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.766055 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=-0.085524 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.459696 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=0,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.985425 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.861884 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.957567 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.410033 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=0,p1=0): OK
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.775315 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.926676 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.934520 FAIL
  1491/1607 tests passed
  Backend Vulkan0: FAIL

Backend 2/2: CPU
  Skipping CPU backend
1/2 backends passed
FAIL

Can you take a look? I'm still busy, but I should be able to take a closer look next week if you don't figure it out until then.

cyzero-kim · 2024-10-16T13:38:52Z

Apologies for the delay.

I don't think this is correct yet, I get this when running the unit tests:

» build_vk/bin/test-backend-ops -o POOL_2D                                                                                                                        130 ↵
ggml_vulkan: Found 1 Vulkan devices:
Vulkan0: Intel(R) UHD Graphics (CML GT2) (Intel open-source Mesa driver) | uma: 1 | fp16: 1 | warp size: 32
Testing 2 devices

Backend 1/2: Vulkan0
  Device description: Intel(R) UHD Graphics (CML GT2)
  Device memory: 7843 MB (7843 MB free)

  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=0,p1=0): OK
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=0,p1=1): [POOL_2D] NMSE = 2.030467935 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=1,p1=0): [POOL_2D] NMSE = 2.268705026 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=1,p1=1): OK
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=0,p1=0): [POOL_2D] NMSE = 3.050673744 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=0,p1=1): [POOL_2D] NMSE = 2.416454361 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=1,p1=0): [POOL_2D] NMSE = 2.688794475 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=1,p1=1): [POOL_2D] NMSE = 3.050729022 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=0,p1=0): [POOL_2D] NMSE = 3.232077375 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=0,p1=1): [POOL_2D] NMSE = 3.399366295 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=1,p1=0): [POOL_2D] NMSE = 2.552616749 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=1,p1=1): [POOL_2D] NMSE = 2.562324992 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=0,p1=0): OK
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=0,p1=1): [POOL_2D] NMSE = 1.935281866 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=1,p1=0): [POOL_2D] NMSE = 2.504371237 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=1,p1=1): OK
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=0,p1=0): [POOL_2D] NMSE = 1.336785832 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=0,p1=1): [POOL_2D] NMSE = 1.277472324 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=1,p1=0): [POOL_2D] NMSE = 2.375843840 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=1,p1=1): [POOL_2D] NMSE = 1.445155213 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=0,p1=0): [POOL_2D] NMSE = 3.140820330 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=0,p1=1): [POOL_2D] NMSE = 2.981867067 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=1,p1=0): [POOL_2D] NMSE = 4.346719475 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=1,p1=1): [POOL_2D] NMSE = 2.721390211 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=0,p1=0): [POOL_2D] NMSE = 2.495996146 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=0,p1=1): [POOL_2D] NMSE = 3.342858313 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=1,p1=0): [POOL_2D] NMSE = 2.575888212 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=1,p1=1): [POOL_2D] NMSE = 2.308845151 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=0,p1=0): [POOL_2D] NMSE = 1.030314006 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=0,p1=1): [POOL_2D] NMSE = 1.689569663 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=1,p1=0): [POOL_2D] NMSE = 1.862128496 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=1,p1=1): [POOL_2D] NMSE = 1.421018250 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=0,p1=0): [POOL_2D] NMSE = 1.248297401 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=0,p1=1): [POOL_2D] NMSE = 2.134703374 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=1,p1=0): [POOL_2D] NMSE = 1.371575790 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=1,p1=1): [POOL_2D] NMSE = 1.685003904 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=0,p1=0): [POOL_2D] NMSE = 2.183317800 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=0,p1=1): [POOL_2D] NMSE = 2.172274261 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=1,p1=0): [POOL_2D] NMSE = 2.871110642 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=1,p1=1): [POOL_2D] NMSE = 3.002149438 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=0,p1=0): [POOL_2D] NMSE = 2.754006227 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=0,p1=1): [POOL_2D] NMSE = 3.897215086 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=1,p1=0): [POOL_2D] NMSE = 2.602448242 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=1,p1=1): [POOL_2D] NMSE = 2.994456405 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=0,p1=0): [POOL_2D] NMSE = 1.570852520 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=0,p1=1): [POOL_2D] NMSE = 2.875054193 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=1,p1=0): [POOL_2D] NMSE = 1.251711103 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=1,p1=1): [POOL_2D] NMSE = 1.622370651 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=0,p1=0): OK
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=0,p1=1): [POOL_2D] NMSE = 1.280032333 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=1,p1=0): [POOL_2D] NMSE = 0.985044499 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=1,p1=1): [POOL_2D] NMSE = 0.042384638 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=0,p1=0): [POOL_2D] NMSE = 1.841743576 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=0,p1=1): [POOL_2D] NMSE = 1.988331853 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=1,p1=0): [POOL_2D] NMSE = 2.954893907 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=1,p1=1): [POOL_2D] NMSE = 3.374220742 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=0,p1=0): [POOL_2D] NMSE = 2.344994872 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=0,p1=1): [POOL_2D] NMSE = 3.964606402 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=1,p1=0): [POOL_2D] NMSE = 2.211845678 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=1,p1=1): [POOL_2D] NMSE = 2.354940776 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=0,p1=0): OK
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=0,p1=1): [POOL_2D] NMSE = 1.641584943 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=1,p1=0): [POOL_2D] NMSE = 1.152639983 > 0.000000100 FAIL
  POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=1,p1=1): [POOL_2D] NMSE = 0.250634787 > 0.000000100 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=0,p1=0): OK
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=0.405328 CPU=-340282346638528859811704183484516925440.000000 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.840913 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=1,p1=1): OK
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=0,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.915980 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=0.177821 CPU=-340282346638528859811704183484516925440.000000 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=-0.180461 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.671645 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=0,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.869008 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-0.463808 CPU=-340282346638528859811704183484516925440.000000 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=-0.365907 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=-0.771264 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=0,p1=0): OK
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-0.835242 CPU=-340282346638528859811704183484516925440.000000 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.385887 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=1,p1=1): OK
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=0,p1=0): [POOL_2D] NMSE = 0.677671239 > 0.000000100 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.804175 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.614433 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.484188 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=0,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.664404 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.638225 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.246217 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.968062 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=0,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.403607 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=-0.531504 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.705453 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.768638 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=0,p1=0): [POOL_2D] NMSE = 0.636545302 > 0.000000100 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.792187 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.891921 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.668178 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=0,p1=0): [POOL_2D] NMSE = 0.643276464 > 0.000000100 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=0.294561 CPU=-340282346638528859811704183484516925440.000000 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.860586 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.777860 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=0,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.946818 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=0.843795 CPU=-340282346638528859811704183484516925440.000000 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.918683 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.807116 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=0,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.661981 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=0.901174 CPU=-340282346638528859811704183484516925440.000000 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.583113 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.947309 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=0,p1=0): [POOL_2D] NMSE = 0.720412981 > 0.000000100 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-0.031811 CPU=-340282346638528859811704183484516925440.000000 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.705607 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.930897 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=0,p1=0): OK
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.951253 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.404126 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.777326 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=0,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.954501 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.766055 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=-0.085524 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.459696 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=0,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.985425 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.861884 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.957567 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.410033 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=0,p1=0): OK
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=0,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.775315 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=1,p1=0): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.926676 FAIL
  POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=1,p1=1): [POOL_2D] inf mismatch: Vulkan0=-340282346638528859811704183484516925440.000000 CPU=0.934520 FAIL
  1491/1607 tests passed
  Backend Vulkan0: FAIL

Backend 2/2: CPU
  Skipping CPU backend
1/2 backends passed
FAIL

Can you take a look? I'm still busy, but I should be able to take a closer look next week if you don't figure it out until then.

It seems that I probably made a mistake during the refactoring process. I will try to resolve it.

fix casting to int. Signed-off-by: Changyeon Kim <[email protected]>

cyzero-kim · 2024-10-27T01:40:48Z

@0cc4m Hello, I solved the problem. It was the wrong order of parameters.
Please review again.

PS C:\work\llm\cyzero\llama.cpp\build\bin\Release> ./test-backend-ops -o POOL_2D
ggml_vulkan: Found 1 Vulkan devices:
Vulkan0: Intel(R) Iris(R) Xe Graphics (Intel Corporation) | uma: 1 | fp16: 1 | warp size: 32
Testing 2 devices

Backend 1/2: Vulkan0
Device description: Intel(R) Iris(R) Xe Graphics
Device memory: 16163 MB (16163 MB free)

POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=0,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=0,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=1,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=1,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=0,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=0,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=1,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=1,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=0,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=0,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=1,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=1,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=0,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=0,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=1,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=1,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=0,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=0,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=1,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=1,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=0,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=0,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=1,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=1,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=0,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=0,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=1,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=1,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=0,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=0,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=1,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=1,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=0,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=0,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=1,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=1,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=0,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=0,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=1,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=1,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=0,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=0,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=1,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=1,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=0,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=0,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=1,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=1,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=0,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=0,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=1,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=1,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=0,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=0,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=1,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=1,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=0,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=0,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=1,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=1,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=0,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=0,p1=1): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=1,p1=0): OK
POOL_2D(pool_type=avg,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=1,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=0,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=0,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=1,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=1,p0=1,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=0,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=0,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=1,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=1,s1=2,p0=1,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=0,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=0,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=1,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=1,p0=1,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=0,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=0,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=1,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=1,s0=2,s1=2,p0=1,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=0,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=0,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=1,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=1,p0=1,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=0,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=0,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=1,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=1,s1=2,p0=1,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=0,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=0,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=1,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=1,p0=1,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=0,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=0,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=1,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=1,k1=3,s0=2,s1=2,p0=1,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=0,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=0,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=1,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=1,p0=1,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=0,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=0,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=1,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=1,s1=2,p0=1,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=0,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=0,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=1,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=1,p0=1,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=0,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=0,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=1,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=1,s0=2,s1=2,p0=1,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=0,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=0,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=1,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=1,p0=1,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=0,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=0,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=1,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=1,s1=2,p0=1,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=0,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=0,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=1,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=1,p0=1,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=0,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=0,p1=1): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=1,p1=0): OK
POOL_2D(pool_type=max,type_input=f32,ne_input=[10,10,3,1],k0=3,k1=3,s0=2,s1=2,p0=1,p1=1): OK
1777/1777 tests passed
Backend Vulkan0: OK

Backend 2/2: CPU
Skipping CPU backend
2/2 backends passed
OK

0cc4m

Thank you, I can confirm the tests go through now. Code looks good, too.

… MobileVLM model. (ggml-org#9763) * ggml: Add POOL2D OP for GPU ACC to the Vulkan. - The MobileVLM model now supports inference acceleration through GPU by utilizing the Vulkan backend. - A GGML_OP_POOL_2D shader has been added. (Pooling) - The encoding performance of the CLIP model improved from 2.8s on the CPU to 0.7s on the GPU. Signed-off-by: Changyeon Kim <[email protected]> * [fix] Correct the incorrect order of the parameters. fix casting to int. Signed-off-by: Changyeon Kim <[email protected]> --------- Signed-off-by: Changyeon Kim <[email protected]>

github-actions bot added Vulkan Issues specific to the Vulkan backend ggml changes relating to the ggml tensor library for machine learning labels Oct 6, 2024

cyzero-kim mentioned this pull request Oct 7, 2024

ggml: Add POOL2D OP for GPU ACC to the Vulkan backend in the MobileVLM model. #9733

Closed

4 tasks

0cc4m self-assigned this Oct 8, 2024

cyzero-kim added 2 commits October 27, 2024 00:56

[fix] Correct the incorrect order of the parameters.

3932fd5

fix casting to int. Signed-off-by: Changyeon Kim <[email protected]>

Merge branch 'ggerganov:master' into mobile_vlm

7b59859

0cc4m approved these changes Oct 29, 2024

View reviewed changes

0cc4m merged commit 8f275a7 into ggml-org:master Oct 29, 2024
53 checks passed

hnfong mentioned this pull request Nov 10, 2024

Bug: missing tensor blk.0.ffn_down_exps.weight when loading mixtral-8x7b-instruct-v0.1.Q5_K_M.gguf #10244

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

ggml: Add POOL2D OP for GPU acceleration to the Vulkan backend in the MobileVLM model. #9763

ggml: Add POOL2D OP for GPU acceleration to the Vulkan backend in the MobileVLM model. #9763

Uh oh!

cyzero-kim commented Oct 6, 2024 •

edited

Loading

Uh oh!

0cc4m commented Oct 16, 2024

Uh oh!

cyzero-kim commented Oct 16, 2024

Uh oh!

cyzero-kim commented Oct 27, 2024

Uh oh!

0cc4m left a comment

Uh oh!

Uh oh!

Uh oh!

ggml: Add POOL2D OP for GPU acceleration to the Vulkan backend in the MobileVLM model. #9763

ggml: Add POOL2D OP for GPU acceleration to the Vulkan backend in the MobileVLM model. #9763

Uh oh!

Conversation

cyzero-kim commented Oct 6, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

master (cpu) :

master (vulkan) :

PR :

test-backend-ops

Full logs:

Uh oh!

0cc4m commented Oct 16, 2024

Uh oh!

cyzero-kim commented Oct 16, 2024

Uh oh!

cyzero-kim commented Oct 27, 2024

Uh oh!

0cc4m left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

cyzero-kim commented Oct 6, 2024 •

edited

Loading