Bug: Vulkan backend unable to allocate memory when run across multiple GPUs for larger models

[richardanaya]

What happened?

I have two 24gb 7900xtx and i've noticed when I try to offload models to them that are definitely within their specs I get OOM errors. @0cc4m

Name and Version

.\server.exe -m ..\gguf_models\Cat-Llama-3-70B-instruct-Q4_K_M.gguf -ngl 100

What operating system are you seeing the problem on?

Windows

Relevant log output

PS Z:\llama_vulkan2> .\server.exe -m ..\gguf_models\Cat-Llama-3-70B-instruct-Q4_K_M.gguf -ngl 100
INFO [                    main] build info | tid="22172" timestamp=1717686385 build=3091 commit="2b338967"
INFO [                    main] system info | tid="22172" timestamp=1717686385 n_threads=8 n_threads_batch=-1 total_threads=16 system_info="AVX = 1 | AVX_VNNI = 0 | AVX2 = 1 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | AVX512_BF16 = 0 | FMA = 1 | NEON = 0 | SVE = 0 | ARM_FMA = 0 | F16C = 1 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | MATMUL_INT8 = 0 | LLAMAFILE = 1 | "
llama_model_loader: loaded meta data with 27 key-value pairs and 723 tensors from ..\gguf_models\Cat-Llama-3-70B-instruct-Q4_K_M.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              = Cat-Llama-3-70B-instruct
llama_model_loader: - kv   2:                          llama.block_count u32              = 80
llama_model_loader: - kv   3:                       llama.context_length u32              = 8192
llama_model_loader: - kv   4:                     llama.embedding_length u32              = 8192
llama_model_loader: - kv   5:                  llama.feed_forward_length u32              = 28672
llama_model_loader: - kv   6:                 llama.attention.head_count u32              = 64
llama_model_loader: - kv   7:              llama.attention.head_count_kv u32              = 8
llama_model_loader: - kv   8:                       llama.rope.freq_base f32              = 500000.000000
llama_model_loader: - kv   9:     llama.attention.layer_norm_rms_epsilon f32              = 0.000010
llama_model_loader: - kv  10:                          general.file_type u32              = 15
llama_model_loader: - kv  11:                           llama.vocab_size u32              = 128258
llama_model_loader: - kv  12:                 llama.rope.dimension_count u32              = 128
llama_model_loader: - kv  13:                       tokenizer.ggml.model str              = gpt2
llama_model_loader: - kv  14:                         tokenizer.ggml.pre str              = llama-bpe
llama_model_loader: - kv  15:                      tokenizer.ggml.tokens arr[str,128258]  = ["!", "\"", "#", "$", "%", "&", "'", ...
llama_model_loader: - kv  16:                  tokenizer.ggml.token_type arr[i32,128258]  = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - kv  17:                      tokenizer.ggml.merges arr[str,280147]  = ["─á ─á", "─á ─á─á─á", "─á─á ─á─á", "...
llama_model_loader: - kv  18:                tokenizer.ggml.bos_token_id u32              = 128000
llama_model_loader: - kv  19:                tokenizer.ggml.eos_token_id u32              = 128257
llama_model_loader: - kv  20:            tokenizer.ggml.padding_token_id u32              = 128001
llama_model_loader: - kv  21:                    tokenizer.chat_template str              = {% for message in messages %}{% if lo...
llama_model_loader: - kv  22:               general.quantization_version u32              = 2
llama_model_loader: - kv  23:                      quantize.imatrix.file str              = /models/Cat-Llama-3-70B-instruct-GGUF...
llama_model_loader: - kv  24:                   quantize.imatrix.dataset str              = /training_data/groups_merged.txt
llama_model_loader: - kv  25:             quantize.imatrix.entries_count i32              = 560
llama_model_loader: - kv  26:              quantize.imatrix.chunks_count i32              = 88
llama_model_loader: - type  f32:  161 tensors
llama_model_loader: - type q4_K:  441 tensors
llama_model_loader: - type q5_K:   40 tensors
llama_model_loader: - type q6_K:   81 tensors
llm_load_vocab: special tokens cache size = 258
llm_load_vocab: token to piece cache size = 0.8000 MB
llm_load_print_meta: format           = GGUF V3 (latest)
llm_load_print_meta: arch             = llama
llm_load_print_meta: vocab type       = BPE
llm_load_print_meta: n_vocab          = 128258
llm_load_print_meta: n_merges         = 280147
llm_load_print_meta: n_ctx_train      = 8192
llm_load_print_meta: n_embd           = 8192
llm_load_print_meta: n_head           = 64
llm_load_print_meta: n_head_kv        = 8
llm_load_print_meta: n_layer          = 80
llm_load_print_meta: n_rot            = 128
llm_load_print_meta: n_embd_head_k    = 128
llm_load_print_meta: n_embd_head_v    = 128
llm_load_print_meta: n_gqa            = 8
llm_load_print_meta: n_embd_k_gqa     = 1024
llm_load_print_meta: n_embd_v_gqa     = 1024
llm_load_print_meta: f_norm_eps       = 0.0e+00
llm_load_print_meta: f_norm_rms_eps   = 1.0e-05
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             = 28672
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  = 500000.0
llm_load_print_meta: freq_scale_train = 1
llm_load_print_meta: n_ctx_orig_yarn  = 8192
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: model type       = 70B
llm_load_print_meta: model ftype      = Q4_K - Medium
llm_load_print_meta: model params     = 70.55 B
llm_load_print_meta: model size       = 39.59 GiB (4.82 BPW)
llm_load_print_meta: general.name     = Cat-Llama-3-70B-instruct
llm_load_print_meta: BOS token        = 128000 '<|begin_of_text|>'
llm_load_print_meta: EOS token        = 128257 '<|im_end|>'
llm_load_print_meta: PAD token        = 128001 '<|end_of_text|>'
llm_load_print_meta: LF token         = 128 'Ä'
llm_load_print_meta: EOT token        = 128257 '<|im_end|>'
ggml_vulkan: Found 2 Vulkan devices:
Vulkan0: AMD Radeon RX 7900 XTX | uma: 0 | fp16: 1 | warp size: 64
Vulkan1: AMD Radeon RX 7900 XTX | uma: 0 | fp16: 1 | warp size: 64
llm_load_tensors: ggml ctx size =    1.10 MiB
llm_load_tensors: offloading 80 repeating layers to GPU
llm_load_tensors: offloading non-repeating layers to GPU
llm_load_tensors: offloaded 81/81 layers to GPU
llm_load_tensors:        CPU buffer size =   563.63 MiB
llm_load_tensors:    Vulkan0 buffer size = 20038.81 MiB
llm_load_tensors:    Vulkan1 buffer size = 19940.68 MiB
...................................................................................................
llama_new_context_with_model: n_ctx      = 8192
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  = 500000.0
llama_new_context_with_model: freq_scale = 1
llama_kv_cache_init:    Vulkan0 KV buffer size =  1312.00 MiB
llama_kv_cache_init:    Vulkan1 KV buffer size =  1248.00 MiB
llama_new_context_with_model: KV self size  = 2560.00 MiB, K (f16): 1280.00 MiB, V (f16): 1280.00 MiB
llama_new_context_with_model: Vulkan_Host  output buffer size =     0.98 MiB
llama_new_context_with_model:    Vulkan0 compute buffer size =  1104.00 MiB
llama_new_context_with_model:    Vulkan1 compute buffer size =  1104.00 MiB
llama_new_context_with_model: Vulkan_Host compute buffer size =    32.01 MiB
llama_new_context_with_model: graph nodes  = 2566
llama_new_context_with_model: graph splits = 3
ggml_vulkan: Device memory allocation of size 939524096 failed.
ggml_vulkan: vk::Device::allocateMemory: ErrorOutOfDeviceMemory

PS Z:\llama_vulkan2> .\server.exe -m ..\gguf_models\dolphin-2.6-mixtral-8x7b.Q6_K.gguf -ngl 100
INFO [                    main] build info | tid="6684" timestamp=1717686213 build=3091 commit="2b338967"
INFO [                    main] system info | tid="6684" timestamp=1717686213 n_threads=8 n_threads_batch=-1 total_threads=16 system_info="AVX = 1 | AVX_VNNI = 0 | AVX2 = 1 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | AVX512_BF16 = 0 | FMA = 1 | NEON = 0 | SVE = 0 | ARM_FMA = 0 | F16C = 1 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | MATMUL_INT8 = 0 | LLAMAFILE = 1 | "
llama_model_loader: loaded meta data with 25 key-value pairs and 995 tensors from ..\gguf_models\dolphin-2.6-mixtral-8x7b.Q6_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              = cognitivecomputations_dolphin-2.6-mix...
llama_model_loader: - kv   2:                       llama.context_length u32              = 32768
llama_model_loader: - kv   3:                     llama.embedding_length u32              = 4096
llama_model_loader: - kv   4:                          llama.block_count u32              = 32
llama_model_loader: - kv   5:                  llama.feed_forward_length u32              = 14336
llama_model_loader: - kv   6:                 llama.rope.dimension_count u32              = 128
llama_model_loader: - kv   7:                 llama.attention.head_count u32              = 32
llama_model_loader: - kv   8:              llama.attention.head_count_kv u32              = 8
llama_model_loader: - kv   9:                         llama.expert_count u32              = 8
llama_model_loader: - kv  10:                    llama.expert_used_count u32              = 2
llama_model_loader: - kv  11:     llama.attention.layer_norm_rms_epsilon f32              = 0.000010
llama_model_loader: - kv  12:                       llama.rope.freq_base f32              = 1000000.000000
llama_model_loader: - kv  13:                          general.file_type u32              = 18
llama_model_loader: - kv  14:                       tokenizer.ggml.model str              = llama
llama_model_loader: - kv  15:                      tokenizer.ggml.tokens arr[str,32002]   = ["<unk>", "<s>", "</s>", "<0x00>", "<...
llama_model_loader: - kv  16:                      tokenizer.ggml.scores arr[f32,32002]   = [0.000000, 0.000000, 0.000000, 0.0000...
llama_model_loader: - kv  17:                  tokenizer.ggml.token_type arr[i32,32002]   = [2, 3, 3, 6, 6, 6, 6, 6, 6, 6, 6, 6, ...
llama_model_loader: - kv  18:                tokenizer.ggml.bos_token_id u32              = 1
llama_model_loader: - kv  19:                tokenizer.ggml.eos_token_id u32              = 32000
llama_model_loader: - kv  20:            tokenizer.ggml.padding_token_id u32              = 0
llama_model_loader: - kv  21:               tokenizer.ggml.add_bos_token bool             = true
llama_model_loader: - kv  22:               tokenizer.ggml.add_eos_token bool             = false
llama_model_loader: - kv  23:                    tokenizer.chat_template str              = {{ bos_token }}{%- set ns = namespace...
llama_model_loader: - kv  24:               general.quantization_version u32              = 2
llama_model_loader: - type  f32:   65 tensors
llama_model_loader: - type  f16:   32 tensors
llama_model_loader: - type q8_0:   64 tensors
llama_model_loader: - type q6_K:  834 tensors
llm_load_vocab: special tokens cache size = 261
llm_load_vocab: token to piece cache size = 0.1637 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          = 32002
llm_load_print_meta: n_merges         = 0
llm_load_print_meta: n_ctx_train      = 32768
llm_load_print_meta: n_embd           = 4096
llm_load_print_meta: n_head           = 32
llm_load_print_meta: n_head_kv        = 8
llm_load_print_meta: n_layer          = 32
llm_load_print_meta: n_rot            = 128
llm_load_print_meta: n_embd_head_k    = 128
llm_load_print_meta: n_embd_head_v    = 128
llm_load_print_meta: n_gqa            = 4
llm_load_print_meta: n_embd_k_gqa     = 1024
llm_load_print_meta: n_embd_v_gqa     = 1024
llm_load_print_meta: f_norm_eps       = 0.0e+00
llm_load_print_meta: f_norm_rms_eps   = 1.0e-05
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             = 14336
llm_load_print_meta: n_expert         = 8
llm_load_print_meta: n_expert_used    = 2
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  = 1000000.0
llm_load_print_meta: freq_scale_train = 1
llm_load_print_meta: n_ctx_orig_yarn  = 32768
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: model type       = 8x7B
llm_load_print_meta: model ftype      = Q6_K
llm_load_print_meta: model params     = 46.70 B
llm_load_print_meta: model size       = 35.74 GiB (6.57 BPW)
llm_load_print_meta: general.name     = cognitivecomputations_dolphin-2.6-mixtral-8x7b
llm_load_print_meta: BOS token        = 1 '<s>'
llm_load_print_meta: EOS token        = 32000 '<|im_end|>'
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: EOT token        = 32000 '<|im_end|>'
ggml_vulkan: Found 2 Vulkan devices:
Vulkan0: AMD Radeon RX 7900 XTX | uma: 0 | fp16: 1 | warp size: 64
Vulkan1: AMD Radeon RX 7900 XTX | uma: 0 | fp16: 1 | warp size: 64
llm_load_tensors: ggml ctx size =    1.25 MiB
llm_load_tensors: offloading 32 repeating layers to GPU
llm_load_tensors: offloading non-repeating layers to GPU
llm_load_tensors: offloaded 33/33 layers to GPU
llm_load_tensors:        CPU buffer size =   102.55 MiB
llm_load_tensors:    Vulkan0 buffer size = 19334.84 MiB
llm_load_tensors:    Vulkan1 buffer size = 17162.72 MiB
....................................................................................................
llama_new_context_with_model: n_ctx      = 32768
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  = 1000000.0
llama_new_context_with_model: freq_scale = 1
llama_kv_cache_init:    Vulkan0 KV buffer size =  2176.00 MiB
llama_kv_cache_init:    Vulkan1 KV buffer size =  1920.00 MiB
llama_new_context_with_model: KV self size  = 4096.00 MiB, K (f16): 2048.00 MiB, V (f16): 2048.00 MiB
llama_new_context_with_model: Vulkan_Host  output buffer size =     0.24 MiB
ggml_vulkan: Device memory allocation of size 2248148992 failed.
ggml_vulkan: vk::Device::allocateMemory: ErrorOutOfDeviceMemory

[slaren]

On a side note, when the alloc_buffer function fails, it should return NULL. It should not crash or throw an exception.

[0cc4m]

@richardanaya This might be a case where your main GPU takes on a little bit more than the other one, and it also handles the UI, so it runs out of VRAM. Try running it with the second one as main gpu -mg 1 or set the environment variable GGML_VK_VISIBLE_DEVICES=1,0.

[richardanaya]

@0cc4m hmm, thanks, i'm going to close this. I was able to get it to run by reducing the number of layers, it was far less than I was expecting. I'll re-open if I can get some more specific vulkan issues.

[0cc4m]

@richardanaya I fixed a VRAM bug that affected llama-3 on Vulkan in #7947, can you check whether that solved your issue? 70B should fit on 48GB VRAM.

[richardanaya]

@0cc4m I think there's been a significant regression, llama-server.exe doesn't seem to detect two 7900xtxs anymore. Older versions of llama.cpp still see two graphics cards. Created a new bug here #7997

PS Z:\llama_vulkan_3> .\llama-server.exe -m  ..\gguf_models\dolphin-2.6-mixtral-8x7b.Q6_K.gguf -ngl 100
INFO [                    main] build info | tid="95340" timestamp=1718758584 build=3181 commit="37bef894"
INFO [                    main] system info | tid="95340" timestamp=1718758584 n_threads=8 n_threads_batch=-1 total_threads=16 system_info="AVX = 1 | AVX_VNNI = 0 | AVX2 = 1 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | AVX512_BF16 = 0 | FMA = 1 | NEON = 0 | SVE = 0 | ARM_FMA = 0 | F16C = 1 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | MATMUL_INT8 = 0 | LLAMAFILE = 1 | "
llama_model_loader: loaded meta data with 25 key-value pairs and 995 tensors from ..\gguf_models\dolphin-2.6-mixtral-8x7b.Q6_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              = cognitivecomputations_dolphin-2.6-mix...
llama_model_loader: - kv   2:                       llama.context_length u32              = 32768
llama_model_loader: - kv   3:                     llama.embedding_length u32              = 4096
llama_model_loader: - kv   4:                          llama.block_count u32              = 32
llama_model_loader: - kv   5:                  llama.feed_forward_length u32              = 14336
llama_model_loader: - kv   6:                 llama.rope.dimension_count u32              = 128
llama_model_loader: - kv   7:                 llama.attention.head_count u32              = 32
llama_model_loader: - kv   8:              llama.attention.head_count_kv u32              = 8
llama_model_loader: - kv   9:                         llama.expert_count u32              = 8
llama_model_loader: - kv  10:                    llama.expert_used_count u32              = 2
llama_model_loader: - kv  11:     llama.attention.layer_norm_rms_epsilon f32              = 0.000010
llama_model_loader: - kv  12:                       llama.rope.freq_base f32              = 1000000.000000
llama_model_loader: - kv  13:                          general.file_type u32              = 18
llama_model_loader: - kv  14:                       tokenizer.ggml.model str              = llama
llama_model_loader: - kv  15:                      tokenizer.ggml.tokens arr[str,32002]   = ["<unk>", "<s>", "</s>", "<0x00>", "<...
llama_model_loader: - kv  16:                      tokenizer.ggml.scores arr[f32,32002]   = [0.000000, 0.000000, 0.000000, 0.0000...
llama_model_loader: - kv  17:                  tokenizer.ggml.token_type arr[i32,32002]   = [2, 3, 3, 6, 6, 6, 6, 6, 6, 6, 6, 6, ...
llama_model_loader: - kv  18:                tokenizer.ggml.bos_token_id u32              = 1
llama_model_loader: - kv  19:                tokenizer.ggml.eos_token_id u32              = 32000
llama_model_loader: - kv  20:            tokenizer.ggml.padding_token_id u32              = 0
llama_model_loader: - kv  21:               tokenizer.ggml.add_bos_token bool             = true
llama_model_loader: - kv  22:               tokenizer.ggml.add_eos_token bool             = false
llama_model_loader: - kv  23:                    tokenizer.chat_template str              = {{ bos_token }}{%- set ns = namespace...
llama_model_loader: - kv  24:               general.quantization_version u32              = 2
llama_model_loader: - type  f32:   65 tensors
llama_model_loader: - type  f16:   32 tensors
llama_model_loader: - type q8_0:   64 tensors
llama_model_loader: - type q6_K:  834 tensors
llm_load_vocab: special tokens cache size = 261
llm_load_vocab: token to piece cache size = 0.1637 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          = 32002
llm_load_print_meta: n_merges         = 0
llm_load_print_meta: n_ctx_train      = 32768
llm_load_print_meta: n_embd           = 4096
llm_load_print_meta: n_head           = 32
llm_load_print_meta: n_head_kv        = 8
llm_load_print_meta: n_layer          = 32
llm_load_print_meta: n_rot            = 128
llm_load_print_meta: n_embd_head_k    = 128
llm_load_print_meta: n_embd_head_v    = 128
llm_load_print_meta: n_gqa            = 4
llm_load_print_meta: n_embd_k_gqa     = 1024
llm_load_print_meta: n_embd_v_gqa     = 1024
llm_load_print_meta: f_norm_eps       = 0.0e+00
llm_load_print_meta: f_norm_rms_eps   = 1.0e-05
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             = 14336
llm_load_print_meta: n_expert         = 8
llm_load_print_meta: n_expert_used    = 2
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  = 1000000.0
llm_load_print_meta: freq_scale_train = 1
llm_load_print_meta: n_ctx_orig_yarn  = 32768
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: model type       = 8x7B
llm_load_print_meta: model ftype      = Q6_K
llm_load_print_meta: model params     = 46.70 B
llm_load_print_meta: model size       = 35.74 GiB (6.57 BPW)
llm_load_print_meta: general.name     = cognitivecomputations_dolphin-2.6-mixtral-8x7b
llm_load_print_meta: BOS token        = 1 '<s>'
llm_load_print_meta: EOS token        = 32000 '<|im_end|>'
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: EOT token        = 32000 '<|im_end|>'
ggml_vulkan: Found 1 Vulkan devices:
Vulkan0: AMD Radeon RX 7900 XTX (AMD proprietary driver) | uma: 0 | fp16: 1 | warp size: 64
llm_load_tensors: ggml ctx size =    0.83 MiB
ggml_vulkan: Device memory allocation of size 1999831040 failed.
ggml_vulkan: vk::Device::allocateMemory: ErrorOutOfDeviceMemory
llama_model_load: error loading model: unable to allocate backend buffer
llama_load_model_from_file: failed to load model
llama_init_from_gpt_params: error: failed to load model '..\gguf_models\dolphin-2.6-mixtral-8x7b.Q6_K.gguf'
 ERR [              load_model] unable to load model | tid="95340" timestamp=1718758588 model="..\\gguf_models\\dolphin-2.6-mixtral-8x7b.Q6_K.gguf"

[richardanaya]

@0cc4m great news! multi-gpu is running with larger models than i've ever been able to run before :) I can get some 70b with 8k context

[richardanaya]

@0cc4m

PS Z:\llama> .\llama-bench.exe -m ..\gguf_models\Cat-Llama-3-70B-instruct-Q4_K_M.gguf

model	size	params	backend	ngl	test	t/s
ggml_vulkan: Found 2 Vulkan devices:
Vulkan0: AMD Radeon RX 7900 XTX (AMD proprietary driver)	uma: 0	fp16: 1	warp size: 64
Vulkan1: AMD Radeon RX 7900 XTX (AMD proprietary driver)	uma: 0	fp16: 1	warp size: 64
llama 70B Q4_K - Medium	39.59 GiB	70.55 B	Vulkan	99	pp512	108.94 ± 0.75
llama 70B Q4_K - Medium	39.59 GiB	70.55 B	Vulkan	99	tg128	14.57 ± 0.01

build: 557b653 (3197)