Datasets:

arm-bench
/

arm-bench-trace

Languages:

Size:

Tags:

License:

Dataset card Files Files and versions

arm-bench-trace / solutions /llama.cpp /reference-scalar /gemm

249 kB

Ctrl+K

Ctrl+K

5 contributors

History: 5 commits

awu12's picture

Llama-3.1-8B e2e set (defs, real-activation workloads, Fable gate-v2 solutions, runs, measurements); Qwen kernel-vs-repack results; fixed baseline-sve2 loop_114/121/219

def2c3e verified 23 days ago

gemm_bf16_n1024_k2048.json

3.22 kB
Add kernel in different precision: 3 months ago
gemm_bf16_n1408_k2048.json

3.22 kB
Add kernel in different precision: 3 months ago
gemm_bf16_n2048_k1024.json

3.22 kB
Add kernel in different precision: 3 months ago
gemm_bf16_n2048_k1408.json

3.22 kB
Add kernel in different precision: 3 months ago
gemm_bf16_n2048_k2048.json

3.22 kB
Add kernel in different precision: 3 months ago
gemm_ggml_q4_K_n1024_k2560.json

9.46 kB
Add Qwen3.5-4B e2e gemm definitions (packed ggml Q4_K/Q5_K/Q6_K ABI): 11 definitions + workloads + reference-scalar + baseline-llamacpp-arm (additive) 28 days ago
gemm_ggml_q4_K_n1024_k4096.json

9.46 kB
Llama-3.1-8B e2e set (defs, real-activation workloads, Fable gate-v2 solutions, runs, measurements); Qwen kernel-vs-repack results; fixed baseline-sve2 loop_114/121/219 23 days ago
gemm_ggml_q4_K_n14336_k4096.json

9.46 kB
Llama-3.1-8B e2e set (defs, real-activation workloads, Fable gate-v2 solutions, runs, measurements); Qwen kernel-vs-repack results; fixed baseline-sve2 loop_114/121/219 23 days ago
gemm_ggml_q4_K_n2560_k4096.json

9.46 kB
Add Qwen3.5-4B e2e gemm definitions (packed ggml Q4_K/Q5_K/Q6_K ABI): 11 definitions + workloads + reference-scalar + baseline-llamacpp-arm (additive) 28 days ago
gemm_ggml_q4_K_n2560_k9216.json

9.46 kB
Add Qwen3.5-4B e2e gemm definitions (packed ggml Q4_K/Q5_K/Q6_K ABI): 11 definitions + workloads + reference-scalar + baseline-llamacpp-arm (additive) 28 days ago
gemm_ggml_q4_K_n4096_k14336.json

9.46 kB
Llama-3.1-8B e2e set (defs, real-activation workloads, Fable gate-v2 solutions, runs, measurements); Qwen kernel-vs-repack results; fixed baseline-sve2 loop_114/121/219 23 days ago
gemm_ggml_q4_K_n4096_k2560.json

9.46 kB
Add Qwen3.5-4B e2e gemm definitions (packed ggml Q4_K/Q5_K/Q6_K ABI): 11 definitions + workloads + reference-scalar + baseline-llamacpp-arm (additive) 28 days ago
gemm_ggml_q4_K_n4096_k4096.json

9.46 kB
Llama-3.1-8B e2e set (defs, real-activation workloads, Fable gate-v2 solutions, runs, measurements); Qwen kernel-vs-repack results; fixed baseline-sve2 loop_114/121/219 23 days ago
gemm_ggml_q4_K_n8192_k2560.json

9.46 kB
Add Qwen3.5-4B e2e gemm definitions (packed ggml Q4_K/Q5_K/Q6_K ABI): 11 definitions + workloads + reference-scalar + baseline-llamacpp-arm (additive) 28 days ago
gemm_ggml_q4_K_n9216_k2560.json

9.46 kB
Add Qwen3.5-4B e2e gemm definitions (packed ggml Q4_K/Q5_K/Q6_K ABI): 11 definitions + workloads + reference-scalar + baseline-llamacpp-arm (additive) 28 days ago
gemm_ggml_q5_K_n2560_k4096.json

9.84 kB
Add Qwen3.5-4B e2e gemm definitions (packed ggml Q4_K/Q5_K/Q6_K ABI): 11 definitions + workloads + reference-scalar + baseline-llamacpp-arm (additive) 28 days ago
gemm_ggml_q5_K_n8192_k2560.json

9.84 kB
Add Qwen3.5-4B e2e gemm definitions (packed ggml Q4_K/Q5_K/Q6_K ABI): 11 definitions + workloads + reference-scalar + baseline-llamacpp-arm (additive) 28 days ago
gemm_ggml_q6_K_n1024_k2560.json

9.05 kB
Add Qwen3.5-4B e2e gemm definitions (packed ggml Q4_K/Q5_K/Q6_K ABI): 11 definitions + workloads + reference-scalar + baseline-llamacpp-arm (additive) 28 days ago
gemm_ggml_q6_K_n1024_k4096.json

9.05 kB
Llama-3.1-8B e2e set (defs, real-activation workloads, Fable gate-v2 solutions, runs, measurements); Qwen kernel-vs-repack results; fixed baseline-sve2 loop_114/121/219 23 days ago
gemm_ggml_q6_K_n128256_k4096.json

9.06 kB
Llama-3.1-8B e2e set (defs, real-activation workloads, Fable gate-v2 solutions, runs, measurements); Qwen kernel-vs-repack results; fixed baseline-sve2 loop_114/121/219 23 days ago
gemm_ggml_q6_K_n248320_k2560.json

9.06 kB
Add Qwen3.5-4B e2e gemm definitions (packed ggml Q4_K/Q5_K/Q6_K ABI): 11 definitions + workloads + reference-scalar + baseline-llamacpp-arm (additive) 28 days ago
gemm_ggml_q6_K_n2560_k9216.json

9.05 kB
Add Qwen3.5-4B e2e gemm definitions (packed ggml Q4_K/Q5_K/Q6_K ABI): 11 definitions + workloads + reference-scalar + baseline-llamacpp-arm (additive) 28 days ago
gemm_ggml_q6_K_n4096_k14336.json

9.06 kB
Llama-3.1-8B e2e set (defs, real-activation workloads, Fable gate-v2 solutions, runs, measurements); Qwen kernel-vs-repack results; fixed baseline-sve2 loop_114/121/219 23 days ago
gemm_q4_k_m_n1024_k2048.json

7.93 kB
Add kernel in different precision: 3 months ago
gemm_q4_k_m_n1408_k2048.json

7.93 kB
Add kernel in different precision: 3 months ago
gemm_q4_k_m_n2048_k1024.json

7.93 kB
Add kernel in different precision: 3 months ago
gemm_q4_k_m_n2048_k1536.json

7.93 kB
Add kernel in different precision: 3 months ago
gemm_q4_k_m_n2048_k2048.json

7.93 kB
Add kernel in different precision: 3 months ago
gemm_q8_0_n1024_k2048.json

4.85 kB
Rewrite llama.cpp q8_0 reference-scalar solution signature, now compatible with bench/dataset/raw.py 3 months ago
gemm_q8_0_n1408_k2048.json

4.85 kB
Rewrite llama.cpp q8_0 reference-scalar solution signature, now compatible with bench/dataset/raw.py 3 months ago
gemm_q8_0_n2048_k1024.json

4.85 kB
Rewrite llama.cpp q8_0 reference-scalar solution signature, now compatible with bench/dataset/raw.py 3 months ago
gemm_q8_0_n2048_k1408.json

4.85 kB
Rewrite llama.cpp q8_0 reference-scalar solution signature, now compatible with bench/dataset/raw.py 3 months ago
gemm_q8_0_n2048_k2048.json

4.85 kB
Rewrite llama.cpp q8_0 reference-scalar solution signature, now compatible with bench/dataset/raw.py 3 months ago