granite-4.1 8b
Q4_K_M·8B params·GGUF
intelligence: see on Artificial Analysis →
checkpoint:
unsloth/granite-4.1-8b-GGUF:Q4_K_Mcommit:
6f9671f73eb0weights 4.98 GiB
All runs (20)
| legacy | stack comparable | GeForce RTX 3090 · 24 GiB450 W maxdrv 590 | llama.cpp cuda-4f13cb7 (cuda) | baseline-pl-450w | chat | 1 | 127.0 | 114.1 | 665.2 | — | 42ms | 7.9 | — | 28 | 100 | 814ms | 0.000 GiB |
| legacy | stack comparable | GeForce RTX 3090 · 24 GiB450 W maxdrv 590 | llama.cpp cuda-4f13cb7 (cuda) | baseline-pl-450w | codegen | 1 | 124.8 | 115.5 | 520.0 | — | 121ms | 8.0 | — | 59 | 735 | 6.37s | 0.000 GiB |
| legacy | stack comparable | GeForce RTX 3090 · 24 GiB450 W maxdrv 590 | llama.cpp cuda-4f13cb7 (cuda) | baseline-pl-450w | rag | 1 | 124.4 | 92.1 | 3118.8 | — | 223ms | 8.0 | — | 695 | 79 | 910ms | 0.000 GiB |
| legacy | stack comparable | GeForce RTX 3090 · 24 GiBcap 350 Wdrv 590 | llama.cpp cuda-4f13cb7 (cuda) | baseline-pl-350w | chat | 1 | 124.1 | 117.7 | 645.1 | — | 45ms | 8.1 | — | 28 | 100 | 838ms | 0.000 GiB |
| legacy | stack comparable | GeForce RTX 3090 · 24 GiB450 W maxdrv 590 | llama.cpp cuda-4f13cb7 (cuda) | baseline-pl-450w | agent | 1 | 123.1 | 112.6 | 2930.7 | — | 189ms | 8.1 | — | 555 | 500 | 4.38s | 0.000 GiB |
| legacy | stack comparable | GeForce RTX 3090 · 24 GiB450 W maxdrv 590 | llama.cpp cuda-4f13cb7 (cuda) | baseline-pl-450w | agent | 4 | 122.8 | 43.2 | 122.3 | — | 6.00s | 8.1 | — | 555 | 500 | 9.49s | 0.000 GiB |
| legacy | stack comparable | GeForce RTX 3090 · 24 GiBcap 350 Wdrv 590 | llama.cpp cuda-4f13cb7 (cuda) | baseline-pl-350w | codegen | 1 | 122.5 | 115.2 | 564.5 | — | 110ms | 8.2 | — | 59 | 735 | 6.60s | 0.000 GiB |
| legacy | stack comparable | GeForce RTX 3090 · 24 GiBcap 350 Wdrv 590 | llama.cpp cuda-4f13cb7 (cuda) | baseline-pl-350w | rag | 1 | 122.1 | 89.3 | 2847.9 | — | 244ms | 8.2 | — | 695 | 79 | 942ms | 0.000 GiB |
| legacy | stack comparable | GeForce RTX 3090 · 24 GiBcap 350 Wdrv 590 | llama.cpp cuda-4f13cb7 (cuda) | baseline-pl-350w | agent | 4 | 120.9 | 46.6 | 97.9 | — | 6.22s | 8.3 | — | 555 | 500 | 9.77s | 0.000 GiB |
| legacy | stack comparable | GeForce RTX 3090 · 24 GiBcap 350 Wdrv 590 | llama.cpp cuda-4f13cb7 (cuda) | baseline-pl-350w | agent | 1 | 120.7 | 109.8 | 3212.3 | — | 173ms | 8.3 | — | 555 | 500 | 4.48s | 0.000 GiB |
| legacy | stack comparable | GeForce RTX 5070 · 12 GiBcap 250 Wdrv 595 | llama.cpp b9174 (vulkan) | baseline | chat | 1 | 100.2 | 97.6 | 766.7 | — | 38ms | 10.0 | — | 28 | 100 | 1.00s | 0.000 GiB |
| legacy | stack comparable | GeForce RTX 5070 · 12 GiBcap 250 Wdrv 595 | llama.cpp b9174 (vulkan) | baseline | codegen | 1 | 99.1 | 98.5 | 1213.0 | — | 55ms | 10.1 | — | 59 | 738 | 7.48s | 0.000 GiB |
| legacy | stack comparable | GeForce RTX 5070 · 12 GiBcap 250 Wdrv 595 | llama.cpp b9174 (vulkan) | baseline | rag | 1 | 98.1 | 91.5 | 5797.6 | — | 95ms | 10.2 | — | 695 | 79 | 1.12s | 0.000 GiB |
| legacy | stack comparable | GeForce RTX 5070 · 12 GiBcap 250 Wdrv 595 | llama.cpp b9174 (vulkan) | baseline | agent | 1 | 96.0 | 94.6 | 2805.2 | — | 159ms | 10.4 | — | 555 | 500 | 5.24s | 0.000 GiB |
| legacy | stack comparable | GeForce RTX 3090 · 24 GiBcap 200 Wdrv 590 | llama.cpp 59778f0 (cuda) | baseline | chat | 1 | 77.2 | 74.4 | 591.9 | — | 49ms | 13.0 | — | 28 | 100 | 1.30s | 0.000 GiB |
| legacy | stack comparable | GeForce RTX 3090 · 24 GiBcap 200 Wdrv 590 | llama.cpp 59778f0 (cuda) | baseline | rag | 1 | 74.6 | 63.8 | 3453.4 | — | 201ms | 13.4 | — | 695 | 79 | 1.22s | 0.000 GiB |
| legacy | stack comparable | GeForce RTX 3090 · 24 GiBcap 200 Wdrv 590 | llama.cpp 59778f0 (cuda) | baseline | codegen | 1 | 72.2 | 70.7 | 581.8 | — | 112ms | 13.8 | — | 59 | 735 | 10.41s | 0.000 GiB |
| legacy | stack comparable | GeForce RTX 3090 · 24 GiBcap 200 Wdrv 590 | llama.cpp 59778f0 (cuda) | baseline | agent | 1 | 71.5 | 66.6 | 1759.5 | — | 254ms | 14.0 | — | 555 | 500 | 7.36s | 0.000 GiB |
| legacy | stack comparable | GeForce RTX 5070 · 12 GiBcap 250 Wdrv 595 | llama.cpp b9174 (vulkan) | baseline | agent | 4 | 59.8 | 36.9 | 135.2 | — | 5.30s | 16.7 | — | 555 | 500 | 11.52s | 0.000 GiB |
| legacy | stack comparable | GeForce RTX 3090 · 24 GiBcap 200 Wdrv 590 | llama.cpp 59778f0 (cuda) | baseline | agent | 4 | 42.3 | 35.0 | 237.7 | — | 2.74s | 23.6 | — | 555 | 500 | 13.85s | 0.000 GiB |
Environment
GeForce RTX 3090 · 24 GiB
cpuAMD EPYC 7302P 16-Core Processor
gpuNVIDIA GeForce RTX 3090
archNVIDIA
vram24 GiB (system 64.0 GiB)
power350 W / 450 W max(78% cap)
pcieGen 4 x16 / Gen 4 x16 max
clocksgfx 1980/2100 MHz · mem 9501 MHz
temp44°C idle · 62°C peak
peak draw326 W
hardware probes
copy 42% of theoryFP16 peak 65.4 TFcopy/math flat across caps
384-bit9751 MHz82 SM/CU
Microbenchmarks for memory copy and tensor math; raw-engine decode and API workload rows measure model-serving speed.
| cap | theory | copy | fp16 | bf16 |
|---|---|---|---|---|
| 200 W | 936 GB/s | 391 GB/s | 65.4 TF | 65.4 TF |
| 300 W | 936 GB/s | 391 GB/s | 65.4 TF | 65.3 TF |
| 450 W | 936 GB/s | 391 GB/s | 65.4 TF | 65.4 TF |
compute: 8.6
backendllama.cpp cuda-4f13cb7 (cuda)
osUbuntu 24.04 LTS
kernel6.17.13-7-pve
driverNVIDIA 590.48.01 + CUDA 13.1
libc2.39
python3.12.3
llama.cppversion: 18 (4f13cb7) built with GNU 13.3.0 for Linux x86_64
build flagsGGML_CUDA=ON CMAKE_BUILD_TYPE=Release
runs/cell5
warmups2
endpoint/v1/chat/completions
streamingtrue
GeForce RTX 3090 · 24 GiB
cpuAMD EPYC 7302P 16-Core Processor
gpuNVIDIA GeForce RTX 3090
archNVIDIA
vram24 GiB (system 64.0 GiB)
power450 W / 450 W max
pcieGen 4 x16 / Gen 4 x16 max
clocksgfx 1980/2100 MHz · mem 9501 MHz
temp44°C idle · 78°C peak
peak draw428 W
backendllama.cpp cuda-4f13cb7 (cuda)
osUbuntu 24.04 LTS
kernel6.17.13-7-pve
driverNVIDIA 590.48.01 + CUDA 13.1
libc2.39
python3.12.3
llama.cppversion: 18 (4f13cb7) built with GNU 13.3.0 for Linux x86_64
build flagsGGML_CUDA=ON CMAKE_BUILD_TYPE=Release
runs/cell5
warmups2
endpoint/v1/chat/completions
streamingtrue
GeForce RTX 3090 · 24 GiB
cpuAMD EPYC 7302P 16-Core Processor
gpuNVIDIA GeForce RTX 3090
archNVIDIA
vram24 GiB (system 64.0 GiB)
power200 W / 450 W max(44% cap)
backendllama.cpp 59778f0 (cuda)
osUbuntu 24.04 LTS
kernel6.17.13-7-pve
driver590.48.01
python3.12.3
runs/cell5
warmups2
endpoint/v1/chat/completions
streamingtrue
GeForce RTX 5070 · 12 GiB
cpuAMD Ryzen 9 7900 12-Core Processor
gpuNVIDIA GeForce RTX 5070
archNVIDIA
vram11.94 GiB (system 30.4 GiB)
power250 W / 300 W max(83% cap)
hardware probes
copy 40% of theoryFP16 peak 69.6 TFcopy/math spread 2.5%
192-bit14001 MHz48 SM/CU
Microbenchmarks for memory copy and tensor math; raw-engine decode and API workload rows measure model-serving speed.
| cap | theory | copy | fp16 | bf16 |
|---|---|---|---|---|
| 200 W | 672 GB/s | 271 GB/s | 67.9 TF | 68.4 TF |
| 250 W | 672 GB/s | 271 GB/s | 69.5 TF | 68.2 TF |
| 300 W | 672 GB/s | 270 GB/s | 69.6 TF | 68.4 TF |
compute: 12
backendllama.cpp b9174 (vulkan)
osCachyOS
kernel7.0.8-1-cachyos
driver595.71.05
python3.14.4
runs/cell5
warmups2
endpoint/v1/chat/completions
streamingtrue