Skip to content

Qwen3.5 27B

Q4_K_M·27B params·GGUF
reasoning
checkpoint: unsloth/Qwen3.5-27B-GGUF:Q4_K_M
commit: 3221f178a6b8
weights 15.59 GiB

All runs (20)

legacystack comparable
GeForce RTX 3090 · 24 GiB450 W maxdrv 590
llama.cpp cuda-4f13cb7 (cuda)baseline-pl-450wchat1
42.7
37.8116.4258ms23.4301002.65s0.000 GiB
legacystack comparable
GeForce RTX 3090 · 24 GiB450 W maxdrv 590
llama.cpp cuda-4f13cb7 (cuda)baseline-pl-450wrag1
42.4
34.11311.6762ms23.68422005.86s0.000 GiB
legacystack comparable
GeForce RTX 3090 · 24 GiB450 W maxdrv 590
llama.cpp cuda-4f13cb7 (cuda)baseline-pl-450wcodegen1
42.2
40.5219.0358ms23.762100024.70s0.000 GiB
legacystack comparable
GeForce RTX 3090 · 24 GiB450 W maxdrv 590
llama.cpp cuda-4f13cb7 (cuda)baseline-pl-450wagent1
42.1
39.11180.7507ms23.759950012.79s0.000 GiB
legacystack comparable
GeForce RTX 3090 · 24 GiB450 W maxdrv 590
llama.cpp cuda-4f13cb7 (cuda)baseline-pl-450wagent4
42.0
16.541.119.29s23.859950031.61s0.000 GiB
legacystack comparable
GeForce RTX 3090 · 24 GiBcap 350 Wdrv 590
llama.cpp cuda-4f13cb7 (cuda)baseline-pl-350wchat1
41.7
37.9127.3236ms24.0301002.64s0.000 GiB
legacystack comparable
GeForce RTX 3090 · 24 GiBcap 350 Wdrv 590
llama.cpp cuda-4f13cb7 (cuda)baseline-pl-350wrag1
41.4
35.21316.8785ms24.18422005.68s0.000 GiB
legacystack comparable
GeForce RTX 3090 · 24 GiBcap 350 Wdrv 590
llama.cpp cuda-4f13cb7 (cuda)baseline-pl-350wcodegen1
41.4
39.8202.4305ms24.262100025.14s0.000 GiB
legacystack comparable
GeForce RTX 3090 · 24 GiBcap 350 Wdrv 590
llama.cpp cuda-4f13cb7 (cuda)baseline-pl-350wagent1
41.3
38.41182.5506ms24.259950013.01s0.000 GiB
legacystack comparable
GeForce RTX 3090 · 24 GiBcap 350 Wdrv 590
llama.cpp cuda-4f13cb7 (cuda)baseline-pl-350wagent4
41.3
16.131.419.67s24.259950032.24s0.000 GiB
legacystack comparable
GeForce RTX 3090 · 24 GiBcap 200 Wdrv 590
llama.cpp 59778f0 (cuda)baselinechat1
22.9
21.2126.7253ms43.7301004.72s0.000 GiB
legacystack comparable
GeForce RTX 3090 · 24 GiBcap 200 Wdrv 590
llama.cpp 59778f0 (cuda)baselinerag1
21.9
19.81037.8912ms45.884220010.12s0.000 GiB
legacystack comparable
GeForce RTX 3090 · 24 GiBcap 200 Wdrv 590
llama.cpp 59778f0 (cuda)baselinecodegen1
21.6
21.4199.6346ms46.362100046.81s0.000 GiB
legacystack comparable
GeForce RTX 3090 · 24 GiBcap 200 Wdrv 590
llama.cpp 59778f0 (cuda)baselineagent1
21.5
20.9980.5611ms46.559950023.89s0.000 GiB
legacystack comparable
Strix Halo · Radeon 8060S · 128 GiB unified (96 GiB VRAM)unified
llama.cpp b1203 (rocm)baselinechat1
12.0
11.696.9330ms83.5311008.59s0.002 GiB
legacystack comparable
Strix Halo · Radeon 8060S · 128 GiB unified (96 GiB VRAM)unified
llama.cpp b1203 (rocm)baselinecodegen1
12.0
11.9172.4417ms83.563100084.02s0.003 GiB
legacystack comparable
Strix Halo · Radeon 8060S · 128 GiB unified (96 GiB VRAM)unified
llama.cpp b1203 (rocm)baselinerag1
12.0
10.9294.61.73s83.6100520018.39s0.005 GiB
legacystack comparable
Strix Halo · Radeon 8060S · 128 GiB unified (96 GiB VRAM)unified
llama.cpp b1203 (rocm)baselineagent1
12.0
11.5279.91.75s83.659950043.53s0.008 GiB
legacystack comparable
GeForce RTX 3090 · 24 GiBcap 200 Wdrv 590
llama.cpp 59778f0 (cuda)baselineagent4
10.9
9.73.91s91.534135.41s0.000 GiB
legacystack comparable
Strix Halo · Radeon 8060S · 128 GiB unified (96 GiB VRAM)unified
llama.cpp b1203 (rocm)baselineagent4
5.6
5.4232.72.12s178.959950093.01s-0.005 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 1965/2100 MHz · mem 9501 MHz
temp45°C idle · 64°C peak
peak draw334 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.
captheorycopyfp16bf16
200 W936 GB/s391 GB/s65.4 TF65.4 TF
300 W936 GB/s391 GB/s65.4 TF65.3 TF
450 W936 GB/s391 GB/s65.4 TF65.4 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
temp40°C idle · 83°C peak
peak draw435 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
Strix Halo · Radeon 8060S · 128 GiB unified (96 GiB VRAM)
cpuAMD RYZEN AI MAX+ 395 w/ Radeon 8060S
gpuAMD Radeon 8060S
archStrix Halo (gfx1151)
vram96 GiB (system 31.1 GiB, unified)
hardware probes
copy 41% of theoryFP16 peak 30.3 TF
256-bit8000 MHz20 SM/CU
Microbenchmarks for memory copy and tensor math; raw-engine decode and API workload rows measure model-serving speed.
captheorycopyfp16bf16
fixed256 GB/s106 GB/s30.3 TF-
compute: 11.5
backendllama.cpp b1203 (rocm)
osUbuntu 24.04.4 LTS
kernel7.0.2-2-pve
python3.12.3
runs/cell3
warmups1
endpoint/v1/chat/completions
streamingtrue