updated Qwen Next Benchmarks

This commit is contained in:
Donato Capitella
2025-12-05 08:32:58 +00:00
parent 62b0e5e173
commit 7584a31548
19 changed files with 371 additions and 253 deletions
@@ -4,7 +4,7 @@ ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | fa | mmap | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | ---: | --------------: | -------------------: |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 245.79 ± 0.39 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 19.30 ± 0.09 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 248.51 ± 1.01 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 25.33 ± 0.52 |
build: c6f7a423c (7189)
build: 03d9a77b8 (7278)
@@ -4,7 +4,7 @@ ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | fa | mmap | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | ---: | --------------: | -------------------: |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 301.76 ± 0.36 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 23.98 ± 0.02 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 302.76 ± 0.59 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 25.42 ± 0.29 |
build: c6f7a423c (7189)
build: 03d9a77b8 (7278)
@@ -4,7 +4,7 @@ ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | fa | mmap | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | ---: | --------------: | -------------------: |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 249.80 ± 4.99 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 19.17 ± 0.01 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 242.26 ± 2.27 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 24.98 ± 0.07 |
build: c6f7a423c (7189)
build: 03d9a77b8 (7278)
@@ -4,7 +4,7 @@ ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | fa | mmap | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | ---: | --------------: | -------------------: |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 305.05 ± 2.25 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 23.99 ± 0.16 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 287.99 ± 2.09 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 26.05 ± 0.16 |
build: c6f7a423c (7189)
build: 03d9a77b8 (7278)
@@ -4,7 +4,7 @@ ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | fa | mmap | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | ---: | --------------: | -------------------: |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 311.74 ± 3.86 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 22.66 ± 0.25 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 318.50 ± 0.98 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 24.75 ± 0.39 |
build: c6f7a423c (7189)
build: 03d9a77b8 (7278)
@@ -4,7 +4,7 @@ ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | fa | mmap | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | ---: | --------------: | -------------------: |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 313.20 ± 3.93 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 23.02 ± 0.30 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 325.56 ± 2.18 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 24.95 ± 0.39 |
build: c6f7a423c (7189)
build: 03d9a77b8 (7278)
@@ -4,7 +4,7 @@ ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | fa | mmap | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | ---: | --------------: | -------------------: |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 320.42 ± 0.35 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 23.06 ± 0.13 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 320.64 ± 0.80 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 24.55 ± 0.01 |
build: c6f7a423c (7189)
build: 03d9a77b8 (7278)
@@ -2,3 +2,9 @@ ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | fa | mmap | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | ---: | --------------: | -------------------: |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 325.06 ± 1.27 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 25.14 ± 0.02 |
build: 03d9a77b8 (7278)
@@ -4,7 +4,7 @@ ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | fa | mmap | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | ---: | --------------: | -------------------: |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 258.30 ± 2.66 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 18.49 ± 0.24 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 264.35 ± 7.13 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 24.89 ± 0.59 |
build: c6f7a423c (7189)
build: 03d9a77b8 (7278)
@@ -4,7 +4,7 @@ ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | fa | mmap | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | ---: | --------------: | -------------------: |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 322.70 ± 2.10 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 23.45 ± 0.19 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 321.93 ± 4.64 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 24.64 ± 1.29 |
build: c6f7a423c (7189)
build: 03d9a77b8 (7278)
@@ -4,7 +4,7 @@ ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | fa | mmap | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | ---: | --------------: | -------------------: |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 253.35 ± 2.88 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 18.25 ± 0.15 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 256.63 ± 1.91 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 25.18 ± 0.03 |
build: c6f7a423c (7189)
build: 03d9a77b8 (7278)
@@ -4,7 +4,7 @@ ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | fa | mmap | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | ---: | --------------: | -------------------: |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 319.09 ± 2.77 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 23.72 ± 0.03 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 324.95 ± 0.42 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 25.54 ± 0.14 |
build: c6f7a423c (7189)
build: 03d9a77b8 (7278)
@@ -4,7 +4,7 @@ ggml_cuda_init: found 1 ROCm devices:
Device 0: AMD Radeon Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | fa | mmap | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | ---: | --------------: | -------------------: |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 254.34 ± 2.60 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 18.06 ± 0.08 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 255.48 ± 2.19 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 24.31 ± 0.65 |
build: c6f7a423c (7189)
build: 03d9a77b8 (7278)
@@ -4,7 +4,7 @@ ggml_cuda_init: found 1 ROCm devices:
Device 0: AMD Radeon Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | fa | mmap | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | ---: | --------------: | -------------------: |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 323.33 ± 1.75 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 23.37 ± 0.17 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 327.95 ± 1.55 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 24.64 ± 1.08 |
build: c6f7a423c (7189)
build: 03d9a77b8 (7278)
@@ -4,7 +4,7 @@ ggml_cuda_init: found 1 ROCm devices:
Device 0: AMD Radeon Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | fa | mmap | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | ---: | --------------: | -------------------: |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 242.43 ± 0.28 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 18.35 ± 0.10 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 264.12 ± 2.61 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 24.94 ± 0.15 |
build: c6f7a423c (7189)
build: 03d9a77b8 (7278)
@@ -4,7 +4,7 @@ ggml_cuda_init: found 1 ROCm devices:
Device 0: AMD Radeon Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | fa | mmap | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | ---: | --------------: | -------------------: |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 305.22 ± 3.10 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 23.60 ± 0.08 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | pp512 | 328.73 ± 4.15 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | ROCm | 99 | 1 | 0 | tg128 | 25.46 ± 0.21 |
build: c6f7a423c (7189)
build: 03d9a77b8 (7278)
@@ -2,7 +2,7 @@ ggml_vulkan: Found 1 Vulkan devices:
ggml_vulkan: 0 = Radeon 8060S Graphics (AMD open-source driver) | uma: 1 | fp16: 1 | bf16: 0 | warp size: 64 | shared memory: 32768 | int dot: 1 | matrix cores: KHR_coopmat
| model | size | params | backend | ngl | fa | mmap | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | ---: | --------------: | -------------------: |
ggml_vulkan: Device memory allocation of size 614888064 failed.
ggml_vulkan: vk::Device::allocateMemory: ErrorOutOfHostMemory
main: error: failed to load model '/home/kyuz0/models/UD-Q8_K_XL/Qwen3-Next-80B-A3B-Instruct-UD-Q8_K_XL-00001-of-00002.gguf'
✖ ! [vulkan_amdvlk] Qwen3-Next-80B-A3B-Instruct-UD-Q8_K_XL-00001-of-00002__fa1 failed (exit 0)
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | Vulkan | 99 | 1 | 0 | pp512 | 417.28 ± 0.97 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | Vulkan | 99 | 1 | 0 | tg128 | 30.31 ± 0.01 |
build: 03d9a77b8 (7278)
@@ -2,7 +2,7 @@ ggml_vulkan: Found 1 Vulkan devices:
ggml_vulkan: 0 = Radeon 8060S Graphics (RADV GFX1151) (radv) | uma: 1 | fp16: 1 | bf16: 0 | warp size: 64 | shared memory: 65536 | int dot: 1 | matrix cores: KHR_coopmat
| model | size | params | backend | ngl | fa | mmap | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | ---: | --------------: | -------------------: |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | Vulkan | 99 | 1 | 0 | pp512 | 324.45 ± 0.42 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | Vulkan | 99 | 1 | 0 | tg128 | 22.31 ± 0.18 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | Vulkan | 99 | 1 | 0 | pp512 | 333.74 ± 1.19 |
| qwen3next ?B Q8_0 | 79.57 GiB | 79.67 B | Vulkan | 99 | 1 | 0 | tg128 | 27.76 ± 0.01 |
build: c6f7a423c (7189)
build: 03d9a77b8 (7278)