julian/amd-strix-halo-toolboxes
1
0
Fork 0
Code Issues Pull Requests 2 Actions Packages Projects Releases Wiki Activity

updted benchmarks

Browse Source
This commit is contained in:
Donato Capitella
2026-03-30 08:37:15 +01:00
parent 614b00af3e
commit a7ace8dba7
595 changed files with 19653 additions and 14511 deletions
Download Patch File Download Diff File Expand all files Collapse all files
benchmark/results/04-02-2026/Devstral-2-123B-Instruct-2512-UD-Q4_K_XL-00001-of-00002__rocm-7_2__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 1 | pp512 | 72.93 ± 0.06 |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 1 | tg128 | 2.95 ± 0.06 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Devstral-2-123B-Instruct-2512-UD-Q4_K_XL-00001-of-00002__rocm-7_2__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 16.57 ± 0.04 |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 2.08 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Devstral-2-123B-Instruct-2512-UD-Q4_K_XL-00001-of-00002__rocm-7_2__hblt0__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 1 | pp512 | 71.58 ± 0.06 |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 1 | tg128 | 2.99 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Devstral-2-123B-Instruct-2512-UD-Q4_K_XL-00001-of-00002__rocm-7_2__hblt0__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 16.48 ± 0.03 |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 2.07 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Devstral-2-123B-Instruct-2512-UD-Q4_K_XL-00001-of-00002__rocm6_4_4__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 1 | pp512 | 79.51 ± 0.07 |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 1 | tg128 | 2.72 ± 0.10 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Devstral-2-123B-Instruct-2512-UD-Q4_K_XL-00001-of-00002__rocm6_4_4__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 17.57 ± 0.05 |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 2.15 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Devstral-2-123B-Instruct-2512-UD-Q4_K_XL-00001-of-00002__rocm6_4_4__hblt0__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 1 | pp512 | 79.24 ± 0.10 |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 1 | tg128 | 2.67 ± 0.11 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Devstral-2-123B-Instruct-2512-UD-Q4_K_XL-00001-of-00002__rocm6_4_4__hblt0__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 17.53 ± 0.03 |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 2.15 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Devstral-2-123B-Instruct-2512-UD-Q4_K_XL-00001-of-00002__rocm7-nightlies__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 1 | pp512 | 78.28 ± 0.06 |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 1 | tg128 | 2.98 ± 0.04 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Devstral-2-123B-Instruct-2512-UD-Q4_K_XL-00001-of-00002__rocm7-nightlies__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 17.18 ± 0.03 |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 2.06 ± 0.02 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Devstral-2-123B-Instruct-2512-UD-Q4_K_XL-00001-of-00002__rocm7-nightlies__hblt0__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 1 | pp512 | 80.59 ± 0.10 |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 1 | tg128 | 2.99 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Devstral-2-123B-Instruct-2512-UD-Q4_K_XL-00001-of-00002__rocm7-nightlies__hblt0__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 17.27 ± 0.02 |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 2.07 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Devstral-2-123B-Instruct-2512-UD-Q4_K_XL-00001-of-00002__vulkan_amdvlk__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | Vulkan | 99 | 1 | pp512 | 17.65 ± 0.01 |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | Vulkan | 99 | 1 | tg128 | 3.00 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Devstral-2-123B-Instruct-2512-UD-Q4_K_XL-00001-of-00002__vulkan_amdvlk__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | Vulkan | 99 | 1 | pp2048 @ d32768 | 4.94 ± 0.01 |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | Vulkan | 99 | 1 | tg32 @ d32768 | 1.69 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Devstral-2-123B-Instruct-2512-UD-Q4_K_XL-00001-of-00002__vulkan_radv__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | Vulkan | 99 | 1 | pp512 | 54.76 ± 11.46 |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | Vulkan | 99 | 1 | tg128 | 3.00 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Devstral-2-123B-Instruct-2512-UD-Q4_K_XL-00001-of-00002__vulkan_radv__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | Vulkan | 99 | 1 | pp2048 @ d32768 | 7.15 ± 0.02 |
| llama ?B Q4_K - Medium | 70.31 GiB | 125.03 B | Vulkan | 99 | 1 | tg32 @ d32768 | 2.27 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-BF16-00001-of-00002__rocm-7_2__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 1 | pp512 | 454.95 ± 1.90 |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 1 | tg128 | 22.26 ± 0.03 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-BF16-00001-of-00002__rocm-7_2__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 94.79 ± 0.56 |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 16.48 ± 0.09 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-BF16-00001-of-00002__rocm-7_2__hblt0__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 1 | pp512 | 425.21 ± 1.79 |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 1 | tg128 | 23.41 ± 0.03 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-BF16-00001-of-00002__rocm-7_2__hblt0__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 93.83 ± 0.40 |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 16.55 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-BF16-00001-of-00002__rocm6_4_4__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 1 | pp512 | 407.15 ± 2.05 |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 1 | tg128 | 21.51 ± 0.03 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-BF16-00001-of-00002__rocm6_4_4__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 101.09 ± 0.37 |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 16.23 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-BF16-00001-of-00002__rocm6_4_4__hblt0__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 1 | pp512 | 414.23 ± 2.09 |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 1 | tg128 | 23.11 ± 0.03 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-BF16-00001-of-00002__rocm6_4_4__hblt0__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 100.06 ± 0.38 |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 15.97 ± 0.45 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-BF16-00001-of-00002__rocm7-nightlies__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 1 | pp512 | 489.62 ± 3.63 |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 1 | tg128 | 23.40 ± 0.02 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-BF16-00001-of-00002__rocm7-nightlies__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 92.48 ± 1.13 |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 16.50 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-BF16-00001-of-00002__rocm7-nightlies__hblt0__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 1 | pp512 | 425.86 ± 2.29 |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 1 | tg128 | 23.41 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-BF16-00001-of-00002__rocm7-nightlies__hblt0__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 92.06 ± 0.08 |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 16.51 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-BF16-00001-of-00002__vulkan_amdvlk__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | Vulkan | 99 | 1 | pp512 | 106.42 ± 0.08 |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | Vulkan | 99 | 1 | tg128 | 10.87 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-BF16-00001-of-00002__vulkan_amdvlk__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | Vulkan | 99 | 1 | pp2048 @ d32768 | 6.09 ± 0.00 |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | Vulkan | 99 | 1 | tg32 @ d32768 | 8.28 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-BF16-00001-of-00002__vulkan_radv__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | Vulkan | 99 | 1 | pp512 | 333.10 ± 6.48 |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | Vulkan | 99 | 1 | tg128 | 9.51 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-BF16-00001-of-00002__vulkan_radv__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | Vulkan | 99 | 1 | pp2048 @ d32768 | 78.99 ± 0.25 |
| deepseek2 30B.A3B BF16 | 55.79 GiB | 29.94 B | Vulkan | 99 | 1 | tg32 @ d32768 | 8.13 ± 0.02 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-UD-Q8_K_XL__rocm-7_2__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 1 | pp512 | 398.34 ± 1.32 |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 1 | tg128 | 35.94 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-UD-Q8_K_XL__rocm-7_2__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 90.22 ± 4.88 |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 22.35 ± 0.04 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-UD-Q8_K_XL__rocm-7_2__hblt0__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 1 | pp512 | 398.87 ± 1.21 |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 1 | tg128 | 36.09 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-UD-Q8_K_XL__rocm-7_2__hblt0__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 92.13 ± 0.15 |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 21.56 ± 1.34 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-UD-Q8_K_XL__rocm6_4_4__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 1 | pp512 | 947.86 ± 2.03 |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 1 | tg128 | 33.77 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-UD-Q8_K_XL__rocm6_4_4__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 84.85 ± 1.04 |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 21.89 ± 0.04 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-UD-Q8_K_XL__rocm6_4_4__hblt0__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 1 | pp512 | 952.84 ± 2.21 |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 1 | tg128 | 35.23 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-UD-Q8_K_XL__rocm6_4_4__hblt0__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 84.01 ± 0.58 |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 21.97 ± 0.04 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-UD-Q8_K_XL__rocm7-nightlies__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 1 | pp512 | 983.72 ± 3.21 |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 1 | tg128 | 36.20 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-UD-Q8_K_XL__rocm7-nightlies__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 80.32 ± 1.28 |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 22.31 ± 0.04 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-UD-Q8_K_XL__rocm7-nightlies__hblt0__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 1 | pp512 | 955.10 ± 4.53 |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 1 | tg128 | 36.16 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-UD-Q8_K_XL__rocm7-nightlies__hblt0__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 81.34 ± 1.80 |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 22.32 ± 0.04 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-UD-Q8_K_XL__vulkan_amdvlk__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | Vulkan | 99 | 1 | pp512 | 368.78 ± 0.17 |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | Vulkan | 99 | 1 | tg128 | 40.80 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-UD-Q8_K_XL__vulkan_amdvlk__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | Vulkan | 99 | 1 | pp2048 @ d32768 | 6.35 ± 0.00 |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | Vulkan | 99 | 1 | tg32 @ d32768 | 18.75 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-UD-Q8_K_XL__vulkan_radv__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | Vulkan | 99 | 1 | pp512 | 877.18 ± 8.15 |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | Vulkan | 99 | 1 | tg128 | 40.07 ± 0.78 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/GLM-4.7-Flash-UD-Q8_K_XL__vulkan_radv__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | Vulkan | 99 | 1 | pp2048 @ d32768 | 90.27 ± 0.42 |
| deepseek2 30B.A3B Q8_0 | 32.70 GiB | 29.94 B | Vulkan | 99 | 1 | tg32 @ d32768 | 23.07 ± 0.03 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Llama-3.3-70B-Instruct-UD-Q8_K_XL-00001-of-00002__rocm-7_2__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 1 | pp512 | 48.83 ± 0.01 |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 1 | tg128 | 2.78 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Llama-3.3-70B-Instruct-UD-Q8_K_XL-00001-of-00002__rocm-7_2__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 29.25 ± 0.17 |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 2.45 ± 0.02 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Llama-3.3-70B-Instruct-UD-Q8_K_XL-00001-of-00002__rocm-7_2__hblt0__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 1 | pp512 | 49.38 ± 0.03 |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 1 | tg128 | 2.79 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Llama-3.3-70B-Instruct-UD-Q8_K_XL-00001-of-00002__rocm-7_2__hblt0__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 29.17 ± 0.18 |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 2.46 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Llama-3.3-70B-Instruct-UD-Q8_K_XL-00001-of-00002__rocm6_4_4__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 1 | pp512 | 146.04 ± 0.21 |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 1 | tg128 | 2.78 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Llama-3.3-70B-Instruct-UD-Q8_K_XL-00001-of-00002__rocm6_4_4__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 36.22 ± 0.16 |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 2.43 ± 0.03 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Llama-3.3-70B-Instruct-UD-Q8_K_XL-00001-of-00002__rocm6_4_4__hblt0__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 1 | pp512 | 146.83 ± 0.25 |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 1 | tg128 | 2.78 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Llama-3.3-70B-Instruct-UD-Q8_K_XL-00001-of-00002__rocm6_4_4__hblt0__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 36.40 ± 0.23 |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 2.46 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Llama-3.3-70B-Instruct-UD-Q8_K_XL-00001-of-00002__rocm7-nightlies__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 1 | pp512 | 155.06 ± 0.11 |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 1 | tg128 | 2.79 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Llama-3.3-70B-Instruct-UD-Q8_K_XL-00001-of-00002__rocm7-nightlies__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 38.36 ± 0.61 |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 2.46 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Llama-3.3-70B-Instruct-UD-Q8_K_XL-00001-of-00002__rocm7-nightlies__hblt0__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 1 | pp512 | 151.70 ± 0.21 |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 1 | tg128 | 2.78 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Llama-3.3-70B-Instruct-UD-Q8_K_XL-00001-of-00002__rocm7-nightlies__hblt0__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 38.35 ± 0.67 |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 2.46 ± 0.02 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Llama-3.3-70B-Instruct-UD-Q8_K_XL-00001-of-00002__vulkan_amdvlk__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | Vulkan | 99 | 1 | pp512 | 21.74 ± 0.01 |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | Vulkan | 99 | 1 | tg128 | 2.81 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Llama-3.3-70B-Instruct-UD-Q8_K_XL-00001-of-00002__vulkan_amdvlk__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | Vulkan | 99 | 1 | pp2048 @ d32768 | 8.35 ± 0.01 |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | Vulkan | 99 | 1 | tg32 @ d32768 | 2.36 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Llama-3.3-70B-Instruct-UD-Q8_K_XL-00001-of-00002__vulkan_radv__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | Vulkan | 99 | 1 | pp512 | 99.39 ± 0.58 |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | Vulkan | 99 | 1 | tg128 | 2.76 ± 0.04 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Llama-3.3-70B-Instruct-UD-Q8_K_XL-00001-of-00002__vulkan_radv__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | Vulkan | 99 | 1 | pp2048 @ d32768 | 11.79 ± 0.02 |
| llama 70B Q8_0 | 75.65 GiB | 70.55 B | Vulkan | 99 | 1 | tg32 @ d32768 | 2.44 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Ministral-3-14B-Instruct-2512-BF16__rocm-7_2__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 1 | pp512 | 800.17 ± 1.72 |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 1 | tg128 | 8.49 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Ministral-3-14B-Instruct-2512-BF16__rocm-7_2__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 169.18 ± 1.16 |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 7.11 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Ministral-3-14B-Instruct-2512-BF16__rocm-7_2__hblt0__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 1 | pp512 | 803.22 ± 2.21 |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 1 | tg128 | 8.49 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Ministral-3-14B-Instruct-2512-BF16__rocm-7_2__hblt0__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 170.11 ± 0.81 |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 7.11 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Ministral-3-14B-Instruct-2512-BF16__rocm6_4_4__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 1 | pp512 | 764.18 ± 1.66 |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 1 | tg128 | 8.48 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Ministral-3-14B-Instruct-2512-BF16__rocm6_4_4__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 166.22 ± 1.20 |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 7.10 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Ministral-3-14B-Instruct-2512-BF16__rocm6_4_4__hblt0__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 1 | pp512 | 766.68 ± 1.07 |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 1 | tg128 | 8.48 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Ministral-3-14B-Instruct-2512-BF16__rocm6_4_4__hblt0__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 164.84 ± 1.99 |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 7.10 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Ministral-3-14B-Instruct-2512-BF16__rocm7-nightlies__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 1 | pp512 | 990.88 ± 3.15 |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 1 | tg128 | 8.50 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Ministral-3-14B-Instruct-2512-BF16__rocm7-nightlies__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 172.42 ± 3.61 |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 7.10 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Ministral-3-14B-Instruct-2512-BF16__rocm7-nightlies__hblt0__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 1 | pp512 | 799.71 ± 2.09 |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 1 | tg128 | 8.49 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Ministral-3-14B-Instruct-2512-BF16__rocm7-nightlies__hblt0__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 170.19 ± 1.69 |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 7.10 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Ministral-3-14B-Instruct-2512-BF16__vulkan_amdvlk__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | Vulkan | 99 | 1 | pp512 | 19.70 ± 0.00 |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | Vulkan | 99 | 1 | tg128 | 8.24 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Ministral-3-14B-Instruct-2512-BF16__vulkan_amdvlk__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | Vulkan | 99 | 1 | pp2048 @ d32768 | 16.69 ± 0.01 |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | Vulkan | 99 | 1 | tg32 @ d32768 | 6.41 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Ministral-3-14B-Instruct-2512-BF16__vulkan_radv__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | Vulkan | 99 | 1 | pp512 | 222.01 ± 0.94 |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | Vulkan | 99 | 1 | tg128 | 7.59 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Ministral-3-14B-Instruct-2512-BF16__vulkan_radv__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | Vulkan | 99 | 1 | pp2048 @ d32768 | 76.47 ± 0.38 |
| mistral3 14B BF16 | 25.16 GiB | 13.51 B | Vulkan | 99 | 1 | tg32 @ d32768 | 6.39 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Nemotron-3-Nano-30B-A3B-UD-Q8_K_XL__rocm-7_2__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 1 | pp512 | 393.61 ± 2.94 |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 1 | tg128 | 42.58 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Nemotron-3-Nano-30B-A3B-UD-Q8_K_XL__rocm-7_2__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 562.85 ± 0.47 |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 40.59 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Nemotron-3-Nano-30B-A3B-UD-Q8_K_XL__rocm-7_2__hblt0__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 1 | pp512 | 388.54 ± 2.76 |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 1 | tg128 | 42.61 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Nemotron-3-Nano-30B-A3B-UD-Q8_K_XL__rocm-7_2__hblt0__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 564.71 ± 0.81 |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 40.60 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Nemotron-3-Nano-30B-A3B-UD-Q8_K_XL__rocm6_4_4__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 1 | pp512 | 1026.87 ± 6.06 |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 1 | tg128 | 41.90 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Nemotron-3-Nano-30B-A3B-UD-Q8_K_XL__rocm6_4_4__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 1042.36 ± 2.24 |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 40.08 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Nemotron-3-Nano-30B-A3B-UD-Q8_K_XL__rocm6_4_4__hblt0__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 1 | pp512 | 1027.41 ± 6.28 |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 1 | tg128 | 42.05 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Nemotron-3-Nano-30B-A3B-UD-Q8_K_XL__rocm6_4_4__hblt0__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 1038.86 ± 3.17 |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 40.04 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Nemotron-3-Nano-30B-A3B-UD-Q8_K_XL__rocm7-nightlies__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 1 | pp512 | 1070.15 ± 5.54 |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 1 | tg128 | 42.56 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Nemotron-3-Nano-30B-A3B-UD-Q8_K_XL__rocm7-nightlies__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 908.79 ± 27.38 |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 40.91 ± 0.03 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Nemotron-3-Nano-30B-A3B-UD-Q8_K_XL__rocm7-nightlies__hblt0__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 1 | pp512 | 1038.67 ± 2.82 |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 1 | tg128 | 42.57 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Nemotron-3-Nano-30B-A3B-UD-Q8_K_XL__rocm7-nightlies__hblt0__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 821.93 ± 29.40 |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 40.92 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Nemotron-3-Nano-30B-A3B-UD-Q8_K_XL__vulkan_amdvlk__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | Vulkan | 99 | 1 | pp512 | 676.59 ± 50.83 |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | Vulkan | 99 | 1 | tg128 | 47.22 ± 0.01 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Nemotron-3-Nano-30B-A3B-UD-Q8_K_XL__vulkan_amdvlk__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | Vulkan | 99 | 1 | pp2048 @ d32768 | 342.52 ± 0.46 |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | Vulkan | 99 | 1 | tg32 @ d32768 | 35.25 ± 0.03 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Nemotron-3-Nano-30B-A3B-UD-Q8_K_XL__vulkan_radv__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | Vulkan | 99 | 1 | pp512 | 951.76 ± 41.03 |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | Vulkan | 99 | 1 | tg128 | 46.68 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Nemotron-3-Nano-30B-A3B-UD-Q8_K_XL__vulkan_radv__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | Vulkan | 99 | 1 | pp2048 @ d32768 | 799.39 ± 0.69 |
| nemotron_h_moe 31B.A3.5B Q8_0 | 37.66 GiB | 31.58 B | Vulkan | 99 | 1 | tg32 @ d32768 | 41.15 ± 0.06 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Qwen3-235B-A22B-Instruct-2507-UD-Q3_K_XL-00001-of-00003__rocm-7_2__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| qwen3moe 235B.A22B Q3_K - Medium | 96.99 GiB | 235.09 B | ROCm | 99 | 1 | pp512 | 202.36 ± 3.50 |
| qwen3moe 235B.A22B Q3_K - Medium | 96.99 GiB | 235.09 B | ROCm | 99 | 1 | tg128 | 15.80 ± 0.00 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Qwen3-235B-A22B-Instruct-2507-UD-Q3_K_XL-00001-of-00003__rocm-7_2__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| qwen3moe 235B.A22B Q3_K - Medium | 96.99 GiB | 235.09 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 41.36 ± 0.87 |
| qwen3moe 235B.A22B Q3_K - Medium | 96.99 GiB | 235.09 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 9.65 ± 0.30 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Qwen3-235B-A22B-Instruct-2507-UD-Q3_K_XL-00001-of-00003__rocm-7_2__hblt0__fa1.log
+8
View File
@@ -0,0 +1,8 @@
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 | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| qwen3moe 235B.A22B Q3_K - Medium | 96.99 GiB | 235.09 B | ROCm | 99 | 1 | pp512 | 200.10 ± 8.37 |
| qwen3moe 235B.A22B Q3_K - Medium | 96.99 GiB | 235.09 B | ROCm | 99 | 1 | tg128 | 16.04 ± 0.06 |
build: e0c93af2a (7938)
benchmark/results/04-02-2026/Qwen3-235B-A22B-Instruct-2507-UD-Q3_K_XL-00001-of-00003__rocm-7_2__hblt0__fa1__longctx32768.log
+8
View File
@@ -0,0 +1,8 @@
ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| qwen3moe 235B.A22B Q3_K - Medium | 96.99 GiB | 235.09 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 41.53 ± 0.41 |
| qwen3moe 235B.A22B Q3_K - Medium | 96.99 GiB | 235.09 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 9.50 ± 0.75 |
build: e0c93af2a (7938)

Some files were not shown because too many files have changed in this diff Show More