=== ncmoe 34 ub 512
| model                          |       size |     params | backend    | ngl |  n_cpu_moe | n_batch | type_k | type_v |  fa |            test |                  t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | ---------: | ------: | -----: | -----: | --: | --------------: | -------------------: |
=== ncmoe 34 ub 2048
| model                          |       size |     params | backend    | ngl |  n_cpu_moe | n_ubatch | type_k | type_v |  fa |            test |                  t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | ---------: | -------: | -----: | -----: | --: | --------------: | -------------------: |
=== ncmoe 38 ub 512
| model                          |       size |     params | backend    | ngl |  n_cpu_moe | n_batch | type_k | type_v |  fa |            test |                  t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | ---------: | ------: | -----: | -----: | --: | --------------: | -------------------: |
| qwen4exp A3B IQ3_XXS - 3.0625 bpw |  70.62 GiB |   176.94 B | CUDA       |  99 |         38 |     512 |   q8_0 |   q8_0 |   1 |          pp2048 |        122.67 ± 0.00 |
| qwen4exp A3B IQ3_XXS - 3.0625 bpw |  70.62 GiB |   176.94 B | CUDA       |  99 |         38 |     512 |   q8_0 |   q8_0 |   1 |            tg64 |         10.75 ± 0.00 |
=== ncmoe 38 ub 2048
| model                          |       size |     params | backend    | ngl |  n_cpu_moe | n_ubatch | type_k | type_v |  fa |            test |                  t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | ---------: | -------: | -----: | -----: | --: | --------------: | -------------------: |
| qwen4exp A3B IQ3_XXS - 3.0625 bpw |  70.62 GiB |   176.94 B | CUDA       |  99 |         38 |     2048 |   q8_0 |   q8_0 |   1 |          pp2048 |        222.23 ± 0.00 |
| qwen4exp A3B IQ3_XXS - 3.0625 bpw |  70.62 GiB |   176.94 B | CUDA       |  99 |         38 |     2048 |   q8_0 |   q8_0 |   1 |            tg64 |         10.65 ± 0.00 |
=== ncmoe 42 ub 512
| model                          |       size |     params | backend    | ngl |  n_cpu_moe | n_batch | type_k | type_v |  fa |            test |                  t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | ---------: | ------: | -----: | -----: | --: | --------------: | -------------------: |
| qwen4exp A3B IQ3_XXS - 3.0625 bpw |  70.62 GiB |   176.94 B | CUDA       |  99 |         42 |     512 |   q8_0 |   q8_0 |   1 |          pp2048 |        118.29 ± 0.00 |
| qwen4exp A3B IQ3_XXS - 3.0625 bpw |  70.62 GiB |   176.94 B | CUDA       |  99 |         42 |     512 |   q8_0 |   q8_0 |   1 |            tg64 |         10.54 ± 0.00 |
=== ncmoe 42 ub 2048
| model                          |       size |     params | backend    | ngl |  n_cpu_moe | n_ubatch | type_k | type_v |  fa |            test |                  t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | ---------: | -------: | -----: | -----: | --: | --------------: | -------------------: |
| qwen4exp A3B IQ3_XXS - 3.0625 bpw |  70.62 GiB |   176.94 B | CUDA       |  99 |         42 |     2048 |   q8_0 |   q8_0 |   1 |          pp2048 |        217.04 ± 0.00 |
| qwen4exp A3B IQ3_XXS - 3.0625 bpw |  70.62 GiB |   176.94 B | CUDA       |  99 |         42 |     2048 |   q8_0 |   q8_0 |   1 |            tg64 |         10.35 ± 0.00 |
TUNE-DONE
