vulkan: extend topk_moe to handle sigmoid w/exp_probs_b for nemotron

[jeffbolznv]

Also handle GGML_OP_SCALE at the end (nemotron, deepseek2).

Fewer pipeline variants and spec constants, just use push constants.

In test_topk_moe, change exp_probs_b to be 1D, matching real networks.

Update test-backend-ops and ggml-backend to allow verifying multiple outputs in a fusion test (topk_moe has two outputs). Previously only the final node was verified.

before:

Z:\github\jeffbolznv\llama.cpp\build\bin\RelWithDebInfo>llama-bench.exe -fa 1 -p 0 -n 128,128,128 -m c:\models\Nemotron-3-Nano-30B-A3B-Q4_K_M.gguf -m c:\models\deepseek-v2-lite-safetensors\deepseek-v2-lite-Q4_K_M.gguf
ggml_vulkan: Found 1 Vulkan devices:
ggml_vulkan: 0 = NVIDIA GeForce RTX 5090 (NVIDIA) | uma: 0 | fp16: 1 | bf16: 1 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: NV_coopmat2
| model                          |       size |     params | backend    | ngl | fa |            test |                  t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| nemotron_h_moe 31B.A3.5B Q4_K - Medium |  22.88 GiB |    31.58 B | Vulkan     |  99 |  1 |           tg128 |       269.32 ± 13.22 |
| nemotron_h_moe 31B.A3.5B Q4_K - Medium |  22.88 GiB |    31.58 B | Vulkan     |  99 |  1 |           tg128 |        260.52 ± 1.17 |
| nemotron_h_moe 31B.A3.5B Q4_K - Medium |  22.88 GiB |    31.58 B | Vulkan     |  99 |  1 |           tg128 |        267.10 ± 5.18 |
| deepseek2 16B Q4_K - Medium    |   9.65 GiB |    15.71 B | Vulkan     |  99 |  1 |           tg128 |       340.67 ± 22.33 |
| deepseek2 16B Q4_K - Medium    |   9.65 GiB |    15.71 B | Vulkan     |  99 |  1 |           tg128 |        356.88 ± 9.24 |
| deepseek2 16B Q4_K - Medium    |   9.65 GiB |    15.71 B | Vulkan     |  99 |  1 |           tg128 |       333.40 ± 12.02 |

after:

Z:\github\jeffbolznv\llama.cpp\build\bin\RelWithDebInfo>llama-bench.exe -fa 1 -p 0 -n 128,128,128 -m c:\models\Nemotron-3-Nano-30B-A3B-Q4_K_M.gguf -m c:\models\deepseek-v2-lite-safetensors\deepseek-v2-lite-Q4_K_M.gguf
ggml_vulkan: Found 1 Vulkan devices:
ggml_vulkan: 0 = NVIDIA GeForce RTX 5090 (NVIDIA) | uma: 0 | fp16: 1 | bf16: 1 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: NV_coopmat2
| model                          |       size |     params | backend    | ngl | fa |            test |                  t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | --------------: | -------------------: |
| nemotron_h_moe 31B.A3.5B Q4_K - Medium |  22.88 GiB |    31.58 B | Vulkan     |  99 |  1 |           tg128 |       288.13 ± 13.10 |
| nemotron_h_moe 31B.A3.5B Q4_K - Medium |  22.88 GiB |    31.58 B | Vulkan     |  99 |  1 |           tg128 |        284.81 ± 2.36 |
| nemotron_h_moe 31B.A3.5B Q4_K - Medium |  22.88 GiB |    31.58 B | Vulkan     |  99 |  1 |           tg128 |        289.09 ± 3.86 |
| deepseek2 16B Q4_K - Medium    |   9.65 GiB |    15.71 B | Vulkan     |  99 |  1 |           tg128 |       343.03 ± 19.78 |
| deepseek2 16B Q4_K - Medium    |   9.65 GiB |    15.71 B | Vulkan     |  99 |  1 |           tg128 |        355.02 ± 4.88 |
| deepseek2 16B Q4_K - Medium    |   9.65 GiB |    15.71 B | Vulkan     |  99 |  1 |           tg128 |        353.27 ± 0.69 |

[ggerganov]

Ack on the ggml-backend changes

[0cc4m]

From my side it looks fine, but the Vulkan Mac CI is reporting an issue. Can you look into that?

[jeffbolznv]

I don't know why the mac system is failing. It's in test cases that should be fused so I don't think it's a fluke, but I can't reproduce it locally on NVIDIA or lavapipe and I can't find anything from code inspection.

While trying I did find that sometimes there are ties that lead to spurious failures, so I've updated the tests to avoid that. I doubt this is related to the mac failures. If it still fails in CI I'll probably need to just disable this fusion for moltenvk.

[jeffbolznv]

I tried a couple experiments through CI, but don't have a workaround for the moltenvk failures. I've disabled the new fusion for moltenvk.