Files
enginex-ascend-910-llama.cpp/ggml/src
Jeff Bolz e6d65fb02d vulkan: support arbitrary KV dimension in flash attention (#16160)
The "Clamp" spec constant is already based on whether KV is a multiple of Bc,
so use that to control whether bounds checking is performed. Add bounds checking
to the scalar and coopmat1 paths. Coopmat2 didn't need any changes (the K/V
tensors are already optionally clamped, nothing else needed to be changed).
2025-09-27 22:43:39 +02:00
..
2025-08-05 22:10:36 +03:00
2025-08-05 22:10:36 +03:00
2025-09-27 02:03:33 +08:00
2025-09-05 11:34:28 +02:00