vulkan: enable flash attention (#12937)

Also adjusts the vulkan windows build pattern to match recent changes in other backends
so incremental builds are faster.

ml/device.go

@@ -432,7 +432,8 @@ func FlashAttentionSupported(l []DeviceInfo) bool {
 		supportsFA := gpu.Library == "cpu" ||
 			gpu.Name == "Metal" || gpu.Library == "Metal" ||
 			(gpu.Library == "CUDA" && gpu.DriverMajor >= 7 && !(gpu.ComputeMajor == 7 && gpu.ComputeMinor == 2)) ||
-			gpu.Library == "ROCm"
+			gpu.Library == "ROCm" ||
+			gpu.Library == "Vulkan"
 
 		if !supportsFA {
 			return false

scripts/build_windows.ps1

@@ -187,11 +187,11 @@ function buildROCm() {
 function buildVulkan(){
     if ($env:VULKAN_SDK) {
         write-host "Building Vulkan backend libraries"
-        & cmake --fresh --preset Vulkan --install-prefix $script:DIST_DIR -DOLLAMA_RUNNER_DIR="vulkan"
+        & cmake -B build\vulkan --preset Vulkan --install-prefix $script:DIST_DIR -DOLLAMA_RUNNER_DIR="vulkan"
         if ($LASTEXITCODE -ne 0) { exit($LASTEXITCODE)}
-        & cmake --build --preset Vulkan  --config Release --parallel $script:JOBS
+        & cmake --build build\vulkan --target ggml-vulkan  --config Release --parallel $script:JOBS
         if ($LASTEXITCODE -ne 0) { exit($LASTEXITCODE)}
-        & cmake --install build --component Vulkan --strip
+        & cmake --install build\vulkan --component Vulkan --strip
         if ($LASTEXITCODE -ne 0) { exit($LASTEXITCODE)}
     }
 }