Cleanup and generation time

flue-core/src/flux/sampling.rs

@@ -8,8 +8,8 @@ pub fn get_noise(
     width: usize,
     device: &Device,
 ) -> Result<Tensor> {
-    let height = (height + 15) / 16 * 2;
-    let width = (width + 15) / 16 * 2;
+    let height = height.div_ceil(16) * 2;
+    let width = width.div_ceil(16) * 2;
     Tensor::randn(0f32, 1., (num_samples, 16, height, width), device)
 }
 
@@ -86,8 +86,8 @@ pub fn get_schedule(num_steps: usize, shift: Option<(usize, f64, f64)>) -> Vec<f
 
 pub fn unpack(xs: &Tensor, height: usize, width: usize) -> Result<Tensor> {
     let (b, _h_w, c_ph_pw) = xs.dims3()?;
-    let height = (height + 15) / 16;
-    let width = (width + 15) / 16;
+    let height = height.div_ceil(16);
+    let width = width.div_ceil(16);
     xs.reshape((b, height, width, c_ph_pw / 4, 2, 2))? // (b, h, w, c, ph, pw)
         .permute((0, 3, 1, 4, 2, 5))? // (b, c, h, ph, w, pw)
         .reshape((b, c_ph_pw / 4, height * 2, width * 2))

flue-server/Cargo.toml

@@ -11,7 +11,7 @@ license.workspace = true
 homepage.workspace = true
 
 [dependencies]
-flue-core = { version = "0.1.0", path = "../flue-core" }
+flue-core = { path = "../flue-core" }
 anyhow = { workspace = true }
 axum = { workspace = true }
 base64 = { workspace = true }
@@ -21,12 +21,3 @@ image = { workspace = true }
 serde = { workspace = true }
 serde_json = { workspace = true }
 tokio = { workspace = true }
-
-[features]
-cuda = ["flue-core/cuda"]
-cudnn = ["flue-core/cudnn"]
-metal = ["flue-core/metal"]
-flash-attn-v2 = ["cuda", "flue-core/flash-attn-v2"]
-flash-attn-v3 = ["cuda", "flue-core/flash-attn-v3"]
-accelerate = ["flue-core/accelerate"]
-mkl = ["flue-core/mkl"]

flue-server/src/main.rs

@@ -46,6 +46,7 @@ fn image_to_base64_png(img: &DynamicImage) -> Result<String> {
 #[derive(Serialize)]
 struct GenerationResponse {
     image: String,
+    gen_time: f64, // Time in seconds
 }
 
 // Application state containing the preloaded models and device settings.
@@ -57,7 +58,11 @@ async fn generate_image_handler(
     Json(req): Json<GenerationRequest>,
 ) -> impl IntoResponse {
     match generate_image(req, &state).await {
-        Ok(img_base64) => Json(GenerationResponse { image: img_base64 }).into_response(),
+        Ok((img_base64, gen_time)) => Json(GenerationResponse {
+            image: img_base64,
+            gen_time,
+        })
+        .into_response(),
         Err(e) => {
             eprintln!("Error generating image: {:?}", e);
             (StatusCode::INTERNAL_SERVER_ERROR, format!("Error: {:?}", e)).into_response()
@@ -66,9 +71,13 @@ async fn generate_image_handler(
 }
 
 /// This function uses the preloaded models from `state` to generate an image (base64).
-async fn generate_image(params: GenerationRequest, state: &AppState) -> Result<String> {
+/// Returns both the base64 image and the generation time in seconds.
+async fn generate_image(params: GenerationRequest, state: &AppState) -> Result<(String, f64)> {
+    let start_time = std::time::Instant::now();
     let image = state.0.run(params)?;
-    image_to_base64_png(&image)
+    let gen_time = start_time.elapsed().as_secs_f64();
+    let base64_image = image_to_base64_png(&image)?;
+    Ok((base64_image, gen_time))
 }
 
 #[tokio::main]