fix: 12 bug fixes from comprehensive review — 51 tok/s verified on RTX 5090

P0 fixes (blocking usability):
- FIX-01: thread-local cuBLAS handle (was creating/destroying per matmul)
- FIX-16: EOS token no longer leaks into API responses
- FIX-17: max_seq_len configurable via --max-seq-len (default 2048, was hardcoded 256)
- FIX-18: max_tokens clamped to available seq space, prompt overflow returns 400

P1 fixes (bugs & performance):
- FIX-07: CachingAllocator wired into all hot paths (to_device, embedding, rope, concat)
- FIX-08: CudaDeviceProp buffer increased to 32KB for CUDA 12.9 safety
- FIX-09: tokenizer byte_fallback graceful degradation (was panic)
- FIX-19: causal mask uses -INFINITY instead of -1e9 (BF16 supports inf)
- FIX-20: LayerNorm rewritten to numerically stable two-pass algorithm
- FIX-21: min block size guard (32 threads) for LayerNorm/RMSNorm launches

P2 fixes (improvements):
- FIX-22: Option<GpuKVCache> + take() eliminates dummy KV cache allocations
- FIX-23: RoPE cache no longer artificially capped at 8192 positions

Verified on dash5 (RTX 5090): 51 tok/s batch=1, 74 tok/s 2-concurrent, 1.7-3.3x HF transformers.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

crates/xserv-server/src/main.rs

@@ -10,13 +10,14 @@ pub struct AppState {
     pub model_name: String,
     pub engine_sender: Mutex<mpsc::Sender<GenerateRequest>>,
     pub engine_tokenizer: Mutex<xserv_tokenizer::Tokenizer>,
+    pub max_seq_len: usize,
 }
 
 #[tokio::main]
 async fn main() {
     let args: Vec<String> = std::env::args().collect();
     if args.len() < 2 {
-        eprintln!("Usage: xserv-server <model-dir> [--port PORT] [--max-batch N]");
+        eprintln!("Usage: xserv-server <model-dir> [--port PORT] [--max-batch N] [--max-seq-len N]");
         std::process::exit(1);
     }
 
@@ -31,6 +32,11 @@ async fn main() {
         .and_then(|i| args.get(i + 1))
         .and_then(|s| s.parse().ok())
         .unwrap_or(4);
+    let max_seq_len: usize = args.iter()
+        .position(|a| a == "--max-seq-len")
+        .and_then(|i| args.get(i + 1))
+        .and_then(|s| s.parse().ok())
+        .unwrap_or(2048);
 
     let model_name = model_dir.file_name()
         .map(|n| n.to_string_lossy().to_string())
@@ -43,7 +49,7 @@ async fn main() {
 
     let model_dir_clone = model_dir.clone();
     std::thread::spawn(move || {
-        let engine = engine::Engine::load(&model_dir_clone, max_batch);
+        let engine = engine::Engine::load(&model_dir_clone, max_batch, max_seq_len);
         engine.run(rx);
     });
 
@@ -51,6 +57,7 @@ async fn main() {
         model_name,
         engine_sender: Mutex::new(tx),
         engine_tokenizer: Mutex::new(tokenizer),
+        max_seq_len,
     });
 
     let app = Router::new()
@@ -60,7 +67,7 @@ async fn main() {
         .layer(Extension(state));
 
     let addr = format!("0.0.0.0:{port}");
-    eprintln!("[server] Listening on {addr} (max_batch={max_batch})");
+    eprintln!("[server] Listening on {addr} (max_batch={max_batch}, max_seq_len={max_seq_len})");
     let listener = tokio::net::TcpListener::bind(&addr).await.unwrap();
     axum::serve(listener, app).await.unwrap();
 }