Release v3.5.0: performance, constants, bilingual docs

- Bump version to 3.5.0
- Performance: hot-path timing only when log_enabled/simulate; execute_parallel takes &[Arc<SwqosClient>]; shared HTTP client constants for SWQoS
- Code quality: validate_protocol_params extracted for buy/sell; BYTES_PER_ACCOUNT, MAX_INSTRUCTIONS_WARN, HTTP timeout constants; prefetch/syscall bypass comments
- Documentation: bilingual (EN + 中文) doc comments in execution, executor, perf, swqos; README/README_CN version and What's new in 3.5.0
- Add release_notes_v3.5.0.md

Co-authored-by: Cursor <cursoragent@cursor.com>

src/common/clock.rs

@@ -0,0 +1,97 @@
+//! High-performance clock (same design as sol-parser-sdk for consistent grpc_recv_us vs "now").
+//!
+//! Uses monotonic clock + base UTC timestamp to avoid frequent syscalls; aligned with sol-parser-sdk
+//! so event-side grpc_recv_us and SDK-side now_micros() share the same time scale.
+
+use std::time::Instant;
+
+/// High-performance clock: monotonic + base UTC microsecond timestamp.
+#[derive(Debug)]
+pub struct HighPerformanceClock {
+    base_instant: Instant,
+    base_timestamp_us: i64,
+    last_calibration: Instant,
+    calibration_interval_secs: u64,
+}
+
+impl HighPerformanceClock {
+    /// Calibrate every 5 minutes by default.
+    pub fn new() -> Self {
+        Self::new_with_calibration_interval(300)
+    }
+
+    /// Sample multiple times and use the lowest-latency baseline to reduce init error.
+    pub fn new_with_calibration_interval(calibration_interval_secs: u64) -> Self {
+        let mut best_offset = i64::MAX;
+        let mut best_instant = Instant::now();
+        let mut best_timestamp = chrono::Utc::now().timestamp_micros();
+
+        for _ in 0..3 {
+            let instant_before = Instant::now();
+            let timestamp = chrono::Utc::now().timestamp_micros();
+            let instant_after = Instant::now();
+            let sample_latency = instant_after.duration_since(instant_before).as_nanos() as i64;
+            if sample_latency < best_offset {
+                best_offset = sample_latency;
+                best_instant = instant_before;
+                best_timestamp = timestamp;
+            }
+        }
+
+        Self {
+            base_instant: best_instant,
+            base_timestamp_us: best_timestamp,
+            last_calibration: best_instant,
+            calibration_interval_secs,
+        }
+    }
+
+    #[inline(always)]
+    pub fn now_micros(&self) -> i64 {
+        let elapsed = self.base_instant.elapsed();
+        self.base_timestamp_us + elapsed.as_micros() as i64
+    }
+
+    /// Recalibrate when needed to prevent drift.
+    pub fn now_micros_with_calibration(&mut self) -> i64 {
+        if self.last_calibration.elapsed().as_secs() >= self.calibration_interval_secs {
+            self.recalibrate();
+        }
+        self.now_micros()
+    }
+
+    fn recalibrate(&mut self) {
+        let current_monotonic = Instant::now();
+        let current_utc = chrono::Utc::now().timestamp_micros();
+        let expected_utc = self.base_timestamp_us
+            + current_monotonic.duration_since(self.base_instant).as_micros() as i64;
+        let drift_us = current_utc - expected_utc;
+        if drift_us.abs() > 1000 {
+            self.base_instant = current_monotonic;
+            self.base_timestamp_us = current_utc;
+        }
+        self.last_calibration = current_monotonic;
+    }
+}
+
+impl Default for HighPerformanceClock {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
+static HIGH_PERF_CLOCK: once_cell::sync::OnceCell<HighPerformanceClock> =
+    once_cell::sync::OnceCell::new();
+
+/// Current time in microseconds (UTC scale); same as sol-parser-sdk clock::now_micros for comparable grpc_recv_us.
+#[inline(always)]
+pub fn now_micros() -> i64 {
+    let clock = HIGH_PERF_CLOCK.get_or_init(HighPerformanceClock::new);
+    clock.now_micros()
+}
+
+/// Elapsed microseconds from start_timestamp_us to now.
+#[inline(always)]
+pub fn elapsed_micros_since(start_timestamp_us: i64) -> i64 {
+    now_micros() - start_timestamp_us
+}

src/common/fast_timing.rs

@@ -174,7 +174,9 @@ mod tests {
         let total_elapsed = start.elapsed();
         let avg_per_call = total_elapsed.as_nanos() / iterations;
 
-        println!("Average fast_now_nanos() call: {}ns", avg_per_call);
+        if crate::common::sdk_log::sdk_log_enabled() {
+            println!("Average fast_now_nanos() call: {}ns", avg_per_call);
+        }
 
         // 快速时间戳应该非常快（< 100ns per call）
         assert!(avg_per_call < 100);
@@ -193,6 +195,8 @@ mod tests {
         let total_elapsed = start.elapsed();
         let avg_per_call = total_elapsed.as_nanos() / iterations;
 
-        println!("Average Instant::now() call: {}ns", avg_per_call);
+        if crate::common::sdk_log::sdk_log_enabled() {
+            println!("Average Instant::now() call: {}ns", avg_per_call);
+        }
     }
 }

src/common/gas_fee_strategy.rs

@@ -354,6 +354,9 @@ impl GasFeeStrategy {
     /// 打印所有策略。
     /// Print all strategies
     pub fn print_all_strategies(&self) {
+        if !crate::common::sdk_log::sdk_log_enabled() {
+            return;
+        }
         for strategy in self.get_strategies(TradeType::Buy) {
             println!("[buy] - {:?}", strategy);
         }

src/common/mod.rs

@@ -1,5 +1,8 @@
+pub mod address_lookup;
 pub mod bonding_curve;
+pub mod clock;
 pub mod fast_fn;
+pub mod sdk_log;
 pub mod fast_timing;
 pub mod gas_fee_strategy;
 pub mod global;
@@ -10,7 +13,6 @@ pub mod spl_token;
 pub mod spl_token_2022;
 pub mod subscription_handle;
 pub mod types;
-pub mod address_lookup;
 
 pub use gas_fee_strategy::*;
 pub use types::*;

src/common/sdk_log.rs

@@ -0,0 +1,19 @@
+//! sol-trade-sdk global log switch
+//!
+//! Controlled by `TradeConfig::log_enabled`, set in `TradingClient::new`.
+//! All SDK logs (timing, SWQOS submit/confirm, WSOL, blacklist, etc.) should check this before output.
+
+use std::sync::atomic::{AtomicBool, Ordering};
+
+static SDK_LOG_ENABLED: AtomicBool = AtomicBool::new(true);
+
+/// Whether SDK logging is enabled (set from TradeConfig.log_enabled in TradingClient::new).
+#[inline(always)]
+pub fn sdk_log_enabled() -> bool {
+    SDK_LOG_ENABLED.load(Ordering::Relaxed)
+}
+
+/// Set the SDK global log switch (only called from TradingClient::new).
+pub fn set_sdk_log_enabled(enabled: bool) {
+    SDK_LOG_ENABLED.store(enabled, Ordering::Relaxed);
+}

src/common/types.rs

@@ -72,6 +72,10 @@ pub struct TradeConfig {
     pub create_wsol_ata_on_startup: bool,
     /// Whether to use seed optimization for all ATA operations (default: true)
     pub use_seed_optimize: bool,
+    /// Whether to pin parallel submit tasks to CPU cores (can reduce latency; set false in containers). Default true.
+    pub use_core_affinity: bool,
+    /// Whether to output all SDK logs (timing, SWQOS submit/confirm, WSOL, blacklist, etc.). Default true.
+    pub log_enabled: bool,
 }
 
 impl TradeConfig {
@@ -80,14 +84,18 @@ impl TradeConfig {
         swqos_configs: Vec<SwqosConfig>,
         commitment: CommitmentConfig,
     ) -> Self {
-        println!("🔧 TradeConfig create_wsol_ata_on_startup default value: true");
-        println!("🔧 TradeConfig use_seed_optimize default value: true");
+        if crate::common::sdk_log::sdk_log_enabled() {
+            println!("🔧 TradeConfig create_wsol_ata_on_startup default: true");
+            println!("🔧 TradeConfig use_seed_optimize default: true");
+        }
         Self {
             rpc_url,
             swqos_configs,
             commitment,
-            create_wsol_ata_on_startup: true,  // 默认：启动时检查并创建
-            use_seed_optimize: true,           // 默认：使用seed优化
+            create_wsol_ata_on_startup: true,  // default: check and create on startup
+            use_seed_optimize: true,           // default: use seed optimization
+            use_core_affinity: true,           // default: pin parallel submit tasks to cores
+            log_enabled: true,                 // default: enable all SDK logs
         }
     }