v2.8.3 新增测试间隔时间：每次采样之间让大模型接口空闲休息，默认5秒可配置(0关闭)，等待期间支持停止

tester.py

@@ -32,6 +32,15 @@ class TestRunner(threading.Thread):
     def log(self, level, msg):
         db.add_log(self.test_id, level, msg)
 
+    def _sleep_interval(self, secs):
+        """测试间隔等待：期间可被用户停止，返回 False 表示已被取消"""
+        deadline = time.time() + secs
+        while time.time() < deadline:
+            if self.should_stop():
+                return False
+            time.sleep(0.2)
+        return not self.should_stop()
+
     # ───────────────────────── 主流程 ─────────────────────────
 
     def run(self):
@@ -61,6 +70,11 @@ class TestRunner(threading.Thread):
         max_tokens = max(1, int(gen.get("max_tokens", 128)))   # 解码输出长度
         avoid_cache = bool(gen.get("avoid_cache"))
         warmup = bool(gen.get("warmup", True))                 # 测试前空转预热
+        # 测试间隔（秒）：每次采样之间让接口空闲休息，默认 5 秒，0 表示不等待
+        try:
+            interval = 5 if gen.get("interval") in (None, "") else max(0, float(gen.get("interval")))
+        except (TypeError, ValueError):
+            interval = 5
 
         # 并发数列表（默认单流 [1]；支持 2/4 及自定义，如 [1,2,4,8]）
         raw_concs = gen.get("concurrency_levels") or []
@@ -73,10 +87,10 @@ class TestRunner(threading.Thread):
         if name:
             self.log("INFO", "测试名称(主题): %s" % name)
         self.log("INFO", "提供商: %s | 模型: %s" % (lp.PROVIDER_LABELS.get(provider, provider), model))
-        self.log("INFO", "上下文长度: %s tokens | 生成长度: %d tokens | 并发数: %s | 每个组合采样: %d 次 | 预热: %s | 避免缓存: %s"
+        self.log("INFO", "上下文长度: %s tokens | 生成长度: %d tokens | 并发数: %s | 每个组合采样: %d 次 | 预热: %s | 避免缓存: %s | 测试间隔: %g 秒"
                  % (" / ".join(str(x) for x in lengths), max_tokens,
                     " / ".join(str(x) for x in concurrency_levels), n,
-                    "开" if warmup else "关", "开" if avoid_cache else "关"))
+                    "开" if warmup else "关", "开" if avoid_cache else "关", interval))
         to = self.gen.get("read_timeout") or "默认(1800)"
         self.log("INFO", "请求超时: 连接 %s s | 等待首字/预处理 %s s（接口慢可在左侧调大）"
                  % (self.gen.get("connect_timeout") or 60, to))
@@ -86,6 +100,7 @@ class TestRunner(threading.Thread):
         self.log("INFO", "校准完成: %.3f tok/字符（%.2f 字符/token）" % (ratio, 1.0 / ratio))
 
         run_seq = 0
+        sample_done = 0   # 已完成的采样数，用于控制测试间隔（首个采样不等待）
         last_conc = None  # 只在实际切换并发档时打印一次表头，避免同一并发数重复刷屏
         for L in lengths:
             if self.should_stop():
@@ -103,7 +118,12 @@ class TestRunner(threading.Thread):
                 for i in range(1, n + 1):
                     if self.should_stop():
                         raise StopRequested()
+                    if sample_done > 0 and interval > 0:
+                        self.log("INFO", "⏳ 接口空闲休息 %g 秒后继续下一采样..." % interval)
+                        if not self._sleep_interval(interval):
+                            raise StopRequested()
                     run_seq += 1
+                    sample_done += 1
                     self.log("INFO", "── [%d tok · 并发%d] 采样 %d/%d 开始 ──" % (L, C, i, n))
                     try:
                         m = self._run_sample(C, base_prompt, max_tokens, avoid_cache)