{"id":592482,"date":"2025-08-14T10:47:27","date_gmt":"2025-08-14T09:47:27","guid":{"rendered":"https:\/\/blog.jetbrains.com\/?post_type=pycharm&p=592482"},"modified":"2025-10-16T11:05:22","modified_gmt":"2025-10-16T10:05:22","slug":"faster-python-unlocking-the-python-global-interpreter-lock","status":"publish","type":"pycharm","link":"https:\/\/blog.jetbrains.com\/zh-hans\/pycharm\/2025\/08\/faster-python-unlocking-the-python-global-interpreter-lock\/","title":{"rendered":"\u66f4\u5feb\u7684 Python\uff1a\u89e3\u5f00 Python \u5168\u5c40\u89e3\u91ca\u5668\u9501"},"content":{"rendered":"

\u4ec0\u4e48\u662f Python \u7684\u5168\u5c40\u89e3\u91ca\u5668\u9501 (GIL)\uff1f<\/h2>\n

\u201c\u5168\u5c40\u89e3\u91ca\u5668\u9501\u201d\uff08\u6216 GIL\uff09\u662f Python \u793e\u533a\u4e2d\u7684\u5e38\u89c1\u672f\u8bed\u3002 \u8fd9\u662f\u4e00\u4e2a\u4f17\u6240\u5468\u77e5\u7684 Python \u529f\u80fd\u3002 \u4f46 GIL \u5230\u5e95\u662f\u4ec0\u4e48\uff1f<\/p>\n

\u5982\u679c\u60a8\u6709\u4f7f\u7528\u5176\u4ed6\u7f16\u7a0b\u8bed\u8a00\uff08\u4f8b\u5982 Rust\uff09\u7684\u7ecf\u9a8c\uff0c\u60a8\u53ef\u80fd\u5df2\u7ecf\u77e5\u9053\u4ec0\u4e48\u662f\u4e92\u65a5\u9501 (mutex)\u3002 \u4e92\u65a5\u9501\u53ef\u4ee5\u786e\u4fdd\u6570\u636e\u6bcf\u6b21\u53ea\u80fd\u7531\u4e00\u4e2a\u7ebf\u7a0b\u8bbf\u95ee\u3002 \u8fd9\u53ef\u4ee5\u9632\u6b62\u6570\u636e\u88ab\u591a\u4e2a\u7ebf\u7a0b\u540c\u65f6\u4fee\u6539\u3002 \u60a8\u53ef\u4ee5\u628a\u5b83\u89c6\u4e3a\u4e00\u79cd\u201c\u9501\u201d\uff0c\u5b83\u4f1a\u963b\u6b62\u6240\u6709\u7ebf\u7a0b\u8bbf\u95ee\u6570\u636e\uff0c\u9664\u4e86\u6301\u6709\u5bc6\u94a5\u7684\u7ebf\u7a0b\u4e4b\u5916\u3002<\/p>\n

GIL \u57fa\u672c\u4e0a\u662f\u4e00\u4e2a\u4e92\u65a5\u9501\u3002 \u5b83\u4e00\u6b21\u53ea\u5141\u8bb8\u4e00\u4e2a\u7ebf\u7a0b\u8bbf\u95ee Python \u89e3\u91ca\u5668\u3002 \u6211\u6709\u65f6\u628a\u5b83\u60f3\u8c61\u6210 Python \u7684\u65b9\u5411\u76d8\u3002 \u60a8\u80af\u5b9a\u4e0d\u4f1a\u60f3\u8ba9\u591a\u4e2a\u4eba\u64cd\u63a7\u65b9\u5411\u76d8\uff01 \u4f46\u8bdd\u8bf4\u56de\u6765\uff0c\u4e00\u7fa4\u4eba\u65c5\u884c\u65f6\u7ecf\u5e38\u4f1a\u6362\u53f8\u673a\u3002 \u8fd9\u5c31\u6709\u70b9\u50cf\u628a\u89e3\u91ca\u5668\u8bbf\u95ee\u6743\u9650\u4ea4\u7ed9\u53e6\u4e00\u4e2a\u7ebf\u7a0b\u3002<\/p>\n

\u7531\u4e8e GIL\uff0cPython \u4e0d\u5141\u8bb8\u771f\u6b63\u7684\u591a\u7ebf\u7a0b\u8fdb\u7a0b\u3002 \u8fd9\u9879\u529f\u80fd\u5728\u8fc7\u53bb\u5341\u5e74\u4e2d\u5f15\u53d1\u4e86\u4e89\u8bae\uff0c\u5e76\u4e14\u6709\u5f88\u591a\u5c1d\u8bd5\u901a\u8fc7\u79fb\u9664 GIL \u548c\u5141\u8bb8\u591a\u7ebf\u7a0b\u8fdb\u7a0b\u6765\u63d0\u9ad8 Python \u7684\u901f\u5ea6\u3002 \u6700\u8fd1\u5728 Python 3.13 \u4e2d\uff0c\u5f15\u5165\u4e86\u4e00\u79cd\u65e0\u9700 GIL \u5373\u53ef\u4f7f\u7528 Python \u7684\u9009\u9879\uff0c\u6709\u65f6\u4e5f\u79f0\u4e3a\u65e0 GIL \u6216\u81ea\u7531\u7ebf\u7a0b Python\u3002 \u7531\u6b64\uff0cPython \u7f16\u7a0b\u7684\u65b0\u65f6\u4ee3\u5f00\u59cb\u4e86\u3002<\/p>\n

\u4e3a\u4ec0\u4e48 GIL \u6700\u5f00\u59cb\u4f1a\u51fa\u73b0\uff1f<\/h2>\n

\u65e2\u7136 GIL \u8fd9\u4e48\u4e0d\u53d7\u6b22\u8fce\uff0c\u90a3\u5f53\u521d\u4e3a\u4ec0\u4e48\u8981\u5b9e\u73b0\u5b83\u5462\uff1f \u62e5\u6709 GIL \u5176\u5b9e\u6709\u5f88\u591a\u597d\u5904\u3002 \u5728\u5176\u4ed6\u5177\u6709\u771f\u6b63\u591a\u7ebf\u7a0b\u5904\u7406\u7684\u7f16\u7a0b\u8bed\u8a00\u4e2d\uff0c\u6709\u65f6\u95ee\u9898\u6e90\u81ea\u591a\u4e2a\u7ebf\u7a0b\u4fee\u6539\u6570\u636e\uff0c\u6700\u7ec8\u7ed3\u679c\u53d6\u51b3\u4e8e\u54ea\u4e2a\u7ebf\u7a0b\u6216\u8fdb\u7a0b\u9996\u5148\u5b8c\u6210\u3002 \u8fd9\u88ab\u79f0\u4e3a\u201c\u7ade\u4e89\u6761\u4ef6\u201d\u3002 \u50cf Rust \u8fd9\u6837\u7684\u8bed\u8a00\u901a\u5e38\u5f88\u96be\u5b66\u4e60\uff0c\u56e0\u4e3a\u7a0b\u5e8f\u5458\u5fc5\u987b\u4f7f\u7528\u4e92\u65a5\u9501\u9632\u6b62\u7ade\u4e89\u6761\u4ef6\u3002<\/p>\n

\u5728 Python \u4e2d\uff0c\u6240\u6709\u5bf9\u8c61\u90fd\u6709\u4e00\u4e2a\u5f15\u7528\u8ba1\u6570\u5668\u6765\u8ddf\u8e2a\u6709\u591a\u5c11\u5176\u4ed6\u5bf9\u8c61\u9700\u8981\u4ece\u5b83\u4eec\u83b7\u53d6\u4fe1\u606f\u3002 \u5982\u679c\u5f15\u7528\u8ba1\u6570\u5668\u8fbe\u5230\u96f6\uff0c\u56e0\u4e3a\u6211\u4eec\u77e5\u9053\u7531\u4e8e GIL\uff0cPython \u4e2d\u4e0d\u5b58\u5728\u7ade\u4e89\u6761\u4ef6\uff0c\u6211\u4eec\u53ef\u4ee5\u653e\u5fc3\u5730\u58f0\u660e\u8be5\u5bf9\u8c61\u4e0d\u518d\u9700\u8981\u5e76\u4e14\u53ef\u4ee5\u4f5c\u4e3a\u5783\u573e\u88ab\u56de\u6536\u3002<\/p>\n

\u5f53 Python \u5728 1991 \u5e74\u9996\u6b21\u53d1\u5e03\u65f6\uff0c\u5927\u591a\u6570\u4e2a\u4eba\u8ba1\u7b97\u673a\u53ea\u6709\u4e00\u4e2a\u6838\u5fc3\uff0c\u5e76\u4e14\u6ca1\u6709\u591a\u5c11\u7a0b\u5e8f\u5458\u8981\u6c42\u591a\u7ebf\u7a0b\u5904\u7406\u652f\u6301\u3002 \u62e5\u6709 GIL \u53ef\u4ee5\u89e3\u51b3\u7a0b\u5e8f\u5b9e\u73b0\u4e2d\u7684\u5f88\u591a\u95ee\u9898\uff0c\u540c\u65f6\u8fd8\u4f7f\u4ee3\u7801\u6613\u4e8e\u7ef4\u62a4\u3002 \u56e0\u6b64\uff0cPython \u7684\u521b\u9020\u8005 Guido van Rossum \u4e8e 1992 \u5e74\u6dfb\u52a0\u4e86 GIL\u3002<\/p>\n

\u5feb\u8fdb\u5230 2025 \u5e74\uff1a\u4e2a\u4eba\u8ba1\u7b97\u673a\u62e5\u6709\u591a\u6838\u5904\u7406\u5668\uff0c\u56e0\u6b64\u8ba1\u7b97\u80fd\u529b\u66f4\u5f3a\u3002 \u6211\u4eec\u53ef\u4ee5\u5229\u7528\u989d\u5916\u7684\u7b97\u529b\u5b9e\u73b0\u771f\u6b63\u7684\u5e76\u53d1\uff0c\u800c\u65e0\u9700\u6446\u8131 GIL\u3002<\/p>\n

\u5728\u672c\u6587\u7684\u540e\u9762\uff0c\u6211\u4eec\u5c06\u5206\u89e3\u8bf4\u660e\u79fb\u9664\u5b83\u7684\u6d41\u7a0b\u3002 \u73b0\u5728\uff0c\u6211\u4eec\u5148\u770b\u770b\u5728 GIL \u5b58\u5728\u7684\u60c5\u51b5\u4e0b\u771f\u6b63\u7684\u5e76\u53d1\u662f\u5982\u4f55\u8fd0\u4f5c\u7684\u3002<\/p>\n

Python \u4e2d\u7684\u591a\u8fdb\u7a0b\u5904\u7406<\/h2>\n

\u5728\u6df1\u5165\u63a2\u7a76\u79fb\u9664 GIL \u7684\u6d41\u7a0b\u4e4b\u524d\uff0c\u6211\u4eec\u5148\u770b\u770b Python \u5f00\u53d1\u8005\u5982\u4f55\u4f7f\u7528 multiprocessing \u5e93\u5b9e\u73b0\u771f\u6b63\u7684\u5e76\u53d1\u3002 multiprocessing \u6807\u51c6\u5e93\u63d0\u4f9b\u672c\u5730\u548c\u8fdc\u7a0b\u5e76\u53d1\uff0c\u4f7f\u7528\u5b50\u8fdb\u7a0b\u800c\u4e0d\u662f\u7ebf\u7a0b\u6709\u6548\u907f\u5f00\u5168\u5c40\u89e3\u91ca\u5668\u9501\u3002 \u8fd9\u6837\u4e00\u6765\uff0cmultiprocessing \u6a21\u5757\u5141\u8bb8\u7a0b\u5e8f\u5458\u5145\u5206\u5229\u7528\u7279\u5b9a\u673a\u5668\u4e0a\u7684\u591a\u4e2a\u5904\u7406\u5668\u3002<\/p>\n

\u4e0d\u8fc7\uff0c\u8981\u6267\u884c\u591a\u8fdb\u7a0b\u5904\u7406\uff0c\u6211\u4eec\u5fc5\u987b\u4ee5\u7a0d\u5fae\u4e0d\u540c\u7684\u65b9\u5f0f\u8bbe\u8ba1\u7a0b\u5e8f\u3002 \u4e0b\u9762\u7684\u793a\u4f8b\u5c55\u793a\u4e86\u5982\u4f55\u5728 Python \u4e2d\u4f7f\u7528 multiprocessing \u5e93\u3002
\u8fd8\u8bb0\u5f97\u6211\u4eec\u535a\u5ba2\u7cfb\u5217\u7b2c 1 \u90e8\u5206<\/a>\u4e2d\u7684\u5f02\u6b65\u6c49\u5821\u5e97\u5417\uff1f<\/p>\n

import asyncio\n\nimport time\n\nasync def make_burger(order_num):\n\n    print(f\"Preparing burger #{order_num}...\")\n\n    await asyncio.sleep(5) # time for making the burger\n\n    print(f\"Burger made #{order_num}\")\n\nasync def main():\n\n    order_queue = []\n\n    for i in range(3):\n\n        order_queue.append(make_burger(i))\n\n    await asyncio.gather(*(order_queue))\n\nif __name__ == \"__main__\":\n\n    s = time.perf_counter()\n\n    asyncio.run(main())\n\n    elapsed = time.perf_counter() - s\n\n    print(f\"Orders completed in {elapsed:0.2f} seconds.\")\n<\/pre>\n
\u6211\u4eec\u53ef\u4ee5\u4f7f\u7528 multiprocessing \u5e93\u6765\u505a\u540c\u6837\u7684\u4e8b\uff0c\u4f8b\u5982\uff1a<\/p>\n
import multiprocessing\n\n\nimport time\n\n\ndef make_burger(order_num):\n\n\n   print(f\"Preparing burger #{order_num}...\")\n\n\n   time.sleep(5) # time for making the burger\n\n\n   print(f\"Burger made #{order_num}\")\n\n\n\n\nif __name__ == \"__main__\":\n\n\n   print(\"Number of available CPU:\", multiprocessing.cpu_count())\n\n\n\n\n   s = time.perf_counter()\n\n\n   all_processes = []\n\n\n   for i in range(3):\n       process = multiprocessing.Process(target=make_burger, args=(i,))\n       process.start()\n       all_processes.append(process)\n\n\n   for process in all_processes:\n       process.join()\n\n\n   elapsed = time.perf_counter() - s\n\n\n   print(f\"Orders completed in {elapsed:0.2f} seconds.\")\n<\/pre>\n
\u60a8\u53ef\u80fd\u8fd8\u8bb0\u5f97\uff0cmultiprocessing \u4e2d\u7684\u8bb8\u591a\u65b9\u6cd5\u4e0e threading \u975e\u5e38\u76f8\u4f3c\u3002 \u4e3a\u4e86\u4e86\u89e3 multiprocessing \u7684\u533a\u522b\uff0c\u6211\u4eec\u6765\u63a2\u7d22\u4e00\u4e2a\u66f4\u590d\u6742\u7684\u7528\u4f8b\uff1a<\/p>\n
import multiprocessing\nimport time\nimport queue\n\n\n\n\ndef make_burger(order_num, item_made):\n   name = multiprocessing.current_process().name\n   print(f\"{name} is preparing burger #{order_num}...\")\n   time.sleep(5)  # time for making burger\n   item_made.put(f\"Burger #{order_num}\")\n   print(f\"Burger #{order_num} made by {name}\")\n\n\n\n\ndef make_fries(order_num, item_made):\n   name = multiprocessing.current_process().name\n   print(f\"{name} is preparing fries #{order_num}...\")\n   time.sleep(2)  # time for making fries\n   item_made.put(f\"Fries #{order_num}\")\n   print(f\"Fries #{order_num} made by {name}\")\n\n\n\n\ndef working(task_queue, item_made, order_num, lock):\n   break_count = 0\n   name = multiprocessing.current_process().name\n   while True:\n       try:\n           task = task_queue.get_nowait()\n       except queue.Empty:\n           print(f\"{name} has nothing to do...\")\n           if break_count > 1:\n               break  # stop if idle for too long\n           else:\n               break_count += 1\n           time.sleep(1)\n       else:\n           lock.acquire()\n           try:\n               current_num = order_num.value\n               order_num.value = current_num + 1\n           finally:\n               lock.release()\n           task(current_num, item_made)\n           break_count = 0\n\n\n\n\nif __name__ == \"__main__\":\n\n\n   print(\"Welcome to Pyburger! Please place your order.\")\n\n\n   burger_num = input(\"Number of burgers:\")\n   fries_num = input(\"Number of fries:\")\n\n\n   s = time.perf_counter()\n\n\n   task_queue = multiprocessing.Queue()\n   item_made = multiprocessing.Queue()\n   order_num = multiprocessing.Value(\"i\", 0)\n   lock = multiprocessing.Lock()\n\n\n   for i in range(int(burger_num)):\n       task_queue.put(make_burger)\n   for i in range(int(fries_num)):\n       task_queue.put(make_fries)\n\n\n   staff1 = multiprocessing.Process(\n       target=working,\n       name=\"John\",\n       args=(\n           task_queue,\n           item_made,\n           order_num,\n           lock,\n       ),\n   )\n   staff2 = multiprocessing.Process(\n       target=working,\n       name=\"Jane\",\n       args=(\n           task_queue,\n           item_made,\n           order_num,\n           lock,\n       ),\n   )\n\n\n   staff1.start()\n   staff2.start()\n\n\n   staff1.join()\n   staff2.join()\n\n\n   print(\"All tasks finished. Closing now.\")\n   print(\"Items created are:\")\n\n\n   while not item_made.empty():\n       print(item_made.get())\n\n\n   elapsed = time.perf_counter() - s\n\n\n   print(f\"Orders completed in {elapsed:0.2f} seconds.\")\n<\/pre>\n
\u8fd9\u662f\u6211\u4eec\u5f97\u5230\u7684\u8f93\u51fa\uff1a<\/p>\n
Welcome to Pyburger! Please place your order.\nNumber of burgers:3\nNumber of fries:2\nJane has nothing to do...\nJohn is preparing burger #0...\nJane is preparing burger #1...\nBurger #0 made by John\nJohn is preparing burger #2...\nBurger #1 made by Jane\nJane is preparing fries #3...\nFries #3 made by Jane\nJane is preparing fries #4...\nBurger #2 made by John\nJohn has nothing to do...\nFries #4 made by Jane\nJane has nothing to do...\nJohn has nothing to do...\nJane has nothing to do...\nJohn has nothing to do...\nJane has nothing to do...\nAll tasks finished. Closing now.\nItems created are:\nBurger #0\nBurger #1\nFries #3\nBurger #2\nFries #4\nOrders completed in 12.21 seconds.\n<\/pre>\n
\u8bf7\u6ce8\u610f\uff0cmultiprocessing \u4e2d\u5b58\u5728\u4e00\u4e9b\u9650\u5236\uff0c\u5bfc\u81f4\u4e0a\u8ff0\u4ee3\u7801\u4ee5\u8fd9\u79cd\u65b9\u5f0f\u8bbe\u8ba1\u3002 \u63a5\u4e0b\u6765\u6211\u4eec\u9010\u4e00\u5206\u6790\u3002<\/p>\n
\u9996\u5148\uff0c\u8bb0\u4f4f\u6211\u4eec\u4e4b\u524d\u6709 make_burger \u548c make_fries \u51fd\u6570\u6765\u751f\u6210\u5177\u6709\u6b63\u786e order_num \u7684\u51fd\u6570\uff1a<\/p>\n
def make_burger(order_num):\n   def making_burger():\n       logger.info(f\"Preparing burger #{order_num}...\")\n       time.sleep(5)  # time for making burger\n       logger.info(f\"Burger made #{order_num}\")\n\n\n   return making_burger\n\n\n\n\ndef make_fries(order_num):\n   def making_fries():\n       logger.info(f\"Preparing fries #{order_num}...\")\n       time.sleep(2)  # time for making fries\n       logger.info(f\"Fries made #{order_num}\")\n\n\n   return making_fries\n<\/pre>\n
\u4f7f\u7528 multiprocessing \u65f6\u6211\u4eec\u4e0d\u80fd\u6267\u884c\u540c\u6837\u7684\u64cd\u4f5c\u3002 \u5982\u679c\u5c1d\u8bd5\uff0c\u4f1a\u5bfc\u81f4\u5982\u4e0b\u9519\u8bef\uff1a<\/p>\n
AttributeError: Can't get local object 'make_burger..making_burger'<\/pre>\n
\u539f\u56e0\u662f multiprocessing \u4f7f\u7528 pickle<\/a>\uff0c\u5b83\u901a\u5e38\u53ea\u80fd\u5e8f\u5217\u5316\u9876\u5c42\u6a21\u5757\u51fd\u6570\u3002 \u8fd9\u662f multiprocessing \u7684\u5c40\u9650\u4e4b\u4e00\u3002<\/p>\n
\u5176\u6b21\uff0c\u6ce8\u610f\u4e0a\u9762\u4f7f\u7528 multiprocessing \u7684\u793a\u4f8b\u4ee3\u7801\u6bb5\uff0c\u6211\u4eec\u6ca1\u6709\u5bf9\u5171\u4eab\u6570\u636e\u4f7f\u7528\u4efb\u4f55\u5168\u5c40\u53d8\u91cf\u3002 \u4f8b\u5982\uff0c\u6211\u4eec\u4e0d\u80fd\u5bf9 item_made \u548c order_num \u4f7f\u7528\u5168\u5c40\u53d8\u91cf\u3002 \u8981\u5728\u4e0d\u540c\u7684\u8fdb\u7a0b\u4e4b\u95f4\u5171\u4eab\u6570\u636e\uff0cmultiprocessing \u5e93\u4e2d\u7684\u7279\u6b8a\u7c7b\u5bf9\u8c61\uff08\u5982 Queue \u548c Value\uff09\u4f1a\u88ab\u4f7f\u7528\u5e76\u4f5c\u4e3a\u5b9e\u53c2\u4f20\u9012\u7ed9\u8fdb\u7a0b\u3002
\u4e00\u822c\u6765\u8bf4\uff0c\u4e0d\u5efa\u8bae\u5728\u4e0d\u540c\u8fdb\u7a0b\u4e4b\u95f4\u5171\u4eab\u6570\u636e\u548c\u72b6\u6001\uff0c\u56e0\u4e3a\u8fd9\u4f1a\u5bfc\u81f4\u66f4\u591a\u95ee\u9898\u3002 \u5728\u4e0a\u9762\u7684\u4f8b\u5b50\u4e2d\uff0c\u6211\u4eec\u5fc5\u987b\u4f7f\u7528 Lock \u786e\u4fdd order_num \u7684\u503c\u6bcf\u6b21\u53ea\u80fd\u88ab\u4e00\u4e2a\u8fdb\u7a0b\u8bbf\u95ee\u548c\u9012\u589e\u3002 \u5982\u679c\u6ca1\u6709 Lock\uff0c\u5546\u54c1\u7684\u8ba2\u5355\u53f7\u53ef\u80fd\u4f1a\u53d8\u6210\u8fd9\u6837\uff1a<\/p>\n
Items created are:\n\nBurger #0\nBurger #0\nFries #2\nBurger #1\nFries #3\n<\/pre>\n
\u4ee5\u4e0b\u662f\u4f7f\u7528\u9501\u6765\u907f\u514d\u9ebb\u70e6\u7684\u65b9\u5f0f\uff1a<\/p>\n
           lock.acquire()\n           try:\n               current_num = order_num.value\n               order_num.value = current_num + 1\n           finally:\n               lock.release()\n           task(current_num, item_made)\n<\/pre>\n
\u8981\u8be6\u7ec6\u4e86\u89e3\u5982\u4f55\u4f7f\u7528 multiprocessing \u6807\u51c6\u5e93\uff0c\u60a8\u53ef\u4ee5\u5728\u8fd9\u91cc<\/a>\u9605\u8bfb\u6587\u6863\u3002<\/p>\n
\u79fb\u9664 GIL<\/h2>\n
\u8fd1\u5341\u5e74\u6765\uff0c\u79fb\u9664 GIL \u4e00\u76f4\u662f\u4e00\u4e2a\u8bdd\u9898\u3002 2016 \u5e74\uff0c\u5728 Python Language Summit \u4e0a\uff0cLarry Hastings \u8bb2\u89e3\u4e86\u4ed6\u5bf9 CPython \u89e3\u91ca\u5668\u8fdb\u884c\u201cGIL \u5207\u9664\u201d\u7684\u60f3\u6cd5\u4ee5\u53ca\u4ed6\u5728\u8fd9\u4e00\u60f3\u6cd5\u4e0a\u53d6\u5f97\u7684\u8fdb\u5c55[1]\u3002 \u8fd9\u662f\u79fb\u9664 Python GIL \u7684\u4e00\u6b21\u5f00\u521b\u6027\u5c1d\u8bd5\u3002 2021 \u5e74\uff0cSam Gross \u91cd\u65b0\u5f15\u53d1\u4e86\u5173\u4e8e\u79fb\u9664 GIL \u7684\u8ba8\u8bba[2]\uff0c\u5e76\u50ac\u751f\u4e86 2023 \u5e74\u53d1\u5e03\u7684 PEP 703 \u2013 Making the Global Interpreter Lock Optional in CPython<\/a>\u3002<\/p>\n
\u53ef\u4ee5\u770b\u5230\uff0c\u79fb\u9664 GIL \u7edd\u4e0d\u662f\u4e00\u4e2a\u4ed3\u4fc3\u7684\u51b3\u5b9a\uff0c\u5e76\u4e14\u5df2\u7ecf\u5728\u793e\u533a\u5185\u5f15\u8d77\u5f88\u591a\u4e89\u8bba\u3002 \u5982\u4e0a\u9762\u7684\u591a\u8fdb\u7a0b\u5904\u7406\u793a\u4f8b\uff08\u4ee5\u53ca\u4e0a\u9762\u94fe\u63a5\u7684 PEP 703\uff09\u6240\u793a\uff0c\u5f53 GIL \u63d0\u4f9b\u7684\u4fdd\u8bc1\u88ab\u79fb\u9664\u65f6\uff0c\u60c5\u51b5\u5f88\u5feb\u5c31\u4f1a\u53d8\u5f97\u590d\u6742\u8d77\u6765\u3002<\/p>\n
[1]: https:\/\/lwn.net\/Articles\/689548\/<\/a><\/p>\n
[2]: https:\/\/lwn.net\/ml\/python-dev\/CAGr09bSrMNyVNLTvFq-h6t38kTxqTXfgxJYApmbEWnT71L74-g@mail.gmail.com\/<\/p>\n
\u5f15\u7528\u8ba1\u6570<\/h3>\n
\u5b58\u5728 GIL \u65f6\uff0c\u5f15\u7528\u8ba1\u6570\u548c\u5783\u573e\u56de\u6536\u66f4\u52a0\u76f4\u89c2\u3002 \u5f53\u4e00\u6b21\u53ea\u6709\u4e00\u4e2a\u7ebf\u7a0b\u53ef\u4ee5\u8bbf\u95ee Python \u5bf9\u8c61\u65f6\uff0c\u6211\u4eec\u53ef\u4ee5\u4f9d\u9760\u7b80\u5355\u7684\u975e\u539f\u5b50\u5f15\u7528\u8ba1\u6570<\/a>\uff0c\u5e76\u5728\u5f15\u7528\u8ba1\u6570\u8fbe\u5230\u96f6\u65f6\u79fb\u9664\u5bf9\u8c61\u3002<\/p>\n
\u79fb\u9664 GIL \u8ba9\u60c5\u51b5\u6709\u4e9b\u68d8\u624b\u3002 \u6211\u4eec\u4e0d\u80fd\u518d\u4f7f\u7528\u975e\u539f\u5b50\u5f15\u7528\u8ba1\u6570\uff0c\u56e0\u4e3a\u8fd9\u4e0d\u80fd\u4fdd\u8bc1\u7ebf\u7a0b\u5b89\u5168<\/a>\u3002 \u5982\u679c\u591a\u4e2a\u7ebf\u7a0b\u540c\u65f6\u5bf9 Python \u5bf9\u8c61\u6267\u884c\u5f15\u7528\u7684\u591a\u6b21\u9012\u589e\u548c\u9012\u51cf\uff0c\u5219\u53ef\u80fd\u51fa\u73b0\u95ee\u9898\u3002 \u7406\u60f3\u60c5\u51b5\u4e0b\uff0c\u539f\u5b50\u5f15\u7528\u8ba1\u6570\u5c06\u7528\u4e8e\u4fdd\u8bc1\u7ebf\u7a0b\u5b89\u5168\u3002 \u4f46\u8fd9\u79cd\u65b9\u6cd5\u5f00\u9500\u8f83\u5927\uff0c\u5f53\u7ebf\u7a0b\u8f83\u591a\u65f6\u6548\u7387\u4f1a\u53d7\u5230\u5f71\u54cd\u3002<\/p>\n
\u89e3\u51b3\u65b9\u6848\u662f\u4f7f\u7528\u504f\u5411\u5f15\u7528\u8ba1\u6570\uff0c\u8fd9\u4e5f\u4fdd\u8bc1\u4e86\u7ebf\u7a0b\u5b89\u5168\u3002 \u8fd9\u91cc\u7684\u60f3\u6cd5\u662f\u5c06\u6bcf\u4e2a\u5bf9\u8c61\u504f\u5411\u4e8e\u6240\u6709\u8005\u7ebf\u7a0b\uff0c\u5373\u5927\u591a\u6570\u65f6\u95f4\u8bbf\u95ee\u8be5\u5bf9\u8c61\u7684\u7ebf\u7a0b\u3002 \u6240\u6709\u8005\u7ebf\u7a0b\u53ef\u4ee5\u5bf9\u5176\u62e5\u6709\u7684\u5bf9\u8c61\u6267\u884c\u975e\u539f\u5b50\u5f15\u7528\u8ba1\u6570\uff0c\u5176\u4ed6\u7ebf\u7a0b\u5219\u9700\u8981\u5bf9\u8fd9\u4e9b\u5bf9\u8c61\u6267\u884c\u539f\u5b50\u5f15\u7528\u8ba1\u6570\u3002 \u8fd9\u4e2a\u65b9\u6cd5\u6bd4\u5355\u7eaf\u7684\u539f\u5b50\u5f15\u7528\u8ba1\u6570\u66f4\u53ef\u53d6\uff0c\u56e0\u4e3a\u5927\u591a\u6570\u5bf9\u8c61\u5927\u591a\u6570\u65f6\u95f4\u4ec5\u7531\u4e00\u4e2a\u7ebf\u7a0b\u8bbf\u95ee\u3002 \u6211\u4eec\u53ef\u4ee5\u901a\u8fc7\u5141\u8bb8\u6240\u6709\u8005\u7ebf\u7a0b\u6267\u884c\u975e\u539f\u5b50\u5f15\u7528\u8ba1\u6570\u6765\u51cf\u5c11\u6267\u884c\u5f00\u9500\u3002<\/p>\n
\"\"<\/figure>\n
\u6b64\u5916\uff0c\u4e00\u4e9b\u5e38\u7528\u7684 Python \u5bf9\u8c61\uff0c\u4f8b\u5982 True\u3001False\u3001<\/em>\u5c0f\u6574\u6570\u548c\u4e00\u4e9b\u6682\u5b58\u5b57\u7b26\u4e32\uff0c\u90fd\u662f\u6301\u4e45\u7684\u3002 \u8fd9\u91cc\u7684\u201c\u6301\u4e45\u201d\u8868\u793a\u5bf9\u8c61\u5c06\u5728\u7a0b\u5e8f\u7684\u6574\u4e2a\u751f\u5b58\u671f\u5185\u4fdd\u7559\uff0c\u56e0\u6b64\u4e0d\u9700\u8981\u5f15\u7528\u8ba1\u6570\u3002<\/p>\n
\u5783\u573e\u56de\u6536<\/h3>\n
\u6211\u4eec\u8fd8\u5fc5\u987b\u4fee\u6539\u5783\u573e\u56de\u6536<\/a>\u5b8c\u6210\u7684\u65b9\u5f0f\u3002 \u5f15\u7528\u88ab\u91ca\u653e\u65f6\uff0c\u4e0d\u662f\u7acb\u5373\u51cf\u5c11\u5f15\u7528\u8ba1\u6570\uff0c\u5f15\u7528\u8ba1\u6570\u8fbe\u5230\u96f6\u65f6\uff0c\u4e5f\u4e0d\u662f\u7acb\u5373\u79fb\u9664\u5bf9\u8c61\uff0c\u800c\u662f\u4f7f\u7528\u4e00\u79cd\u540d\u4e3a\u201c\u5ef6\u8fdf\u5f15\u7528\u8ba1\u6570\u201d\u7684\u6280\u672f\u3002\u00a0<\/p>\n
\u9700\u8981\u51cf\u5c11\u5f15\u7528\u8ba1\u6570\u65f6\uff0c\u5bf9\u8c61\u88ab\u5b58\u50a8\u5728\u4e00\u4e2a\u8868\u4e2d\uff0c\u8868\u540e\u7eed\u5c06\u63a5\u53d7\u53cc\u91cd\u68c0\u67e5\u4ee5\u786e\u5b9a\u5f15\u7528\u8ba1\u6570\u7684\u9012\u51cf\u662f\u5426\u51c6\u786e\u3002 \u8fd9\u907f\u514d\u4e86\u5728\u5176\u88ab\u5f15\u7528\u65f6\u8fc7\u65e9\u79fb\u9664\u5bf9\u8c61\uff0c\u8fd9\u79cd\u60c5\u51b5\u5728\u6ca1\u6709 GIL \u7684\u60c5\u51b5\u4e0b\u53ef\u80fd\u4f1a\u53d1\u751f\uff0c\u56e0\u4e3a\u5f15\u7528\u8ba1\u6570\u5e76\u4e0d\u50cf GIL \u90a3\u6837\u76f4\u89c2\u3002 \u8fd9\u4f7f\u5783\u573e\u56de\u6536\u8fc7\u7a0b\u66f4\u4e3a\u590d\u6742\uff0c\u56e0\u4e3a\u5783\u573e\u56de\u6536\u53ef\u80fd\u9700\u8981\u904d\u5386\u6bcf\u4e2a\u7ebf\u7a0b\u7684\u5806\u6808<\/a>\u4ee5\u83b7\u53d6\u6bcf\u4e2a\u7ebf\u7a0b\u81ea\u5df1\u7684\u5f15\u7528\u8ba1\u6570\u3002<\/p>\n
\u53e6\u4e00\u4ef6\u9700\u8981\u8003\u8651\u7684\u4e8b\uff1a\u5783\u573e\u56de\u6536\u671f\u95f4\u5f15\u7528\u8ba1\u6570\u9700\u8981\u4fdd\u6301\u7a33\u5b9a\u3002 \u5982\u679c\u4e00\u4e2a\u5bf9\u8c61\u5373\u5c06\u88ab\u4e22\u5f03\u4f46\u7a81\u7136\u88ab\u5f15\u7528\uff0c\u8fd9\u5c06\u5bfc\u81f4\u4e25\u91cd\u7684\u95ee\u9898\u3002 \u56e0\u6b64\uff0c\u5728\u5783\u573e\u56de\u6536\u5468\u671f\u4e2d\uff0c\u5b83\u5fc5\u987b\u201c\u505c\u6b62\u6574\u4e2a\u4e16\u754c\u201d\u6765\u63d0\u4f9b\u7ebf\u7a0b\u5b89\u5168\u4fdd\u8bc1\u3002<\/p>\n
\u5185\u5b58\u5206\u914d<\/h3>\n
\u6709 GIL \u4fdd\u8bc1\u7ebf\u7a0b\u5b89\u5168\u65f6\uff0c\u5c06\u4f7f\u7528 Python \u5185\u90e8\u5185\u5b58\u5206\u914d\u5668 pymalloc\u3002 \u4f46\u662f\u6ca1\u6709 GIL\uff0c\u6211\u4eec\u5c06\u9700\u8981\u4e00\u4e2a\u65b0\u7684\u5185\u5b58\u5206\u914d\u5668\u3002 Sam Gross \u5728 PEP \u4e2d\u63d0\u51fa\u4e86 mimalloc<\/a>\uff0c\u8fd9\u662f\u7531 Daan Leijen \u521b\u5efa\u5e76\u7531 Microsoft \u7ef4\u62a4\u7684\u901a\u7528\u5206\u914d\u5668\u3002 \u8fd9\u662f\u4e00\u4e2a\u5f88\u597d\u7684\u9009\u62e9\uff0c\u56e0\u4e3a\u5b83\u7ebf\u7a0b\u5b89\u5168\uff0c\u5e76\u4e14\u5728\u5c0f\u5bf9\u8c61\u4e0a\u6027\u80fd\u826f\u597d\u3002<\/p>\n
Mimalloc \u4f7f\u7528\u9875\u9762\u586b\u5145\u5176\u5806\uff0c\u4f7f\u7528\u5757\u586b\u5145\u9875\u9762\u3002 \u6bcf\u4e2a\u9875\u9762\u90fd\u5305\u542b\u5757\uff0c\u5e76\u4e14\u6bcf\u4e2a\u9875\u9762\u5185\u7684\u5757\u5927\u5c0f\u5747\u76f8\u540c\u3002 \u5bf9\u5217\u8868\u548c\u5b57\u5178\u8bbf\u95ee\u6dfb\u52a0\u4e00\u4e9b\u9650\u5236\u540e\uff0c\u5783\u573e\u56de\u6536\u5668\u4e0d\u5fc5\u7ef4\u62a4\u5173\u8054\u5217\u8868<\/a>\u5373\u53ef\u67e5\u627e\u6240\u6709\u5bf9\u8c61\uff0c\u5e76\u4e14\u8fd8\u53ef\u4ee5\u5728\u4e0d\u83b7\u53d6\u9501\u7684\u60c5\u51b5\u4e0b\u5bf9\u5217\u8868\u548c\u5b57\u5178\u8fdb\u884c\u8bfb\u53d6\u8bbf\u95ee\u3002<\/p>\n
\"\"<\/figure>\n
\u5173\u4e8e\u79fb\u9664 GIL \u8fd8\u6709\u66f4\u591a\u7ec6\u8282\uff0c\u4f46\u8fd9\u91cc\u4e0d\u53ef\u80fd\u5168\u90e8\u6db5\u76d6\u3002 \u60a8\u53ef\u4ee5\u67e5\u770b PEP 703 \u2013 Making the Global Interpreter Lock Optional in CPython<\/a> \u83b7\u5f97\u5b8c\u6574\u5206\u6790\u3002<\/p>\n
\u6709\u65e0 GIL \u7684\u6027\u80fd\u5dee\u5f02<\/h2>\n
\u7531\u4e8e Python 3.13 \u63d0\u4f9b\u4e86\u81ea\u7531\u7ebf\u7a0b\u9009\u9879\uff0c\u6211\u4eec\u53ef\u4ee5\u5c06 Python 3.13 \u6807\u51c6\u7248\u672c\u7684\u6027\u80fd\u4e0e\u81ea\u7531\u7ebf\u7a0b\u7248\u672c\u8fdb\u884c\u6bd4\u8f83\u3002<\/p>\n
\u5b89\u88c5\u81ea\u7531\u7ebf\u7a0b Python<\/h3>\n
\u6211\u4eec\u5c06\u4f7f\u7528 pyenv<\/a> \u5b89\u88c5\u4e24\u4e2a\u7248\u672c\uff1a\u6807\u51c6\u7248\uff08\u4f8b\u5982 3.13.5\uff09\u548c\u81ea\u7531\u7ebf\u7a0b\u7248\uff08\u4f8b\u5982 3.13.5t\uff09\u3002\u00a0<\/p>\n
\u6216\u8005\uff0c\u60a8\u4e5f\u53ef\u4ee5\u4f7f\u7528 Python.org<\/a> \u4e0a\u7684\u5b89\u88c5\u7a0b\u5e8f\u3002 \u786e\u4fdd\u5728\u5b89\u88c5\u8fc7\u7a0b\u4e2d\u9009\u62e9 Customize<\/em>\uff08\u81ea\u5b9a\u4e49\uff09\u9009\u9879\uff0c\u5e76\u9009\u4e2d\u53ef\u9009\u6846\u4ee5\u5b89\u88c5\u81ea\u7531\u7ebf\u7a0b Python\uff08\u53c2\u89c1\u8fd9\u7bc7\u535a\u6587\u4e2d\u7684\u793a\u4f8b<\/a>\uff09\u3002<\/p>\n
\u5b89\u88c5\u4e24\u4e2a\u7248\u672c\u540e\uff0c\u6211\u4eec\u53ef\u4ee5\u5c06\u5b83\u4eec\u6dfb\u52a0\u4e3a PyCharm \u9879\u76ee\u4e2d\u7684\u89e3\u91ca\u5668\u3002<\/p>\n
\u9996\u5148\uff0c\u70b9\u51fb\u53f3\u4e0b\u89d2\u7684 Python \u89e3\u91ca\u5668\u7684\u540d\u79f0\u3002<\/p>\n
\"\"<\/figure>\n
\u5728\u83dc\u5355\u4e2d\u9009\u62e9 Add New Interpreter<\/em>\uff08\u6dfb\u52a0\u65b0\u89e3\u91ca\u5668\uff09\uff0c\u7136\u540e\u9009\u62e9 Add Local Interpreter<\/em>\uff08\u6dfb\u52a0\u672c\u5730\u89e3\u91ca\u5668\uff09\u3002<\/p>\n
\"\"<\/figure>\n
\u9009\u62e9 Select existing<\/em>\uff08\u9009\u62e9\u73b0\u6709\uff09\uff0c\u7b49\u5f85\u89e3\u91ca\u5668\u8def\u5f84\u52a0\u8f7d\uff08\u5982\u679c\u60a8\u50cf\u6211\u4e00\u6837\u6709\u5f88\u591a\u89e3\u91ca\u5668\uff0c\u8fd9\u53ef\u80fd\u9700\u8981\u4e00\u6bb5\u65f6\u95f4\uff09\uff0c\u7136\u540e\u4ece\u4e0b\u62c9\u83dc\u5355 Python path<\/em>\uff08Python \u8def\u5f84\uff09\u4e2d\u9009\u62e9\u521a\u521a\u5b89\u88c5\u7684\u65b0\u89e3\u91ca\u5668\u3002<\/p>\n
\"\"<\/figure>\n
\u70b9\u51fb OK<\/em>\uff08\u786e\u5b9a\uff09\u6765\u6dfb\u52a0\u3002 \u5bf9\u53e6\u4e00\u4e2a\u89e3\u91ca\u5668\u91cd\u590d\u76f8\u540c\u7684\u6b65\u9aa4\u3002 \u73b0\u5728\uff0c\u518d\u6b21\u70b9\u51fb\u53f3\u4e0b\u89d2\u7684\u89e3\u91ca\u5668\u540d\u79f0\u65f6\uff0c\u60a8\u5c06\u770b\u5230\u591a\u4e2a Python 3.13 \u89e3\u91ca\u5668\uff0c\u5c31\u50cf\u4e0a\u56fe\u4e00\u6837\u3002<\/p>\n
\u4f7f\u7528\u53d7 CPU \u9650\u5236\u7684\u8fdb\u7a0b\u8fdb\u884c\u6d4b\u8bd5<\/h3>\n
\u63a5\u4e0b\u6765\uff0c\u6211\u4eec\u9700\u8981\u4e00\u4e2a\u811a\u672c\u6765\u6d4b\u8bd5\u4e0d\u540c\u7684\u7248\u672c\u3002 \u6211\u4eec\u5728\u672c\u535a\u6587\u7cfb\u5217\u7684\u7b2c 1 \u90e8\u5206<\/a>\u4e2d\u89e3\u91ca\u8fc7\uff0c\u4e3a\u4e86\u52a0\u5feb\u53d7 CPU \u9650\u5236\u7684\u8fdb\u7a0b\u7684\u901f\u5ea6\uff0c\u6211\u4eec\u9700\u8981\u771f\u6b63\u7684\u591a\u7ebf\u7a0b\u5904\u7406\u3002 \u4e3a\u4e86\u67e5\u770b\u79fb\u9664 GIL \u662f\u5426\u80fd\u591f\u5b9e\u73b0\u771f\u6b63\u7684\u591a\u7ebf\u7a0b\u5904\u7406\u5e76\u4f7f Python \u66f4\u5feb\uff0c\u6211\u4eec\u53ef\u4ee5\u5728\u591a\u4e2a\u7ebf\u7a0b\u4e0a\u4f7f\u7528\u53d7 CPU \u9650\u5236\u7684\u8fdb\u7a0b\u8fdb\u884c\u6d4b\u8bd5\u3002 \u8fd9\u662f\u6211\u8ba9 Junie<\/a> \u751f\u6210\u7684\u811a\u672c\uff08\u6211\u505a\u4e86\u4e00\u4e9b\u6700\u7ec8\u8c03\u6574\uff09\uff1a<\/p>\n
import time\nimport multiprocessing  # Kept for CPU count\nfrom concurrent.futures import ThreadPoolExecutor\nimport sys\n\n\n\n\ndef is_prime(n):\n   \"\"\"Check if a number is prime (CPU-intensive operation).\"\"\"\n   if n <= 1:\n       return False\n   if n <= 3:\n       return True\n   if n % 2 == 0 or n % 3 == 0:\n       return False\n   i = 5\n   while i * i <= n:\n       if n % i == 0 or n % (i + 2) == 0:\n           return False\n       i += 6\n   return True\n\n\n\n\ndef count_primes(start, end):\n   \"\"\"Count prime numbers in a range.\"\"\"\n   count = 0\n   for num in range(start, end):\n       if is_prime(num):\n           count += 1\n   return count\n\n\n\n\ndef run_single_thread(range_size, num_chunks):\n   \"\"\"Run the prime counting task in a single thread.\"\"\"\n   chunk_size = range_size \/\/ num_chunks\n   total_count = 0\n\n\n   start_time = time.time()\n\n\n   for i in range(num_chunks):\n       start = i * chunk_size + 1\n       end = (i + 1) * chunk_size + 1 if i < num_chunks - 1 else range_size + 1\n       total_count += count_primes(start, end)\n\n\n   end_time = time.time()\n\n\n   return total_count, end_time - start_time\n\n\n\n\ndef thread_task(start, end):\n   \"\"\"Task function for threads.\"\"\"\n   return count_primes(start, end)\n\n\n\n\ndef run_multi_thread(range_size, num_threads, num_chunks):\n   \"\"\"Run the prime counting task using multiple threads.\"\"\"\n   chunk_size = range_size \/\/ num_chunks\n   total_count = 0\n\n\n   start_time = time.time()\n\n\n   with ThreadPoolExecutor(max_workers=num_threads) as executor:\n       futures = []\n       for i in range(num_chunks):\n           start = i * chunk_size + 1\n           end = (i + 1) * chunk_size + 1 if i < num_chunks - 1 else range_size + 1\n           futures.append(executor.submit(thread_task, start, end))\n\n\n       for future in futures:\n           total_count += future.result()\n\n\n   end_time = time.time()\n\n\n   return total_count, end_time - start_time\n\n\n\n\ndef main():\n   # Fixed parameters\n   range_size = 1000000  # Range of numbers to check for primes\n   num_chunks = 16       # Number of chunks to divide the work into\n   num_threads = 4       # Fixed number of threads for multi-threading test\n\n\n   print(f\"Python version: {sys.version}\")\n   print(f\"CPU count: {multiprocessing.cpu_count()}\")\n   print(f\"Range size: {range_size}\")\n   print(f\"Number of chunks: {num_chunks}\")\n   print(\"-\" * 60)\n\n\n   # Run single-threaded version as baseline\n   print(\"Running single-threaded version (baseline)...\")\n   count, single_time = run_single_thread(range_size, num_chunks)\n   print(f\"Found {count} primes in {single_time:.4f} seconds\")\n   print(\"-\" * 60)\n\n\n   # Run multi-threaded version with fixed number of threads\n   print(f\"Running multi-threaded version with {num_threads} threads...\")\n   count, thread_time = run_multi_thread(range_size, num_threads, num_chunks)\n   speedup = single_time \/ thread_time\n   print(f\"Found {count} primes in {thread_time:.4f} seconds (speedup: {speedup:.2f}x)\")\n   print(\"-\" * 60)\n\n\n   # Summary\n   print(\"SUMMARY:\")\n   print(f\"{'Threads':<10} {'Time (s)':<12} {'Speedup':<10}\")\n   print(f\"{'1 (baseline)':<10} {single_time:<12.4f} {'1.00x':<10}\")\n   print(f\"{num_threads:<10} {thread_time:<12.4f} {speedup:.2f}x\")\n\n\nif __name__ == \"__main__\":\n   main()\n<\/pre>\n
\u4e3a\u4e86\u66f4\u8f7b\u677e\u5730\u4f7f\u7528\u4e0d\u540c\u7684 Python \u89e3\u91ca\u5668\u8fd0\u884c\u811a\u672c\uff0c\u6211\u4eec\u53ef\u4ee5\u5411 PyCharm \u9879\u76ee\u6dfb\u52a0\u81ea\u5b9a\u4e49\u8fd0\u884c\u811a\u672c\u3002<\/p>\n
\u5728\u9876\u90e8\uff0c\u4ece Run<\/em>\uff08\u8fd0\u884c\uff09\u6309\u94ae () \u65c1\u8fb9\u7684\u4e0b\u62c9\u83dc\u5355\u4e2d\u9009\u62e9 Edit Configurations\u2026<\/em>\uff08\u7f16\u8f91\u914d\u7f6e\u2026\uff09\u3002<\/p>\n
<\/p>\n
\u70b9\u51fb\u5de6\u4e0a\u89d2\u7684 + \u6309\u94ae\uff0c\u7136\u540e\u4ece Add New Configuration<\/em>\uff08\u6dfb\u52a0\u65b0\u914d\u7f6e\uff09\u4e0b\u62c9\u83dc\u5355\u4e2d\u9009\u62e9 Python\u3002<\/p>\n
\"\"<\/figure>\n
\u9009\u62e9\u4e00\u4e2a\u540d\u79f0\uff0c\u4ee5\u4fbf\u81ea\u5df1\u77e5\u9053\u6b63\u5728\u4f7f\u7528\u5177\u4f53\u54ea\u4e2a\u89e3\u91ca\u5668\uff0c\u4f8b\u5982 3.13.5 \u4e0e 3.15.3t\u3002 \u9009\u62e9\u6b63\u786e\u7684\u89e3\u91ca\u5668\u5e76\u6dfb\u52a0\u6d4b\u8bd5\u811a\u672c\u7684\u540d\u79f0\uff0c\u5982\u4e0b\u6240\u793a\uff1a<\/p>\n
\"\"<\/figure>\n
\u6dfb\u52a0\u4e24\u4e2a\u914d\u7f6e\uff0c\u6bcf\u4e2a\u89e3\u91ca\u5668\u4e00\u4e2a\u3002 \u7136\u540e\uff0c\u70b9\u51fb OK<\/em>\uff08\u786e\u5b9a\uff09\u3002<\/p>\n
\u73b0\u5728\uff0c\u6211\u4eec\u53ef\u4ee5\u9009\u62e9\u914d\u7f6e\u5e76\u70b9\u51fb\u9876\u90e8\u7684 Run<\/em>\uff08\u8fd0\u884c\uff09\u6309\u94ae () \u8f7b\u677e\u9009\u62e9\u5e76\u8fd0\u884c\u5e26\u6709\u6216\u4e0d\u5e26\u6709 GIL \u7684\u6d4b\u8bd5\u811a\u672c\u3002<\/p>\n
\u6bd4\u8f83\u7ed3\u679c<\/h3>\n
\u8fd9\u662f\u6211\u8fd0\u884c\u5e26\u6709 GIL \u7684\u6807\u51c6\u7248\u672c 3.13.5 \u65f6\u5f97\u5230\u7684\u7ed3\u679c\uff1a<\/p>\n
Python version: 3.13.5 (main, Jul 10 2025, 20:33:15) [Clang 17.0.0 (clang-1700.0.13.5)]\nCPU count: 8\nRange size: 1000000\nNumber of chunks: 16\n------------------------------------------------------------\nRunning single-threaded version (baseline)...\nFound 78498 primes in 1.1930 seconds\n------------------------------------------------------------\nRunning multi-threaded version with 4 threads...\nFound 78498 primes in 1.2183 seconds (speedup: 0.98x)\n------------------------------------------------------------\nSUMMARY:\nThreads    Time (s)     Speedup   \n1 (baseline) 1.1930       1.00x     \n4          1.2183       0.98x\n<\/pre>\n
\u53ef\u4ee5\u770b\u5230\uff0c\u4e0e\u5355\u7ebf\u7a0b\u57fa\u7ebf\u76f8\u6bd4\uff0c\u8fd0\u884c 4 \u4e2a\u7ebf\u7a0b\u7684\u7248\u672c\u65f6\u901f\u5ea6\u6ca1\u6709\u663e\u8457\u53d8\u5316\u3002 \u6211\u4eec\u770b\u770b\u8fd0\u884c\u81ea\u7531\u7ebf\u7a0b\u7248\u672c 3.13.5t \u65f6\u4f1a\u5f97\u5230\u4ec0\u4e48\uff1a<\/p>\n
Python version: 3.13.5 experimental free-threading build (main, Jul 10 2025, 20:36:28) [Clang 17.0.0 (clang-1700.0.13.5)]\nCPU count: 8\nRange size: 1000000\nNumber of chunks: 16\n------------------------------------------------------------\nRunning single-threaded version (baseline)...\nFound 78498 primes in 1.5869 seconds\n------------------------------------------------------------\nRunning multi-threaded version with 4 threads...\nFound 78498 primes in 0.4662 seconds (speedup: 3.40x)\n------------------------------------------------------------\nSUMMARY:\nThreads    Time (s)     Speedup   \n1 (baseline) 1.5869       1.00x     \n4          0.4662       3.40x\n<\/pre>\n
\u8fd9\u6b21\u7684\u901f\u5ea6\u662f\u4e4b\u524d\u7684 3 \u500d\u591a\u3002 \u8bf7\u6ce8\u610f\uff0c\u4e24\u79cd\u60c5\u51b5\u4e0b\u90fd\u4f1a\u6709\u591a\u7ebf\u7a0b\u5904\u7406\u7684\u5f00\u9500\u3002 \u56e0\u6b64\uff0c\u5373\u4f7f\u662f\u771f\u6b63\u7684\u591a\u7ebf\u7a0b\u5904\u7406\uff0c\u8fd0\u884c 4 \u4e2a\u7ebf\u7a0b\u65f6\u7684\u901f\u5ea6\u4e5f\u4e0d\u4f1a\u662f 4 \u500d\u3002<\/p>\n
\u7ed3\u8bba<\/h2>\n
\u5728\u300a\u66f4\u5feb\u7684 Python\u300b\u535a\u6587\u7cfb\u5217\u7684\u7b2c 2 \u90e8\u5206\u4e2d\uff0c\u6211\u4eec\u8ba8\u8bba\u4e86\u8fc7\u53bb\u4f7f\u7528 Python GIL \u7684\u539f\u56e0\u3001\u4f7f\u7528 multiprocessing \u89c4\u907f GIL \u7684\u9650\u5236\uff0c\u4ee5\u53ca\u79fb\u9664 GIL \u7684\u6d41\u7a0b\u548c\u6548\u679c\u3002<\/p>\n
\u622a\u81f3\u8fd9\u7bc7\u535a\u6587\u53d1\u5e03\uff0cPython \u7684\u81ea\u7531\u7ebf\u7a0b\u7248\u672c\u4ecd\u7136\u4e0d\u662f\u9ed8\u8ba4\u7248\u672c\u3002 \u4e0d\u8fc7\uff0c\u968f\u7740\u793e\u533a\u548c\u7b2c\u4e09\u65b9\u5e93\u7684\u91c7\u7528\uff0c\u793e\u533a\u9884\u8ba1 Python \u7684\u81ea\u7531\u7ebf\u7a0b\u7248\u672c\u5c06\u5728\u672a\u6765\u6210\u4e3a\u6807\u51c6\u3002 \u6839\u636e\u516c\u5e03\u7684\u6d88\u606f\uff0cPython 3.14 \u5c06\u5305\u542b\u4e00\u4e2a\u81ea\u7531\u7ebf\u7a0b\u7248\u672c\uff0c\u8be5\u7248\u672c\u5c06\u5ea6\u8fc7\u5b9e\u9a8c\u9636\u6bb5\uff0c\u4f46\u4ecd\u4e3a\u53ef\u9009\u3002<\/p>\n
PyCharm<\/a> \u63d0\u4f9b\u4e86\u4e00\u6d41\u7684 Python \u652f\u6301\uff0c\u53ef\u4ee5\u786e\u4fdd\u901f\u5ea6\u548c\u51c6\u786e\u6027\u3002 \u53d7\u76ca\u4e8e\u6700\u667a\u80fd\u7684\u4ee3\u7801\u8865\u5168\u3001PEP 8 \u5408\u89c4\u6027\u68c0\u67e5\u3001\u667a\u80fd\u91cd\u6784\u548c\u5404\u79cd\u68c0\u67e5\uff0c\u6ee1\u8db3\u6240\u6709\u7f16\u7801\u9700\u6c42\u3002 \u5982\u672c\u6587\u6240\u793a\uff0cPyCharm \u4e3a Python \u89e3\u91ca\u5668\u548c\u8fd0\u884c\u914d\u7f6e\u63d0\u4f9b\u4e86\u81ea\u5b9a\u4e49\u8bbe\u7f6e\uff0c\u53ea\u9700\u70b9\u51fb\u51e0\u4e0b\u5373\u53ef\u5728\u89e3\u91ca\u5668\u4e4b\u95f4\u5207\u6362\uff0c\u8fd9\u4f7f\u5176\u9002\u7528\u4e8e\u5404\u79cd Python \u9879\u76ee\u3002<\/p>\n
\n
\u7acb\u5373\u4e0b\u8f7d PyCharm<\/a><\/div>\n<\/div>\n\n\n
<\/p>\n\n\n\n
\u672c\u535a\u6587\u82f1\u6587\u539f\u4f5c\u8005\uff1a<\/p>\n\n\n    
\n        
\n            
\n                
\n                    \"Cheuk\n                <\/div>\n                
\n                                            
Cheuk Ting Ho<\/h4>\n                                                        <\/div>\n            <\/div>\n        <\/div>\n    <\/div>\n","protected":false},"author":1297,"featured_media":592626,"comment_status":"closed","ping_status":"closed","template":"","categories":[1401,2347],"tags":[8056,5377],"cross-post-tag":[8851],"acf":[],"_links":{"self":[{"href":"https:\/\/blog.jetbrains.com\/zh-hans\/wp-json\/wp\/v2\/pycharm\/592482"}],"collection":[{"href":"https:\/\/blog.jetbrains.com\/zh-hans\/wp-json\/wp\/v2\/pycharm"}],"about":[{"href":"https:\/\/blog.jetbrains.com\/zh-hans\/wp-json\/wp\/v2\/types\/pycharm"}],"author":[{"embeddable":true,"href":"https:\/\/blog.jetbrains.com\/zh-hans\/wp-json\/wp\/v2\/users\/1297"}],"replies":[{"embeddable":true,"href":"https:\/\/blog.jetbrains.com\/zh-hans\/wp-json\/wp\/v2\/comments?post=592482"}],"version-history":[{"count":5,"href":"https:\/\/blog.jetbrains.com\/zh-hans\/wp-json\/wp\/v2\/pycharm\/592482\/revisions"}],"predecessor-version":[{"id":650356,"href":"https:\/\/blog.jetbrains.com\/zh-hans\/wp-json\/wp\/v2\/pycharm\/592482\/revisions\/650356"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/blog.jetbrains.com\/zh-hans\/wp-json\/wp\/v2\/media\/592626"}],"wp:attachment":[{"href":"https:\/\/blog.jetbrains.com\/zh-hans\/wp-json\/wp\/v2\/media?parent=592482"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blog.jetbrains.com\/zh-hans\/wp-json\/wp\/v2\/categories?post=592482"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blog.jetbrains.com\/zh-hans\/wp-json\/wp\/v2\/tags?post=592482"},{"taxonomy":"cross-post-tag","embeddable":true,"href":"https:\/\/blog.jetbrains.com\/zh-hans\/wp-json\/wp\/v2\/cross-post-tag?post=592482"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}