500万行单表查询，sum over 跑不出结果

[anafin (0106)]

为提高效率，请提供以下信息，问题描述清晰能够更快得到解决：
【 TiDB 使用环境】5.1，tiflash也试过
SELECT t1.id,
ROW_NUMBER() OVER(PARTITION BY t1.storetype ORDER BY t1.sale desc) as rk,
sum(t1.sale) OVER(PARTITION BY t1.storetype ORDER BY t1.sale desc rows between unbounded preceding and current row) as saleamt_order
from Temp_2021pool t1

---

若提问为性能优化、故障排查类问题，请下载脚本运行。终端输出的打印结果，请务必全选并复制粘贴上传。

[anafin (0106)]

注销掉 sum(t1.sale) OVER(PARTITION BY t1.storetype ORDER BY t1.sale desc rows between unbounded preceding 执行，10多秒就完成，主要时间在网络传输数据，sum() over() 有什么坑么？：
SELECT t1.id,
ROW_NUMBER() OVER(PARTITION BY t1.storetype ORDER BY t1.sale desc) as rk,
– sum(t1.sale) OVER(PARTITION BY t1.storetype ORDER BY t1.sale desc rows between unbounded preceding and current row) as saleamt_order
from Temp_2021pool t1

[songxuecheng (小白)]

麻烦发一下执行计划

[数据小黑]

顶楼上，此类问题带explain analyze的执行计划，可能就分析出原因了。
盲猜一下，可能会有内存不足，溢出磁盘的动作。