[CINN] Adjust tile config parameters based on model benchmarks

[lshpku]

PR Category

CINN

PR Types

Performance

Description

Adjust two key parameters in TileConfig based on recent end-to-end model benchmark results.

Change 1:
Add one more condition for using WarpReduce. This improves performance for huge-S-small-R cases, such as:

reduce(shape=[128, 197, 768] axis=2)
reduce(shape=[32, 512, 56, 56], axis=(2, 3))

Change 2:
Set tighter limit on spatial inner loop number (from 32 to 4). This prevents performance degradation in Transpose and certain element-wise operations.
Note: this change doesn't imply that 32 is inherently a worse choice than 4; rather, it is due to the lack of support for scheduling Transpose, and 4 is currently the optimal choice for nvcc to do just acceptable optimization.

Pcard-85711

[paddle-bot]

你的PR提交成功，感谢你对开源项目的贡献!
请关注后续CI自动化测试结果，详情请参考Paddle-CI手册。
Your PR has been submitted. Thanks for your contribution!
Please wait for the result of CI firstly. See Paddle CI Manual for details.