Fix max/min other value when int dtype and add int dtype tests

[0x45f]

PR Category

Operator

Type of Change

Bug Fix

Description

The following code can reproduce the bug:

import torch
import flag_gems
import numpy as np

torch.set_default_device("cuda")

x = torch.tensor([1056, 1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024,
        1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024,
        1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024,
        1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024,
        1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024, 1024,
        1024, 1024, 1024, 1024, 1024], dtype=torch.int64)
# print(x.max())
with flag_gems.use_gems():
    print(x.max())

Issue

Progress

• Change is properly reviewed (1 reviewer required, 2 recommended).
• Change is responded to an issue.
• Change is fully covered by a UT.

Performance