[Test] Add/update GEMM tests to be consistent for performance.

[mshahneo]

Updates the GEMM test cases to have:

• Input range from 0.0 to 1.0
• Use large register file
• Remove the gpu.barrier to sync threads, this barrier is only needed so that some threads do not go too far forward with the prefetching that other threads necessary prefetched data is thrown out of the cache. However, doing gpu.barrier is very expensive. Until we have something of a named_barrier type available, skipping this barrier proved to be more performant.

Please review these guidelines to help with the review process:

• Have you provided a meaningful PR description?
• Have you added a test, a reproducer, or a reference to an issue with a reproducer?
• Have you tested your changes locally for CPU and GPU devices?
• Have you made sure that new changes do not introduce compiler warnings?
• If this PR is a work in progress, are you filing the PR as a draft?
• Have you organized your commits logically and ensured each can be built by itself?