Sample on GPU

[EricLBuehler]

Refs #66.

[EricLBuehler]

It appears that the sorting takes the longest amount of time.

[github-actions]

Code Metrics Report

  ───────────────────────────────────────────────────────────────────────────────
Language                 Files     Lines   Blanks  Comments     Code Complexity
───────────────────────────────────────────────────────────────────────────────
Rust                        46     15509     1090       638    13781        781
───────────────────────────────────────────────────────────────────────────────
Total                       46     15509     1090       638    13781        781
───────────────────────────────────────────────────────────────────────────────
Estimated Cost to Develop 24,463
Estimated Schedule Effort 9.931007 months
Estimated People Required 3.797193
───────────────────────────────────────────────────────────────────────────────
Processed 524223 bytes, 0.524 megabytes (SI)
───────────────────────────────────────────────────────────────────────────────
  

[EricLBuehler]

Developments on master have made this obsolete, and there is a <2 T/s improvement over master on an A10 (mistral GGUF Q4_K_M).