Todo List Final Presentation

[carlfriess]

• [x] Check on input sizes for each kernel -> report to Alex (@carlfriess, @krantikiran68, @mezhang111)
• [x] Make sure to use float32 (@carlfriess)

Absolute latency evaluation

• [x] Get latency for each kernel from NPBench from Alex (@adityamanglik)
• [x] Tag versions that we will benchmarks (@carlfriess, @krantikiran68)
• [ ] Merger everything into main (@carlfriess, @krantikiran68, @fabianlandwehr1, @mezhang111)
• [x] Measure latency for each version of kernels (@carlfriess, @krantikiran68, @fabianlandwehr1, @mezhang111)
• [x] Make latency plots for each kernel (@adityamanglik)
• [x] (Measure latency for each version of kernels for different input sizes) (@carlfriess, @krantikiran68, @fabianlandwehr1, @mezhang111)
• [x] (Make latency plots for different sizes for each kernel) (@adityamanglik)

Roofline model

• [x] Practical memory bandwidth (@adityamanglik)
• [x] Practical peak performance (@adityamanglik)
  • [x] Get guidance from Alex on how to measure (@adityamanglik)
  • [x] Do what Alex says (@adityamanglik)
• [x] Calculate computational intensity for each kernel version (@carlfriess, @krantikiran68, @fabianlandwehr1, @mezhang111)
• [x] Make roofline model plots (@adityamanglik)

Presentation

• [ ] Put everything in presentation
• [ ] Make it beautiful (@fabianlandwehr1)

[krantikiran68]

The spreadsheet for adding data: (@carlfriess @adityamanglik @mezhang111 @fabianlandwehr1 @krantikiran68) https://docs.google.com/spreadsheets/d/1yp5Pj-bTi_LkkpL6jKLOupz5uDRKITiMpUgAuj_17fw/edit#gid=377228479