semify-eda / tristan

1 stars 0 forks source link

Design specification for demonstrator #6

Closed kstrohmayer closed 1 year ago

kstrohmayer commented 1 year ago

More comprehensive report.

Add link to document on sharepoint

kstrohmayer commented 1 year ago

Indicate how to visualize the improvement of the custom instruction

adam-hrvth commented 1 year ago

The "in-progress" report has been added to our Google Drive.

adam-hrvth commented 1 year ago

Hi @kstrohmayer and @mole99, I have finished the report for the demonstrator. Please review it and provide some feedback. If no changes are required, I will upload it to SharePoint.

kstrohmayer commented 1 year ago

Thanks for the report. Looks fine in general.

Some questions for my side:

adam-hrvth commented 1 year ago

Thanks for the report. Looks fine in general.

Some questions for my side:

  • Are we able to provide already some numbers for the achievable compression ratio

  • Are we able to provide already some numbers for the achievable performance speed-up

@mole99, you've been making progress on running the encoding on hardware. Do you have any numerical data we could include in the report?

mole99 commented 1 year ago

We can measure the time it takes for a full encoding + decoding with a soft cntb and then a second time using the current hard cntb. The ratio will give us the speedup. We can do this together on Thursday morning. For the compression ratio I think there are some numbers in Marios thesis.

adam-hrvth commented 1 year ago

Sure, next Thursday is good for me. But just to let you know, I've got to submit the report by the end of Tuesday, so I'll have to skip the performance speed-up part. I'll check Mario's thesis for the compression ratio.

adam-hrvth commented 1 year ago

I've added the report to SharePoint. I've included Mario's results on the compression/decompression improvements. Currently, it doesn't include the results on the achievable performance speed-up, but we can add that during the actual demonstration.