|
Last change
on this file was 810c2c5, checked in by Michael Brooks <mlbrooks@…>, 8 days ago |
|
address two string-section oustanding items: remove confusing sentences; add attribution plot overlay for separate compilation overhead
|
-
Property mode
set to
100644
|
|
File size:
1.1 KB
|
| Rev | Line | |
|---|
| [810c2c5] | 1 | The overhead-amount numbers are produced in MS Excel, to fast-track a data exploration result "into production."
|
|---|
| 2 | The python script overlays them onto the data that gives the location at which they should be drawn.
|
|---|
| 3 |
|
|---|
| 4 | [string-separate-compilation-overhead.xlsx]
|
|---|
| 5 | 'Raw' tab
|
|---|
| 6 | - side-by-side pastes of
|
|---|
| 7 | A-I, b.*: result-allocate-speed-cfa.csv
|
|---|
| 8 | J-R, ns.*: result-allocate-speed-cfa-NOSEPCOMP.csv
|
|---|
| 9 | - with calculated columns in S+
|
|---|
| 10 | - cmp.chk verifies the pastes line up
|
|---|
| 11 | - cmp.DeltaOpDur is the key output: nanoseconds by which NoSepCmp beats Baseline
|
|---|
| 12 | 'Agg' tab
|
|---|
| 13 | - power query on raw: group by {corp-len-grp, tun}; each group has 5 repeat runs
|
|---|
| 14 | - a group's mean cmp.DeltaOpDur is the NoSepCmp advantage at this corpus length and tuning choice
|
|---|
| 15 | last tab
|
|---|
| 16 | - pick just the two overhead cases that the graph uses; project only the overhead delta
|
|---|
| 17 | - the groups (50, 0.2) and (200, 0.2) correspond to the pair of CFA configurations in plot-string-allocn-attrib
|
|---|
| 18 | - export as %.csv
|
|---|
| 19 |
|
|---|
| 20 | [string-separate-compilation-overhead.csv]
|
|---|
| 21 | - export of %.xlsx, last tab
|
|---|
| 22 | - input to the "usual" Python script
|
|---|
Note:
See
TracBrowser
for help on using the repository browser.
