1.23
2.0
main
test-branch
| Line | |
|---|
| 1 | Implement the following examples:
|
|---|
| 2 | life
|
|---|
| 3 | reduce
|
|---|
| 4 | vecadd
|
|---|
| 5 | matmult
|
|---|
| 6 |
|
|---|
| 7 | Other:
|
|---|
| 8 | [x]Work on wiki page
|
|---|
| 9 | [x]Figure out what to put on the wiki page
|
|---|
| 10 | [x]Make it so that local workgroups can be larger than 1
|
|---|
| 11 | clGetDeviceIDs as a regular method
|
|---|
| 12 | Change names of variables so there probably won't be any conflicts
|
|---|
| 13 | figure out how to use the queueCL for clCreateCommandQueue
|
|---|
| 14 |
|
|---|
| 15 | finished examples:
|
|---|
| 16 | square.cvl
|
|---|
| 17 |
|
|---|
| 18 | Worked on the wiki page for the transformation into openCL,
|
|---|
| 19 | which goes into detail on what the current equivalent in our code for a method is.
|
|---|
| 20 | Square.cvl was "finished" earlier, but did not have a way to accomodate for workgroup sizes greater than one.
|
|---|
| 21 | This has been fixed. Planning on a version 2.1, which has a pre-written part for methods that are the same in all translations.
|
|---|
| 22 |
|
|---|
| 23 | Work on the next example, life.cvl. Plan out what exactly 2.1 entails. Possibly work on another example.
|
|---|
| 24 |
|
|---|
| 25 | This less literal interpretation uses more abstraction and reduces openCL methods into much smaller pieces.
|
|---|
| 26 | However, I had to think about the logistics of what the parser would do about and how it would get the information.
|
|---|
| 27 | Somewhat more complicated to think about than having every method of openCL "implemented" in terms of string parsing.
|
|---|
| 28 | However, things are much quicker to convert by hand. And some methods aren't even needed anymore, which is nice. |
|---|
Note:
See
TracBrowser
for help on using the repository browser.
