Firmware size savings #3236
Merged
Firmware size savings #3236
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
jepler
force-pushed
the
firmware-size-gemma-trinket
branch
from
c1f97b5 to
930000e
Compare
|
Many of these strings are already in the qstr pool. Do you want to switch these to |
This is a slight trade-off with code size, in places where a "_varg" mp_raise variant is now used. The net savings on trinket_m0 is just 32 bytes. It also means that the translation will include the original English text, and cannot be translated. These are usually names of Python types such as int, set, or dict or special values such as "inf" or "Nan".
This saves a very small amount of flash, 8 bytes on trinket_m0
jepler
changed the title
jepler
force-pushed
the
firmware-size-gemma-trinket
branch
from
63bfaec to
d3fb6ff
Compare
|
I've heavily revised this PR and now it contains a couple of substantial size savers. |
This removes runtime allocations of the cstring version of the qstring. It is not a size improvement
This saves nearly 200 bytes. Curiously, it also saves RAM.
The missing second "const" made these mutable arrays pointing to const string data.
This function computes the remainder of a value `x` modulo pi/2, to high precision. It does this by dividing the flotaing point values into several ranges by magnitude, and applies successively slower but more accurate algorithms. The last two steps, one covering values up to around 2^7 * pi/2 (called "medium size") and a final one covering all possible float values, require big tables. By eliminating the "medium size" case, a table and some code are removed from the binary. This makes some cases take longer, but saves hundreds of bytes. It does _NOT_ affect the result, only the speed. ``` [desktop python] >>> sum(math.sin(2.**i) for i in range(21)) 1.4206898748939305 [trinket m0, before change to ef_rem_pio2.c] >>> sum(math.sin(2.**i) for i in range(21)) 1.42069 [trinket m0, after change to ef_rem_pio2.c] >>> sum(math.sin(2.**i) for i in range(21)) 1.42069 ```
This array was of 32-bit values, but the entries were only ever in the 0-255 range. Convert to uint8_t. Testing performed: The result of the sum-of-sin was unchanged >>> import math; sum(math.sin(2.**i) for i in range(21)) 1.42069
jepler
force-pushed
the
firmware-size-gemma-trinket
branch
from
d3fb6ff to
6669ced
Compare
|
dhalbert
previously approved these changes
Aug 4, 2020
tannewt
reviewed
Aug 4, 2020
Saves 12 bytes code on trinket m0
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
I ended up finding a range of firmware size (flash utilization) improvements. One item is a performance trade-off, and one item is a feature removal for the most constrainted boards.
A small amount of RAM is also saved.
Baseline: c394af4 (origin/main) Merge pull request #3241 from jepler/translation-percent-space-fixes
Binary: trinket_m0 en_US
Flash free: +1404 bytes
RAM free: +76 bytes
Total translation strings: -17