Builtin Array example #7092
Builtin Array example #7092
Conversation
Unisay
added
Do not merge
No Changelog Required
Unisay
force-pushed
the
yura/run-compiled-code
branch
4 times, most recently
from
e7bc91d to
0edc4ee
Compare
Unisay
force-pushed
the
yura/builtin-array-example
branch
from
3ce5ccf to
68761e8
Compare
Unisay
force-pushed
the
yura/run-compiled-code
branch
6 times, most recently
from
6022c7a to
310556f
Compare
Unisay
force-pushed
the
yura/run-compiled-code
branch
from
310556f to
c575c94
Compare
Unisay
force-pushed
the
yura/builtin-array-example
branch
from
68761e8 to
f4e3f38
Compare
Unisay
force-pushed
the
yura/builtin-array-example
branch
from
f4e3f38 to
0f4f5c9
Compare
Unisay
force-pushed
the
yura/builtin-array-example
branch
from
0f4f5c9 to
9c4bb4a
Compare
|
Unisay
force-pushed
the
yura/builtin-array-example
branch
from
9c4bb4a to
5c3251c
Compare
There was a problem hiding this comment.
It should be clarified that this is an upcoming feature - not usable yet. And I think it deserves more publicity than being a part of "Other Optimization Techniques".
I suggest creating a category of pages, called "Upcoming Features". Under it, add a page called "Builtin Arrays", and put the content there. You can reference it in "Other Optimization Techniques".
|
|
||
| For instance, a single lookup at index 99 of a 100-element list demonstrates this trade-off: | ||
| * **Lookup Cost:** The CPU cost for the lookup on a standard Plinth list (a sum-of-products type) is 206 times higher than on a `BuiltinArray`. | ||
| * **Creation Cost:** However, creating the `BuiltinArray` (by first creating a list and then converting it) is 11 times more expensive than creating the list alone. |
There was a problem hiding this comment.
Really? Creating a list is linear, creating an array is also linear, how can it be 11x worse?
Are you possibly trying to say: if you already have a list, then performing a single lookup is 11x faster than converting the entire list to an array, and then performing a single lookup?
|
|
||
| For multiple lookups by index, converting a list to a `BuiltinArray` can be more cost-effective than performing lookups on the list directly. The initial cost of converting the list to an array can be offset by the significantly faster lookups. | ||
|
|
||
| As a rule of thumb, if you only need to perform a single lookup, it is likely not worth the conversion cost. The benefits of using a `BuiltinArray` become apparent when you need to perform several lookups on the same data structure. |
There was a problem hiding this comment.
it is likely not worth the conversion cost.
What about picking list vs. array to use in the first place? If you use array in your data type in the first place, then you won't need the conversion. So should one generally prefer using array or list in the first place, and why?
Closes https://github.com/IntersectMBO/plutus-private/issues/1590
Preview: https://plutus.cardano.intersectmbo.org/pr-preview/pr-7092/working-with-scripts/other-optimization-techniques#using-builtinarray-for-index-based-lookups