Merged
Conversation
Creates a `ROUND_TO` function that rounds it's input to one of the
provided values. Like so:
```
ROUND_TO(v, 0, 5000, 10000, 20000, 40000, 100000)
v | ROUND_TO
0 | 0
100 | 0
6000 | 5000
45001 | 40000
999999 | 100000
```
For some sequences of numbers you could do this with the `/` operator -
but for arbitrary sequences of numbers you needed `CASE` which is quite
slow. And hard to read!
Rewriting the example above would look like:
```
CASE (
v < 5000, 0,
v < 10000, 5000,
v < 20000, 10000,
v < 40000, 20000,
v < 100000, 40000,
100000
)
```
Even better, this is *fast*:
```
(operation) Mode Cnt Score Error Units
round_to_4_via_case avgt 7 138.124 ± 0.738 ns/op
round_to_4 avgt 7 0.805 ± 0.011 ns/op
round_to_3 avgt 7 0.739 ± 0.011 ns/op
round_to_2 avgt 7 0.651 ± 0.009 ns/op
date_trunc avgt 7 2.425 ± 0.018 ns/op
```
I've included a comparison to `DATE_TRUNC` above because we should be
able to rewrite `DATE_TRUNC` into `ROUND_TO` when we know the date range
of the index. This doesn't do it now, but it should be possible.
elasticsearchmachine
added
the
Team:Analytics
Collaborator
|
Pinging @elastic/es-analytical-engine (Team:Analytics) |
Collaborator
|
Hi @nik9000, I've created a changelog YAML for you. |
nik9000
commented
May 22, 2025
Member
Author
nik9000
left a comment
nik9000
left a comment
There was a problem hiding this comment.
So this is fun: I realized that I was linearly searching a list and I ask binary search questions in interviews all the time. Gives us an excuse to talk about O with a topic everyone understand. You know. Super obvious.
Well, any guess how big the array has to be before binary search gets faster?
Anyway, no we have a linear search and a binary search.
Member
Author
|
FWIW, I suspect https://www.felixcloutier.com/x86/pcmpgtq but have no decompiled. |
Member
Author
|
Also, the magic isn't entirely building this fast function - it's for when we can merge it into lucene's structures like doc-value-skipper. |
idegtiarenko
approved these changes
May 23, 2025
nik9000
added
the
auto-backport
Member
Author
|
Thanks @idegtiarenko ! |
Collaborator
💔 Backport failed
You can use sqren/backport to manually backport by running |
nik9000
added a commit
to nik9000/elasticsearch
that referenced
this pull request
May 23, 2025
Creates a `ROUND_TO` function that rounds it's input to one of the
provided values. Like so:
```
ROUND_TO(v, 0, 5000, 10000, 20000, 40000, 100000)
v | ROUND_TO
0 | 0
100 | 0
6000 | 5000
45001 | 40000
999999 | 100000
```
For some sequences of numbers you could do this with the `/` operator -
but for arbitrary sequences of numbers you needed `CASE` which is quite
slow. And hard to read!
Rewriting the example above would look like:
```
CASE (
v < 5000, 0,
v < 10000, 5000,
v < 20000, 10000,
v < 40000, 20000,
v < 100000, 40000,
100000
)
```
Even better, this is *fast*:
```
(operation) Mode Cnt Score Error Units
round_to_4_via_case avgt 7 138.124 ± 0.738 ns/op
round_to_4 avgt 7 0.805 ± 0.011 ns/op
round_to_3 avgt 7 0.739 ± 0.011 ns/op
round_to_2 avgt 7 0.651 ± 0.009 ns/op
date_trunc avgt 7 2.425 ± 0.018 ns/op
```
I've included a comparison to `DATE_TRUNC` above because we should be
able to rewrite `DATE_TRUNC` into `ROUND_TO` when we know the date range
of the index. This doesn't do it now, but it should be possible.
Member
Author
|
Backport is #128397 |
elasticsearchmachine
pushed a commit
that referenced
this pull request
May 23, 2025
Creates a `ROUND_TO` function that rounds it's input to one of the
provided values. Like so:
```
ROUND_TO(v, 0, 5000, 10000, 20000, 40000, 100000)
v | ROUND_TO
0 | 0
100 | 0
6000 | 5000
45001 | 40000
999999 | 100000
```
For some sequences of numbers you could do this with the `/` operator -
but for arbitrary sequences of numbers you needed `CASE` which is quite
slow. And hard to read!
Rewriting the example above would look like:
```
CASE (
v < 5000, 0,
v < 10000, 5000,
v < 20000, 10000,
v < 40000, 20000,
v < 100000, 40000,
100000
)
```
Even better, this is *fast*:
```
(operation) Mode Cnt Score Error Units
round_to_4_via_case avgt 7 138.124 ± 0.738 ns/op
round_to_4 avgt 7 0.805 ± 0.011 ns/op
round_to_3 avgt 7 0.739 ± 0.011 ns/op
round_to_2 avgt 7 0.651 ± 0.009 ns/op
date_trunc avgt 7 2.425 ± 0.018 ns/op
```
I've included a comparison to `DATE_TRUNC` above because we should be
able to rewrite `DATE_TRUNC` into `ROUND_TO` when we know the date range
of the index. This doesn't do it now, but it should be possible.
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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.
Creates a
ROUND_TOfunction that rounds it's input to one of the provided values. Like so:For some sequences of numbers you could do this with the
/operator - but for arbitrary sequences of numbers you neededCASEwhich is quite slow. And hard to read!Rewriting the example above would look like:
Even better, this is fast:
I've included a comparison to
DATE_TRUNCabove because we should be able to rewriteDATE_TRUNCintoROUND_TOwhen we know the date range of the index. This doesn't do it now, but it should be possible.