[jira] [Commented] (ARROW-15571) [C++] Add min/max/sqrt scalar kernels to execution engine
[ https://issues.apache.org/jira/browse/ARROW-15571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17488129#comment-17488129 ] Yaron Gvili commented on ARROW-15571: - Since plenty of NaN values exist, I find returning an input NaN value more user-friendly then returning a null, which just drops information, which can always be done in a later step. However, if there is a convention that other operations conform to on this issue, then it should probably be followed on. > [C++] Add min/max/sqrt scalar kernels to execution engine > - > > Key: ARROW-15571 > URL: https://issues.apache.org/jira/browse/ARROW-15571 > Project: Apache Arrow > Issue Type: Improvement > Components: C++ >Reporter: Yaron Gvili >Priority: Major > Labels: kernel, pull-request-available > Time Spent: 50m > Remaining Estimate: 0h > > The list of execution engine's scalar kernels currently available in > `cpp/src/arrow/compute/kernels/scalar_arithmetic.cc` does not cover the > common minimum, maximum, and square-root functions. -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Commented] (ARROW-15571) [C++] Add min/max/sqrt scalar kernels to execution engine
[ https://issues.apache.org/jira/browse/ARROW-15571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17488098#comment-17488098 ] David Li commented on ARROW-15571: -- There are no operations that return NaN instead of null in this case. It could perhaps be added as an option (that would also be a separate JIRA) > [C++] Add min/max/sqrt scalar kernels to execution engine > - > > Key: ARROW-15571 > URL: https://issues.apache.org/jira/browse/ARROW-15571 > Project: Apache Arrow > Issue Type: Improvement > Components: C++ >Reporter: Yaron Gvili >Priority: Major > Labels: kernel, pull-request-available > Time Spent: 50m > Remaining Estimate: 0h > > The list of execution engine's scalar kernels currently available in > `cpp/src/arrow/compute/kernels/scalar_arithmetic.cc` does not cover the > common minimum, maximum, and square-root functions. -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Commented] (ARROW-15571) [C++] Add min/max/sqrt scalar kernels to execution engine
[ https://issues.apache.org/jira/browse/ARROW-15571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17487692#comment-17487692 ] Yaron Gvili commented on ARROW-15571: - I'm good with splitting the issue; I'll get to that a bit later. In the meantime, something I just noticed: "min_element_wise" is equivalent to "minimum_checked", both ignoring a single NaN input, but not to "minimum", which returns a single NaN input (the same applies to the corresponding max functions). Are there existing min/max operations that have this latter behavior? > [C++] Add min/max/sqrt scalar kernels to execution engine > - > > Key: ARROW-15571 > URL: https://issues.apache.org/jira/browse/ARROW-15571 > Project: Apache Arrow > Issue Type: Improvement > Components: C++ >Reporter: Yaron Gvili >Priority: Major > Labels: kernel, pull-request-available > Time Spent: 50m > Remaining Estimate: 0h > > The list of execution engine's scalar kernels currently available in > `cpp/src/arrow/compute/kernels/scalar_arithmetic.cc` does not cover the > common minimum, maximum, and square-root functions. -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Commented] (ARROW-15571) [C++] Add min/max/sqrt scalar kernels to execution engine
[ https://issues.apache.org/jira/browse/ARROW-15571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17487648#comment-17487648 ] Yibo Cai commented on ARROW-15571: -- Better to separate "sqrt" and "elementwise min/max" to two PRs. They are very different. Maybe rename this Jira (and related PR) to "sqrt" only, and create a new Jira for "improve elementwise min/max for special cases"? > [C++] Add min/max/sqrt scalar kernels to execution engine > - > > Key: ARROW-15571 > URL: https://issues.apache.org/jira/browse/ARROW-15571 > Project: Apache Arrow > Issue Type: Improvement > Components: C++ >Reporter: Yaron Gvili >Priority: Major > Labels: kernel, pull-request-available > Time Spent: 50m > Remaining Estimate: 0h > > The list of execution engine's scalar kernels currently available in > `cpp/src/arrow/compute/kernels/scalar_arithmetic.cc` does not cover the > common minimum, maximum, and square-root functions. -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Commented] (ARROW-15571) [C++] Add min/max/sqrt scalar kernels to execution engine
[ https://issues.apache.org/jira/browse/ARROW-15571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17487569#comment-17487569 ] Yaron Gvili commented on ARROW-15571: - Sounds good. Please advise on what I should add to the PR, like what to cover in unit tests. To keep the PR tight, could the dispatching be deferred to a separate one? > [C++] Add min/max/sqrt scalar kernels to execution engine > - > > Key: ARROW-15571 > URL: https://issues.apache.org/jira/browse/ARROW-15571 > Project: Apache Arrow > Issue Type: Improvement > Components: C++ >Reporter: Yaron Gvili >Priority: Major > Labels: kernel, pull-request-available > Time Spent: 50m > Remaining Estimate: 0h > > The list of execution engine's scalar kernels currently available in > `cpp/src/arrow/compute/kernels/scalar_arithmetic.cc` does not cover the > common minimum, maximum, and square-root functions. -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Commented] (ARROW-15571) [C++] Add min/max/sqrt scalar kernels to execution engine
[ https://issues.apache.org/jira/browse/ARROW-15571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17487565#comment-17487565 ] David Li commented on ARROW-15571: -- That also makes a lot of sense! We should definitely have sqrt. > [C++] Add min/max/sqrt scalar kernels to execution engine > - > > Key: ARROW-15571 > URL: https://issues.apache.org/jira/browse/ARROW-15571 > Project: Apache Arrow > Issue Type: Improvement > Components: C++ >Reporter: Yaron Gvili >Priority: Major > Labels: kernel, pull-request-available > Time Spent: 40m > Remaining Estimate: 0h > > The list of execution engine's scalar kernels currently available in > `cpp/src/arrow/compute/kernels/scalar_arithmetic.cc` does not cover the > common minimum, maximum, and square-root functions. -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Commented] (ARROW-15571) [C++] Add min/max/sqrt scalar kernels to execution engine
[ https://issues.apache.org/jira/browse/ARROW-15571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17487562#comment-17487562 ] Yaron Gvili commented on ARROW-15571: - A [comment in the PR|https://github.com/apache/arrow/pull/12338#issuecomment-1030529616] reports on a benchmark showing sqrt(x ) being much faster than pow(x, 0.5). > [C++] Add min/max/sqrt scalar kernels to execution engine > - > > Key: ARROW-15571 > URL: https://issues.apache.org/jira/browse/ARROW-15571 > Project: Apache Arrow > Issue Type: Improvement > Components: C++ >Reporter: Yaron Gvili >Priority: Major > Labels: kernel, pull-request-available > Time Spent: 40m > Remaining Estimate: 0h > > The list of execution engine's scalar kernels currently available in > `cpp/src/arrow/compute/kernels/scalar_arithmetic.cc` does not cover the > common minimum, maximum, and square-root functions. -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Commented] (ARROW-15571) [C++] Add min/max/sqrt scalar kernels to execution engine
[ https://issues.apache.org/jira/browse/ARROW-15571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17487561#comment-17487561 ] David Li commented on ARROW-15571: -- Thanks for the code. That's not surprising, considering the elementwise min/max are variadic. We could dispatch to the specialized case for when we have only two arguments, though (and we could add this benchmark to the upstream benchmarks to track the improvement). > [C++] Add min/max/sqrt scalar kernels to execution engine > - > > Key: ARROW-15571 > URL: https://issues.apache.org/jira/browse/ARROW-15571 > Project: Apache Arrow > Issue Type: Improvement > Components: C++ >Reporter: Yaron Gvili >Priority: Major > Labels: kernel, pull-request-available > Time Spent: 40m > Remaining Estimate: 0h > > The list of execution engine's scalar kernels currently available in > `cpp/src/arrow/compute/kernels/scalar_arithmetic.cc` does not cover the > common minimum, maximum, and square-root functions. -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Commented] (ARROW-15571) [C++] Add min/max/sqrt scalar kernels to execution engine
[
https://issues.apache.org/jira/browse/ARROW-15571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17487559#comment-17487559
]
Yaron Gvili commented on ARROW-15571:
-
Running the benchmark code below on my CPU "Intel(R) Core(TM) i7-10875H CPU @
2.30GHz" showed the minimum/maximum operations in my PR are faster when
operating on a single array/column. For an array of size 1M the speed-up factor
was ~1.5 and even higher for shorter arrays.
{code:java}
#include
#include
#include
#include
#include
#include
uint64_t get_time() {
return std::chrono::high_resolution_clock::now().time_since_epoch().count();
}
arrow::Status RunMain(int argc, char **argv) {
constexpr int n = 100;
arrow::DoubleBuilder builder;
builder.Reserve(n);
for (int i=0; i array = *maybe_array;
std::shared_ptr low_threshold =
std::make_shared(0.1);
std::shared_ptr high_threshold =
std::make_shared(0.8);
uint64_t t;
t = get_time();
ARROW_ASSIGN_OR_RAISE(arrow::Datum max0,
arrow::compute::CallFunction("max_element_wise", {array, low_threshold}));
ARROW_ASSIGN_OR_RAISE(arrow::Datum min0,
arrow::compute::CallFunction("min_element_wise", {max0, high_threshold}));
uint64_t t0 = get_time() - t;
t = get_time();
ARROW_ASSIGN_OR_RAISE(arrow::Datum max1,
arrow::compute::CallFunction("maximum", {array, low_threshold}));
ARROW_ASSIGN_OR_RAISE(arrow::Datum min1,
arrow::compute::CallFunction("minimum", {max1, high_threshold}));
uint64_t t1 = get_time() - t;
if (!arrow::ArrayEquals(*min0.make_array(), *min1.make_array())) {
return arrow::Status::Invalid("result arrays are not equal");
}
std::cout << "timing: " << t0 << " vs " << t1 << " (speed-up factor " << (1.0
* t0 / t1) << ")" << std::endl;
return arrow::Status::OK();
}
int main(int argc, char **argv) {
arrow::Status st = RunMain(argc, argv);
if (!st.ok()) {
std::cerr << st << std::endl;
return 1;
}
return 0;
} {code}
> [C++] Add min/max/sqrt scalar kernels to execution engine
> -
>
> Key: ARROW-15571
> URL: https://issues.apache.org/jira/browse/ARROW-15571
> Project: Apache Arrow
> Issue Type: Improvement
> Components: C++
>Reporter: Yaron Gvili
>Priority: Major
> Labels: kernel, pull-request-available
> Time Spent: 40m
> Remaining Estimate: 0h
>
> The list of execution engine's scalar kernels currently available in
> `cpp/src/arrow/compute/kernels/scalar_arithmetic.cc` does not cover the
> common minimum, maximum, and square-root functions.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
[jira] [Commented] (ARROW-15571) [C++] Add min/max/sqrt scalar kernels to execution engine
[ https://issues.apache.org/jira/browse/ARROW-15571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17487214#comment-17487214 ] Yaron Gvili commented on ARROW-15571: - OK, I'll try to set up and report back on a benchmark comparing the suggested operations and the ones in my PR. > [C++] Add min/max/sqrt scalar kernels to execution engine > - > > Key: ARROW-15571 > URL: https://issues.apache.org/jira/browse/ARROW-15571 > Project: Apache Arrow > Issue Type: Improvement > Components: C++ >Reporter: Yaron Gvili >Priority: Major > Labels: kernel, pull-request-available > Time Spent: 0.5h > Remaining Estimate: 0h > > The list of execution engine's scalar kernels currently available in > `cpp/src/arrow/compute/kernels/scalar_arithmetic.cc` does not cover the > common minimum, maximum, and square-root functions. -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Commented] (ARROW-15571) [C++] Add min/max/sqrt scalar kernels to execution engine
[ https://issues.apache.org/jira/browse/ARROW-15571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17487207#comment-17487207 ] Rok Mihevc commented on ARROW-15571: Python maps directly to c++ kernels here and it's less verbose. Perhaps take a look at [https://github.com/apache/arrow/blob/master/cpp/src/arrow/compute/kernels/scalar_compare_test.cc#L1228] and [https://github.com/apache/arrow/blob/master/cpp/src/arrow/compute/kernels/scalar_arithmetic_test.cc#L932] to get c++ implementation start. > [C++] Add min/max/sqrt scalar kernels to execution engine > - > > Key: ARROW-15571 > URL: https://issues.apache.org/jira/browse/ARROW-15571 > Project: Apache Arrow > Issue Type: Improvement > Components: C++ >Reporter: Yaron Gvili >Priority: Major > Labels: kernel, pull-request-available > Time Spent: 0.5h > Remaining Estimate: 0h > > The list of execution engine's scalar kernels currently available in > `cpp/src/arrow/compute/kernels/scalar_arithmetic.cc` does not cover the > common minimum, maximum, and square-root functions. -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Commented] (ARROW-15571) [C++] Add min/max/sqrt scalar kernels to execution engine
[ https://issues.apache.org/jira/browse/ARROW-15571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17487204#comment-17487204 ] David Li commented on ARROW-15571: -- I think we're missing some terminology here. Note the Python functions are the exact same as the C++ ones. Rok's example can be translated to C++ and they both use the same implementation here. Scalar functions operate on array and scalar values. "Scalar" just means that they operate elementwise. Aggregate functions reduce multiple values to one. It seems they are not what you are looking for here. However, again, note that min/max both have aggregate versions (min(array) -> scalar) and scalar versions (min(array, array) -> array or min(scalar, array) -> array or ...). So I think Rok's suggestion is what you are looking for. If performance is not up to par in benchmarks, then there is opportunity to optimize. (The scalar min/max kernels do some extra work because they take any number of arguments, for instance - though I don't recall if they have specializations for binary inputs.) > [C++] Add min/max/sqrt scalar kernels to execution engine > - > > Key: ARROW-15571 > URL: https://issues.apache.org/jira/browse/ARROW-15571 > Project: Apache Arrow > Issue Type: Improvement > Components: C++ >Reporter: Yaron Gvili >Priority: Major > Labels: kernel, pull-request-available > Time Spent: 0.5h > Remaining Estimate: 0h > > The list of execution engine's scalar kernels currently available in > `cpp/src/arrow/compute/kernels/scalar_arithmetic.cc` does not cover the > common minimum, maximum, and square-root functions. -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Commented] (ARROW-15571) [C++] Add min/max/sqrt scalar kernels to execution engine
[ https://issues.apache.org/jira/browse/ARROW-15571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17487199#comment-17487199 ] Yaron Gvili commented on ARROW-15571: - The code seems reasonable for a Python setting, but I'm targeting the C++ execution engine (specifically, projection using an expression), and would like to get best performance. My guess is scalar operations are faster than aggregate ones for scalar use cases of the kind I described. Otherwise, it's not clear to me why some operation, such as Add, is implemented as both a scalar and an aggregate one. > [C++] Add min/max/sqrt scalar kernels to execution engine > - > > Key: ARROW-15571 > URL: https://issues.apache.org/jira/browse/ARROW-15571 > Project: Apache Arrow > Issue Type: Improvement > Components: C++ >Reporter: Yaron Gvili >Priority: Major > Labels: kernel, pull-request-available > Time Spent: 0.5h > Remaining Estimate: 0h > > The list of execution engine's scalar kernels currently available in > `cpp/src/arrow/compute/kernels/scalar_arithmetic.cc` does not cover the > common minimum, maximum, and square-root functions. -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Commented] (ARROW-15571) [C++] Add min/max/sqrt scalar kernels to execution engine
[
https://issues.apache.org/jira/browse/ARROW-15571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17487186#comment-17487186
]
Rok Mihevc commented on ARROW-15571:
In pyarrow 7.0.0 that would be something like this? :
{code:python}
import pyarrow as pa
import pyarrow.compute as pc
low_threshold = 1
high_threshold = 3
arr = pa.array([1, 2, 3])
pc.min_element_wise(pc.max_element_wise(pc.power(arr, 0.5), low_threshold),
high_threshold)
{code}
> [C++] Add min/max/sqrt scalar kernels to execution engine
> -
>
> Key: ARROW-15571
> URL: https://issues.apache.org/jira/browse/ARROW-15571
> Project: Apache Arrow
> Issue Type: Improvement
> Components: C++
>Reporter: Yaron Gvili
>Priority: Major
> Labels: kernel, pull-request-available
> Time Spent: 0.5h
> Remaining Estimate: 0h
>
> The list of execution engine's scalar kernels currently available in
> `cpp/src/arrow/compute/kernels/scalar_arithmetic.cc` does not cover the
> common minimum, maximum, and square-root functions.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
[jira] [Commented] (ARROW-15571) [C++] Add min/max/sqrt scalar kernels to execution engine
[
https://issues.apache.org/jira/browse/ARROW-15571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17487181#comment-17487181
]
Yaron Gvili commented on ARROW-15571:
-
OK regarding power working on floating point values. This
[SO|https://stackoverflow.com/questions/17417490/difference-between-sqrtx-and-powx-0-5]
suggests sqrt(x) would be faster and a bit more accurate than pow(x, 0.5), so
there is room for a sqrt operation.
Regarding my goal, it is to apply the operations to a single column with best
performance in Arrow. Here's an example for how I'm currently invoking the code
in my PR:
{code:java}
ProjectNodeOptions(
{call("minimum",
{call("maximum",
{call("square_root",
{field_ref("column_0")}),
literal(low_threshold)}),
literal(high_threshold)})},
{"result_0"}){code}
Note that the above is `using namespace arrow::compute` and that "column_0" is
of type double. How should I achieve the same functionality with the existing
operations? If the solution involves aggregate operations, do they provide best
performance, as compared to (the missing) scalar operations?
> [C++] Add min/max/sqrt scalar kernels to execution engine
> -
>
> Key: ARROW-15571
> URL: https://issues.apache.org/jira/browse/ARROW-15571
> Project: Apache Arrow
> Issue Type: Improvement
> Components: C++
>Reporter: Yaron Gvili
>Priority: Major
> Labels: kernel, pull-request-available
> Time Spent: 0.5h
> Remaining Estimate: 0h
>
> The list of execution engine's scalar kernels currently available in
> `cpp/src/arrow/compute/kernels/scalar_arithmetic.cc` does not cover the
> common minimum, maximum, and square-root functions.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
[jira] [Commented] (ARROW-15571) [C++] Add min/max/sqrt scalar kernels to execution engine
[ https://issues.apache.org/jira/browse/ARROW-15571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17487164#comment-17487164 ] David Li commented on ARROW-15571: -- Looking at the PR you're looking for a row-wise min/max? In which case that is implemented on floating point values, yes. (It will also work given scalar values.) Power also works on floating point values. > [C++] Add min/max/sqrt scalar kernels to execution engine > - > > Key: ARROW-15571 > URL: https://issues.apache.org/jira/browse/ARROW-15571 > Project: Apache Arrow > Issue Type: Improvement > Components: C++ >Reporter: Yaron Gvili >Priority: Major > Labels: kernel, pull-request-available > Time Spent: 0.5h > Remaining Estimate: 0h > > The list of execution engine's scalar kernels currently available in > `cpp/src/arrow/compute/kernels/scalar_arithmetic.cc` does not cover the > common minimum, maximum, and square-root functions. -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Commented] (ARROW-15571) [C++] Add min/max/sqrt scalar kernels to execution engine
[ https://issues.apache.org/jira/browse/ARROW-15571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17487158#comment-17487158 ] Yaron Gvili commented on ARROW-15571: - Sorry, I should have been cleared in the title that I'm referring to floating point scalars. Are the aggregate operations good for operating on floating point scalars, rather than arrays? Is power good for operating on floating points, rather than integers? I'm relatively new to Arrow, so I may have missed a couple of things. > [C++] Add min/max/sqrt scalar kernels to execution engine > - > > Key: ARROW-15571 > URL: https://issues.apache.org/jira/browse/ARROW-15571 > Project: Apache Arrow > Issue Type: Improvement > Components: C++ >Reporter: Yaron Gvili >Priority: Major > Labels: kernel, pull-request-available > Time Spent: 0.5h > Remaining Estimate: 0h > > The list of execution engine's scalar kernels currently available in > `cpp/src/arrow/compute/kernels/scalar_arithmetic.cc` does not cover the > common minimum, maximum, and square-root functions. -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Commented] (ARROW-15571) [C++] Add min/max/sqrt scalar kernels to execution engine
[ https://issues.apache.org/jira/browse/ARROW-15571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17487143#comment-17487143 ] David Li commented on ARROW-15571: -- Min/max are available as aggregations: [https://arrow.apache.org/docs/dev/cpp/compute.html#aggregations] and there are also element-wise min/max: [https://arrow.apache.org/docs/dev/cpp/compute.html#comparisons] Square root is indeed not available, though power is: [https://arrow.apache.org/docs/dev/cpp/compute.html#arithmetic-functions] (presumably, a dedicated sqrt would be faster, though) > [C++] Add min/max/sqrt scalar kernels to execution engine > - > > Key: ARROW-15571 > URL: https://issues.apache.org/jira/browse/ARROW-15571 > Project: Apache Arrow > Issue Type: Improvement > Components: C++ >Reporter: Yaron Gvili >Priority: Major > > The list of execution engine's scalar kernels currently available in > `cpp/src/arrow/compute/kernels/scalar_arithmetic.cc` does not cover the > common minimum, maximum, and square-root functions. -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Commented] (ARROW-15571) [C++] Add min/max/sqrt scalar kernels to execution engine
[ https://issues.apache.org/jira/browse/ARROW-15571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17487142#comment-17487142 ] Rok Mihevc commented on ARROW-15571: Min and max are covered in https://github.com/apache/arrow/blob/master/cpp/src/arrow/compute/kernels/aggregate_basic.cc. > [C++] Add min/max/sqrt scalar kernels to execution engine > - > > Key: ARROW-15571 > URL: https://issues.apache.org/jira/browse/ARROW-15571 > Project: Apache Arrow > Issue Type: Improvement > Components: C++ >Reporter: Yaron Gvili >Priority: Major > > The list of execution engine's scalar kernels currently available in > `cpp/src/arrow/compute/kernels/scalar_arithmetic.cc` does not cover the > common minimum, maximum, and square-root functions. -- This message was sent by Atlassian Jira (v8.20.1#820001)
