[SPARK-30754][SQL] Reuse results of floorDiv in calculations of floorMod in DateTimeUtils

[MaxGekk]

What changes were proposed in this pull request?

In the case of back-to-back calculation of floorDiv and floorMod with the same arguments, the result of foorDiv can be reused in calculation of floorMod. The floorMod method is defined as the following in Java standard library:

    public static int floorMod(int x, int y) {
        int r = x - floorDiv(x, y) * y;
        return r;
    }

If floorDiv(x, y) has been already calculated, it can be reused in x - floorDiv(x, y) * y.

I propose to modify 2 places in DateTimeUtils:

1. microsToInstant which is widely used in many date-time functions. Math.floorMod(us, MICROS_PER_SECOND) is just replaced by its definition from Java Math library.
2. truncDate: Math.floorMod(oldYear, divider) == 0 is replaced by Math.floorDiv(oldYear, divider) * divider == oldYear where floorDiv(...) * divider is pre-calculated.

Why are the changes needed?

This reduces the number of arithmetic operations, and can slightly improve performance of date-time functions.

Does this PR introduce any user-facing change?

No

How was this patch tested?

By existing test suites DateTimeUtilsSuite, DateFunctionsSuite and DateExpressionsSuite.

[SparkQA]

Test build #118031 has finished for PR 27491 at commit 363d8a2.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[srowen]

It's pretty trivially faster, I'd imagine, but the logic looks sound and it's not more complex

[kiszk]

Can we use a kind of this optimization in HotSpot?

[MaxGekk]

@kiszk The ops are not regular arithmetic / and %, they are lib calls. So, they present in byte code:

public microsToInstant(J)Ljava/time/Instant;
    // parameter final  us
   L0
    LINENUMBER 339 L0
    LLOAD 1
    LDC 1000000
    INVOKESTATIC java/lang/Math.floorDiv (JJ)J
   L1
    LSTORE 3
   L2
    LINENUMBER 340 L2
    LLOAD 1
    LDC 1000000
    INVOKESTATIC java/lang/Math.floorMod (JJ)J

[maropu]

The fix looks fine to me. If you have performance numbers, could you put them in the PR decription?

[kiszk]

@MaxGekk Got it. This is not a pair of div and mod.

[SparkQA]

Test build #118122 has finished for PR 27491 at commit 38cd6da.

• This patch fails due to an unknown error code, -9.
• This patch merges cleanly.
• This patch adds no public classes.

[MaxGekk]

jenkins, retest this, please

[maropu]

Looks fine to me. I'll leave this to @srowen and the other committers. cc: @dongjoon-hyun @HyukjinKwon

[SparkQA]

Test build #118140 has finished for PR 27491 at commit 38cd6da.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[srowen]

Merged to master (leaving out of 3.0)