With PQ, Performance degradation observed for range predicates returning fewer rows

Issue:

-------------
Table has 10L rows with column "a" ranging from 1 to 10L
select * from query where a<1000 ;
With PQ (100) , performance degradation seen when query returns fewer rows.

Note:

Perf improvement seen with a>1000 condition

Query:

analyze select * from t2 where a<1000;

serial: 0.001 (1ms)
parallel(100) : 0.009 (9ms)
perf degradation: 800% (9 times slower)

For analyze output refer:
serial : analyze-json.txt
Parallel : analyze-json-PQ100.txt

Note:

parallel(2,4,8) : 2ms – analyze-json-PQ2.txt
parallel(16) : 3ms – analyze-json-PQ16.txt
parallel(32) : 4ms – analyze-json-PQ32.txt
parallel(64) : 6ms – analyze-json-PQ64.txt
parallel(100) : 9ms --analyze-json-PQ100.txt

Version details:

----------------------
Branch: bb-13.2-MDEV-40012-refactored-input3
Commit:1648b516083ccdbddeb123dc6cf2c2f62d7b0328
Version:13.1.0

How to repro:

create table t2( a int);

create unique index i1 on t2(a);

insert into t2 select seq from seq_1_to_1000000;

//Run query with 0 and 100 threads and note the difference

set parallel_worker_threads=0;

analyze select * from t2 where a<1000;