Github user zuyu commented on the issue:
https://github.com/apache/incubator-quickstep/pull/98
> For the efficiency part:
1-) Using new comparison operators obviously are faster than old
counterparts using if else statements. (Of course the overall effect might not
be fast as we expected in the TPCH queries.) Thank you.
I agree there would be some improvements in the theory, but do you have the
number?
> 2-) No matter why TypedValue is used in query processing , the cutting
the size of struct representation by 50% improves the memory usage:
a-) A page can contain 50% more values.
b-) Recent x86 architectures favor using the values 32 bits or 64 bits, not
48 bits.
I don't think so. The existing represent of one `uint32_t` and two
`uint8_t`s would take 64 bits in total due to memory alignment requirements,
instead of 48 bits as what you thought. Furthermore, as we use `TypeValue`
instead of `DateLit` directly, I don't see much space improvements after
refactoring using one single `uint32_t`.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
