RFR: 8352075: Perf regression accessing fields [v28]

[Radim Vansa]

On Tue, 10 Jun 2025 13:02:15 GMT, Coleen Phillimore <coleenp at openjdk.org> wrote:

>> src/hotspot/share/utilities/packedTable.hpp line 56:
>> 
>>> 54:     // Packed table does NOT support duplicate keys.
>>> 55:     virtual bool next(uint32_t* key, uint32_t* value) = 0;
>>> 56:   };
>> 
>> Does it make sense to take the cost of an indirect call for each kv pair? You can't inline it, so the stack frame needs to be popped and pushed, and you're taking 2 registers (16 bytes) to give 8 bytes and 1 bit of information.
>> 
>>  We can amortize the cost by implementing this signature instead:
>> 
>> 
>> virtual uint32_t next(Pair<uint32_t, uint32_t>* kvs, uint32_t kvs_size);
>
> This was done this way with a "Supplier" because this package would be useful for other Unsigned5 packed types.

But then you'd need create an array of `Pair` in `create_search_table` and copy the data into that. You wouldn't need a Supplier at all, just pass the array and its length to the `fill` method. If you are worried about the virtual calls, I could do that.

Alternatively, we could replace virtual calls with templated `fill` method. In that case the Comparator should be a template method as well, since that one is even more perf-critical?

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/24847#discussion_r2137952293