* Switch mapRelay to use shared_ptr<CTransaction>
* Switch the relay code to copy mempool shared_ptr's, rather than copying
the transaction itself.
* Change vRelayExpiration to store mapRelay iterators rather than hashes
(smaller and faster).
Saves about 10% of application memory usage once the mempool warms up. Since the
mempool is DynamicUsage-regulated, this will translate to a larger mempool in
the same amount of space.
Map value type: eliminate the vin index; no users of the map need to know which
input of the transaction is spending the prevout.
Map key type: replace the COutPoint with a pointer to a COutPoint. A COutPoint
is 36 bytes, but each COutPoint is accessible from the same map entry's value.
A trivial DereferencingComparator functor allows indirect map keys, but the
resulting syntax is misleading: `map.find(&outpoint)`. Implement an indirectmap
that acts as a wrapper to a map that uses a DereferencingComparator, supporting
a syntax that accurately reflect the container's semantics: inserts and
iterators use pointers since they store pointers and need them to remain
constant and dereferenceable, but lookup functions take const references.
269a440 Add test for dbwrapper iterators with same-prefix keys. (Matt Corallo)
6030625 test: Add more thorough test for dbwrapper iterators (Wladimir J. van der Laan)
84c13e7 chain: Add assertion in case of missing records in index db (Wladimir J. van der Laan)
a886dbf Use std::atomic for fRequestShutdown and fReopenDebugLog (Pieter Wuille)
16cf85f Revert "Include signal.h for sig_atomic_t in WIN32" (Pieter Wuille)
This reduces the rate of not founds by better matching the far
end expectations, it also improves privacy by removing the
ability to use getdata to probe for a node having a txn before
it has been relayed.
Previously the benchmark code used an integer division (%) with
a non-constant in the inner-loop. This is quite slow on many
processors, especially ones like ARM that lack a hardware divide.
Even on fairly recent x86_64 like haswell an integer division can
take something like 100 cycles-- making it comparable to the
runtime of siphash.
This change avoids the division by using bitmasking instead. This
was especially easy since the count was only increased by doubling.
This change also restarts the timing when the execution time was
very low this avoids mintimes of zero in cases where one execution
ends up below the timer resolution. It also reduces the impact of
the overhead on the final result.
The formatting of the prints is changed to not use scientific
notation make it more machine readable (in particular, gnuplot
croaks on the non-fixedpoint, and it doesn't sort correctly).
This also hoists out all the floating point divisions out of the
semi-hot path because it was easy to do so.
It might be prudent to break out the critical test into a macro
just to guarantee that it gets inlined. It might also make sense
to just save out the intermediate counts and times and get the
floating point completely out of the timing loop (because e.g.
on hardware without a fast hardware FPU like some ARM it will
still be slow enough to distort the results). I haven't done
either of these in this commit.
Fixing formatting
Adding test case into automatically generated test case set
Clean up commits
removing extra whitespace from eol
Removing extra whitespace on macro line
Pieter Wuille