real_or_random has quit [Server closed connection]
real_or_random has joined #bitcoin-core-dev
<phantomcircuit>
sliv3r__, i did the test that gmaxwell is talking about and it slightly improved performance but only in the extremely happy path where every transaction in the block is in the mempool
hacker4web3bitco has joined #bitcoin-core-dev
Cory65 has joined #bitcoin-core-dev
Cory67 has joined #bitcoin-core-dev
Cory49 has quit [Ping timeout: 250 seconds]
Cory65 has quit [Ping timeout: 250 seconds]
<phantomcircuit>
sliv3r__, the unhappy path it was like 2-3x slower
<phantomcircuit>
so overall i think not a good change, at least not at this time
<phantomcircuit>
sipa is working on a mempool clustering thing and we should reassess after that, maybe it'll be faster
danielabrozzoni has quit [Server closed connection]
danielabrozzoni has joined #bitcoin-core-dev
<sliv3r__>
@phantomcircuit do you have the experiment documented to take a look?
<phantomcircuit>
sliv3r__, uh
<phantomcircuit>
looks like no
<sliv3r__>
:'( it's ok, thanks!
<phantomcircuit>
i think i just changed PartiallyDownloadedBlock::InitData to iterate over one of the mempool boost multindexes instead of the txns_randomized vector
<phantomcircuit>
sliv3r__, i seem to recall it was annoying to figure out how the multi index container thing works but once i had that it was trivial and of course i cant remember anymore
<sliv3r__>
noted, will check, thx
<gmaxwell>
the decode speed appeared to be clearly problematic though, I had assumed that it was all due to the fact that it had to iterate everything now due to missing transactions, hadn't considered the fact that more recent changes may have just made it much slower in general
<gmaxwell>
(and I guess while I say clearly problematic the decode time is still kinda of irrelevant compared to missed txn)
<bitcoin-git>
[bitcoin] frankomosh opened pull request #33252: fuzz: add target for `DifferenceFormatter` and invariant check in `net_processing.cpp` (master...fuzz-differenceFormatter) https://github.com/bitcoin/bitcoin/pull/33252
cmirror has quit [Remote host closed the connection]
cmirror has joined #bitcoin-core-dev
robszarka has joined #bitcoin-core-dev
szarka has quit [Ping timeout: 256 seconds]
Cory12 has joined #bitcoin-core-dev
Cory67 has quit [Ping timeout: 250 seconds]
<phantomcircuit>
gmaxwell, i seem to recall my basic thought was that it didn't actually matter that much compared to reducing tx misses
<phantomcircuit>
and because sipa was changing stuff there anyways any benchmark would be outdated soon enough
<_aj_>
gmaxwell: doing a cache optimisation and not having either a comment in the code as to what it's for, nor a benchmark to measure its effectiveness or catch regressions seems like the bigger issue than C++...
shiza has quit [Server closed connection]
shiza has joined #bitcoin-core-dev
Guest42 has joined #bitcoin-core-dev
Guest42 has quit [Client Quit]
michaelfolkson has quit [Ping timeout: 258 seconds]
Lightsword_ has joined #bitcoin-core-dev
JTL2 has joined #bitcoin-core-dev
oribellow_ has joined #bitcoin-core-dev
JTL has quit [Killed (NickServ (GHOST command used by JTL2))]
michaelfolkson has joined #bitcoin-core-dev
JTL2 is now known as JTL
Earnestly has quit [Ping timeout: 258 seconds]
oribellow has quit [Ping timeout: 258 seconds]
Lightsword has quit [Ping timeout: 258 seconds]
entropyx has quit [Ping timeout: 258 seconds]
Lightsword_ is now known as Lightsword
entropyx has joined #bitcoin-core-dev
Earnestly has joined #bitcoin-core-dev
oribellow_ is now known as oribellow
<gmaxwell>
_aj_: "a cache optimization" -- come on. Any time you add more pointer chasing it's a potentially performance hit, unless none of it is in a performance relevant path. Are you suggesting the involved contributors didn't know block reconstruction was performance relevant? I think it's a lot more likely that no one considered that the change would have a performance consequence at all.
<_aj_>
gmaxwell: i don't think those two scenarios are meaningfully different, or that it being in a language other than C++ would've helped in either case
<_aj_>
gmaxwell: prioritising clean/simple code over performance optimisations is extremely common; so (a) not being aware of a technique that's likely to be important to performance; (b) not having a good judgement of where an improvement is important; and (c) prioritising other things even knowing there could be some effect on performance all seems like things to expect to see from open source
<_aj_>
contributors, unless you have a closed repo like sqlite or something i guess
<gmaxwell>
I withdraw the comment, because that change wasn't actually obfscuated by an interface in this case.
<_aj_>
haha, okay i get where you were coming from now
w0xlt has joined #bitcoin-core-dev
stringintech has joined #bitcoin-core-dev
stringintech has quit [Ping timeout: 250 seconds]
takinbo has quit [Server closed connection]
takinbo has joined #bitcoin-core-dev
gnusha_ has quit [Server closed connection]
gnusha has joined #bitcoin-core-dev
PaperSword has joined #bitcoin-core-dev
pinheadmz has quit [Server closed connection]
pinheadmz has joined #bitcoin-core-dev
yuvicc has joined #bitcoin-core-dev
abubakarsadiq has joined #bitcoin-core-dev
yuvicc has quit [Quit: yuvicc]
<bitcoin-git>
[bitcoin] maflcko closed pull request #33252: fuzz: add target for `DifferenceFormatter` and invariant check in `net_processing.cpp` (master...fuzz-differenceFormatter) https://github.com/bitcoin/bitcoin/pull/33252
w0xlt has quit [Ping timeout: 248 seconds]
kerm|t has quit [Server closed connection]
kerm|t has joined #bitcoin-core-dev
jerryf has joined #bitcoin-core-dev
jerryf_ has quit [Ping timeout: 272 seconds]
f321x has joined #bitcoin-core-dev
<bitcoin-git>
[bitcoin] maflcko reopened pull request #33252: fuzz: add target for `DifferenceFormatter` and invariant check in `net_processing.cpp` (master...fuzz-differenceFormatter) https://github.com/bitcoin/bitcoin/pull/33252
f321x has quit [Ping timeout: 272 seconds]
f321x has joined #bitcoin-core-dev
Guyver2 has joined #bitcoin-core-dev
Talkless has joined #bitcoin-core-dev
rodarmor has quit [Server closed connection]
rodarmor has joined #bitcoin-core-dev
___nick___ has joined #bitcoin-core-dev
purpleKarrot has joined #bitcoin-core-dev
___nick___ has quit [Client Quit]
___nick___ has joined #bitcoin-core-dev
jerryf_ has joined #bitcoin-core-dev
jerryf has quit [Ping timeout: 272 seconds]
FelixWeis__ has quit [Server closed connection]
FelixWeis__ has joined #bitcoin-core-dev
_durandal has joined #bitcoin-core-dev
emcy__ has quit [Ping timeout: 260 seconds]
<bitcoin-git>
[bitcoin] frankomosh closed pull request #33252: fuzz: add target for `DifferenceFormatter` and invariant check in `net_processing.cpp` (master...fuzz-differenceFormatter) https://github.com/bitcoin/bitcoin/pull/33252
<bitcoin-git>
[bitcoin] frankomosh reopened pull request #33252: fuzz: add target for `DifferenceFormatter` and invariant check in `net_processing.cpp` (master...fuzz-differenceFormatter) https://github.com/bitcoin/bitcoin/pull/33252
_aj_ has quit [Server closed connection]
maxedw has quit [Server closed connection]
maxedw has joined #bitcoin-core-dev
jetpack has quit [Quit: ZNC 1.8.2+deb2+deb11u1 - https://znc.in]
jetpack has joined #bitcoin-core-dev
f321x has quit [Remote host closed the connection]
hacker4web3bitco has quit [Ping timeout: 244 seconds]
Cory67 has quit [Ping timeout: 250 seconds]
tqfoley2 has joined #bitcoin-core-dev
Cory11 has joined #bitcoin-core-dev
Cory5 has quit [Ping timeout: 250 seconds]
aleggg has quit [Ping timeout: 248 seconds]
aleggg has joined #bitcoin-core-dev
w0xlt has quit [Ping timeout: 248 seconds]
eugenesiegel has joined #bitcoin-core-dev
<_aj_>
instagibbs: re: 33253, the benchmark doesn't exercise a8203e9 -- it just passes an empty vector of extra transactions
w0xlt has joined #bitcoin-core-dev
<instagibbs>
pasted on the PR thanks
w0xlt has quit [Ping timeout: 258 seconds]
zeropoint has joined #bitcoin-core-dev
Ademan has joined #bitcoin-core-dev
f321x has quit [Quit: f321x]
Cory61 has joined #bitcoin-core-dev
Cory11 has quit [Ping timeout: 250 seconds]
nanotube has quit [Ping timeout: 276 seconds]
jonatack has joined #bitcoin-core-dev
jon_atack has quit [Ping timeout: 255 seconds]
jon_atack has joined #bitcoin-core-dev
jonatack has quit [Ping timeout: 256 seconds]
jonatack has joined #bitcoin-core-dev
<jonatack>
If I understand GitHub protocol correctly, "Drahtbot locked and limited conversation to collaborators" just after each merge seems a little aggressive? I cannot provide a post-merge review.
jon_atack has quit [Ping timeout: 256 seconds]
<jonatack>
IIUC "collaborators" does not seem to include repository members, only those with write access.
<fanquake>
Which PR?
<fanquake>
Drahtbot generally only batch locks historical PRs, on some schedule
<fanquake>
jonatack?
jon_atack has joined #bitcoin-core-dev
jonatack has quit [Ping timeout: 248 seconds]
<jon_atack>
power went off, the off-grid setup here needs better failover