does monero blockchain import tool support symlink, it seems not: file not found

> because p2pool can generate transactions with much more than 16 outputs

Isn't the number of outputs limited to 16, or is that only for non-coinbase transactions?

non-coinbase

the limit was imposed due to how bulletproofs are constructed

I got AES working with jamtis address tags, but... my test show it is 10x slower than blowfish for decrypting a single block (blowfish is a 64-bit block, AES is 128-bit, so it's 5x slower per bit).

> the limit was imposed due to how bulletproofs are constructed

Interesting, I always thought it was so that decoy flooding attacks were harder to pull off

Ok I tried tiny AES ( github.com/kokke/tiny-AES-c ) but only got ~20% faster. Looking at mbedtls and openssl but their code looks really annoying to extract... hmm

Is twofish on table or no?

Sorry if this discussion was already had

*on the

Or cha cha?

AES is definitely the most battle hardened

twofish is a possibility, the code just looks like a pita so I've been avoiding it lol

all this c makes my head hurt

(I found better code with a license that can actually maybe be used... pending)

> (I found better code with a license that can actually maybe be used... pending)

For twofish?

> all this c makes my head hurt

I really felt that

yeah twofish

jeffro256[m]: whoo twofish is way faster than aes (and probably a lot of optimizations on the table in this code I found)

Yay! Twofish is a "newer" (1998) cipher but it has stood up pretty well so far

Should blow AES out of the water in terms of performace. Also it's very tunable!

I don't think we have any twofish code in the code tree though

But didn't you mention that we didn't have AES encryption code anyways either?

we have oaes (open aes)

that's what I tested first, then tiny aes which was faster

I have already added blake2b, blowfish, siphash, tiny aes, and now twofish to my branch (most of those can be removed). Getting good at this lol

Is your branch on your GitHub (I can test out performace later if you'd like)

How much better was Twofish? I'm kind of surpised Twofish did so well since a lot of hardware has acceleration for AES

well the aes I was using is just raw c

the perf test I run: ./build/Darwin/seraphis_lib/release/tests/performance_tests/performance_tests --filter=\*test_remote_scanner_client_scan_sp\* --stats

implementation of the ciphers (check cipher context): github.com/UkoeHB/monero/blob/serap…raphis/jamtis_address_tag_utils.cpp

I'm not doing a direct performance test of the ciphers (that would probably be more robust).

jeffro256[m]: twofish was ~5x faster than my fastest run with tiny aes (using a 16-byte key)

Wow that's pretty good!

The real test would be to compare it against a heavy-hitter native library like OpenSSL's implementation

For reference, I compared the speed of our in-tree MD5 hash function to OpenSSL's on my machine and OpenSSL was 4x faster

I'll definittely take a look later. right now I'm trying to finish this homework before midnight lol

no worries and no rush :) it's pretty plug-and-play for this usecase

Great job on the GUI, can confirm long close-times are fixed.

UkoeHB AES can't be slower, CPUs have dedicated instructions for it. You probably tested software implementation

jberman[m]: 8046, do you want the first commit removed then if it seldom applies ?

this person keeps opening nonsense issues and then complains when I close them lol monero-project/monero #8303

<selsta> "this person keeps opening..." <- Tell them to go back to Bitcoin if they hate updates so much

Any time you think things can't possibly get any crazier, something like this comes along.

Watch 5 new coins getting created on the uncontested v18 :P

s/getting/get/

can we ban web-sharp from the repo please

-- Doesn't understand how to run tests

-- Wants a release rollback

-- doesn't read carefully, even his own writing

-- Accuses selsta of a "51%" attack by *checks notes* tagging a new release

dunno what a dumbshit like that is doing paying for a pro github acct

-- Doesn't understand make targets

-- Doubles down

-- Yet still wants to distribute their binaries as "official"

Im surprised he didnt ask to speak with the manager..

Does anyone here have any expereince with valgrind's massif tool?

I've used it a few times. Got me to kill the ugly spike in a game when switching levels. A bit hard to see what's going on for sure though. There are a couple GUIs for its outputs.

The GUIs are real helpful there, you can see the evolution of the main hogs with time.

Much more helpful than for, say, perf.

nice, i want sync/relay over clear&tor idgaf,... (full message at libera.ems.host/_matrix/media/r0/do…242462a921800a809cda94d1b2fd23fa3ab)

i will not punch port holes, so only outgoing

how?

> <@flashd:matrix.org> nice, i want sync/relay over clear&tor idgaf,... (full message at libera.ems.host/_matrix/media/r0/do…51057842d2e67d9471e9a00892357cca335)

Whoops yes jeffro, meet me over there

moneromooo Nice! I'm going to try to use to see what is taking up so much RAM with with monerod, but I keep getting invalid callstack errors when running with massif. Isn't there a callstack library that the monero code is using, and is it possible to turn if off?

unwind?

* Whoops yes dr flash*****, meet me over there

jeffro256[m]: github.com/monero-project/monero/bl…91da9770/src/common/stack_trace.cpp

do you mean that?

Hmm. Not sure. I have a custom patch for stack trace, never got PR'd. Might help, but IIRC these things tend to stop in a huff very easily...

If you look in my crash branch, git log and grep for "stack" and you'll find it.

It uses boost's stack trace system IIRC.

Make sure you build with -g, don't use -s at link time, -fno-omit-frame-pointer might help, though I think it's off at -O2 and below...

-g should not be needed per se, but doesn't hurt.

I had to do a non -static build for the tag though

so it would link with libunbound from packages

catering to liberssl

yet master has unbound removed afaik (submodule)

libressl*

(note that I dunno whether this patch would help, I'm not sure why boost and/or unwind and/or builtin give good traces or not)

E exceptin in main! failed to init p2p server... why?

--anonymous-inbound= is not eaten it seems but why

how to get verbose logging

* verbose logging (where is monerod.log file?)

Anonymous inbound have to set an unused port, not 18080

Use 18084

18083 is used for p2pool

Log file should be in .bitmonero folder in home directory

w[m]: no more output comapred to console print = bad!

Hey, wrong chat room!

i use 18083 p2p only (no 18081 for wallet rpc for now)

This is -dev

hey dev, why --tx-proxy= mandatory for p2pserver?

<moneromooo> "jberman: 8046, do you want the..." <- responded, didn't consider the case of reusing across attempts, approved as is