sipa changed the topic of #bitcoin-wizards to: This channel is for discussing theoretical ideas with regard to cryptocurrencies, not about short-term Bitcoin development | http://bitcoin.ninja/ | This channel is logged. | For logs and more information, visit http://bitcoin.ninja
Chris_Stewart_5 has quit [Ping timeout: 244 seconds]
alpalp has joined #bitcoin-wizards
rusty2 has joined #bitcoin-wizards
Chris_Stewart_5 has joined #bitcoin-wizards
alpalp has quit [Ping timeout: 250 seconds]
alpalp has joined #bitcoin-wizards
alpalp has quit [Changing host]
alpalp has joined #bitcoin-wizards
toffoo has joined #bitcoin-wizards
roidster has joined #bitcoin-wizards
roidster is now known as Guest77060
Guest77060 is now known as roidster
jannes has quit [Quit: Leaving]
Chris_Stewart_5 has quit [Ping timeout: 258 seconds]
NewLiberty_ has joined #bitcoin-wizards
afk11 has quit [Ping timeout: 260 seconds]
NewLiberty has quit [Ping timeout: 250 seconds]
Ylbam has quit [Quit: Connection closed for inactivity]
afk11 has joined #bitcoin-wizards
afk11 has joined #bitcoin-wizards
afk11 has quit [Changing host]
NewLiberty_ has quit [Ping timeout: 260 seconds]
Chris_Stewart_5 has joined #bitcoin-wizards
dEBRUYNE has quit [Quit: Leaving]
waxwing has quit [Ping timeout: 276 seconds]
waxwing has joined #bitcoin-wizards
EtherPro has joined #bitcoin-wizards
<EtherPro>
Slockit released the Ethereum Computer.. with the ETHOS seems cool check this out: http://ether-computer.com/
Burrito has quit [Quit: Leaving]
belcher has quit [Quit: Leaving]
N0S4A2 has quit [Quit: WeeChat 1.5]
<nsh>
lol
Noldorin has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<smooth>
i wish i were brave enough to click that link and see what lulz await
_anthony_ has joined #bitcoin-wizards
stqism has quit [Quit: Like 3 fire emojis lit rn
coup_de_shitlord has joined #bitcoin-wizards
<bsm1175321>
smooth: curl is your friend
<gmaxwell>
curl has had remotely explotable vulnerablities in the past.
<bsm1175321>
wget then? <!-- Mirrored from slock.it/ethereum_computer.html by HTTrack Website Copier/3.x [XR&CO'2014], Mon, 20 Jun 2016 03:09:56 GMT -->
drzw has quit []
King_Rex has quit [Remote host closed the connection]
N0S4A2 has joined #bitcoin-wizards
ghtdak has quit [Ping timeout: 276 seconds]
coup_de_shitlord is now known as stqism
stqism has quit [Changing host]
stqism has joined #bitcoin-wizards
b-itcoinssg has quit [Quit: Connection closed for inactivity]
waxwing has quit [Ping timeout: 276 seconds]
stqism has quit [Quit: Like 3 fire emojis lit rn
coup_de_shitlord has joined #bitcoin-wizards
coup_de_shitlord is now known as coup
coup is now known as stqism
stqism has quit [Changing host]
stqism has joined #bitcoin-wizards
hashtag has quit [Ping timeout: 260 seconds]
alpalp has quit [Ping timeout: 249 seconds]
Chris_Stewart_5 has quit [Ping timeout: 244 seconds]
bit2017 has joined #bitcoin-wizards
priidu has quit [Ping timeout: 244 seconds]
adamg has joined #bitcoin-wizards
waxwing has joined #bitcoin-wizards
zooko has joined #bitcoin-wizards
CrazyTruthYakDDS has joined #bitcoin-wizards
PERSIAN has joined #bitcoin-wizards
netsinn has joined #bitcoin-wizards
ghtdak has joined #bitcoin-wizards
c0rw1n- has quit [Ping timeout: 244 seconds]
tunafizz has joined #bitcoin-wizards
pro has quit [Quit: Leaving]
mrpocoyo has quit [Ping timeout: 244 seconds]
nuke1989 has quit [Remote host closed the connection]
<smooth>
1
_anthony_ has quit [Quit: WeeChat 1.3]
Sosumi has joined #bitcoin-wizards
netsinn has quit [Remote host closed the connection]
freekevin has joined #bitcoin-wizards
PERSIAN has quit [Ping timeout: 276 seconds]
Jeremy_Rand_2 has joined #bitcoin-wizards
Jeremy_Rand_2 has left #bitcoin-wizards ["Konversation terminated!"]
maaku_ has joined #bitcoin-wizards
maaku has quit [Ping timeout: 244 seconds]
frankenmint has joined #bitcoin-wizards
mrpocoyo has joined #bitcoin-wizards
mrpocoyo has quit [Ping timeout: 244 seconds]
bramc has joined #bitcoin-wizards
TheSeven has quit [Ping timeout: 260 seconds]
TheSeven has joined #bitcoin-wizards
<bramc>
I just had the rather embarrassing realization that I've been assuming that memory lookups act like disk lookups but faster, with the normal behavior being that whenever you pull a memory location into cache its whole neighborhood gets pulled with it. Reading over how memory caches work on wikipedia now, I'm not sure if that's true.
<bramc>
Or rather, I'm not sure how close to true it is.
<bramc>
The thing I've been working on makes perfect sense as an on-disk data structure
<bsm1175321>
4kb for disk, 64 bytes for L1, according to this.
<katu_>
bramc: usually, the kernel speculatively reads ahead
<katu_>
between 128k and several megabytes, depending on io pressure and free buffer cache
<bramc>
For disks the local affinity is even stronger because the whole read or write can be done at once, even if it spans multiple pages
<bsm1175321>
bramc: Yes adjacent reads on spinning rust are faster. That shouldn't be true on RAM or flash though.
<bsm1175321>
According to /proc/cpuinfo, ache_alignment is 64 which I think is the relevant value for me.
<katu_>
well, cachelines work slightly differently
<katu_>
depends if the dram channel is free or not
<katu_>
if it is, they're fetched in parallel, otherwise cacheline waits
<katu_>
ie a lot like raid stripes
<bramc>
My whole thing is based on there being a 'block', which is an allocated chunk of memory which generally has the property that if you some of it into cache then usually most of it gets pulled into cache
<katu_>
with ram, that can be assumed only on cacheline granularity
<bramc>
And cachelines are extremely small
<katu_>
not necessarily
<bsm1175321>
Meanwhile /me just wrote several paragraphs about how to make block-validation time irrelevant...
<katu_>
for example gpus do large fetches, up to 64kb
<katu_>
because high the massive bandwidth they have can afford it
<katu_>
and this avoids contention for channel between threads
<bramc>
What about CPUs? wikipedia says their cache lines are about 128 bytes
<katu_>
bramc: can be anything between 16 and 256 bytes
<bramc>
katu_: I'd call that extremely small
<katu_>
64 or 128 is usually assumed as average for desktop cpus.
<bsm1175321>
Nvidia Fermi and Kepler have 128-byte cache lines.
<bsm1175321>
I'm guessing this is the bus width...
<bramc>
Although maybe for in-memory stuff the cache misses don't matter so much and my stuff should be assumed to be optimized for on-disk performance and high memory efficiency.
<katu_>
bsm1175321: nah, there actually is cacheline this small at lowest level
<katu_>
but it majority of latency is the thread/global cache contention (a bit like l2, except l2 on cpus does not have radically higher granularity), not l1
<katu_>
bramc: depends entirely how much you stream. if its random probing accesses like scrypt, depends on cache a lot
<katu_>
if its some proof of space scheme spanning several gigabytes, cachelines wont matter that much, imo (assuming you read in sequence most of the time)
netsinn has joined #bitcoin-wizards
<bramc>
My favored proof of space scheme involves very simple localized reads, but that's not what I'm working on at the moment. At the moment I'm working on a merkle tree.
<katu_>
depends on how you organize it
<bsm1175321>
bramc: There's an optimization here that maybe you're not considering... We can group the contents of the mempool into the same page (or nearby pages) so that when a block comes, there's a very high chance the majority of it will be in the mempool-optimized bucket of the hash, and have few cache misses.
netsinn has quit [Remote host closed the connection]
<katu_>
some sort of binary heap, skiplist, btree
<katu_>
skiplist would be probably worst
netsinn has joined #bitcoin-wizards
<bramc>
katu_: I'm working on a merkle tree. These things suck.
<katu_>
no idea what nature of lookups you do there so its hard to say
<katu_>
but i'd not rely on random access having high latencies much
<bsm1175321>
katu_: It's basically the worst possible data structure for caching.
<katu_>
that can be easily defeated by high bandwidth+high latency+high parallelism
<katu_>
ie just like it happened for scrypt
<bramc>
katu_: This is for utxo commitments, not proofs of work
<katu_>
oh
<bramc>
So in this case latencies are all bad.
<bsm1175321>
bramc I think you can get a whole long way by organizing the UTXO set using a probabilistic model: In mempool is highest priority. Known re-use addresses as second priority. Satoshi's coins as lowest priority.
<bsm1175321>
Use that to assign buckets in your hash table.
<bramc>
bsm1175321: I'm assuming everything isn't in the mempool, trying to make the worst case not a problem.
<bsm1175321>
bramc: Ok, dig your own grave. ;-)
<bsm1175321>
Meanwhile I'm working on the time required to verify a commitment not matter. ;-)
<bsm1175321>
Also, grammar.
<katu_>
i dont see this that bad, assuming you have reasonable index of whatever you look up often
<bramc>
Somebody else can worry about mempool optimizations. They should already be in there.
<bramc>
I want to assume a totally random set of stuff in the utxo set gets spent every block
<bsm1175321>
Why assume totally random? You've just created the problem...
<bramc>
Because if it works in the worst case, it works in every case
<bsm1175321>
FWIW, I had hoped that by allowing blocks to have multiple parents and moving to a DAG instead of a linked list, I could allow a "block" to be a transaction. (the extreme of random access) However I've found that without extreme gymnastics, the tx rate cannot be more than about 100x faster than Bitcoin (e.g. ~ 6s block time) in order to achieve consensus.
<bsm1175321>
In other words, at today's transaction rates of ~1000 tx/block, you're already in the regime where doing away with a mempool and the p2p layer is impossible, for consensus reasons.
<bsm1175321>
As you and I have discussed before, this is fundamentally related to the size of the Earth, and a ~few second round trip time. So optimizing for the mempool, or whatever p2p tells you, is a shortcut that's relevant now and will become more relevant in the future.
<bramc>
Right now I'm very narrowly focused on Bitcoin as it exists today
<bsm1175321>
Can't make caches faster with software. We need to fix Satoshi's design flaw. The network is not synchronous.
mrpocoyo has joined #bitcoin-wizards
<bsm1175321>
You're optimizing for an operation that needs to be zero time, because the assumption is that the network is synchronous, and it only becomes synchronous in the limit that (validation time) -> 0. It's the assumption that's flawed, and Satoshi's original design. There are other ways to fix it.
roidster has quit [Ping timeout: 240 seconds]
ThomasV has joined #bitcoin-wizards
N0S4A2 has quit [Quit: WeeChat 1.5]
arubi has quit [Ping timeout: 240 seconds]
arubi has joined #bitcoin-wizards
ThomasV has quit [Ping timeout: 250 seconds]
bramc has quit [Ping timeout: 250 seconds]
netsinn has quit [Remote host closed the connection]
netsinn has joined #bitcoin-wizards
bustd_soket has quit [Read error: Connection reset by peer]
laurentmt has joined #bitcoin-wizards
laurentmt has quit [Client Quit]
Ylbam has joined #bitcoin-wizards
netsinn has quit []
Ylbam has quit [Ping timeout: 260 seconds]
lmatteis has quit [Ping timeout: 260 seconds]
jl2012 has quit [Ping timeout: 260 seconds]
bustd_soket has joined #bitcoin-wizards
CrazyTruthYakDDS has quit [Quit: Connection closed for inactivity]
paveljanik has quit [Quit: Leaving]
Guyver2 has joined #bitcoin-wizards
rusty2 has quit [Ping timeout: 260 seconds]
frankenmint has quit [Remote host closed the connection]
lmatteis has joined #bitcoin-wizards
damethos has joined #bitcoin-wizards
ThomasV has joined #bitcoin-wizards
frankenmint has joined #bitcoin-wizards
lmatteis has quit [Remote host closed the connection]
lmatteis has joined #bitcoin-wizards
Alanius_ is now known as Alanius
bustd_soket has quit [Ping timeout: 260 seconds]
<JackH>
so CSV has locked in now
bustd_soket has joined #bitcoin-wizards
Ylbam has joined #bitcoin-wizards
Giszmo has quit [Quit: Leaving.]
dEBRUYNE has joined #bitcoin-wizards
jl2012 has joined #bitcoin-wizards
ManfredMacx has joined #bitcoin-wizards
ThomasV has quit [Ping timeout: 276 seconds]
Davasny has joined #bitcoin-wizards
cyphase has joined #bitcoin-wizards
frankenmint has quit []
MoALTz has joined #bitcoin-wizards
Guyver2 has quit [Remote host closed the connection]
ThomasV has joined #bitcoin-wizards
toffoo has quit []
nuke1989 has joined #bitcoin-wizards
rusty2 has joined #bitcoin-wizards
mrpocoyo has quit [Quit: leaving]
lmatteis has quit [Quit: Connection closed for inactivity]
afk11 has quit [Ping timeout: 276 seconds]
afk11 has joined #bitcoin-wizards
afk11 has quit [Changing host]
afk11 has joined #bitcoin-wizards
go1111111 has quit [Ping timeout: 264 seconds]
pro has joined #bitcoin-wizards
rusty2 has quit [Ping timeout: 240 seconds]
King_Rex has joined #bitcoin-wizards
hashtag has joined #bitcoin-wizards
rubensayshi has quit [Ping timeout: 240 seconds]
dEBRUYNE has quit [Quit: Leaving]
ThomasV has quit [Ping timeout: 260 seconds]
PERSIAN has joined #bitcoin-wizards
PERSIAN has quit [Client Quit]
rubensayshi has joined #bitcoin-wizards
Chris_Stewart_5 has joined #bitcoin-wizards
hashtagg has joined #bitcoin-wizards
hashtag_ has quit [Ping timeout: 260 seconds]
ThomasV has joined #bitcoin-wizards
rubensayshi has quit [Ping timeout: 240 seconds]
ruby32 has joined #bitcoin-wizards
rubensayshi has joined #bitcoin-wizards
laurentmt has joined #bitcoin-wizards
laurentmt has quit [Client Quit]
PERSIAN has joined #bitcoin-wizards
Chris_Stewart_5 has quit [Ping timeout: 272 seconds]
bsm1175321 has quit [Remote host closed the connection]
PaulCapestany has quit [Read error: Connection reset by peer]
<amiller>
kanzure, lol, i didn't remember that at all
PaulCapestany has joined #bitcoin-wizards
<amiller>
january 2012
Chris_Stewart_5 has quit [Ping timeout: 244 seconds]
Chris_Stewart_5 has joined #bitcoin-wizards
laurentmt has joined #bitcoin-wizards
laurentmt has quit [Client Quit]
Chris_Stewart_5 has quit [Ping timeout: 244 seconds]
ThomasV has quit [Ping timeout: 244 seconds]
Noldorin has joined #bitcoin-wizards
bit2017 has joined #bitcoin-wizards
murch has joined #bitcoin-wizards
dEBRUYNE has quit [Quit: Leaving]
atgreen has joined #bitcoin-wizards
ThomasV has joined #bitcoin-wizards
<bsm1175321>
I'm writing up my slides for the "On Chain Scaling" conference. They're given me one of the first talk slots, so I'd like to carefully frame the problem in my intro. What are some good references for the consequences of scaling, or points you'd like to see discussed? I intend to hit: larger blocksize, segwit, orphans, selfish mining.
<bsm1175321>
One thing I'd like some references on is the argument that larger block sizes "causes centralization".
<bsm1175321>
I'll also hit SPV mining...
<katu_>
i doubt its a primary cause people sometimes peddle it as
ratoder has quit [Read error: Connection reset by peer]
alferz has quit [Ping timeout: 240 seconds]
ratoder has joined #bitcoin-wizards
alferz has joined #bitcoin-wizards
ratoder has quit [Remote host closed the connection]
nuke1989 has joined #bitcoin-wizards
paveljanik has joined #bitcoin-wizards
paveljanik has quit [Changing host]
paveljanik has joined #bitcoin-wizards
priidu has joined #bitcoin-wizards
ratoder has joined #bitcoin-wizards
priidu has quit [Quit: Leaving]
<JackH>
!seen maaku_
<gribble>
maaku_ was last seen in #bitcoin-wizards 3 weeks, 2 days, 1 hour, 21 minutes, and 34 seconds ago: <maaku_> bsm117532: not everyone is double-digit percentages of the network
<JackH>
!seen maaku
<gribble>
maaku was last seen in #bitcoin-wizards 1 day, 1 hour, 42 minutes, and 44 seconds ago: <maaku> kanzure: thanks, great link
ThomasV has quit [Ping timeout: 240 seconds]
priidu has joined #bitcoin-wizards
dEBRUYNE has joined #bitcoin-wizards
Chris_Stewart_5 has joined #bitcoin-wizards
blackwraith has joined #bitcoin-wizards
priidu has quit [Ping timeout: 276 seconds]
PeterR has joined #bitcoin-wizards
<PeterR>
"One thing I'd like some references on is the argument that larger block sizes 'causes centralization'" -- bsm1175321
<PeterR>
Me too. I've never understood the arguments behind this.
<PeterR>
Another one is "Large miners will make big blocks which propagate slowly in order to disadvantage smaller miners."
Sosumi has quit [Quit: Bye]
go1111111 has joined #bitcoin-wizards
gmaxwell has left #bitcoin-wizards [#bitcoin-wizards]
zooko has joined #bitcoin-wizards
Chris_Stewart_5 has quit [Ping timeout: 250 seconds]
<maaku_>
JackH: looking for me?
HostFat has joined #bitcoin-wizards
ruby32d has joined #bitcoin-wizards
Cory has quit [Ping timeout: 264 seconds]
Inaltoasinistra has quit [Ping timeout: 264 seconds]
bustd_soket has quit [Ping timeout: 264 seconds]
ruby32 has quit [Ping timeout: 264 seconds]
face has quit [Ping timeout: 264 seconds]
espes___ has quit [Ping timeout: 264 seconds]
BonyM has quit [Ping timeout: 264 seconds]
ybit_ has quit [Ping timeout: 264 seconds]
bustd_soket has joined #bitcoin-wizards
pro has quit [Ping timeout: 264 seconds]
comboy_ has quit [Ping timeout: 264 seconds]
pro has joined #bitcoin-wizards
espes__ has joined #bitcoin-wizards
comboy has joined #bitcoin-wizards
ybit has joined #bitcoin-wizards
Cory has joined #bitcoin-wizards
Inaltoasinistra has joined #bitcoin-wizards
blackwraith has quit [Read error: Connection reset by peer]
blackwraith has joined #bitcoin-wizards
Chris_Stewart_5 has joined #bitcoin-wizards
face has joined #bitcoin-wizards
BonyM has joined #bitcoin-wizards
<Anduck>
PeterR: bigger blocks = more resources needed. -> those who don't have enough resources can't use bitcoin 100%-equal to others
<Anduck>
more resources by an individual. because you need the node for *yourself* to be part of bitcoin. you're not supposed to be happy to just be able to connect to some "trusted" node
<PeterR>
Yes agreed. And a reference for this would be the Cornell paper where the showed that at 4MB, 10% of the nodes would be unable to "keep up."
<PeterR>
And 50% of the nodes would be unable to keep up at 37MB.
<midnightmagic>
PeterR: Since it seems it's going to fall to me, I just wanted to ask you personally. Given the fairly significant provocation from your direction (e.g. the creation and/or promotion of cartoons where people are crushed to death; the insults and belittling; the general uncivility) I'm going to ask you to please leave here of your own accord.
<Anduck>
keep up is not enough. you need to be able to share information. if you can't share like others can, you're not 100% equal to them
<PeterR>
(midnightmagic: personal comments are off topic here)
<Anduck>
well, tbh your question would be more suitable for #bitcoin, too.. :)
<PeterR>
It was Bob's question.
<midnightmagic>
PeterR: I mean no personal disrespect. I just think that everyone else's hands appear to be politically tied as a result of the extreme provocation and oppotunistic criminality surrounding the issue. Please leave.
rusty2 has joined #bitcoin-wizards
Guyver2 has joined #bitcoin-wizards
grashopr_ has joined #bitcoin-wizards
dnaleor has quit [Quit: Leaving]
<PeterR>
Anduck: anyways, I agree that ceteris paribus bigger blocks requires more resources (centralizing). But bigger blocks is also correlated with increased usage (decentralizing).
dnaleor has joined #bitcoin-wizards
zooko has quit [Ping timeout: 240 seconds]
<fluffypony>
bsm1175321: I'm three talks after you, last one on that day
<fluffypony>
I mention braids in my talk, albeit briefly :)
<Anduck>
yes, but the model where there are lots of people paying little fees (like $0.05 per transaction) still cost the same amount of bytes. $0.05 doesn't pay for those bytes.
<midnightmagic>
PeterR: Given that you responded to my first request, I'm currently assuming you are in fact reading my requests. Your past venom and provocation have created an atmosphere of distrust, anger, resentment, and the people using your comments as weapons are in some cases opportunistically criminal. I'm trying to ask nicely out of respect for civil discourse.
<midnightmagic>
Oh well.
PeterR has left #bitcoin-wizards [requested by midnightmagic (PeterR)]
ratoder has quit [Remote host closed the connection]
<bsm1175321>
Let me count the ways...don't make an abstract state machine. Use a proof-oriented language like ML or Prolog. I really don't want to dig through the assembly language generated for your bizarre never-before-seen state machine to figure out if there are exploits. No one has time for that shit, especially not lawyers.
<bsm1175321>
OTOH how to recruit people to your project: create a never-before-seen abstract state machine that needs compiler writers (hey I took a class in that!!!)