I've posted updates on puzzles, CryptoNight, and community opportunities posted at CN forum.
Guys, what would you like to see covered in this FAQ? I'm quite stuck at the moment.
Topic: [BCN] Uncovering CryptoNote technology and Bytecoin BCN FAQ - page 3. (Read 13771 times)
And here I am again, confused. Another bot?
Most likely a bad translation.
Quote
bra with detectors of affection
I do think this is the best Idea I have seen on the site
Good reasons. I agree.
I personally prefer Hy-quolity products whoever developed by.
Fame can execute several tasks but it's just a tool nevertheless.
I personally prefer Hy-quolity products whoever developed by.
Fame can execute several tasks but it's just a tool nevertheless.
Can you sum up for not_so_smart people? Is Bytecoin super ASIC proof or ?
We can only talk about ASIC resistance.
To be "ASIC-proof" would mean that no ASIC implementation is possible,
because the algorithm requires way more memory than can fit on an ASIC
in the foreseeable future. Even then, an ASIC solution is possible, but it
just wouldn't be self-contained and still need to be hooked up to some DIMMs.
So I would say Bytecoin is ASIC-resistant, and significantly more so than scrypt.
I'm not a big fan of the word super, which gets abused alot.
Like all these hash-mish-mash algorithms being called "super-secure".
That's just "super-laughable"...
1) Cache sizes slowly grow over time (Moore's law). Currently, high-end x86 has 2.5MB / core.
The proof-of-work should use as much of this as possible.
You want each core running its own instance while fully utilizing its cache.
Ideally, the dynamic difficulty adjustment should be able to increase the memory requirement,
so as to keep up with hardware improvements.
The proof-of-work should use as much of this as possible.
You want each core running its own instance while fully utilizing its cache.
Ideally, the dynamic difficulty adjustment should be able to increase the memory requirement,
so as to keep up with hardware improvements.
It's interesting though. RAM is not that much slower than L3 cache,* and in fact this algorithm runs faster with two instances per core (with hyperthreading), suggesting that cache doesn't really matter here.
* RAM = 60 cycles, L3 unshared - 40 cycles. See page 22 at https://software.intel.com/sites/products/collateral/hpc/vtune/performance_analysis_guide.pdf
Thanks for the reference. That's a much smaller difference than I thought.
I just ran some tests with the simple cuckoo miner, and here's how the runtime scales with memory
on a Xeon with a total of 12MB L3 cache:
1M 0.075s
2M 0.15s
4M 0.32s
8M 0.7s
16M 1.9s
32M 5.5s
64M 13.7s
128M 31s
So we do see an extra slowdown when it crosses the L3 cache size, but it's indeed not very dramatic.
This suggests that a memory-bound proof-of-work should just try to use as much memory as possiible
to frustrate GPUs/ASICs rather than try to optimize of L3 cache size.
Can you sum up for not_so_smart people? Is Bytecoin super ASIC proof or ?
I'm not an expert either, but his main explanation was in https://bitcointalksearch.org/topic/m.6249426
and it indicates that it is harder/more expensive to implement ASIC for bytecoin pow than it was for Scrypt, and it also indicates that the ratio of how much more efficient it would be to GPU/ASIC mine it (after such things are developed) instead of CPU will also be smaller in relation to the Scrypt case.
1) Cache sizes slowly grow over time (Moore's law). Currently, high-end x86 has 2.5MB / core.
The proof-of-work should use as much of this as possible.
You want each core running its own instance while fully utilizing its cache.
Ideally, the dynamic difficulty adjustment should be able to increase the memory requirement,
so as to keep up with hardware improvements.
The proof-of-work should use as much of this as possible.
You want each core running its own instance while fully utilizing its cache.
Ideally, the dynamic difficulty adjustment should be able to increase the memory requirement,
so as to keep up with hardware improvements.
It's interesting though. RAM is not that much slower than L3 cache,* and in fact this algorithm runs faster with two instances per core (with hyperthreading), suggesting that cache doesn't really matter here.
* RAM = 60 cycles, L3 unshared - 40 cycles. See page 22 at https://software.intel.com/sites/products/collateral/hpc/vtune/performance_analysis_guide.pdf
Thanks for the reference. That's a much smaller difference than I thought.
I just ran some tests with the simple cuckoo miner, and here's how the runtime scales with memory
on a Xeon with a total of 12MB L3 cache:
1M 0.075s
2M 0.15s
4M 0.32s
8M 0.7s
16M 1.9s
32M 5.5s
64M 13.7s
128M 31s
So we do see an extra slowdown when it crosses the L3 cache size, but it's indeed not very dramatic.
This suggests that a memory-bound proof-of-work should just try to use as much memory as possiible
to frustrate GPUs/ASICs rather than try to optimize of L3 cache size.
Can you sum up for not_so_smart people? Is Bytecoin super ASIC proof or ?
1) Cache sizes slowly grow over time (Moore's law). Currently, high-end x86 has 2.5MB / core.
The proof-of-work should use as much of this as possible.
You want each core running its own instance while fully utilizing its cache.
Ideally, the dynamic difficulty adjustment should be able to increase the memory requirement,
so as to keep up with hardware improvements.
The proof-of-work should use as much of this as possible.
You want each core running its own instance while fully utilizing its cache.
Ideally, the dynamic difficulty adjustment should be able to increase the memory requirement,
so as to keep up with hardware improvements.
It's interesting though. RAM is not that much slower than L3 cache,* and in fact this algorithm runs faster with two instances per core (with hyperthreading), suggesting that cache doesn't really matter here.
* RAM = 60 cycles, L3 unshared - 40 cycles. See page 22 at https://software.intel.com/sites/products/collateral/hpc/vtune/performance_analysis_guide.pdf
Thanks for the reference. That's a much smaller difference than I thought.
I just ran some tests with the simple cuckoo miner, and here's how the runtime scales with memory
on a Xeon with a total of 12MB L3 cache:
1M 0.075s
2M 0.15s
4M 0.32s
8M 0.7s
16M 1.9s
32M 5.5s
64M 13.7s
128M 31s
So we do see an extra slowdown when it crosses the L3 cache size, but it's indeed not very dramatic.
This suggests that a memory-bound proof-of-work should just try to use as much memory as possiible
to frustrate GPUs/ASICs rather than try to optimize for L3 cache size.
1) Cache sizes slowly grow over time (Moore's law). Currently, high-end x86 has 2.5MB / core.
The proof-of-work should use as much of this as possible.
You want each core running its own instance while fully utilizing its cache.
Ideally, the dynamic difficulty adjustment should be able to increase the memory requirement,
so as to keep up with hardware improvements.
The proof-of-work should use as much of this as possible.
You want each core running its own instance while fully utilizing its cache.
Ideally, the dynamic difficulty adjustment should be able to increase the memory requirement,
so as to keep up with hardware improvements.
It's interesting though. RAM is not that much slower than L3 cache,* and in fact this algorithm runs faster with two instances per core (with hyperthreading), suggesting that cache doesn't really matter here.
* RAM = 60 cycles, L3 unshared - 40 cycles. See page 22 at https://software.intel.com/sites/products/collateral/hpc/vtune/performance_analysis_guide.pdf
I don't see why anonymity needs to be attributed to illegality or shadiness. It's clearly a choice that was made, and should probably be respected.
You shouldn't need a face to promote this, and it's pretty well documented that you don't need one to get a 5 billion market cap.
Sure bitcoin has a visible core team now, but that's not how it started.
Someone provided something great, and it was accepted. The reason for this is because it's open source, and because it had something potentially world-changing to offer.
It wasn't accepted because I can sit around and think about what a great man Satoshi Nakamoto is, but instead what was provided by someone who's still not known until this day . . because for all I know Satoshi was just the name given so that people would stop asking questions.
If one was to try to take an unbiased, unadulterated approach to something that could very well perform on par with bitcoin . . then an approachable personal identity would only serve to get in the way of that.
You shouldn't need a face to promote this, and it's pretty well documented that you don't need one to get a 5 billion market cap.
Sure bitcoin has a visible core team now, but that's not how it started.
Someone provided something great, and it was accepted. The reason for this is because it's open source, and because it had something potentially world-changing to offer.
It wasn't accepted because I can sit around and think about what a great man Satoshi Nakamoto is, but instead what was provided by someone who's still not known until this day . . because for all I know Satoshi was just the name given so that people would stop asking questions.
If one was to try to take an unbiased, unadulterated approach to something that could very well perform on par with bitcoin . . then an approachable personal identity would only serve to get in the way of that.
Good reasons. I agree.
I personally prefer Hy-quolity products whoever developed by.
Fame can execute several tasks but it's just a tool nevertheless.
Our conversation as not such invention like bra with detectors of affection, it's about new currency. Cheap boom is unnecessarily thing for this.
Arguing for ASIC-resistance due to perceived high cost of 1MB is silly though.
(gridseed LTC ASICs already have 512KB on board)
The original claim doesn't even mention cost. It should elaborate on what they
mean by ASIC pipeline though, as that has me confused...
(gridseed LTC ASICs already have 512KB on board)
The original claim doesn't even mention cost. It should elaborate on what they
mean by ASIC pipeline though, as that has me confused...
There is something to be said for a cache-oriented proof-of-work though,
of which there are not many examples. There are several concerns:
1) Cache sizes slowly grow over time (Moore's law). Currently, high-end x86 has 2.5MB / core.
The proof-of-work should use as much of this as possible.
You want each core running its own instance while fully utilizing its cache.
Ideally, the dynamic difficulty adjustment should be able to increase the memory requirement,
so as to keep up with hardware improvements.
2) There must be no easy memory-time trade-off (like scrypt has).
Memory accesses must be dependent on earlier ones, so that must necessarily be done
in sequential order. Otherwise, a GPU/FPGA/ASIC will gain a lot from increased parallellism.
3) The amount of computation per memory access must be minimized in order for the core
to have a very high frequency of (cache) memory accesses. The point is that ASICs can
always do computation much faster, but they cannot do random memory access much faster.
If you get these things right; then a GPU will not be competitive.
Certainly, CryptoNight is an improvement on scrypt on all 3 counts above
(for item 3, mostly due to its smaller block size, 16 bytes vs 128 bytes).
Do you wonder why people who develop things like BCN or CN are staying in the shadows? It's not like it's not legal or something...
Maybe they are afraid of getting CIA'd since, clearly, they are smart. Plus not everyone is looking for a fame, even if it's a fame amongst small amount of people.
I think you must have special mindset for this kind of things - not everyone can develop something new. And probably this particular mindset is not looking for a fame.
Maybe but it's not like CIA is eating developers for breakfast/dinner/etc. they are paying salaries and not a small ones I imagine. So this is still a mystery for me. That's not a logic thing to do.
Do you wonder why people who develop things like BCN or CN are staying in the shadows? It's not like it's not legal or something...
Maybe they are afraid of getting CIA'd since, clearly, they are smart. Plus not everyone is looking for a fame, even if it's a fame amongst small amount of people.
I think you must have special mindset for this kind of things - not everyone can develop something new. And probably this particular mindset is not looking for a fame.
Do you wonder why people who develop things like BCN or CN are staying in the shadows? It's not like it's not legal or something...
Quote
Oh. Okay then. 512KB is nothing for the Bytecoin according to your calculations.
512KB is nothing in general now days. Not only for Bytecoin - there is no CPU with this amount of memory.
Arguing for ASIC-resistance due to perceived high cost of 1MB is silly though.
(gridseed LTC ASICs already have 512MB on board)
The original claim doesn't even mention cost. It should elaborate on what they
mean by ASIC pipeline though, as that has me confused...
(gridseed LTC ASICs already have 512MB on board)
The original claim doesn't even mention cost. It should elaborate on what they
mean by ASIC pipeline though, as that has me confused...
Did gridseed announce the cost of this ASIC?
Also, will CPU get bigger (in physical size) in order to have 512MB cache on board?
Sorry for annoying you, I really don't know much about it. Curiousity
Sorry; I meant 512KB, not MB:)
512KB cache is very modest for a modern CPU, but could be a large fraction
of a scrypt ASIC.
Oh. Okay then. 512KB is nothing for the Bytecoin according to your calculations.
Cache is by far the largest part of the die of any modern CPU.
Not by far. Cache area is usually close in size to Core area.
Yes I see that is the case. Thank you for the correction. My understanding was out of date or just wrong.
Cache is by far the largest part of the die of any modern CPU.
Not by far. Cache area is usually close in size to Core area.
There is no "most expensive part of the CPU". Cost is dominated by die size and fault rate.
It's true that 37.5MB takes a lot of die space, so that certainly contributes to cost.
It's true that 37.5MB takes a lot of die space, so that certainly contributes to cost.
Cache is by far the largest part of the die of any modern CPU. I don't know how that contributes to failure rate though, since individual bad cache cells might not render the die unusable.
Arguing for ASIC-resistance due to perceived high cost of 1MB is silly though.
(gridseed LTC ASICs already have 512MB on board)
The original claim doesn't even mention cost. It should elaborate on what they
mean by ASIC pipeline though, as that has me confused...
(gridseed LTC ASICs already have 512MB on board)
The original claim doesn't even mention cost. It should elaborate on what they
mean by ASIC pipeline though, as that has me confused...
Did gridseed announce the cost of this ASIC?
Also, will CPU get bigger (in physical size) in order to have 512MB cache on board?
Sorry for annoying you, I really don't know much about it. Curiousity
Sorry; I meant 512KB, not MB:)
512KB cache is very modest for a modern CPU, but could be a large fraction
of a scrypt ASIC.
yo ppl!
All this secrecy is giving me a headache seriously... All deepweb projects are so subtle ?
The only thing I heard about the deepweb (from my newbie-Bitcoin shore) is Silkroad. And it wasn't secret and subtle like, at all.
All this secrecy is giving me a headache seriously... All deepweb projects are so subtle ?
The only thing I heard about the deepweb (from my newbie-Bitcoin shore
) is Silkroad. And it wasn't secret and subtle like, at all. not a long time ago there was a thread on bitcointalk about the deepweb and it's wiki, you can try to find it.
you'll be amazed what you can find there
Well you were right. I was amazed. But not in a good way. Now I understand why people hide it.
Jump to: