Topic: These new EFFICIENT x11 algos everyone is talking about ?? BULLSHIT or real? - page 15. (Read 16291 times)

I believe the reason why x11 is rising in popularity is simply the fact it uses less heat and power. All algos are asic resistant, until someone makes a ASIC for that algo, so that's for me is irrelevant. It hashes 3-4x faster compared to scrypt, but again irrelevant, as everyone else is hashing "faster" as well.

the problem here is that the persons able to build a far more efficient miner that uses the cards full capacity have no reason to release it. It hasn't been proven either way yet. However i think more time is required before we see a mass exodus from scrypt to x11.

Is x11 more open to asics than say QRK for instance since theirs is random not sequential.

Let's examine the differences between QRK and x11, does the mining the qrk algo seem to use a lot less electricity and produce a lot less heat?

I mean what exactly does x11 provide in terms of security and efficiency over and above qrks ??

Explain it to me so i can see exactly why everyone is wishing to jump to x11?  

Also why has scrypt jane lost favour to x11?  

If you were a developer now which algos would you be going for? and why?

Has x11 really had enough time to be tested?  and if there is nothing clearly superior about it over those that came before what is the point of it exactly.

Would tagging on a few more algos for x15 make it even better?

If we are really talking asic resistance x11 is nothing, i have heard countless times it would have been a lot easier to produce an x11 asic than even a scrypt asic. It has been said it would be far easier for an individual or small group of individuals to produce and x11 asic ?

So for now we really only have those with high N factor that seem resistant for now...

I won't! I will probably PM him/her if he/she is still interested when I get home from school tomorrow.

Please don't talk about sgminer optimizations here. This topic is about the inner working of X11 whether it is architectural limited in ressource consumption or if this is just what the masses believe.

Here's the right thread for such questions: https://bitcointalksearch.org/topic/ann-sph-sgminer-multi-coin-multi-algorithm-gpu-miner-added-marucoin-475795

No offense, just don't want to see this interesting thread going offtopic
Thanks!

I'm away from my computer, do not remember. Will try to post ASAP though.

What are your parameters ?

Sgminer does not work with me. I assume its same parameters as cgminer, doesn't work. Put the same stratum info into darkcoins miner, and bloop! Mining hiro xD

+1

GPU miner and the source if somebody is up for some "optimizations" : https://bitcointalksearch.org/topic/ann-sph-sgminer-multi-coin-multi-algorithm-gpu-miner-added-marucoin-475795

Every algorithm in X11 are able to create by ASIC since day one, and it is not high memory constraint like Scrypt. The darkcoin dev also said.
Here is my thread : https://bitcointalksearch.org/topic/why-i-say-x11-and-sha3-are-not-asic-resistent-540160

I'm too tired atm, perhaps I didn't understand what you mean ...
Well, it could be hashing "at full capacity" with less heat, because the hash code is different. There are parts of the chip left unused, the code itself doesn't demand operations from what's memory related. There are not random addressing jumps like scrypt. Those parts just sit there because they just have nothing to do, while the arithmetic core is already at 100% usage.

I'll try to give a counter-example; the so called "CPU only" Heavycoin where the devs struggled to destroy paralellism and prevent GPU mining. It took 2 weeks(?) for the first heavycoin GPU miner to be mentioned. cgminer-heavy work-in-progress (that one needed to compile from source) could only reach 7(?) Mh/s on a 280x in the very beginning, then it raised to 11, to 15 Mh/s and I don't know how much it is now. Christian's ccminer could reach 13 Mh/s on a 750Ti and for a few hours it beat a R9-290. Reorder had to program, operate the pool, go on with his life, etc... Was someone holding the R9-290 speed vs a simpler 750Ti due to a conspiracy? No, of course not. It takes massive brain power, time, skill and personal effort to optimize miner code. Now, go and mine Heavycoin at the maxed out performance, and look at the temperatures.

Of course, I'm not trying to distract from the possibility that there are secret X11 miners out there and a few guys are mining at much higher speed than the rest of us. It could be! There have been lots of suspicions (or even confirmations) regarding private miners or optimizations during last months. However, that's a different discussion from what I'm trying to get at here.

I agree with some of what has been said. As an example, GPUs mining SHA-256 doesn't run as hot as those running scrypt (and also uses a smidge less electricity).

With that said, does it mean that the miner isn't optimized? In the sense that "significant" improvements can be made, part of me says yes; there may be quicker opportunities to tweak the vectorizing. But this may be limited to an extent based on how the architecture interacts (I may be missing something regarding threading, so I'm not entirely sure.). Does this mean an already more efficient miner exists and only a select few know about it and use it? Wouldn't necessarily say yes, but I would say that's more likely with X11 than scrypt.

[I also want to call shenanigans on the lower draw; it seems much lower than what would make sense. However, this is from what I can gather, haven't had the chance to peruse everything in the source (and admittedly won't take all of the effort in doing so).]

well this is exactly the point.... that miner gives the best widely KNOWN performace.............. however it seems your gpu is not working all that hard. Almost like driving a car at 1000rpm.

What if some people know how to unlock the rev limiter and step up the gears.... yes they may burn 2x the fuel but may be able to travel 10x as far.

This is what we want to find out.

Also .... what if x11 is not as secure as we all think against asics... what if there are already asics for it.

Would asics be far harder to build for scryptn, scrypt jane high n, and even qrk.  I mean what if x11 was less secure and more open to asics as well as not being any more efficient either???

That would kind of take the shine off of x11 wouldn't it. Actually x11 would look like a terrible idea.

Let's hope this is not the case.... but we need to look into it all the same. Imagine all the hard forking of all these coins having to swap back to scrypt N , scrypt jane, scrypt, QRK...

that would be terrible, let's explore x11 very very deeply before we all hop on board.

Interesting.... since i'm not "clued up" with the design and workings of gpus... let's explore that idea further. I like to learn things...

I say nothing.... i am asking.

So just that i may understand what you are saying..... scrypt is stressing parts of the card harder than these other algos. The memory side of things in simple terms. Scrypt is memory intensive so it uses full memory bandwidth etc and can force the card to it's full limits. More heat and electricity used.

You are saying perhaps these other algos because they are not memory hard and do not stress the memory to it's full potential..... however the gpu could still not process these algos any faster not because the memory is being fully used but because other parts of the card are already stressed to the max. Therefore unlikely any more optimisation is possible regardless of the mining software?

Is that what you mean? This is what i was asking about when i was saying is there a bottleneck...........i mean surely if you could increase the cards calculating potential in all other areas except the memory parts then eventually the memory would be saturated by solving these new algos right? there is some part of the card holding back it having all of it's memory resources being fully exhausted?

Is that what you mean?

But that's the point. Imagine if I optimized a game to push my GPU as hard as Furmark does just to squeeze some extra work out of it. Both of us run the game and while you get 32 FPS using the exact same hardware as I am, I'm getting 48 FPS due to my secret optimizations under the hood. I know this is a very terrible analogy, but think about it.

None of this would be a problem if pushing the GPU or limiting the GPU is a user choice, much like scrypt. Let's say I mine a Scrypt coin and get 500Kh/s on my GPU stock, but if I OC and OV some I can push it to 585Kh/s; now durring the summer months this might be a problem where you live due in part to extreme heat, increased electricity price and increased power consumption, so I go ahead and underclock 50% and undervolt to about 60% and now get 255Kh/s. This makes sense to me; what doesn't make sense is claiming 50% less heat and power consumption while still hashing at full capacity.

I think that the best miner out there at the moment is sgminer when it comes to best performance overall!

Correct me if i am wrong please.

I don't see where's a serious bottleneck or some conspiracy. Since X11 (and hefty, keccak and those others from sph-sgminer) is not memory hard, you're not stressing the memory controller, L2 caches and the ram chips.

You're used to scrypt, see it as a "reference" and say that others must not be optimized due to low power use and thermals. However, scrypt is the odd one out in the first place, a complete card fuck like Furmark or some videocard stress test.

I personally don't care about ASIC resistance, as it's going to happen eventually. It seems ASIC resistance is a balancing act, if the algo is too "hard" then the GPU will be churning at max getting like no kh/s (in comparison to other algo's). I think the best way to go with this is just chain another algo to it every so often, so naming scheme could either be x11 to x12 or x11.1 or something like that....

who can say, these kind of things with way better optimisation can often yield very high gains...

and even this is NOT the point.... if everyone was then given the new miner that could utilise the full potential of your gpu... then it would be pointless because then everyone will mine using the same power and temps as scrypt? and there goes  any temps and efficiency argument for x11 or the other less memory hard algos.... this bullshit about efficient algos is just a myth?

when people are screaming about more efficient algos...... are they actually screaming these algos leave it open for average miners to get raped by those who know how to make better mining software.

So if we debunk the efficiency claims... we are left with the security claims...


Is x11 more secure than other chained algos?  is having the sequential just as secure as having them random?

What about we tag on some more algos make it x15.... would that be more secure.

If we are talking asic resistant is there point tagging more algos or just go with scrytn or scrypt jane with high N factor?

So far i am starting to see zero reason for any coin to use x11. I hope i am wrong because i see a lot of coin thread including some i have previously even suggested they should add x11.... now i see i was probably repeating a load of crap that i just heard and assumed since everyone was saying it and there were no people coming forwards strongly to say that this new mishmash was all marketing hype and no substance at best and at worst an opportunity for normal miners to be put at a gross disadvantage.

Yeah I get it now, as GPUZ claims on Avg. 89% loads, but WU in cgminer for me stays the same. If it could be better coded, what do you think the hash rate improvements would be? Also as stated before, what would the consumption increase be like?

+1 We feel the public miner is not utilizing 100% of the GPU, hence the lower power usage. We are positive the miner can be better coded to use more GPU power, if it hasn't already.