have you been able to find what's wrong with CN heavy ?

I tested with XMR stak and it also take advantage of 8gb card for heavy thus increasing the speed compared to CNv7...

I double checked with your code...... and Xmrstak ones  ... but I was unable to find anything.

Damn, I am not a good Dev!



it's a good time to introduce profit-based optimization and algorithm switching. ? is it ?

Isn't easier to send a pm to Nplus, Nemos,Tutulfo, Megaminer, Forage   and ask them how to integrate your miner in their autoswitch software ?
They normally need a command line ... and Api...

the italian translation is crappy as well.

Fatta con Google translate eh ? Mah

Well, guys, if you are in the Euro zone, that change a lot.

I 'm interested in a sizable number.
I will fly in anywhere in Europe to see samples

Kindly inform

As long as it as well other from C/N heavy will not appear in the calculation of the GPU page, they are a little useless.
Having monero as reference, it means that I will have to move to cryptunit to have a better picture.

Etash & CN are the most mined with AMD.

Does really have sense not being able to compare the two because a lot of coins of the second group aren't present in the calculation?
I understand not collecting all the data from all the coins, but having the data and not using them is weird.

So what's the point for AMD miners of visiting that page?

Better say: ok, thanks, I know you are working to improve.

You know what will be really useful ?

A portfolio.
Everthing I found online is full of token and with few coins.
I will be happy with a portfolio of mineable coins

Btw, Haven protocol isn' t listed as well as most CN heavy coins

Ok, you need comparison ?

YOUR RX480 8GB SAMSUNG.... Ubermix timing, stock for the rest with 9 hours optimize+memory timing+watchdog running)
As reference, I have 28.4 Mh/s in etash.( Optimize + memorytiming NOT extended)

GGS MoneroV7: 1034 h/s
GGS SumokoinV7:670H/s
Srbminer Monero v7 1thread 976 h/s 2threads 1033 H/s
Srbminer Sumokoinv7 1 thread 980 h/s 2 threads 988 h/s

That is what an average guy can do

but:

Rx 580 8gb Micron, PBE 1 click
29.5 Etash vs 31.9 Claymore

GGS SumokoinV7:670H/s
SRBMiner Sumokoin 1060 ( and it work at 1300 core!)

So, it seems, that V7 is in par but Heavy is not when is something YOU optimize ( or you have had in hand).
Heavy in SRB is weard, it keep moving between 980 1030 1060. Not always the same card.
Cpu kicks ?  it happened to me in GGS when using ODD numbers of threads/worksize as well.

When something isn't preoptimized, the gap is bigger in Heavy but exist also in v7 smaller

Not Bad Mr Zawawa

https://imgur.com/a/amyvaCH

Just a 9Mh/s X16r RVN mining with an RX 480 8gb Samsung..... that is 1 Mh/s more than the other miners

Hopefully the new Kernels will speed it up to a GTX 1070 Level






 

@Iamtutut.

I am not saying that the stock value are different then the one GGS place.
I am saying that the cards rx 580 micron memory that I have does give plenty (PLENTY) of errors above the value I indicated. In fact they work , and then they suddendly stop working after a while at 1300.

I honestly don't care if they give me worst performance compared to others, I will be just happy to be able to decide to myself what is best for the the card I have.

Zawawa can place a button , a thick or sign to say: go with  stock (or not in my case) if he can.

If not, in the algos that are problematic ( X16s in that case), I will use something else.

And in some case,  I will use something else ANIWAY because running a card at 1300 and not at 1130 for mining Etash will consume too much energy.
 
He have to make the best product for the majority of people, I understand that and I don't believe there are plenty of RX 580 8gb Asus Dual with Micron around ( and crappy yeld..) aniway.

Oh, last, cards are "bios modded" in the sense that I just used 1 click with polaris, nothing more.

PLEASE do not put as default 1300/2200 core/mem for some algos.
My micron rx580 doesn't work to that level, most can't go higher than 1174/2050!
Or simply  let the user choose somehow.
Right now when I launched x16s it start, reach 7M, and all card start stopping and just one out of 4 / rig keep working.




on other hand, I gave a look to github.

when you start cleaning you really do!

sorry to disturb

I can get  some (SOME full rack) CPU: 2x xeon E5420 ( 4 cores each ,12 mb l2) 2.5GHz servers reasonably priced.

Does this miner work with windows server ?
any idea +- of the potential speed with those kind of Xeon ?

So, if I understood well. Etash doesn't use on the core a lot  but its memory intensive.  In fact all the core frequency can be lowered as well as Mv applied to it.

Finding an algos that is core intensive and use low memory will optimize the resources:

So, besides what Claymore does, which other algos can fit that description ? Neoscrypt for sure but is a little bit too much energy consuming.

CN litev7 ? I will be happy with 300 h/s Cn and 28 Mhs etash.
Also in most of the 8gb card, etash doesn't use all the available memory, normally leave 4gb free..

Better choices ?

Its in function of which card you bought and how much you are paying energy and how good you are at setting up the card. So you need to study, at least a little.

In order to understand what to use and what you have to learn to setup use go to whattomine and cryptunit.


If you don't pay energy, Neoscript is also a good choice. Gobyte can be a safe bet

If you pay energy a lot , mine Cryptonight, I will say Sumokoin if you have an 8gb card and Monero if you have an 4Gb. ( don't ask me why, try and see by yourself:) )

If you want to follow what everybody does, simply mine Ethereum + Xvg with Claymore as a miner or Phoenix if you decide to mine only ethereum. Just Ethereum will bring you (TODAY!) 1$ net/day/card.

Uncle, I hope that Zawawa doesn't spend his time  on improving Etash.

Even if he reach the same level, Claymore have DM with XVG & Keccak.
At pair with the two( well, Claymore is 1.5 dev fee and Phonex its .65 %) how many people will switch ?
I personally will not. I am not mining Etash besides Pirl ( that BTW its 0 % dev fee from Claymore) but I feel like the speed with Etash Ellesmere have reached a plafond. Vega mmmm.
Isn't much easier to beat an old sgminer on Lyra 2z ?
Or retake the CN crownd ? Or "virtually mod" a stock card ?

Aniway, I was not speaking of my wishes, I was just trying to make a recap of what Zawawa said previouslyhere /hint/publish on github.

So, we do expect:
a new installer
a new etash kernel (may be for Baffin ?) I saw something on github but I'm not sure
a new x16r kernel ?
a good for all memory timing split by memory vendor ?
a lira2z miner

in the upcoming release ?

Those are good numbers for hynix

the baffin rx 550 should  give around 500 h/s with intensity 56/58 on CN and 14.3Mhs with etash, shaders unlocked , with claymore with less than 50 watt

I have not Rx 560 but its not so different just more expensive.

I am aware of some other people with rx 550 rigs besides me, I don't know if someone have a rigs with that card so ,I believe its not that THE reason  but just ONE reason.

"shaders unlocked" give 1 Mh/s more on etash" with Claymore.
I don't think that you card screening recognize it or not, but I suspect not

There is a nice 3d by branko about the RX560 here in mining altcoins where all the people its reporting their settings, if you would like to avoid starting from scratch.

If you are referring to CN heavy, I can say that your speed are good but not great most because those are 4gb card. in my little experience ONLY 8gb card go over 1000 easily.

Different story if you are speaking of normal v7 algo.


Why don't take one out of the rig, put in a pc with a watchdog and run optimize with ggs ?

I am normally lauching Nplus when I don't know WTM or I don't have time to follow (I have 4 GTX 1070 asus )

Today, first time , I looked at the calculations from 2 pm till 9 pm while I was preparing and installing another rig.

While I almost saw on the benchmark page it was mining always at >1 mbtc/day the number displayed in green is 0.6 mbtc.
The Pool cann't help as I have other 3 similar  rigs, same wallet, pointing at ahashpool all together.

I was wondering if its the benchmarking that is wrong( speed is correct) its the API coming from Ahashpool or its the fact that more than one rig share the same wallet that is creating this dicotomy.
Or its a bug, but I doubt, as once I week the total amount from the 4 rigs is meeting the expectations, Ideas ?

you know I'm not a dev and not english native.

But I can read:


The size of the SIMD VGPR files introduce an important limit: VGPRs of the SIMD are evenly divided between threads of the active waves. If a shader requires more VGPRs than are available, the SIMD will not be able to execute the optimal number of waves. Occupancy, a measure of the parallel work that the GPU could perform at a given time, will suffer as a result.

https://gpuopen.com/optimizing-gpu-occupancy-resource-usage-large-thread-groups/

So, in my limited understanding, the result will be a better occupancy thus an increase in performance from 2 % up to 10 % depends of the kernel.

and because of this:

GCN was very VGPR limited, I managed to get 3 out of a maximum of 10
threads to fit by isolating bits of code in dummy branches. Since the
algorithm is memory heavy any improvements here would help a lot, but
unfortunately although I think VGPR use could be substantially lower, the
AMD compiler likes to go crazy and use as many as possible.

I think you basically rewrite a part of the AMD compiler ?

Do not forget to query for number of (unlocked) shaders and not refer at the standard number ,as most of the Baffin series can be unlocked