Where can I find genoil ethminer 1.1.6 source code?
I cant find it anywhere on github.
I cant find it anywhere on github.
https://bitcointalksearch.org/topic/ethminer-0941-genoil-11-1368785
Topic: ethminer-0.9.41-genoil-1.1 - page 28. (Read 397363 times)
https://bitcointalksearch.org/topic/ethminer-0941-genoil-11-1368785
Where can I find genoil ethminer 1.1.6 source code?
I cant find it anywhere on github.
I cant find it anywhere on github.
Thanks @Wolf0
I am now using DwarfPool and their Stratum Proxy (found nothing else useable?) and it's working.
I am now using DwarfPool and their Stratum Proxy (found nothing else useable?) and it's working.
You can connect directly to dwarf using stratum support of genoil miner, I personally don't recommend it due to issue with connection and rejected shares as discussed previously in this thread. But if u want to try it;
ethminer -U -S exp-us.dwarfpool.com:8018 -SP 1 -O 0x1497684e0740d41cd05f5a0d2f2f51953b940510.gtx970
thats my command line for expanse u just need to sub in the info for eth.
Thanks @Wolf0
I am now using DwarfPool and their Stratum Proxy (found nothing else useable?) and it's working.
I am now using DwarfPool and their Stratum Proxy (found nothing else useable?) and it's working.
Mine always drops a little bit when submitting a share or when there's a new block but never to zero.
Thanks @Wolf0
I am now using DwarfPool and their Stratum Proxy (found nothing else useable?) and it's working.
I am now using DwarfPool and their Stratum Proxy (found nothing else useable?) and it's working.
Thanks! I was saving that for when I would get bored sometime. But I had something else in mind to make that faster anyway 
I do, too, but didn't finish it. Short attention span.
I am currently mining Eth on a Radeon RX 480 on Win 10 and wanted to ask if it's normal that my GPU load seems to jump from 100 to ~0 when a share gets submitted.
Is there any workaround? Do I need to change my settings? Or is this normal?`
Thanks for an Answer
Is there any workaround? Do I need to change my settings? Or is this normal?`
Thanks for an Answer
Its not wddm - ist just how win 10 vs 7/8 are dealing with memory(maxwell doesnt win from wddm 2.1 - for maxwell it will be lower than win7/8) - for neoscrypt it needs optimizations on win 10 ppwpddm 2.1 as well as 7/8
Based on what I've read about WDDM, I am pretty certain that is what drivers use to deal with the OS virtual memory.
The supposed 'fix' for W10 was the WDDM update, which wont be back ported to W7/8. I don't really get what you're saying though as your posts are really fragmented. I already understand the miners are broken and need to be fixed. Not exactly certain as to why they're broken though, which is what I was looking for here.
So we're back to square one two weeks ago, when people were bringing this up, which is Dagger and NeoS are still broken on W7/8 and NeoS is broken on W10, so switching to W10 wont fix these issues (switching OS's isn't a 'fix' to buggy software anyway).
Ask a administrator to switch OS's because a piece of his software is broken and he'll laugh you out of his tiny little office. This really just looks like people are trying to shift responsibility around here so things don't get fixed though... As to why? I'm not exactly certain.
Hopefully Genoil will be able to start tidying this up once he gets his 1070 as will the other developers.
41 on what as opposed to ether? Do we need to install cuda?
Quote from: Amph
link=topic=1368785.msg15499345#msg15499345 date=1467883237
link=topic=1368785.msg15499345#msg15499345 date=1467883237
it work good on windows 10, you can get 41MH there with a single 1070 and 27MH for etheruem, all with a very little consumption, around 140w
sooner or later will be forced to move to win 10 anyway, it's better to not be stuck with win 7 forever
sooner or later will be forced to move to win 10 anyway, it's better to not be stuck with win 7 forever
41MH on lyra2v2, which is impressive, but you need to oc to 1900 core
i think you just need win 10 but maybe you need also one of the insider preview build
Genoil - you'll need to adapt this slightly to yours, but this is smaller and gets another wave in flight for your DAG gen. I got bored.
Also a little sloppy, you might want to clean it up a tad.
Code:
typedef union _Node
{
uint dwords[16];
uint2 qwords[8];
uint4 dqwords[4];
} Node;
static void SHA3_512(uint2 *s, uint isolate)
{
uint2 st[25];
for(uint i = 0; i < 8; ++i) st[i] = s[i];
for (uint i = 8; i != 25; ++i)
{
st[i] = (uint2){ 0, 0 };
}
((uint2 *)st)[8].x = 0x00000001;
((uint2 *)st)[8].y = 0x80000000;
//s[8] = 0x8000000000000001UL;
//keccak_f1600_no_absorb(s, 8, isolate);
KECCAK_PROCESS(st, 8, 8, isolate);
for(uint i = 0; i < 8; ++i) s[i] = st[i];
}
__kernel void ethash_calculate_dag_item(uint start, __global const Node *Cache, __global Node *DAG, uint LIGHT_SIZE, uint isolate)
{
uint NodeIdx = start + get_global_id(0);
//if (NodeIdx > DAG_SIZE) return;
Node DAGNode = Cache[NodeIdx % LIGHT_SIZE];
DAGNode.dwords[0] ^= NodeIdx;
SHA3_512(DAGNode.qwords, isolate);
for(uint i = 0; i < 256; ++i)
{
uint ParentIdx = fnv(NodeIdx ^ i, DAGNode.dwords[i & 15]) % LIGHT_SIZE;
__global const Node *ParentNode = Cache + ParentIdx;
#pragma unroll
for(uint x = 0; x < 4; ++x)
{
DAGNode.dqwords[x] *= (uint4)(FNV_PRIME);
DAGNode.dqwords[x] ^= ParentNode->dqwords[x];
}
}
SHA3_512(DAGNode.qwords, isolate);
DAG[NodeIdx] = DAGNode;
}
Also a little sloppy, you might want to clean it up a tad.
41 on what as opposed to ether? Do we need to install cuda?
Quote from: Amph
link=topic=1368785.msg15499345#msg15499345 date=1467883237
link=topic=1368785.msg15499345#msg15499345 date=1467883237
it work good on windows 10, you can get 41MH there with a single 1070 and 27MH for etheruem, all with a very little consumption, around 140w
sooner or later will be forced to move to win 10 anyway, it's better to not be stuck with win 7 forever
sooner or later will be forced to move to win 10 anyway, it's better to not be stuck with win 7 forever
Its not wddm - ist just how win 10 vs 7/8 are dealing with memory(maxwell doesnt win from wddm 2.1 - for maxwell it will be lower than win7/8) - for neoscrypt it needs optimizations on win 10 ppwpddm 2.1 as well as 7/8
Forgot to implement a +2% speed increase (7950) that @Wolf0 hinted me at a while back. Will likely make it to 1.1.8
Well done! Which was it?
The bfe. I'd done it on CUDA but it didn't help, but on AMD 7950 I went a few kH up. Also finally taking the time now to streamline the keccak stages. Also worth about 1kH I think. Also going to try a 4-way. Just because I can
4-way is only really useful in one case: ds_shuffle_b32.
EDIT: I did do a 2-way in OCL, though. Maybe I should revisit it.
uh i remember i said the same about win 7, i don't care if it work there or not, i don't like linux because of gaming, and i'm also mining with a gaming rig(why i should leave 1 card there doing nothing when i'm not doing gaming?)
if there were no other option, then i would end up using linux for the mining rig, but since on win 10 it work fine i'll use that when ready
i'm just waiting for someone to fix neos at least on win 10, which is the most profitable for me now(dunno but i'm getting more than what whattomine display...), before buying 1070...
if there were no other option, then i would end up using linux for the mining rig, but since on win 10 it work fine i'll use that when ready
i'm just waiting for someone to fix neos at least on win 10, which is the most profitable for me now(dunno but i'm getting more than what whattomine display...), before buying 1070...
The new Nvidia 368.69 drivers have been released, can someone test a 1070/1080 on Windows7/8/10 with these?
win 7 need genoil work, because the drivers will not fix anything with pascal
Holy shit, what I said over a week ago and you told me to suck it up and go to W10 because everything was working peachy keen.
That aside, in replies that followed Genoil and Vaulter said that there would be a driver update that updates WDDM in W7/8. Although I'm almost 100% certain this isn't true and Microsoft has no interest in bringing newer versions of WDDM back to W7/8. That's one of the defining features of W10. Microsoft is in control of WDDM, not Nvidia or AMD. So a driver update wont fix this.
Once again we're back to the problem with bugs. And 'your friend' which is a clever way of saying yourself, I'm sure will be happy when this bug finally gets fixed.
You would've figured this out all over a week ago if you actually had a conversation with me, instead of telling me that I'm wrong and reassuring me of your non-existant fixes, which you still have no idea WTF you were talking about.
I'll just leave this here -
LINUX, LINUX, LINUX, LINUX, LINUX, LINUX, LINUX, LINUX, LINUX, LINUX, LINUX,
LINUX, LINUX, LINUX, LINUX, LINUX, LINUX, LINUX, LINUX, LINUX, LINUX, LINUX,
LINUX, LINUX, LINUX and LINUX.
I have better uses for my time then debugging kernels and recompiling for the sheer masochistic fun.
It's comments like this that let me know this fool actually probably tried Linux for all of ten seconds, and got the rest of his info from rumors and bullshit.
Because you don't know anything about hyperbole as a rhetorical device? I thought we were past the point of actually making meaningful points. You know, with the whole 'one bug and you should abandon your OS' thing.
The new Nvidia 368.69 drivers have been released, can someone test a 1070/1080 on Windows7/8/10 with these?
win 7 need genoil work, because the drivers will not fix anything with pascal
Holy shit, what I said over a week ago and you told me to suck it up and go to W10 because everything was working peachy keen.
That aside, in replies that followed Genoil and Vaulter said that there would be a driver update that updates WDDM in W7/8. Although I'm almost 100% certain this isn't true and Microsoft has no interest in bringing newer versions of WDDM back to W7/8. That's one of the defining features of W10. Microsoft is in control of WDDM, not Nvidia or AMD. So a driver update wont fix this.
Once again we're back to the problem with bugs. And 'your friend' which is a clever way of saying yourself, I'm sure will be happy when this bug finally gets fixed.
You would've figured this out all over a week ago if you actually had a conversation with me, instead of telling me that I'm wrong and reassuring me of your non-existant fixes, which you still have no idea WTF you were talking about.
no you are talking shit as usual, it was not my friend and it was not myself, it just another user that asked for help , i don't have any pascal yet, i'm not so stupid to buy something that can mine on one good algo only(lyra2v2), if you still think otherwise it's your problem
actually i don't care about win 7 not working, as long as it work on at least one windows and not only on linux i'm fine
That wasn't what you were saying roughly a month ago when I said exactly the same thing... OMG people that remember.
are you saying the new driver itsself doesn't fix the issue? Does the tech preview + new driver fix windows 10 hashing?
All pascal gpus (10xx) are good to go over 2gb
And today there is 368.69 driver availible which enables to mine eth at full speed for 10xx series in win 10 insider preview - just install this driver and go
but what is the limit ?
https://bitcointalksearch.org/topic/assessing-the-impact-of-tlb-trashing-on-memory-hard-algorhitms-1268355
https://github.com/Genoil/dagSimCL
There is no cliff drop as with maxwell - its gradual drop after 4.2gb for 1070/1080
As for eth and wddm - its wddm 2.1 change on win 10 rs1 (release on 2nd of august, availible today with insider preview to everyone) and just a driver change (no wddm) for 368.69 - but needs work from developers to enable linux/win10 performance for eth
All pascal gpus (10xx) are good to go over 2gb
And today there is 368.69 driver availible which enables to mine eth at full speed for 10xx series in win 10 insider preview - just install this driver and go
but what is the limit ?
https://bitcointalksearch.org/topic/assessing-the-impact-of-tlb-trashing-on-memory-hard-algorhitms-1268355
https://github.com/Genoil/dagSimCL
As for eth and wddm - its wddm 2.1 change on win 10 rs1 (release on 2nd of august, availible today with insider preview to everyone) and just a driver change (no wddm) for 368.69 - but needs work from developers to enable linux/win10 performance for eth
So things are still broken with Windows 7/8 for NeoS and Eth on Pascal, since WDDM will not be backported to W10 and needs to be fixed by developers.
are you saying the new driver itsself doesn't fix the issue? Does the tech preview + new driver fix windows 10 hashing?
All pascal gpus (10xx) are good to go over 2gb
And today there is 368.69 driver availible which enables to mine eth at full speed for 10xx series in win 10 insider preview - just install this driver and go
but what is the limit ?
https://bitcointalksearch.org/topic/assessing-the-impact-of-tlb-trashing-on-memory-hard-algorhitms-1268355
https://github.com/Genoil/dagSimCL
There is no cliff drop as with maxwell - its gradual drop after 4.2gb for 1070/1080
As for eth and wddm - its wddm 2.1 change on win 10 rs1 (release on 2nd of august, availible today with insider preview to everyone) and just a driver change (no wddm) for 368.69 - but needs work from developers to enable linux/win10 performance for eth
The new driver fix win 10 insider preview by itself and needs work on win7/8 (its os specific - devs need work on win7/8 kernel) All pascal gpus (10xx) are good to go over 2gb
And today there is 368.69 driver availible which enables to mine eth at full speed for 10xx series in win 10 insider preview - just install this driver and go
but what is the limit ?
https://bitcointalksearch.org/topic/assessing-the-impact-of-tlb-trashing-on-memory-hard-algorhitms-1268355
https://github.com/Genoil/dagSimCL
As for eth and wddm - its wddm 2.1 change on win 10 rs1 (release on 2nd of august, availible today with insider preview to everyone) and just a driver change (no wddm) for 368.69 - but needs work from developers to enable linux/win10 performance for eth
no you are talking shit as usual, it was not my friend and it was not myself, it just another user that asked for help , i don't have any pascal yet, i'm not so stupid to buy something that can mine on one good algo only(lyra2v2), if you still think otherwise it's your problem
actually i don't care about win 7 not working, as long as it work on at least one windows and not only on linux i'm fine
I think now Pascal have more algo than AMD good for mining
yeah amd is not optimized outside of eth(which is trash at the moment and will die soon and i highly doubt that expanse and shift will take its place), which is really bad, is populated with shitty private algo, until all the algo are properly optimized i'm not touching amd ever
no you are talking shit as usual, it was not my friend and it was not myself, it just another user that asked for help , i don't have any pascal yet, i'm not so stupid to buy something that can mine on one good algo only(lyra2v2), if you still think otherwise it's your problem
actually i don't care about win 7 not working, as long as it work on at least one windows and not only on linux i'm fine
I think now Pascal have more algo than AMD good for mining
it work good on windows 10, you can get 41MH there with a single 1070 and 27MH for etheruem, all with a very little consumption, around 140w
sooner or later will be forced to move to win 10 anyway, it's better to not be stuck with win 7 forever
sooner or later will be forced to move to win 10 anyway, it's better to not be stuck with win 7 forever
Forgot to implement a +2% speed increase (7950) that @Wolf0 hinted me at a while back. Will likely make it to 1.1.8
Well done! Which was it?
Jump to: