NiceHash EQM Zcash NVIDIA optimized miner [Maxwell/Pascal] + CPU mining v1.0.4c

[Amph]

getting 300 sol at the same wattage as before, now that is an improvement, as usual 55% tdp 120 core 400 mem, still only slightly better than the free pool miner

[NiceHashSupport]

Forgot to mention, there is hidden secret:

Use two instances per GPU device: -cd 0 0

Or to launch two instances for every GPU device: -ca -ca

[reb0rn21]

Forgot to mention, there is hidden secret:

Use two instances per GPU device: -cd 0 0

Or to launch two instances for every GPU device: -ca -ca

yeah now fixed (GPu usage 100% with 0 0 1 1,.... , but if there is no EU and USA server failover i will run EU + USA server exe

[mo35]

i must admit , impressive work

1x 1080 +100/+400
EWBF 190w 320sols
EQM 160w 345sols (double instances)

2x 1070 3x 1060 +100/+400
EWBF 800w 1100sols
EQM 720w 1440 sols (double instances)

[rednoW]

Here is link for Win version, Linux is coming very shortly (matter of minutes, hour or two).

https://www.nicehash.com/tools/eqm_v1.0.3a_Win64.zip

Now, lets see who can pull 400 Sol out of his card  

Only 370 sols (((
palit sjs gtx 1070 (samsung)

what kind of OC settings? any special options when running the miner?
I'm only getting ~280 sols

90%tdp +118 gpu +1000mem +gpu mem overvoltage in palit thundermaster

yields 2150/2400 gpu/mem

2 threads on gpu

[ioglnx]

Running -ca -ca instantly gave me BSOD - can be reproduced.
If I run two seperate ones it runs without problem.

[reb0rn21]

Running -ca -ca instantly gave me BSOD - can be reproduced.
If I run two seperate ones it runs without problem.

-ca one is 2x instances, -ca -ca will be 3x, but worked for me on 8GB system and 5x 1070 win10

so run just -ca

[Mugatu]

Here is link for Win version, Linux is coming very shortly (matter of minutes, hour or two).

https://www.nicehash.com/tools/eqm_v1.0.3a_Win64.zip

Now, lets see who can pull 400 Sol out of his card  

Only 370 sols (((
palit sjs gtx 1070 (samsung)

what kind of OC settings? any special options when running the miner?
I'm only getting ~280 sols

reboot fixed it....

[ioglnx]

Running -ca -ca instantly gave me BSOD - can be reproduced.
If I run two seperate ones it runs without problem.

-ca one is 2x instances, -ca -ca will be 3x, but worked for me on 8GB system and 5x 1070 win10

so run just -ca

Looks like 2 instances ... -ca -ca

[19:12:22][0x00000dbc][info] =====> 967.56 I/s, 1795.73 Sols/s
[19:12:22][0x00000dbc][debug] miner#0: 100 I/s, 185.74 Sols/s
[19:12:22][0x00000dbc][debug] miner#1: 95.6833 I/s, 177.337 Sols/s
[19:12:22][0x00000dbc][debug] miner#2: 94.57 I/s, 175.887 Sols/s
[19:12:22][0x00000dbc][debug] miner#3: 94.82 I/s, 176.51 Sols/s
[19:12:22][0x00000dbc][debug] miner#4: 98.7 I/s, 183.26 Sols/s
[19:12:22][0x00000dbc][debug] miner#5: 100 I/s, 185.617 Sols/s
[19:12:22][0x00000dbc][debug] miner#6: 95.6967 I/s, 177.403 Sols/s
[19:12:22][0x00000dbc][debug] miner#7: 94.5533 I/s, 175.953 Sols/s
[19:12:22][0x00000dbc][debug] miner#8: 94.8333 I/s, 176.223 Sols/s
[19:12:22][0x00000dbc][debug] miner#9: 98.7033 I/s, 181.803 Sols/s

[NiceHashSupport]

-ca adds one instance per device, so if you do -ca -ca (and no -cd) it will create 2 instances per device.

If you are having any issues with it, let me know what kind of OS, driver ver, GPUs, overclocks etc. We had no issues and tested several system, there were no crashes or unexpected behaviour.

[reb0rn21]

Not here under win10 -ca is doing 2x initializing per GPU at start of the miner... I have 5x GPU per rig

yeah I used -ca and -cd in same config so yeah that why I have it dubble

3x ca works but start very fast 400sol per 1070 (rig 5x card sum over 2000sol!) and then slowly drop to ~1800 after 4-5 min

[bmdrka]

Here is link for Win version, Linux is coming very shortly (matter of minutes, hour or two).

https://www.nicehash.com/tools/eqm_v1.0.3a_Win64.zip

Now, lets see who can pull 400 Sol out of his card  

Only 370 sols (((
palit sjs gtx 1070 (samsung)

what kind of OC settings? any special options when running the miner?
I'm only getting ~280 sols

90%tdp +118 gpu +1000mem +gpu mem overvoltage in palit thundermaster

yields 2150/2400 gpu/mem

2 threads on gpu

Something has gone terribly wrong with that options. Rig went down

[NiceHashSupport]

So, anyone with 400 sol so we can show these AMD users to go living back in to the caves? 

[reb0rn21]

its a bit hard to play with OC and low TDP limits, it depend on GPU usage, temperature, miner and card bios (each have different power limits per TDP)

[Kompik]

Can confirm that with -ca settings you can get over 400 sols, but only temporarily since it slows down.

[reb0rn21]

So, anyone with 400 sol so we can show these AMD users to go living back in to the caves?  

If 3x -sa would hold 400sol would be easy, duno why hash drop slowly, freq. power usage, temp is same, maybe hash calculations are off or 1070 memory get siturated or something

after 5min from 400 to 370, after 10min down to 364 with 3x -sa, must be some way to fix in future and break 400 for sure!

[NiceHashSupport]

So, anyone with 400 sol so we can show these AMD users to go living back in to the caves?  

If 3x -sa would hold 400sol would be easy, duno why hash drop slowly, freq. power usage, temp is same, maybe hash calculations are off or 1070 memory get siturated or something

after 5min from 400 to 370, after 10min down to 364 with 3x -sa, must be some way to fix in future and break 400 for sure!

Could be a bug in speed measuring subsystem. Is there same hash drop on pool side?

EDIT: Also, there is no point in running more than 2 instances. There are context switches needed when using multiple instances. For some reason, CUDA doesn't handle well kernel launches (there are some unused resources) and this algorithm is special - with other algorithms, there are only few kernel launches with lot of work. Here we have multiple kernel launches with small work. Using concurrent instances can help and load the GPU 100%. But with multiple contexts, context switching is happening which also takes resource, the less of them you have, the better. So I would say using 2 is the best.

Try to achieve higher speeds with overclocking; higher core clock should have more effect with this version, because we put more processing load onto shared memory, which runs with the cores freq.

[nicehash]

Here is now the complete set of latest Windows and Linux releases:

DOWNLOAD WINDOWS version (Windows 7+)

DOWNLOAD Linux Ubuntu 14.04 version

DOWNLOAD Linux Ubuntu 16.04 version

IMPORTANT: to achieve maximum speed you should enable two threads per GPU:

Code:

-cd [GPU0] [GPU0] [GPU1] [GPU1]...

for example, when using two threads per GPU while having three GPUs in the system:

Code:

-cd 0 0 1 1 2 2

or you can simply use -ca parameter to run on all cuda devices:

Code:

-ca -ca

Keep on mining! 

Best regards,
NiceHash team.

[reb0rn21]

I don`t know, will try to see, but pool side is never stable in short time

[ioglnx]

Could you please also add an option to set static diff? This would help also. The big drop comes when reaching the 512Diff.
As long as I have 256 its all running fast but when i get 512 or 1024 its goes down a lot.