Open Source ZEC (ZCash) GPU Miner AMD & NVidia (up to 45 sol/s on RX480)

[Za1n]

MSI RX470 x2 (1050 core/1850 mem)

Share accepted.
[GPU 0] T=-1C A=-1% sols=16.513
[GPU 1] T=-1C A=-1% sols=16.378
(ST/INV/DUP): 18x 0ch(0/0/0)

Haven't experimented too much with changing the clock speeds yet, I just used default settings from my ETH mining, so switching between the two coins is pretty much effortless. The pleasant surprise was this "test" rig was drawing ~290 watts at-the-wall while mining ETH and this has dropped to 199 watts mining ZCash. This suggests the ability to potentially push the clocks higher with this algorithm.

[eXtremal]

With the kernel im getting 14 sol/s on 290 give it a shot.

I got same results (15sols/s on 290 with 'threads'=11264 in config.txt file). It seems this miner a bit faster others on RX470/RX480, but don't show good perfornance on R9 2xx.
But it's not a final release!

It will not work with suprnova stratum?
Just do not want to install ZCash wallet.

Can't do before mainnet launch.

[navydude]

OpenCL error: -11 at /build/xpmclient/src/opencl.cpp:153

can help me?

I get the same:(made sure cards were default settings)


found platform[0] name = 'AMD Accelerated Parallel Processing'
<info> found 5 devices
Using device 0 as GPU 0
Using device 1 as GPU 1
Using device 2 as GPU 2
Using device 3 as GPU 3
Using device 4 as GPU 4
threads: 8192, work size: 256
ZCash GPU miner thread 0 started
threads: 8192, work size: 256
ZCash GPU miner thread 1 started
threads: 8192, work size: 256
ZCash GPU miner thread 2 started
threads: 8192, work size: 256
ZCash GPU miner thread 3 started
threads: 8192, work size: 256
ZCash GPU miner thread 4 started
Connecting to frontend: coinsforall.io:6668 ...
Connecting to bitcoin: coinsforall.io:60200 ...
Connecting to signals: coinsforall.io:60201 ...
Work received: height=1284 diff=6974.7629 latency=921ms
OpenCL error: -45 at /build/xpmclient/src/zcash/zcashgpuclient.cpp:112
OpenCL error: -45 at /build/xpmclient/src/zcash/zcashgpuclient.cpp:370
OpenCL error: -45 at /build/xpmclient/src/zcash/zcashgpuclient.cpp:112
OpenCL error: -45 at /build/xpmclient/src/zcash/zcashgpuclient.cpp:370
Work received: height=1285 diff=6958.8853 latency=140ms

GPU 4 found share
Share accepted.
[GPU 0] T=-1C A=-1% sols=7.898
[GPU 1] T=-1C A=-1% sols=7.119
[GPU 2] crashed!
[GPU 3] crashed!
[GPU 4] T=-1C A=-1% sols=8.068
GPU 0: core=-1MHz mem=-1MHz powertune=-1 fanspeed=-1
GPU 1: core=-1MHz mem=-1MHz powertune=-1 fanspeed=-1
GPU 2: core=-1MHz mem=-1MHz powertune=-1 fanspeed=-1
GPU 3: core=-1MHz mem=-1MHz powertune=-1 fanspeed=-1
GPU 4: core=-1MHz mem=-1MHz powertune=-1 fanspeed=-1
(ST/INV/DUP): 3x 0ch(0/0/0)

[ivcelmik]

With the kernel im getting 14 sol/s on 290 give it a shot.

I got same results (15sols/s on 290 with 'threads'=11264 in config.txt file). It seems this miner a bit faster others on RX470/RX480, but don't show good perfornance on R9 2xx.
But it's not a final release!

It will not work with suprnova stratum?
Just do not want to install ZCash wallet.

Can't do before mainnet launch.

Great work how u can get 8.5 Sol on a Nvidia 750 Ti i get only 1.5Sol

Thx in advance

[jiggytom]

How difficult would it be to make a Cuda version for nvidia cards?

[Za1n]

OpenCL error: -11 at /build/xpmclient/src/opencl.cpp:153

can help me?

I get the same:(made sure cards were default settings)


found platform[0] name = 'AMD Accelerated Parallel Processing'
<info> found 5 devices
Using device 0 as GPU 0
Using device 1 as GPU 1
Using device 2 as GPU 2
Using device 3 as GPU 3
Using device 4 as GPU 4
threads: 8192, work size: 256
ZCash GPU miner thread 0 started
threads: 8192, work size: 256
ZCash GPU miner thread 1 started
threads: 8192, work size: 256
ZCash GPU miner thread 2 started
threads: 8192, work size: 256
ZCash GPU miner thread 3 started
threads: 8192, work size: 256
ZCash GPU miner thread 4 started
Connecting to frontend: coinsforall.io:6668 ...
Connecting to bitcoin: coinsforall.io:60200 ...
Connecting to signals: coinsforall.io:60201 ...
Work received: height=1284 diff=6974.7629 latency=921ms
OpenCL error: -45 at /build/xpmclient/src/zcash/zcashgpuclient.cpp:112
OpenCL error: -45 at /build/xpmclient/src/zcash/zcashgpuclient.cpp:370
OpenCL error: -45 at /build/xpmclient/src/zcash/zcashgpuclient.cpp:112
OpenCL error: -45 at /build/xpmclient/src/zcash/zcashgpuclient.cpp:370
Work received: height=1285 diff=6958.8853 latency=140ms

GPU 4 found share
Share accepted.
[GPU 0] T=-1C A=-1% sols=7.898
[GPU 1] T=-1C A=-1% sols=7.119
[GPU 2] crashed!
[GPU 3] crashed!
[GPU 4] T=-1C A=-1% sols=8.068
GPU 0: core=-1MHz mem=-1MHz powertune=-1 fanspeed=-1
GPU 1: core=-1MHz mem=-1MHz powertune=-1 fanspeed=-1
GPU 2: core=-1MHz mem=-1MHz powertune=-1 fanspeed=-1
GPU 3: core=-1MHz mem=-1MHz powertune=-1 fanspeed=-1
GPU 4: core=-1MHz mem=-1MHz powertune=-1 fanspeed=-1
(ST/INV/DUP): 3x 0ch(0/0/0)

Try lowering your worksize and/or threads. I got that same error when trying to push my worksize up.

[mogs]

What dependencies needed for this?

[IOTUSA]

How difficult would it be to make a Cuda version for amd cards?

CUDA is per definition for nVidia cards.

[ShopemNL]

Weird, the first card seems to be slow (only 2 sol) rest is 14 sol.

[ps_jb]

Is any way to get an address without installing Ubuntu, compiling wallet etc?

[Za1n]

MSI RX470 x2 (1050 core/1850 mem)

Share accepted.
[GPU 0] T=-1C A=-1% sols=16.513
[GPU 1] T=-1C A=-1% sols=16.378
(ST/INV/DUP): 18x 0ch(0/0/0)

Haven't experimented too much with changing the clock speeds yet, I just used default settings from my ETH mining, so switching between the two coins is pretty much effortless. The pleasant surprise was this "test" rig was drawing ~290 watts at-the-wall while mining ETH and this has dropped to 199 watts mining ZCash. This suggests the ability to potentially push the clocks higher with this algorithm.

A little update.

I have been playing around a bit more, it crashed immediately (console windows closes) if I push the memory clocks much more. However, it loves the core clock increases. I got my core set at 1300 MHz now, or 50 MHz past default, and I am cranking along at:

Work received: height=1301 diff=8694.1578 latency=156ms
[GPU 0] T=-1C A=-1% sols=18.712
[GPU 1] T=-1C A=-1% sols=18.068

I also tried adjusting the worksize and threads a bit as well and it seemed to have a slight improvement moving the worksize down to 192 and increasing threads to 11264, but not enough difference to be definitive yet.

[IOTUSA]

MSI RX470 x2 (1050 core/1850 mem)

Share accepted.
[GPU 0] T=-1C A=-1% sols=16.513
[GPU 1] T=-1C A=-1% sols=16.378
(ST/INV/DUP): 18x 0ch(0/0/0)

Haven't experimented too much with changing the clock speeds yet, I just used default settings from my ETH mining, so switching between the two coins is pretty much effortless. The pleasant surprise was this "test" rig was drawing ~290 watts at-the-wall while mining ETH and this has dropped to 199 watts mining ZCash. This suggests the ability to potentially push the clocks higher with this algorithm.

A little update.

I have been playing around a bit more, it crashed immediately (console windows closes) if I push the memory clocks much more. However, it loves the core clock increases. I got my core set at 1300 MHz now, or 50 MHz past default, and I am cranking along at:

Work received: height=1301 diff=8694.1578 latency=156ms
[GPU 0] T=-1C A=-1% sols=18.712
[GPU 1] T=-1C A=-1% sols=18.068

I also tried adjusting the worksize and threads a bit as well and it seemed to have a slight improvement moving the worksize down to 192 and increasing threads to 11264, but not enough difference to be definitive yet.

It seems memory bus/frequency is a far more important metric than clock-speed. Try cranking up the memory as high as you can, and downlclocking your cores.

[Za1n]

Is any way to get an address without installing Ubuntu, compiling wallet etc?

For testing you can just use the address in the config file (that's what I and it looks like a lot of others are doing) but for mainnet launch you will need your own address. Depends on if an exchange trades on day 1 or not if you can use that for an address, but it is probably worth your while to download and compile your own wallet.

Remember this is launching with a slow-start, so even if you get in a few days later you will not be missing out on a lot of coins as that was pretty much the intention to give everyone time to get kinks worked out etc. before full block mining appears. I am sure within a week or so there will be Windows wallets available and maybe even sooner you will have exchanges with addresses you can mine to.

[dianasta]

3 x Sapphire 290X  ~ 51 sols/s


[GPU 0] T=-1C A=-1% sols=16.829
[GPU 1] T=-1C A=-1% sols=17.005
[GPU 2] T=-1C A=-1% sols=17.040

[jiggytom]

How difficult would it be to make a Cuda version for amd cards?

CUDA is per definition for nVidia cards.

Whoops I meant for nvidia.  As this miner is opencl, right?  But it's based on Cuda code?

[fishbone]

R9 Nano. All default settings

[GPU 0] T=-1C A=-1% sols=19.758
 
6x Rx 470 8gb 1750 strap

[GPU 0] T=-1C A=-1% sols=14.228
[GPU 1] T=-1C A=-1% sols=16.658
[GPU 2] T=-1C A=-1% sols=16.507
[GPU 3] T=-1C A=-1% sols=16.750
[GPU 4] T=-1C A=-1% sols=16.885
[GPU 5] T=-1C A=-1% sols=16.568

[ShopemNL]

6x MSI RX 470 8GB running at 15 sol per card, using 520W @ wall

[IOTUSA]

How difficult would it be to make a Cuda version for amd cards?

CUDA is per definition for nVidia cards.

Whoops I meant for nvidia.  As this miner is opencl, right?  But it's based on Cuda code?

I would assume so. But then again, you know what they say about assumptions.

[Gaglam]

First of all, thanks for this miner.
I'm having 5 R9 280x with the lastest catalyst on Win7.

[GPU 0] T=-1C A=-1% sols=10.103
[GPU 1] T=-1C A=-1% sols=9.186
[GPU 2] T=-1C A=-1% sols=9.288
[GPU 3] T=-1C A=-1% sols=8.833
[GPU 4] T=-1C A=-1% sols=9.417

Cards on 1050 Clock and 1500 Memory

[Genoil]

nice! clocking ~18Sols/s on the same R9 390X that does ~22 using my own brew. your miner has one major advantage though...it actually comes up with solutions that are accepted by the pool