EWBF's CUDA Zcash miner

[philipma1957]

will be doing a 1080 ti rig

1 card is on order.

I will post photos and results on weekend .

I hope as parts are not in.

[Kompik]

will be doing a 1080 ti rig

1 card is on order.

I will post photos and results on weekend .

I hope as parts are not in.

Great, will be eagerly waiting for the reports . I plan to buy 1080ti for testing as well, but I am waiting for the EVGA one with three fans since it looks so cool!

[Diod-ionica]

will be doing a 1080 ti rig

1 card is on order.

I will post photos and results on weekend .

I hope as parts are not in.

Great, will be eagerly waiting for the reports . I plan to buy 1080ti for testing as well, but I am waiting for the EVGA one with three fans since it looks so cool!

Great idea and successful execution so that everything turned out and our esteemed developer to be able to adapt his algorithm to the new card as best as possible! All the maximum profit!

[kapsel]

http://cryptomining-blog.com/8572-looking-for-the-efficiency-sweet-spot-of-gtx-1080-ti/

(NiceHash Excavator v1.1.4a)

Efficiency on 1080ti doesn't look too great, i mined zec with 4 gtx 1080s FE and my peak efficiency was around 4.2-4.3 sol/s per W using ewbf 0.3.3b(65% power), haven't tried excavator myself.

[philipma1957]

http://cryptomining-blog.com/8572-looking-for-the-efficiency-sweet-spot-of-gtx-1080-ti/

(NiceHash Excavator v1.1.4a)

Efficiency on 1080ti doesn't look too great, i mined zec with 4 gtx 1080s FE and my peak efficiency was around 4.2-4.3 sol/s per W using ewbf 0.3.3b(65% power), haven't tried excavator myself.

hash rate is very high  on the 1080ti  but it may be a bit power hungry compared to the 1080 or the 1070

I am looking forward to running  my build  if I get 3.5 sols  It is better then my amd cards are for zec.

as they get 2 sols per watt.

it would allow me to mine nvidia @ zec and amd @ eth

[NiceHashSupport]

http://cryptomining-blog.com/8572-looking-for-the-efficiency-sweet-spot-of-gtx-1080-ti/

(NiceHash Excavator v1.1.4a)

Efficiency on 1080ti doesn't look too great, i mined zec with 4 gtx 1080s FE and my peak efficiency was around 4.2-4.3 sol/s per W using ewbf 0.3.3b(65% power), haven't tried excavator myself.

hash rate is very high  on the 1080ti  but it may be a bit power hungry compared to the 1080 or the 1070

I am looking forward to running  my build  if I get 3.5 sols  It is better then my amd cards are for zec.

as they get 2 sols per watt.

it would allow me to mine nvidia @ zec and amd @ eth

philip, highest efficiency is 4x1080Ti giving 2250 sol/s at 600 W (wall) - that is 3.75 sol/W (real, not what SMI shows - these values are 20% less).

1080 Ti are the fastest and the most efficient cards on the market. AMD will not reach that for another 3 years.

[philipma1957]

http://cryptomining-blog.com/8572-looking-for-the-efficiency-sweet-spot-of-gtx-1080-ti/

(NiceHash Excavator v1.1.4a)

Efficiency on 1080ti doesn't look too great, i mined zec with 4 gtx 1080s FE and my peak efficiency was around 4.2-4.3 sol/s per W using ewbf 0.3.3b(65% power), haven't tried excavator myself.

hash rate is very high  on the 1080ti  but it may be a bit power hungry compared to the 1080 or the 1070

I am looking forward to running  my build  if I get 3.5 sols  It is better then my amd cards are for zec.

as they get 2 sols per watt.

it would allow me to mine nvidia @ zec and amd @ eth

philip, highest efficiency is 4x1080Ti giving 2250 sol/s at 600 W (wall) - that is 3.75 sol/W (real, not what SMI shows - these values are 20% less).

1080 Ti are the fastest and the most efficient cards on the market. AMD will not reach that for another 3 years.

my build will be 2 1080 ti  with 1 zotac 1080 mini
I have this empty board.

it is a biostar z170

goal would be 610 + 610 + 480 = 1700 at 540-575 watts

[Biodom]

The card (Zotac 1060 6gb mini) is probably locked, darn it.

You can OC these cards using nvOC.  I run multiple Zotac 1060 6gb mini 6x rigs with 100 powerlimit 150 cc and 600 mc for ~1900 sol/s poolside using 670 watts.

Cool...thanks, i might try it.
Why a need to connect monitor to the second card, specifically, if I understood it correctly?

[agente]

The card (Zotac 1060 6gb mini) is probably locked, darn it.

You can OC these cards using nvOC.  I run multiple Zotac 1060 6gb mini 6x rigs with 100 powerlimit 150 cc and 600 mc for ~1900 sol/s poolside using 670 watts.

I have inno3d mini 1060 and I have a lot of problem with temp. What about Zotac? Your "farm" is in a cold place?

[upsidedown75]

Anyone tried the new Razers on mining a new gpu coin?

I think if you could get couple of those you could mine the very first blocks of any new coin, getting to a level of early bird which would help your income a lot when it hits exchanges.

[painmaker]

(further quotes SNIPPed, see #2032)

while i agree with all of the supplied information about OC-ing nvidia-cards under linux i just found my system (GTX970 + GTX980) to be stuck in performance-level 2 (out of a range of 0 - 3) and ignoring GPUMemoryTransferRateOffset -commands.

running

Code:

nvidia-settings -q GPUPerfModes

confirms that memTransferRateeditable=0 and memclockeditable=0 for perf < 3 for my GTX980 and =0 in all perf-levels for my GTX970.

i'm not sure about the GTX970 but i am somewhat confident that i've set a custom GPUMemoryTransferRateOffset on the GTX980 in the past but cannot remember how.
can anybody hint me to what i might need to do to enable/set performance-level 3 on my cards?

many thanks in advance,
painmaker

For performance mode

Code:

nvidia-settings -a [gpu:0]/GPUPowerMizerMode=1 

Valid values for 'GPUPowerMizerMode' are: 0, 1 and 2.

If you have many cards replace [gpu:0] with [gpu:N]

sorry to bring this up again, didn't have the time to fully test this out earlier but it looks like my problem of being stuck in perf-level 2 of 3 (and therefore unable to adjust fan/gutRates/memoryRates) as soon as i put some decent work on my GPUs still hasn't been solved.  

just to be sure, is anyone running a setup similar to mine who is able to OC (all) the nvidia-cards present in the system? here are some specs:
- debian 8
- nvidia-driver 375.39 and 378.13 (dkms-install)
- cuda 8.0.44 installed (without driver)
- gtx980 + gtx970
- aiming for multi-user.target (running no X-server), but problem also present when in graphical.target (=desktop-mode)
- in xorg.conf, option CoolBits is set to 12, RegistryDwords set to 'PowerMizerEnable=0x1; PerfLevelSrc=0x2222; PowerMizerDefault=0x1'

if there is someone with a roughly similar setup that allows OCing, what did you do to get there?

btw, here's another thread mentioning some nvidia OC steps for linux. sounds like its also recommending using ubuntu (16.04 in this case).  
EDIT: might take a look at fullzeros nvOC which looks promising...

[murasame2003]

EWBF,

Will it be possible to add another switch under --eexit?
Like, exit the miner if total speed reaches 0 sol/s.

I second this, I will run for a good 2 days and then start hashing 0.  For me I don't think it is OC issues cause I already had to go down from when I ran 3b.  3b was the most stable at highest clocks.  I dont mind the lower clocks but frustrated with the stability issue so having a command to restart the miner if it reaches 0 would be great.

Yes, but i think it's better to fix this. On the side of the pool speed also falls? And can you show the log file?

Everything falls to 0 sol/sec after 3~5 minutes for some reason.
Here is the log. I'm not sure how much of anything this yields.

INFO: Target: 0001ffffffffffff...
INFO: Detected new work: 1491313323_47506
INFO: Detected new work: 1491313323_47506
CUDA: Device: 0 GeForce GTX 970, 4096 MB PCI: 0000:01:00.0
CUDA: Device: 1 GeForce GTX 970, 4096 MB PCI: 0000:04:00.0
CUDA: Device: 2 GeForce GTX 970, 4096 MB PCI: 0000:05:00.0
CUDA: Device: 3 GeForce GTX 950, 2048 MB PCI: 0000:08:00.0
CUDA: Device: 0 Selected solver: 0
CUDA: Device: 1 Selected solver: 0
CUDA: Device: 2 Selected solver: 0
CUDA: Device: 3 Selected solver: 0
INFO 09:42:37: GPU2 Accepted share 427ms [A:1, R:0]
Temp: GPU0 64C GPU1 58C GPU2 61C GPU3 49C
GPU0: 275 Sol/s GPU1: 285 Sol/s GPU2: 279 Sol/s GPU3: 143 Sol/s
Total speed: 982 Sol/s
INFO: Detected new work: 1491313369_47507
Temp: GPU0 65C GPU1 64C GPU2 65C GPU3 54C
GPU0: 275 Sol/s GPU1: 279 Sol/s GPU2: 283 Sol/s GPU3: 144 Sol/s
Total speed: 981 Sol/s
INFO 09:43:18: GPU1 Accepted share 437ms [A:1, R:0]
INFO 09:43:23: GPU2 Accepted share 422ms [A:2, R:0]
INFO 09:43:28: GPU2 Accepted share 437ms [A:3, R:0]
INFO: Detected new work: 1491313424_47508
Temp: GPU0 65C GPU1 66C GPU2 66C GPU3 55C
GPU0: 279 Sol/s GPU1: 278 Sol/s GPU2: 279 Sol/s GPU3: 142 Sol/s
Total speed: 978 Sol/s
INFO 09:43:48: GPU1 Accepted share 438ms [A:2, R:0]
INFO 09:44:15: GPU1 Accepted share 437ms [A:3, R:0]
Temp: GPU0 65C GPU1 66C GPU2 67C GPU3 57C
GPU0: 275 Sol/s GPU1: 279 Sol/s GPU2: 278 Sol/s GPU3: 146 Sol/s
Total speed: 978 Sol/s
INFO 09:44:20: GPU2 Accepted share 438ms [A:4, R:0]
INFO 09:44:22: GPU0 Accepted share 465ms [A:1, R:0]
INFO 09:44:27: GPU1 Accepted share 453ms [A:4, R:0]
INFO: Detected new work: 1491313479_47509
Temp: GPU0 66C GPU1 66C GPU2 67C GPU3 57C
GPU0: 279 Sol/s GPU1: 285 Sol/s GPU2: 285 Sol/s GPU3: 142 Sol/s
Total speed: 991 Sol/s
INFO 09:44:49: GPU0 Accepted share 1352ms [A:2, R:0]
Temp: GPU0 65C GPU1 66C GPU2 68C GPU3 57C
GPU0: 275 Sol/s GPU1: 270 Sol/s GPU2: 284 Sol/s GPU3: 143 Sol/s
Total speed: 972 Sol/s
INFO: Detected new work: 1491313534_47510
Temp: GPU0 65C GPU1 66C GPU2 67C GPU3 57C
GPU0: 275 Sol/s GPU1: 277 Sol/s GPU2: 292 Sol/s GPU3: 141 Sol/s
Total speed: 985 Sol/s
INFO 09:46:01: GPU2 Accepted share 438ms [A:5, R:0]
Temp: GPU0 66C GPU1 66C GPU2 67C GPU3 57C
GPU0: 271 Sol/s GPU1: 280 Sol/s GPU2: 281 Sol/s GPU3: 146 Sol/s
Total speed: 978 Sol/s
INFO 09:46:24: GPU1 Accepted share 437ms [A:5, R:0]
INFO: Detected new work: 1491313589_47511
INFO 09:46:33: GPU2 Accepted share 436ms [A:6, R:0]
Temp: GPU0 50C GPU1 56C GPU2 51C GPU3 49C
GPU0: 276 Sol/s GPU1: 272 Sol/s GPU2: 283 Sol/s GPU3: 141 Sol/s
Total speed: 972 Sol/s
INFO: Detected new work: 1491313625_47512
Temp: GPU0 46C GPU1 49C GPU2 46C GPU3 44C
GPU0: 0 Sol/s GPU1: 0 Sol/s GPU2: 0 Sol/s GPU3: 0 Sol/s
Total speed: 0 Sol/s
Temp: GPU0 45C GPU1 46C GPU2 44C GPU3 41C
GPU0: 0 Sol/s GPU1: 0 Sol/s GPU2: 0 Sol/s GPU3: 0 Sol/s
Total speed: 0 Sol/s
INFO: Target: 0003ffffffffffff...
INFO: Detected new work: 1491313680_47513
Temp: GPU0 44C GPU1 44C GPU2 43C GPU3 40C
GPU0: 0 Sol/s GPU1: 0 Sol/s GPU2: 0 Sol/s GPU3: 0 Sol/s
Total speed: 0 Sol/s

[k0stas]

(further quotes SNIPPed, see #2032[/i])

while i agree with all of the supplied information about OC-ing nvidia-cards under linux i just found my system (GTX970 + GTX980) to be stuck in performance-level 2 (out of a range of 0 - 3) and ignoring GPUMemoryTransferRateOffset -commands.

running

Code:

nvidia-settings -q GPUPerfModes

confirms that memTransferRateeditable=0 and memclockeditable=0 for perf < 3 for my GTX980 and =0 in all perf-levels for my GTX970.

i'm not sure about the GTX970 but i am somewhat confident that i've set a custom GPUMemoryTransferRateOffset on the GTX980 in the past but cannot remember how.
can anybody hint me to what i might need to do to enable/set performance-level 3 on my cards?

many thanks in advance,
painmaker

For performance mode

Code:

nvidia-settings -a [gpu:0]/GPUPowerMizerMode=1 

Valid values for 'GPUPowerMizerMode' are: 0, 1 and 2.

If you have many cards replace [gpu:0] with [gpu:N]

sorry to bring this up again, didn't have the time to fully test this out earlier but it looks like my problem of being stuck in perf-level 2 of 3 (and therefore unable to adjust fan/gutRates/memoryRates) as soon as i put some decent work on my GPUs still hasn't been solved.  

just to be sure, is anyone running a setup similar to mine who is able to OC (all) the nvidia-cards present in the system? here are some specs:
- debian 8
- nvidia-driver 375.39 and 378.13 (dkms-install)
- cuda 8.0.44 installed (without driver)
- gtx980 + gtx970
- aiming for multi-user.target (running no X-server), but problem also present when in graphical.target (=desktop-mode)
- in xorg.conf, option CoolBits is set to 12, RegistryDwords set to 'PowerMizerEnable=0x1; PerfLevelSrc=0x2222; PowerMizerDefault=0x1'

if there is someone with a roughly similar setup that allows OCing, what did you do to get there?

btw, here's another thread mentioning some nvidia OC steps for linux. sounds like its also recommending using ubuntu (16.04 in this case).  
EDIT: might take a look at fullzeros nvOC which looks promising...

Change coolbits from 12 to 28

Code:

cd /etc/X11

sudo nvidia-xconfig -a --cool-bits=28  --allow-empty-initial-configuration

and remove this line from xorg.conf

Code:

Option  "RegistryDwords" "PowerMizerEnable=0x1; PerfLevelSrc=0x2222; PowerMizerDefaultAC=0x1"

and add this lines to autostart script

Code:

nvidia-settings -a '[gpu:0]/GPUPowerMizerMode=1'

nvidia-settings -a '[gpu:0]/GPUFanControlState=1'

nvidia-settings -a '[fan:0]/GPUTargetFanSpeed=70'

and reboot

[painmaker]

@k0stas: thanks for your reply!
done as you told (changed per default to graphical.target to make sure nvidia-settings has some Xserver), noticed fans still in auto-mode right after startup, found that powermizer still goes back to perfLevel 2 right after launching ewbf.
any other ideas?
can you post some config-specs of your system (OS, driver, ...)?

cheers, pm

EDIT: btw, i think --cool-bits=12 ( =4+8 ) really should suffice as bit5 ( =16 ) seems to enable voltage offsets; AFAIK bit4 ( =8 ) enables freq-adjusts and bit3 ( =4 ) enables fan-control.

[k0stas]

This settings working for me on suse leap 42.2 nvidia driver 375.39.

[PovertyByte]

http://cryptomining-blog.com/8572-looking-for-the-efficiency-sweet-spot-of-gtx-1080-ti/

(NiceHash Excavator v1.1.4a)

Efficiency on 1080ti doesn't look too great, i mined zec with 4 gtx 1080s FE and my peak efficiency was around 4.2-4.3 sol/s per W using ewbf 0.3.3b(65% power), haven't tried excavator myself.

That is very high efficiency for a 1080 at 65% power. I peak at 3.7 if I move the power there and I got full memory OC and a fair core OC. I need to drop to 55% for that efficiency you got. Is that your best card?

[philipma1957]

http://cryptomining-blog.com/8572-looking-for-the-efficiency-sweet-spot-of-gtx-1080-ti/

(NiceHash Excavator v1.1.4a)

Efficiency on 1080ti doesn't look too great, i mined zec with 4 gtx 1080s FE and my peak efficiency was around 4.2-4.3 sol/s per W using ewbf 0.3.3b(65% power), haven't tried excavator myself.

That is very high efficiency for a 1080 at 65% power. I peak at 3.7 if I move the power there and I got full memory OC and a fair core OC. I need to drop to 55% for that efficiency you got. Is that your best card?

It is a1080 ti



Back to this software.  I am up and running my first  nvidia miner.

I loaded nvidia 378.78 driver

and 3.3b of this software

and windows 7 pro





it is this single card


mini sized zotac 1080 just came in

http://www.ebay.com/itm/381941454927?

short small lightweight  only has a 8 pin  for power.


here it is in red at nicehash

[painmaker]

(quotes SNIPPed, see #2287)

btw, here's another thread mentioning some nvidia OC steps for linux. sounds like its also recommending using ubuntu (16.04 in this case).  
EDIT: might take a look at fullzeros nvOC which looks promising...

so i tried fullzeros nvOC which is basically an ubuntu 16.04 image with autologin automatically starting a terminal-window which per default runs a shellscript that does the magic. nice idea, quite convenient to use but sadly, no overclocking possible with this as well as my cards go straight to perfLevel2 after launching a miner (tried several, behaviour is all the same) :/

so it seems not to be an issue of the OS itself neither the configuration as i suspect fullzero is able to adjust clocks when running nvOC.
then it might be a matter of a certain card-bios behaviour? whats odd is that i have two different cards in my system from different vendors/manufacturers.
the gtx 980 is a rather sh***y one from Inno3d (HerculeZ X3 Air Boss), the gtx 970 is an asus strix oc one which seems to be of decent quality.
both cards show the same behaviour with being stuck in performance level 2 when under load thux loosing their OC-setting and failing on re-setting it while not in perfLevel 3 (read-only in pl2).
wondering that i haven't come across more ppl with the same issue, might not appear when running windows possibly...

[7s4wju]

Hello! I'm very comfortable with the log file with the parameter --log 1, look for errors, but in this mode there is no date and time errors - can this be fixed?

[philipma1957]

I have a 1080  and I can do

160 watts at wall on kwatt meter  and hash  about 492h



do the numbers below  reflect at the wall numbers?







edit that is a software number used below  so it does not include the rig just the card.