NiceHash EQM Zcash NVIDIA optimized miner [Maxwell/Pascal] + CPU mining v1.0.4c - page 6.

m1n1ngP4d4w4n

full member

Activity: 224

Merit: 100

CryptoLearner

Quote from: tbearhere on January 05, 2017, 05:06:27 AM

Quote from: NiceHashSupport on January 04, 2017, 03:44:58 PM

Yes, this is used in NHM right now.

Regarding crashes. We will not deny that it cannot happen. We used some unsafe shared memory to gain extra speed. The crash possibility is very low. That is why there are auto-restart scripts included. If crash happens, it will not break your driver, so restart fixes it and you keep on mining.

It would be nice if you can share the script that won't break the driver if crash occurs. Thx

I personnaly use a script that check both API local & nicehash web, presence of EQM process, presence of cards (eqm -ci will fail to show cards if drivers is crashed, you can also use nvidia-smi to do this check), autoloop with token check, network and so on and do action accordingly (restart miner, restart rig hard or soft) if you want the best monitoring that's what you need to do

reb0rn21

legendary

Activity: 1901

Merit: 1024

with c version speed up is ~1% or so 1.7% Grin

tbearhere

legendary

Activity: 3164

Merit: 1003

Quote from: NiceHashSupport on January 05, 2017, 05:13:10 AM

Quote from: tbearhere on January 05, 2017, 05:06:27 AM

Quote from: NiceHashSupport on January 04, 2017, 03:44:58 PM

Yes, this is used in NHM right now.

Regarding crashes. We will not deny that it cannot happen. We used some unsafe shared memory to gain extra speed. The crash possibility is very low. That is why there are auto-restart scripts included. If crash happens, it will not break your driver, so restart fixes it and you keep on mining.

It would be nice if you can share the script that won't break the driver if crash occurs. Thx

Just usual loop script. The trick it how the program crashes - if it is due to too high overclock, it would crash the driver alltogether. If it is an error in cuda app such as invalid memory access, it will crash without driver crash thus simple restart is possible.

NiceHashSupport....yes I see thank you.

tbearhere

legendary

Activity: 3164

Merit: 1003

Quote from: Mitja1998 on January 05, 2017, 05:05:20 AM

Hi, can someone tell me if a gigabyte gtx 1070 windforce oc is a good card and a good deal for 400€? I don' have much experience with nvidia cards. I have asus p5q-e mobo and 2x rx470 and 1x hd7950 and dont know if it is going to be compatible with nvidia card.

Check your pm.

NiceHashSupport

hero member

Activity: 588

Merit: 501

Quote from: tbearhere on January 05, 2017, 05:06:27 AM

Quote from: NiceHashSupport on January 04, 2017, 03:44:58 PM

Yes, this is used in NHM right now.

Regarding crashes. We will not deny that it cannot happen. We used some unsafe shared memory to gain extra speed. The crash possibility is very low. That is why there are auto-restart scripts included. If crash happens, it will not break your driver, so restart fixes it and you keep on mining.

It would be nice if you can share the script that won't break the driver if crash occurs. Thx

Just usual loop script. The trick it how the program crashes - if it is due to too high overclock, it would crash the driver alltogether. If it is an error in cuda app such as invalid memory access, it will crash without driver crash thus simple restart is possible.

tbearhere

legendary

Activity: 3164

Merit: 1003

Quote from: NiceHashSupport on January 04, 2017, 03:44:58 PM

Yes, this is used in NHM right now.

Regarding crashes. We will not deny that it cannot happen. We used some unsafe shared memory to gain extra speed. The crash possibility is very low. That is why there are auto-restart scripts included. If crash happens, it will not break your driver, so restart fixes it and you keep on mining.

It would be nice if you can share the script that won't break the driver if crash occurs. Thx

Mitja1998

newbie

Activity: 56

Merit: 0

Hi, can someone tell me if a gigabyte gtx 1070 windforce oc is a good card and a good deal for 400€? I don' have much experience with nvidia cards. I have asus p5q-e mobo and 2x rx470 and 1x hd7950 and dont know if it is going to be compatible with nvidia card.

xPwnK

newbie

Activity: 41

Merit: 0

Quote from: nexitorned on January 05, 2017, 04:36:54 AM

Quote from: datBTC on January 04, 2017, 10:28:35 PM

Quote from: NiceHashSupport on January 04, 2017, 11:50:57 AM

We were able to reach 490 on overclocked GTX 1080. We also noticed that temperature has some effect on all pascal cards - the higher it is, the lower it would clock and thus slower it would mine. But the hashing rate was always consistent, never more or less, as long as all clocks were fine. Also, if there is a driver crash due to too intensive clock, it may put your card into locked low clock and reboot is needed. That is all I can help you with. You may observe GPU load, TDP, memory load and see how these numbers compare to your first run of 500+.

Yeah, maybe i just have a freak card or something. I stopped/started the miner today and boom the 1080 is hashing 530+ SOL/sec. Image below

https://s27.postimg.org/rp95lmckj/1080.png

Lol nice Hashrate, with GTX 1060 took me like 20 days to reach minimun payment with NiceHash, i had to move mining ZEC directly and seems more profitable. 1$ per day with 250SOL/sec. It's weid with NiceHash EQM give to me 210-250Sol/Sec 0 stability, but with another pool 250Sol/sec.

The minimum payment is 0.001 btc. Shouldn't take too long with gtx 1060.

xPwnK

newbie

Activity: 41

Merit: 0

Quote from: Dr_Victor on January 05, 2017, 01:31:32 AM

Quote from: m1n1ngP4d4w4n on January 04, 2017, 02:51:38 PM

Latest EQM seems to have stability issues, what about you all others fellow miners ? Im not able to keep it from crashing with same level even 5% lower OC than previous EQM version. Keep randomly crashing every 3-4h, happily i got my nice monitoring scripts to pick-up so mining downtime isn't that much but still annoying Sad

by the way nicehash guys, how do you use the monitoring/api port you can start with -a xxxx is that feature working ? i don't see any datas or port even being opened.

For me - the latest version is very unstable. I've got crashes a lot in compare with previous one.

Have you tried using the auto restart script?

nexitorned

member

Activity: 116

Merit: 10

Quote from: datBTC on January 04, 2017, 10:28:35 PM

Quote from: NiceHashSupport on January 04, 2017, 11:50:57 AM

We were able to reach 490 on overclocked GTX 1080. We also noticed that temperature has some effect on all pascal cards - the higher it is, the lower it would clock and thus slower it would mine. But the hashing rate was always consistent, never more or less, as long as all clocks were fine. Also, if there is a driver crash due to too intensive clock, it may put your card into locked low clock and reboot is needed. That is all I can help you with. You may observe GPU load, TDP, memory load and see how these numbers compare to your first run of 500+.

Yeah, maybe i just have a freak card or something. I stopped/started the miner today and boom the 1080 is hashing 530+ SOL/sec. Image below

Lol nice Hashrate, with GTX 1060 took me like 20 days to reach minimun payment with NiceHash, i had to move mining ZEC directly and seems more profitable. 1$ per day with 250SOL/sec. It's weid with NiceHash EQM give to me 210-250Sol/Sec 0 stability, but with another pool 250Sol/sec.

Dr_Victor

sr. member

Activity: 954

Merit: 250

Quote from: m1n1ngP4d4w4n on January 04, 2017, 02:51:38 PM

Latest EQM seems to have stability issues, what about you all others fellow miners ? Im not able to keep it from crashing with same level even 5% lower OC than previous EQM version. Keep randomly crashing every 3-4h, happily i got my nice monitoring scripts to pick-up so mining downtime isn't that much but still annoying Sad

by the way nicehash guys, how do you use the monitoring/api port you can start with -a xxxx is that feature working ? i don't see any datas or port even being opened.

For me - the latest version is very unstable. I've got crashes a lot in compare with previous one.

martanFTM

newbie

Activity: 17

Merit: 0

http://

Quote from: rednoW on January 04, 2017, 07:06:24 AM

Quote from: giagge on January 04, 2017, 06:58:15 AM

Quote from: rednoW on January 04, 2017, 06:32:21 AM

Quote from: martanFTM on January 03, 2017, 11:45:53 PM

488 sol/s on GTX 1070 // +200core / + 950mem - No TDP limit

kinda fake ... or more then 2.2ghz on gpu?

Power comsumption Huh

.

It doesn't matter. I think it is impossible without hardcore hardware mode overclock.

Thing is, this is really nice piece of silicon. I have it for 1600p Gaming. I have a rig with 4x 1070, all of them are identical to this one (even same batch), but they are some 5 percent worse - power/stability wise.

MSI Gaming X - Samsung memory (all of them)
running around 2167core (sometimes boosting over 2200mhz) and 1000 mem - memory bandwith is doing the numbers. On other cards i can reach some 850 - 900mem 24/7 stable, this one was running 1000 for 2 months. Now i slowed it down to 950, i think i will fry memory in some time.
So, below screenshot is just for show, that was just a max overclock clean run for you as a proof. I can assure you it is stable indeed, but not worth it long term.

So, this is standard BIOS gpu, air cooled, Windows 10 1607, 375.95 drivers, default NiceHash setup.

One last thing...This computer has IGP connected to LCD. If i connect Nvidia output - hash drops some 10 - 15 sols....So, this is maybe the answer what you looking for - no magic invoved

Good hashing guys.

hash screenshot:
http://i67.tinypic.com/16jfosn.png

PS: GPU power shows 88, which is rather lot - as i said, this is just for show. If i drop OC just a bit, I`m around 80 and still achieving 475-480 sol/s
So, this card is also achieving better temps and power consumption than the other identical MSI. Question is, if it is a GPU quality, or Nvidia is doing bad job in applying thermal paste consistently on all cards....I will not tear it apart to just look, but someone could disassemble, put quality thermal paste and try...

datBTC

newbie

Activity: 16

Merit: 0

Quote from: NiceHashSupport on January 04, 2017, 11:50:57 AM

We were able to reach 490 on overclocked GTX 1080. We also noticed that temperature has some effect on all pascal cards - the higher it is, the lower it would clock and thus slower it would mine. But the hashing rate was always consistent, never more or less, as long as all clocks were fine. Also, if there is a driver crash due to too intensive clock, it may put your card into locked low clock and reboot is needed. That is all I can help you with. You may observe GPU load, TDP, memory load and see how these numbers compare to your first run of 500+.

Yeah, maybe i just have a freak card or something. I stopped/started the miner today and boom the 1080 is hashing 530+ SOL/sec. Image below

https://s27.postimg.org/rp95lmckj/1080.png

bensam1231

legendary

Activity: 1764

Merit: 1024

Quote from: NiceHashSupport on January 04, 2017, 01:41:46 PM

There will be another update with 1-2% speed increase. Then we will work on to prepare public source release. We will put solver cuda_djezo into existing nheqminer: https://github.com/nicehash/nheqminer and bump the version up to 0.5. We made this decision, because nheqminer is widely popular (200 times forked) and will serve good as a base for this new solver (with possibility of other solvers being added in near future - something faster for AMD perhaps?).

Cryptonote?

m1n1ngP4d4w4n

full member

Activity: 224

Merit: 100

CryptoLearner

Quote from: NiceHashSupport on January 04, 2017, 05:41:38 PM

Quote from: m1n1ngP4d4w4n on January 04, 2017, 04:08:29 PM

Nicehash guys would be nice to be able to bind the admin port to a network card for remote query Wink

Since it was only 2 min of work, we added option to choose API bind IP. Will be in next 4c release.

Awesome, thx Shocked

i will be able to add to my little monitoring app/script local API in addition to NH web API for more accurate & responsive monitoring

NiceHashSupport

hero member

Activity: 588

Merit: 501

Quote from: m1n1ngP4d4w4n on January 04, 2017, 04:08:29 PM

Nicehash guys would be nice to be able to bind the admin port to a network card for remote query Wink

Since it was only 2 min of work, we added option to choose API bind IP. Will be in next 4c release.

m1n1ngP4d4w4n

full member

Activity: 224

Merit: 100

CryptoLearner

Quote from: felixbrucker on January 04, 2017, 05:35:04 PM

Quote from: m1n1ngP4d4w4n on January 04, 2017, 04:08:29 PM

Quote from: NiceHashSupport on January 04, 2017, 03:44:58 PM

Yes, this is used in NHM right now.

Regarding crashes. We will not deny that it cannot happen. We used some unsafe shared memory to gain extra speed. The crash possibility is very low. That is why there are auto-restart scripts included. If crash happens, it will not break your driver, so restart fixes it and you keep on mining.

Yeah that's why when i saw the script i was telling myself "hummm maybe this one isn't as stable since they put this script" ahah

Quote from: felixbrucker on January 04, 2017, 02:59:03 PM

Quote from: m1n1ngP4d4w4n on January 04, 2017, 02:51:38 PM

Latest EQM seems to have stability issues, what about you all others fellow miners ? Im not able to keep it from crashing with same level even 5% lower OC than previous EQM version. Keep randomly crashing every 3-4h, happily i got my nice monitoring scripts to pick-up so mining downtime isn't that much but still annoying Sad

by the way nicehash guys, how do you use the monitoring/api port you can start with -a xxxx is that feature working ? i don't see any datas or port even being opened.

for me it works as in nheqminer: you query "status\n" on a raw socket on the port and get a json result back (though the last version i tested it with was a while back, but if nothing broke it it should still work)

Ah ok got it, the port is only listening on localhost, so you can't remotely query it, how silly of me it was actually in plain sight in the help (-a [port] Local API port) . It indeed work on localhost, thx for sharing.

edit :

echo "status" |nc localhost 4444 |jq "."
{
  "method": "status",
  "result": {
   "connected": true,
   "interval_seconds": 300,
   "speed_ips": 1274.77,
   "speed_sps": 2380.54,

work, very nice

status was enough no need for the newline \n or carriage return \r

Nicehash guys would be nice to be able to bind the admin port to a network card for remote query Wink

"echo" automatically appends newline for you Wink

Oh right

echo -n indeed doesn't work, used to have echo -n as default for echo because of an alias but i don't have it on this cygwin ^^;;

NiceHashSupport

hero member

Activity: 588

Merit: 501

Quote from: m1n1ngP4d4w4n on January 04, 2017, 04:54:04 PM

Quote from: sp_ on January 04, 2017, 04:49:13 PM

Quote from: NiceHashSupport on January 04, 2017, 11:50:57 AM

We were able to reach 490 on overclocked GTX 1080. We also noticed that temperature has some effect on all pascal cards - the higher it is, the lower it would clock and thus slower it would mine. But the hashing rate was always consistent, never more or less, as long as all clocks were fine. Also, if there is a driver crash due to too intensive clock, it may put your card into locked low clock and reboot is needed. That is all I can help you with. You may observe GPU load, TDP, memory load and see how these numbers compare to your first run of 500+.

Did you try to remove compute61 from the project file. Use compute 52 or 60 instead. In my latest skein mod the pascal cards trottle too easy when compiled with 61.

Interesting

would it be why some pascal optimized miners seems to max out card better ?

Interesting idea, but did not improve anything.

felixbrucker

hero member

Activity: 700

Merit: 500

Quote from: m1n1ngP4d4w4n on January 04, 2017, 04:08:29 PM

Quote from: NiceHashSupport on January 04, 2017, 03:44:58 PM

Yes, this is used in NHM right now.

Regarding crashes. We will not deny that it cannot happen. We used some unsafe shared memory to gain extra speed. The crash possibility is very low. That is why there are auto-restart scripts included. If crash happens, it will not break your driver, so restart fixes it and you keep on mining.

Yeah that's why when i saw the script i was telling myself "hummm maybe this one isn't as stable since they put this script" ahah

Quote from: felixbrucker on January 04, 2017, 02:59:03 PM

Quote from: m1n1ngP4d4w4n on January 04, 2017, 02:51:38 PM

Latest EQM seems to have stability issues, what about you all others fellow miners ? Im not able to keep it from crashing with same level even 5% lower OC than previous EQM version. Keep randomly crashing every 3-4h, happily i got my nice monitoring scripts to pick-up so mining downtime isn't that much but still annoying Sad

by the way nicehash guys, how do you use the monitoring/api port you can start with -a xxxx is that feature working ? i don't see any datas or port even being opened.

for me it works as in nheqminer: you query "status\n" on a raw socket on the port and get a json result back (though the last version i tested it with was a while back, but if nothing broke it it should still work)

Ah ok got it, the port is only listening on localhost, so you can't remotely query it, how silly of me it was actually in plain sight in the help (-a [port] Local API port) . It indeed work on localhost, thx for sharing.

edit :

echo "status" |nc localhost 4444 |jq "."
{
  "method": "status",
  "result": {
   "connected": true,
   "interval_seconds": 300,
   "speed_ips": 1274.77,
   "speed_sps": 2380.54,

work, very nice

status was enough no need for the newline \n or carriage return \r

Nicehash guys would be nice to be able to bind the admin port to a network card for remote query Wink

"echo" automatically appends newline for you Wink

m1n1ngP4d4w4n

full member

Activity: 224

Merit: 100

CryptoLearner

Quote from: sp_ on January 04, 2017, 04:49:13 PM

Quote from: NiceHashSupport on January 04, 2017, 11:50:57 AM

We were able to reach 490 on overclocked GTX 1080. We also noticed that temperature has some effect on all pascal cards - the higher it is, the lower it would clock and thus slower it would mine. But the hashing rate was always consistent, never more or less, as long as all clocks were fine. Also, if there is a driver crash due to too intensive clock, it may put your card into locked low clock and reboot is needed. That is all I can help you with. You may observe GPU load, TDP, memory load and see how these numbers compare to your first run of 500+.

Did you try to remove compute61 from the project file. Use compute 52 or 60 instead. In my latest skein mod the pascal cards trottle too easy when compiled with 61.

Interesting

would it be why some pascal optimized miners seems to max out card better ?

Topic: NiceHash EQM Zcash NVIDIA optimized miner [Maxwell/Pascal] + CPU mining v1.0.4c - page 6. (Read 238052 times)