excavator by NiceHash - multi-algorithm NVIDIA GPU miner [1.5.6a] - page 61.

NiceHashSupport

hero member

Activity: 588

Merit: 501

Quote from: jwarren81 on March 20, 2017, 12:55:41 PM

Can you post the command line you are using for overclocking and lowering the TDP for the 1070s? Just want to see an example to work from.

Thanks.

If your cards ID is 0 then the example for setting +100 core, +700 mem and 60 tdp would be:

Quote from: kalhohan on March 20, 2017, 02:03:30 PM

Quote from: NiceHashSupport on March 20, 2017, 12:05:07 PM

If you are so affraid of "frying" your card, then perhaps you shouldn't be mining at all and just let the card be in idle mode

I don't think it needed sarcasm, but it was fun, so it's ok Wink

Again I'm no expert in mining on cuda, and I see no one on gaming forum who overclock this high the memory of a 1080. So +800 feels like possible bacon smell to me, maybe not.

And yes, when I'm mining with my AMD rigs or my this nvidia gaming pc, I consider all the costs even the burning overkill of a card, risks cost compared to profits. Correct me if I'm wrong but don't see that as being frightened by the outcome of an extreme oc.

Overclocking AMD of 10% with modified bios for 200€ (in France) 480s is ok because the cards are "cheap" compared to the nvidia ones.

Haven't seen a Pascal card to die because of high overclock. We did destroy one due to water leakage and some stuff was burnt on it, but never due to high memory clock and we are running them 24/7 mining. Just make sure you have adequate cooling, which you already need if you are mining 24/7.

kalhohan

newbie

Activity: 17

Merit: 0

Quote from: NiceHashSupport on March 20, 2017, 12:05:07 PM

If you are so affraid of "frying" your card, then perhaps you shouldn't be mining at all and just let the card be in idle mode

I don't think it needed sarcasm, but it was fun, so it's ok Wink

Again I'm no expert in mining on cuda, and I see no one on gaming forum who overclock this high the memory of a 1080. So +800 feels like possible bacon smell to me, maybe not.

And yes, when I'm mining with my AMD rigs or my this nvidia gaming pc, I consider all the costs even the burning overkill of a card, risks cost compared to profits. Correct me if I'm wrong but don't see that as being frightened by the outcome of an extreme oc.

Overclocking AMD of 10% with modified bios for 200€ (in France) 480s is ok because the cards are "cheap" compared to the nvidia ones.

jwarren81

full member

Activity: 211

Merit: 100

Can you post the command line you are using for overclocking and lowering the TDP for the 1070s? Just want to see an example to work from.

Thanks.

NiceHashSupport

hero member

Activity: 588

Merit: 501

Quote from: kalhohan on March 20, 2017, 11:41:38 AM

Quote from: NiceHashSupport on March 20, 2017, 10:45:23 AM

When you overclock memory, your power consuption does not increase much, maybe for up to few % only. But if algorithm is coded properly and relies more on memory, you will gain more speed at the end with lower TDP.

In fact, you can reach 510 sol with 5500 mem at 126W TDP (also doing core OC). And you get full 510 sols at the pool because there is no fee.

With the risk of frying the memory? how is it better coded when you need faster device (with overclocking, nearly 20%) to get the same result with 20 watt less?
I just don't see what we have to gain, in term of profit, by overclocking 600$ card when it's obviously not needed.

and for the devfee at 2% it's still higher sol/s .

For one card, minus pc wattage (around 160 w) taken with the prob on the wall:

excavator (with +800 memory )
power cost
per day 0.43
per month 12.9

ewbf (without oc)
power cost
per day 0.53
per month 15.9

profit 3 $ per month even with a big farm the risk doesn't seem to match the profit, taken that oc can fry your card memory, but maybe I 'm wrong.

If you are so affraid of "frying" your card, then perhaps you shouldn't be mining at all and just let the card be in idle mode

ExploitAgency

member

Activity: 70

Merit: 11

https://github.com/wine-compholio/wine-staging/wiki/Installation

Closest to running on linux is using wine-staging but it throws constant errors.

Maybe someone with more time can figure it out if a Linux binary/source doesn't come out soon. I am not a Wine expert by any means. In a similar bit with GPU's I've only ran scallion through mono.

Code:

fixme:msvcrt:_CurrentScheduler_Id stub
fixme:msvcrt:_CurrentScheduler_Id stub
fixme:msvcrt:_CurrentScheduler_Id stub
fixme:msvcrt:_CurrentScheduler_Id stub
fixme:msvcrt:_CurrentScheduler_Id stub

Startup

Code:

wine excavator.exe -a pascal -s pascal.eu.nicehash.com:3358 -u 1LHVh4ZraCMRfKQvtvJx7u3TT7gYvE462F.1070 -cd 0
fixme:winediag:start_process Wine Staging 2.3 is a testing version containing experimental patches.
fixme:winediag:start_process Please mention your exact version when filing bug reports on winehq.org.
fixme:heap:RtlSetHeapInformation 0x330000 0 0x23fd30 4 stub
fixme:msvcp:_Mtx_init_in_situ unknown flags ignored: 2
fixme:msvcp:_Locinfo__Locinfo_ctor_cat_cstr (0x23f840 1 C) semi-stub

==================== www.nicehash.com ====================
         Excavator v1.1.3a GPU Miner for NiceHash.
    Copyright (C) 2017 NiceHash. All rights reserved.
  Developed by djeZo with help and contributors of code:
    pallas, Vorksholk, bitbandi, ocminer
==================== www.nicehash.com ====================

fixme:msvcrt:_getptd returns undocumented/not fully filled data
fixme:msvcrt:_getptd returns undocumented/not fully filled data
[12:02:44][0x0000002f][info] Log started
fixme:msvcp:_Mtx_init_in_situ unknown flags ignored: 2
fixme:msvcrt:_getptd returns undocumented/not fully filled data
fixme:msvcrt:_getptd returns undocumented/not fully filled data
[12:02:44][0x0000002f][info] Miner starting up, press [CTRL + C] to quit
^Cfixme:msvcrt:_getptd returns undocumented/not fully filled data
fixme:msvcrt:_getptd returns undocumented/not fully filled data
[12:02:44][0x0000002f][info] net | Connecting to pascal.eu.nicehash.com:3358
fixme:msvcp:_Mtx_init_in_situ unknown flags ignored: 2
fixme:msvcrt:_getptd returns undocumented/not fully filled data
fixme:msvcrt:_getptd returns undocumented/not fully filled data
[12:02:44][0x0000002f][info] Shutting down
fixme:msvcrt:_getptd returns undocumented/not fully filled data
fixme:msvcp:_Locinfo__Locinfo_ctor_cat_cstr (0xe0de00 1 C) semi-stub
fixme:msvcp:_Locinfo__Locinfo_ctor_cat_cstr (0xe0d8d0 1 C) semi-stub
fixme:msvcrt:_getptd returns undocumented/not fully filled data
[12:02:44][0x00000033][info] device#0 | Initializing thread #0 GPU#0 GeForce GTX 1070
fixme:msvcrt:_getptd returns undocumented/not fully filled data
fixme:msvcrt:_getptd returns undocumented/not fully filled data
[12:02:44][0x00000030][info] net | Connected!
fixme:msvcrt:_getptd returns undocumented/not fully filled data
fixme:msvcrt:_getptd returns undocumented/not fully filled data
[12:02:44][0x00000033][info] device#0 | Algorithm: CUDA-pascal parameters: B=122835,TPB=512
fixme:msvcrt:_getptd returns undocumented/not fully filled data
fixme:msvcrt:_getptd returns undocumented/not fully filled data
[12:02:44][0x00000033][info] device#0 | Thread #0 ended (GPU#0 GeForce GTX 1070)

kalhohan

newbie

Activity: 17

Merit: 0

Quote from: NiceHashSupport on March 20, 2017, 10:45:23 AM

When you overclock memory, your power consuption does not increase much, maybe for up to few % only. But if algorithm is coded properly and relies more on memory, you will gain more speed at the end with lower TDP.

In fact, you can reach 510 sol with 5500 mem at 126W TDP (also doing core OC). And you get full 510 sols at the pool because there is no fee.

With the risk of frying the memory? how is it better coded when you need faster device (with overclocking, nearly 20%) to get the same result with 20 watt less?
I just don't see what we have to gain, in term of profit, by overclocking 600$ card when it's obviously not needed.

and for the devfee at 2% it's still higher sol/s .

For one card, minus pc wattage (around 160 w) taken with the prob on the wall:

excavator (with +800 memory )
power cost
per day 0.43
per month 12.9

ewbf (without oc)
power cost
per day 0.53
per month 15.9

profit 3 $ per month even with a big farm the risk doesn't seem to match the profit, taken that oc can fry your card memory, but maybe I 'm wrong.

NiceHashSupport

hero member

Activity: 588

Merit: 501

Quote from: kalhohan on March 20, 2017, 08:48:28 AM

Quote from: NiceHashSupport on March 20, 2017, 08:04:05 AM

Up mem for +700.

Here, with mem upped +800 (5500), 500 sol @ 60% TDP (126W). Make sure you up the mem with excavator's OC functionality and also set '-or'. Use '-od' if your PC is a bit slow and two threads aren't initialized BEFORE overclocks are applied. If you use third party tools for OCing, you will not be able to reach so high memory as for some reason, when CUDA is initialized/deinitialized, cards go into p0 state for short period of time and with very high memory overclock this can result in crash (memory clock p2 state is always lower).

Well pardon me, maybe I'm missing something important, but what's the point to overclock up to 800 memory for 500 sols/s, if I obtain the same speed with no overclock on another soft? How can it be safer?

When you overclock memory, your power consuption does not increase much, maybe for up to few % only. But if algorithm is coded properly and relies more on memory, you will gain more speed at the end with lower TDP.

In fact, you can reach 510 sol with 5500 mem at 126W TDP (also doing core OC). And you get full 510 sols at the pool because there is no fee.

ExploitAgency

member

Activity: 70

Merit: 11

^{+1 for LINUX VERSION PLEASE!}

reb0rn21

legendary

Activity: 1901

Merit: 1024

@kalhohan
Its different kernel and miner...... they don`t use the memory with same utilization

kalhohan

newbie

Activity: 17

Merit: 0

Quote from: NiceHashSupport on March 20, 2017, 08:04:05 AM

Up mem for +700.

Here, with mem upped +800 (5500), 500 sol @ 60% TDP (126W). Make sure you up the mem with excavator's OC functionality and also set '-or'. Use '-od' if your PC is a bit slow and two threads aren't initialized BEFORE overclocks are applied. If you use third party tools for OCing, you will not be able to reach so high memory as for some reason, when CUDA is initialized/deinitialized, cards go into p0 state for short period of time and with very high memory overclock this can result in crash (memory clock p2 state is always lower).

Well pardon me, maybe I'm missing something important, but what's the point to overclock up to 800 memory for 500 sols/s, if I obtain the same speed with no overclock on another soft? How can it be safer?

NiceHashSupport

hero member

Activity: 588

Merit: 501

Quote from: kalhohan on March 20, 2017, 07:56:18 AM

Quote from: du44 on March 20, 2017, 07:21:41 AM

Try -ca -ca or -ca -ca -ca

It was already -ca -ca
I tried the triple -ca and the result is 413 sols/s

For information, I'm on windows 10 pro with driver 378.78 for nvidia.
And by the way msi afterburner shows that the card is running at 73% instead of 84 % with EWBF ( tdp at 85%)

Up mem for +700.

Here, with mem upped +800 (5500), 500 sol @ 60% TDP (126W). Make sure you up the mem with excavator's OC functionality and also set '-or'. Use '-od' if your PC is a bit slow and two threads aren't initialized BEFORE overclocks are applied. If you use third party tools for OCing, you will not be able to reach so high memory as for some reason, when CUDA is initialized/deinitialized, cards go into p0 state for short period of time and with very high memory overclock this can result in crash (memory clock p2 state is always lower).

kalhohan

newbie

Activity: 17

Merit: 0

Quote from: du44 on March 20, 2017, 07:21:41 AM

Try -ca -ca or -ca -ca -ca

It was already -ca -ca
I tried the triple -ca and the result is 413 sols/s

For information, I'm on windows 10 pro with driver 378.78 for nvidia.
And by the way msi afterburner shows that the card is running at 73% instead of 84 % with EWBF ( tdp at 85%)

du44

full member

Activity: 251

Merit: 100

Quote from: kalhohan on March 20, 2017, 06:45:16 AM

Hi from my side with my 1080, i don't have the expected results:

without overclocking, and with powerlimit at 85%
With EWBF : 500 sols/s
with escavator 1.1.3a: 400 sols/s

Am I doing something wrong?

Many thanks.

Try -ca -ca or -ca -ca -ca

kalhohan

newbie

Activity: 17

Merit: 0

Hi from my side with my 1080, i don't have the expected results:

without overclocking, and with powerlimit at 85%
With EWBF : 500 sols/s
with escavator 1.1.3a: 400 sols/s

Am I doing something wrong?

Many thanks.

Newton90

sr. member

Activity: 463

Merit: 256

have 85 - 87 hash per thread on asus gtx1060 o3g (65% tdp,core +170),total 255-261 hash.

on the same sets have 26x - 27x hash/s with ewbf.

without fee ewbf speed is faster.need improvements.

minerx117

sr. member

Activity: 728

Merit: 256

NemosMiner-v3.8.1.3

Hi Ive been testing your new version 1.1.3a been having great results with the 1070's at -30-40 tdp very similar speeds to EWBF now (well done) and with no fee(thank-you)

im trying to add excavator to my multi algo Miner.. it is working but as my Miner switches algos by stratum disconnect.. excavator will not close and instead it say's stratum disconnect retrying in 12 seconds and just stays on loop doing that..

my question is there a command a can add to the bat file to make excavator close when stratum is disconnected?

i have tried excavator.exe -h for a full list of current commands but there doesn't seem to be anything related to stratum disconnect or close if connection lost..

your help would be much appreciated

Regards
Nemo

Kompik

sr. member

Activity: 463

Merit: 250

Quote from: reb0rn21 on March 19, 2017, 02:58:23 PM

you have two options:
afterburner freq/voltage graph edit (ctrl+f) + locking with "L" the desired freq/volt (I keep my MSI at 1980Mhz at 0.95v)

setting low TDP + huge OC like 100-200Mhz offset which at the end produce desired frequency at lower voltage, thise setting depend on card bios (gainward phoenic can do 80% TDP limit with +200Mhz and +800 on memory) which give you 1900Mhz on GPU at 0.91v and ~120w usage

Yes those things I know, thanks!

reb0rn21

legendary

Activity: 1901

Merit: 1024

you have two options:
afterburner freq/voltage graph edit (ctrl+f) + locking with "L" the desired freq/volt (I keep my MSI at 1980Mhz at 0.95v)

setting low TDP + huge OC like 100-200Mhz offset which at the end produce desired frequency at lower voltage, thise setting depend on card bios (gainward phoenic can do 80% TDP limit with +200Mhz and +800 on memory) which give you 1900Mhz on GPU at 0.91v and ~120w usage

Kompik

sr. member

Activity: 463

Merit: 250

Quote from: reb0rn21 on March 19, 2017, 02:43:05 PM

It depend how you limit freq, excaviator use less TDP so card will boost more in same TDP...
for me GTx1070 giv me same on pool with TDP od 120w some ~440sol/s per card (heavy undervolt +OC with TDP limit)

How do you even undervolt the 1070? It just does absolutely nothing?

reb0rn21

legendary

Activity: 1901

Merit: 1024

It depend how you limit freq, excaviator use less TDP so card will boost more in same TDP...
for me GTx1070 giv me same on pool with TDP od 120w some ~440sol/s per card (heavy undervolt +OC with TDP limit)

Topic: excavator by NiceHash - multi-algorithm NVIDIA GPU miner [1.5.6a] - page 61. (Read 179268 times)