[ANN] ccminer 2.3 - opensource - GPL (tpruvot)

[Epsylon3]

no rie

[rednoW]

I must admit that problem "nonce xxxxxxxxx was already sent 14 seconds ago" was due to too much intensity.
sp_ when he was advertising his private crap recommended to use -i 28 or even -i 29 that causes high absolute numbers on the miner but this nonce errors and extreme unstable and low poolside hashrate.
Now I use -i 26 settings and problem seems to vanish still I'm a bit disappointed with poolside hashrate. The Wolf amd miner is better in this sense.

Bellow is the chart of 1 gtx1070 hashing 300mhs
To the left - with sp_ recommended intensity - to the right - normal intensity.
pool side hashrate drop is observed both in sp_private3 and epsylon's 1.8.2 if setting intensity to high (28-29)
 

[BasementRot]

So any idea why I only get 280MHs with actual running core clock of 2090mhz and uncapped wattage?

I tried -i 23, 28 and 29 and it didn't make hardly any difference.

[bit815792215]

So any idea why I only get 280MHs with actual running core clock of 2090mhz and uncapped wattage?

I tried -i 23, 28 and 29 and it didn't make hardly any difference.

maybe change riser, reinstall a few older drivers, see if card is reaching thermal limits

[BasementRot]

So any idea why I only get 280MHs with actual running core clock of 2090mhz and uncapped wattage?

I tried -i 23, 28 and 29 and it didn't make hardly any difference.

maybe change riser, reinstall a few older drivers, see if card is reaching thermal limits

not the risers or thermal limits, i think I'm already using the oldest drivers with pascal chip support

I'm hoping someone who gets 300+ knows why. What version of cuda and nvidia drivers are you using?

[bucketofsocks]

when passing --gpu-clock=1949 in my batch file I end up with a clock of 2075 as read by gpu-z, can anybody tell me why? is there a ratio to know what its going to be?

[Epsylon3]

its an offset issue, the public nvidia api doesnt know the reference clock once its changed :/ well.. there is a trick to find to get them correctly...

in conclusion, dont mess with clocks prefer tlimit and plimit

its only correct when the card use defaults clocks. Can be done once manually on boot, but for now... its better to remove that from batches

[tbearhere]

Does anyone know the default intensities on the 970 gtx and 980ti on windows 8.1 on 1.8.2   plz...... thx
Is there away to find these with a command?

[crysx]

Does anyone know the default intensities on the 970 gtx and 980ti on windows 8.1 on 1.8.2   plz...... thx
Is there away to find these with a command?

hehehe ...

ive been asking for this for years now mate ...

alexis78 has come through with one of my requests for default intensity notification at miner start up - but no one has put down a table of intensities that are default to a specific version of a specific miner ( as they change per miner ) ...

alexis78 may do this - if we 'push' him enough ( ie - donate maybe? ) ...

or epsylon3? ... or ...

...

#crysx

[Epsylon3]

its already commited, will be in 1.8.3

else... check yiimp benchs, its often the default

[giagge]

its already commited, will be in 1.8.3

else... check yiimp benchs, its often the default

Thanks 1000 .

1.8.3 when released ? .

[Epsylon3]

dont ask for ETA, im not a company

[TomDD]

Apologies if I've missed this somewhere in the thread but I've noticed an interesting behaviour of my 1070 GTX. Basically I normally get around 272MH/s with 1.8.2 but have noticed sometimes it has been dipping to around 225MH/s and I couldn't figure out why. Anyway I did a bit a digging and noticed if the Memory Usage on the GPU (monitored with GPU-Z) drops below around 512MB the GPU Core Clock drops down to a default speed of around 1531MHz. If I open a web browser or something which increases the GPU memory usage it pushes up the GPU core clock up to max and I get 272MH/s again.

Anybody else seen this? I'm using the latest NVIDIA drivers 372.90

TomDD

[TomDD]

Apologies if I've missed this somewhere in the thread but I've noticed an interesting behaviour of my 1070 GTX. Basically I normally get around 272MH/s with 1.8.2 but have noticed sometimes it has been dipping to around 225MH/s and I couldn't figure out why. Anyway I did a bit a digging and noticed if the Memory Usage on the GPU (monitored with GPU-Z) drops below around 512MB the GPU Core Clock drops down to a default speed of around 1531MHz. If I open a web browser or something which increases the GPU memory usage it pushes up the GPU core clock up to max and I get 272MH/s again.

Anybody else seen this? I'm using the latest NVIDIA drivers 372.90

TomDD

I've investigated this a little more and it may not be memory but something to do with the driver and how it turns on the GPU boost. As a test I opened a browser (Edge) and the clock was at 1531MHz and the Memory Load 0.57GB. Load up this site to type this reply and the clock was still at 1531MHz. I then opened a new tab and simply doing this pushes the core clock to 1873MHz, exactly the same Memory Load. Closing the new tab the Clock drops to 1531MHz??? This is repeatable. I've a ton of things running the background but none of these make any difference. I tried a few other programs, playing a video with the built in Windows 10 player pushes it up to 1873MHz again but playing a video with VLC doesn't? Anyway to keep the core clock high I have to leave Edge open with two tabs, which isn't a problem but am interested why this is happening.

Run for a while now and it is stable with 69-72C temp range and the clock fluctuates between 1847.5Mhz and 1873MHz.

TomDD

[Epsylon3]

Yep.. as often with nvidia... the card is not shipped with a complete doc And for now the pascal drivers seems to be a features/bug lottery.

Too much AI i guess :p With random bugs to make it more "human"

[pallas]

My pascal rig is using much more cpu than the maxwell one (so it wastes a bit of power).
Probably a driver or cuda issue, maybe linux only.

[ioglnx]

Well I noticed that on my GTX1080 too..temp side they are good and the boost keeps at both on 2100MHz.
But the MH/s is dropping from 820MH/s to 780ish even the difficulty is the same. For now I have a little powershell watchdog that monitors the hash and restarted ccminer if it drops below 790MH/s.

So I made in 24h 10more lbry coins.

[rednoW]

My pascal rig is using much more cpu than the maxwell one (so it wastes a bit of power).
Probably a driver or cuda issue, maybe linux only.

I have very slow cpus (amd athlon 64 x2) in my rigs and no problem with pascal card in windows.

[go6ooo1212]

My pascal rig is using much more cpu than the maxwell one (so it wastes a bit of power).
Probably a driver or cuda issue, maybe linux only.

I have very slow cpus (amd athlon 64 x2) in my rigs and no problem with pascal card in windows.

Mine Pascal-rigs don't consume additional CPU - power also. I'm on win7 and 8.1 x64...

[Amph]

it's not sitting on 0% either, my cpu is at 30-50% all the time, so there is definitely consumption coming from the cpu