lolMiner 1.53: Dual mining Eth+Kaspa and more efficient AMD kernels

[Lolliedieb]

hi its possible to run in background lolminer with some parameter in json file?

Not yet - well on Linux you can just append & on the cmd to run it completely in background.

Memory: Frequency - 1100Mhz, 905 mV.
GPU: P5 1212 Mhz, 1100mV. P6 1307Mhz, 905 mV. P7 1472 Mhz, 925 mV.
Radeon 19.12.1.
CPU Intel i3-4170
teamredminer-v0.6.1 - work.
lolMiner-0.9.3 don't work.
"Stuck device detected, invoking emergency script."

Lifting my virtual hat then for trm (I like the guys anyways - they are great and made a great product!). I must admit I did not try that high memclk - afaik only Samsung mem allows to clock that high and also that often requite quite high core voltage.
I tested - and optimized - the miner more for more watts saving clocks like 1280 / 1000 which runs often with voltages low as 825mV (some cards may need 850 or 862 ...). At least on Linux, 19.30 the 0.9.4 miner then still mines at ~1.15 g/s (rocm 1.2 when you got no risers) but also power draw is <160W then (chip reported even <125, but always need to add some for board and memory).

With P7 at 1470 / 1100 I am sure you can not end up below  230? (more a guess ^^)

[1660720350]

1660720350

[1660720350]

1660720350

[1660720350]

1660720350

[1660720350]

1660720350

[kidus]

With P7 at 1470 / 1100 I am sure you can not end up below  230? (more a guess ^^)

I started with low frequency and voltage. Then lifted above.
https://yadi.sk/i/pj9PGJFXOafMEg
https://yadi.sk/i/Wg0OH_zF9MU4ag
https://yadi.sk/i/UBrqhdqPqM-2xA
https://yadi.sk/i/3y8nXCxyEjgTxQ
https://yadi.sk/i/GXRQu7x2onCWmA

[Lolliedieb]

Looks like healthy settings - but remember that the 190W displayed are chip power only - need to add memory and board for AMDs (else I could claim my miner would do 1.15 g/s at 100W - which is chip power only, but obviously cheating ^^)

[kidus]

I have two VEGA-64 cards that can 't work on such high settings.
Settings of these cards:
GPU 1032 Mhz 905 mV. Memory: 1020 Mhz, 905 mV.
Consumes 170W GPU power.
0,97 g/s

[MATHReX]

I have two VEGA-64 cards that can 't work on such high settings.
Settings of these cards:
GPU 1032 Mhz 905 mV. Memory: 1020 Mhz, 905 mV.
Consumes 170W GPU power.
0,97 g/s

Use AMD Mem Tweaker to apply weak settings on Vega Cards.
That should provide a bit boost to the hash rate.
Also, undervolt the core more as undervolting memory doesn't do anything.

[laik2]

I have two VEGA-64 cards that can 't work on such high settings.
Settings of these cards:
GPU 1032 Mhz 905 mV. Memory: 1020 Mhz, 905 mV.
Consumes 170W GPU power.
0,97 g/s

Use AMD Mem Tweaker to apply weak settings on Vega Cards.
That should provide a bit boost to the hash rate.
Also, undervolt the core more as undervolting memory doesn't do anything.

CuckaToo31 is more of compute intensive algorithm, amdmemorytweak will mostly raise power consumtion rather than increase hashrate drasticly...

[zeef]

hi its possible to run in background lolminer with some parameter in json file?

Not yet - well on Linux you can just append & on the cmd to run it completely in background.

Memory: Frequency - 1100Mhz, 905 mV.
GPU: P5 1212 Mhz, 1100mV. P6 1307Mhz, 905 mV. P7 1472 Mhz, 925 mV.
Radeon 19.12.1.
CPU Intel i3-4170
teamredminer-v0.6.1 - work.
lolMiner-0.9.3 don't work.
"Stuck device detected, invoking emergency script."

Lifting my virtual hat then for trm (I like the guys anyways - they are great and made a great product!). I must admit I did not try that high memclk - afaik only Samsung mem allows to clock that high and also that often requite quite high core voltage.
I tested - and optimized - the miner more for more watts saving clocks like 1280 / 1000 which runs often with voltages low as 825mV (some cards may need 850 or 862 ...). At least on Linux, 19.30 the 0.9.4 miner then still mines at ~1.15 g/s (rocm 1.2 when you got no risers) but also power draw is <160W then (chip reported even <125, but always need to add some for board and memory).

With P7 at 1470 / 1100 I am sure you can not end up below  230? (more a guess ^^)

Thanks for answer,
Cant u do in next releases the background windows function ?
Thanks

[Lolliedieb]

I have two VEGA-64 cards that can 't work on such high settings.
Settings of these cards:
GPU 1032 Mhz 905 mV. Memory: 1020 Mhz, 905 mV.
Consumes 170W GPU power.
0,97 g/s

With 1032 mhz core you can go much lower in voltage. This way its inbalanced because
-The very low core loweres performance
- The voltage needs to be high because mem clock is high
- The high voltage causes too high consumption


Rather go to 1200 / 920 / 835mV ... that should be (0.9.4) slightly over 1.1g/s at much lower consumption.
In mmpos chanel I saw one guy running 6x V64 on 1120 / 920 / 760mv (! Not all cards will alloe that). He had 6.3 g/s on all cards and 920W at wall (6card + cpu + mobo + ram...)

Thanks for answer,
Cant u do in next releases the background windows function ?
Thanks

Can try, sure
But before need to solve the miracle how to port the 0.9.4 speed improvement to Windows :/

[ctsbitcoin]

this is the announcement thread for lolMiner 0.9.4 a multi algorithm Equihash / Beam & Grin miner with focus on AMD GPUs (OpenCL based).
Note it will run on most Nvidia GPUs, too but the speed and stability may vary.

Just wanted to say that 0.9.3 for Windows has been awesome!  Ever since I upgraded I have not had to babysit the rigs for crashing GPUs.  Really looking forward to the hashing increases of 0.9.4! 

[joeykolb]

u cant implement the x16v2 for lolminer?

[zeef]

I have two VEGA-64 cards that can 't work on such high settings.
Settings of these cards:
GPU 1032 Mhz 905 mV. Memory: 1020 Mhz, 905 mV.
Consumes 170W GPU power.
0,97 g/s

With 1032 mhz core you can go much lower in voltage. This way its inbalanced because
-The very low core loweres performance
- The voltage needs to be high because mem clock is high
- The high voltage causes too high consumption


Rather go to 1200 / 920 / 835mV ... that should be (0.9.4) slightly over 1.1g/s at much lower consumption.
In mmpos chanel I saw one guy running 6x V64 on 1120 / 920 / 760mv (! Not all cards will alloe that). He had 6.3 g/s on all cards and 920W at wall (6card + cpu + mobo + ram...)

Thanks for answer,
Cant u do in next releases the background windows function ?
Thanks

Can try, sure
But before need to solve the miracle how to port the 0.9.4 speed improvement to Windows :/

hehe thanks

[kaiokenx10]

So is it more profitable to mine c32 or c31 now ?

[WhyMe]

Any chance to have BFC variant ?

[tromp]

So is it more profitable to mine c32 or c31 now ?

Until the C31 phaseout starts around Jan 15, C31 will certainly be more profitable.
It depends on your hardware when C32 becomes more profitable, but likely not until April...

[Lolliedieb]

Until the C31 phaseout starts around Jan 15, C31 will certainly be more profitable.
It depends on your hardware when C32 becomes more profitable, but likely not until April...

Well I would guess a little earlier - lets assume a Vega64 does 1.3 g/s C31 and 0.4 g/s C32. So 30.7% of the speed. Considering the higher scale of C32 the profitability is right now 63% of C31. So at current speeds it will take until end of March until the profitability flips. But C32 is barely optimized yet

Ps: currently on travel, sry for the silence. Also still waiting for reply from AMD about the Windows issue :/

[Lolliedieb]

Good and bad news from AMD (Link if anyone wants to read themselves why)

It seems I will be able to port 0.9.4 speedup for Navi on Windows, but not for Vega 56/64 and VII. "Stupid" driver limitation imao - well at least they are aware of the issue now. Will need to figure out if I see a way to speed it up without using the extra memory (in that case also 470/480/570 and 580 would profit ... I will try...)

[tale01]

Can you support MoneroV (XMV) Algorithm: Cuckarood29v  ?
Official Miner https://github.com/xmvdev/MoneroVMiner/releases

I need to support 4G 

[xMindx163]

Can you support MoneroV (XMV) Algorithm: Cuckarood29v  ?
Official Miner https://github.com/xmvdev/MoneroVMiner/releases

I need to support 4G 

As soon as there is support with manners, the coin will be rolled into the bottom. So you don't need anything. Be happy with what you have.

[Lolliedieb]

Happy new Year you all

Can you support MoneroV (XMV) Algorithm: Cuckarood29v  ?
Official Miner https://github.com/xmvdev/MoneroVMiner/releases

I need to support 4G 

As soon as there is support with manners, the coin will be rolled into the bottom. So you don't need anything. Be happy with what you have.

Just hearing of MoneroV the first time. Wonder what this is about. Well it seems there are only changes in the  cycle length and some small stuff in the stratum (Hell, why people always fail to document that stuff when they fork a coin)... but seriously: There is not even 1k g/s using my C29D code although its the only 4G implementation for AMDs ... how much can I expect from this - the network not even is 1/100 the size of grin. That likely won't roi the time for me changing the code

[pbfarmer]

HNY!

Just wanted to post some c31 settings (for vega64) which may be more interesting to people w/ prices so low these days.  In general, it seems that 1150cclk/800mclk @ 800mv is pretty safe - for about 1.03-1.05gps.  Only one of my 8 Nitros needed a tiny bump down on cclk (to 1140MHz) to be stable.  Some can go up quite a bit while still maintaining 800mv - the last two in this screenshot are running w/ a setting of 1250+ cclk for ~1.1gps (I haven't tuned the others yet.)  I'm seeing an avg of about 150-153w/GPU ATW on a 1600 plat PSU @ 120v, which should be significantly more efficient than any of the settings provided in the ann.