Topic: ccminer 1.2.0/sgminer 0.1.3 for MTP: Fastest MTP miner for nvidia cards - page 8. (Read 4614 times)

I have a few rigs with 3 or 4 1080ti, they work perfectly. But for rigs with 7x 1080ti, ram usage spikes to 99% within 10s of running the miner, hanging the whole system. Do I need more ram or page size, or is the issue due to something else?

Specs:

7x 1080ti FE at 70% PL
G4560
4GB Ddr4
67gb virtual ram

1600w psu


The other 3/4 1080ti rigs are doing 3.9 - 4.1mh/s each.

I've got 4.2mhs at 220 watt (gtx1080ti). Nice!

The default one seems to give the best result (-i 20).
Regarding OC, you'll get a substantial hash increase by increasing core clock (or power limit... but obviously it depends a lot on electricity cost)
+150 on core does nicely

What intensity to use for 1080ti and what oc?

driver is 430.64

The one I use is called "OhGodAnETHlargementPill-r2.exe"
You can probably google it (I prefer not to give a link because there are quite a few and none of them are linked to OhGodAGirl...)

I currently use drivers 430.86 the gaming version

It seems no effect if I use the ohmygod pill or I'm using a wrong pill. I have old ethereum pill from ohmygod from github before it's working in ethereum but when I try to use ohmygodpill in Zcoin mining it doesn't seem to work?

This is the complete name "OhGodAnETHlargementPill-master" can you provide the working for MTP?


What version of driver you currently installed with 1080ti?

and 2.2 MH/s on 1070 @ 85% power, also better than other miners & previous version

good improvement vs previous version, I get 4 MHh/s with 1080 ti and pill @85% power on mintpond, much higher than others...
(would be good to have nicehash, they sometimes pay more than regular mining!)

sorry but for the moment there is no support for nicehash

Hi djm34,

I appreciate your contributions; here is some feedback for your consideration.
I haven't been able to mine successfully with this release.
Hashrate appears competitive with t-rex, but all my shares are rejected (Nicehash pool).

- 7x 1070ti
- 8GB RAM, 80GB virtual RAM
- Afterburner settings: PL 72%, Core +200, Mem +0

My .BAT string:

ccminer -a mtp -o stratum+tcp://mtp.usa.nicehash.com:3374 -u WALLET.USER -p x

wow...
seriously... if it is related to trust rating, it is a random guy 4 years ago, who put a -1 trust over a sarcastic comment I made on some shitcoin  

what warning ? (don't see any)

Why warning msg at the top of page.Djm one of the best miner coder i ever seen.Will try miner soon.

I don't think there is a cuda-schedule command in ccminer... what are you trying to attempt ?
if you mean to set the intensity, -i 20 is probably your example.

This is the report of your Miner, when it detects a 2080.

x miner threads startet, using the ´mtp´ algorithm.
Gpu #0: Intensity set to 23.3477, 11304960 cuda threads number of.....
Gpu #1: Intensity set to 23.3477, 11304960 cuda threads number of.....

So the autodetect might not be optimized for the 2080.
Can i set it manual?

--cuda-schedule 12000000

didnt work for me.

Yes for the moment, T-rex is still faster on the 20x serie (which I don't own, will change when I will get my hand on one)
But as far as the 10x serie is concerned it is the fastest (especially with the help of OhGodAPill for the models using gdd5x).

In the latest release, you can vary the intensity.

I don't understand the 46Threads per block thing... can you elaborate ?

I love your miner and its nice watching how you improved it, but its not the fastest.

I didnt check the 1xxx Series, but on the 2080 (running just under 1800 Mhz) the

T-Rex gets 4.173 Mhash and CCMiner only 3.5 Mhash.

CCMiner detects 11304960 Cuda Threads, 46 Threads per Block 320

Any improvement with manual settings possible?

Hello,

currently ccminer is the fastest miner for MTP and the cheapest

* 2.5MH/s on 1070
* 3.9MH/s on 1080ti with OhGodAPill (3.55MH/s without)

* 5.5MH/s on 2080ti

latest release:
new sgminer version released:
(improved stability)
https://github.com/zcoinofficial/sgminer/releases/tag/0.1.3

new ccminer version released:

https://github.com/zcoinofficial/ccminer/releases/tag/1.2.0

*better handling of miner/pool connection and packet losses in case of bad connection
solve various crash problem due to pool sending empty data

sliglty improved speed on 20x cards
This release should solve most of the crash which was encountered on the previous version.

performance:

2080ti gets now 5.5MH/s

1080ti 3.55MH/s
3.9MH/s with OhGodAPill

1070 at 2.5MH/s
1660 at 2.06MH/s

implement a non mandatory donation system (default 0.25% 1 share over 400)
This donation can be turned off by using --no-donation flag
%age donated can be changed by using flag --donation N (in per cent)

you can find binaries for windows (ccminer.exe)
and for linux:
* ccminer-linux-cuda101 (binaries for linux compiled with cuda 10.1
* ccminer-linux-cuda100 (binaries for linux compiled with cuda 10.0
* ccminer-linux-cuda92 (binaries for linux compiled with cuda 9.2

example of bat file:  RUN-ZCOIN-MTP.cmd


new version of sgminer has been released:
(no fee... too painfull to implement in sgminer  , I'll probably add one, if I beat teamred on vega )

https://github.com/zcoinofficial/sgminer/releases/tag/0.1.2

Improved hashrate on both mtp.cl and mtp_vega.cl

mtp.cl for RX5xx and RX4xx

1.69MH @1250MH/z on RX580
do not use this kernel for vega performance are ridiculous Grin

mtp_vega.cl
1.685MH @1250MHz on RX580
3.1MH/s @1350MHz on Vega 56
currently this is on my RX580  around 400kh/s faster than teamred and it run at a lower frequency (100MHz lower for some reason...)
and 70kh/s slower than teamred on vega (same frequency)
use sgminer.conf to pass kernels/pool etc...

cheers,
djm34