VEGA 64 mining FREE WISDOM - page 4.

bandana121

newbie

Activity: 50

Merit: 0

Quote from: oblione on April 16, 2018, 10:41:19 AM

Howdy People!!

Just wondering if anyone can help me with this.

I've already got a few rigs of vega 64's & 56's (sapphire reference cards) that run excellent at the lowe power rates (~160w.)

I got hold of a few Gigabyte vega 56's Gaming OC editiion cards and was wondering if I can apply the same regedit tables etc to them as I would with a reference card. They draw way too much power just over clocked but I'm not sure how safe it is to use the powerplay tables / regedit on them.

Any one got any experience with these?

cheers

Hey,

Personally I'm not sure but there a section on this tutorial about powerplay tables that should hopefully clarify that.

http://vega.miningguides.com/

oblione

newbie

Activity: 71

Merit: 0

Howdy People!!

Just wondering if anyone can help me with this.

I've already got a few rigs of vega 64's & 56's (sapphire reference cards) that run excellent at the lowe power rates (~160w.)

I got hold of a few Gigabyte vega 56's Gaming OC editiion cards and was wondering if I can apply the same regedit tables etc to them as I would with a reference card. They draw way too much power just over clocked but I'm not sure how safe it is to use the powerplay tables / regedit on them.

Any one got any experience with these?

cheers

Biggums

newbie

Activity: 45

Merit: 0

Quote from: CryptoPlay on April 12, 2018, 08:33:34 PM

xmr-stak is hanging a lot my rig.

Are u using it ? What miner are u on?

Yes I'm using Stak. What are your intensity settings? Are you also using cpu, or just gpu alone?

CryptoPlay

jr. member

Activity: 202

Merit: 2

xmr-stak is hanging a lot my rig.

Are u using it ? What miner are u on?

RIGED

member

Activity: 182

Merit: 12

Quote from: Biggums on April 11, 2018, 07:27:34 PM

Quote from: RIGED on April 11, 2018, 12:13:35 PM

qms software now has an all gpu disable/enable button
sourced from device id post detection and edit of devcon, to make it automatic for all systems with no manual setup

if user thinks an easy manual refresh will help system stability

On my rig (with 7 Vega's at present), I have a bat file that disables and then enables all 7 cards. This is the number one time that my rig gets blue screen or hangs. I would estimate that it crashes 1 in 5 times when in this process, and 3 out of 5 times when I am running the enable disable process while managing it remotely from another device.

Personally IMO, it would be better if QMS had a way to just enable the hbm2 directly in a similar way to how the AMD software does instead of using the devcon enable/disable trick method.

If the software could just see if the hbm memory is enabled on each card the way the AMD software does, and then enable it if desired, and have an option to have it always enabled would be better than completely disabling(and re-enabling) the entire device in Windows just to turn on one feature of the GPU. Then if the software doesn't see it enabled it just enables it without having to disable the cards in Windows. AMD's software does this so somehow it's possible.

Hi!

we are on it, but ofcourse easier said than done

the AMD Driver disable/enable BUG and ALL other BUGS are planned to be fixed by Q4 2018 the latest with release of VGA VegaComputeDrivers v01.0.0
we have been clear from the beggining,
the VGA project has primary focus on Development of NEW COMPUTE DRIVER packages, and this is the next VGA community task

The QMS v01.0.0 is a purely mining support program which will develop in its own right as an indespensable tool for the Vega owner

in other words when you participate in the VGA Community its not to just get the initial QMS package, its a running investment through time with constant developments

Ed has been busy adding features to the QMS Grin

https://imgur.com/CrKYf8p

Biggums

newbie

Activity: 45

Merit: 0

Quote from: RIGED on April 11, 2018, 12:13:35 PM

qms software now has an all gpu disable/enable button
sourced from device id post detection and edit of devcon, to make it automatic for all systems with no manual setup

if user thinks an easy manual refresh will help system stability

On my rig (with 7 Vega's at present), I have a bat file that disables and then enables all 7 cards. This is the number one time that my rig gets blue screen or hangs. I would estimate that it crashes 1 in 5 times when in this process, and 3 out of 5 times when I am running the enable disable process while managing it remotely from another device.

Personally IMO, it would be better if QMS had a way to just enable the hbm2 directly in a similar way to how the AMD software does instead of using the devcon enable/disable trick method.

If the software could just see if the hbm memory is enabled on each card the way the AMD software does, and then enable it if desired, and have an option to have it always enabled would be better than completely disabling(and re-enabling) the entire device in Windows just to turn on one feature of the GPU. Then if the software doesn't see it enabled it just enables it without having to disable the cards in Windows. AMD's software does this so somehow it's possible.

RIGED

member

Activity: 182

Merit: 12

qms software now has an all gpu disable/enable button
sourced from device id post detection and edit of devcon, to make it automatic for all systems with no manual setup

if user thinks an easy manual refresh will help system stability

RIGED

member

Activity: 182

Merit: 12

Quote from: Biggums on April 10, 2018, 08:12:59 PM

Quote from: RIGED on April 09, 2018, 08:48:03 AM

IM NOT MINING JUST FOR PROFITS

IM TRYING TO CHANGE THE WORLD

so if you are too stupid and/or psychologicaly/financialy complexed to comprehend that, stand out of the way and watch
(or go back to your amd sponsored cubicle)

Your comment made me think of this video. You can skip the first 5 minutes or so of this video... but as he nears the end he gets into a passionate talk that shares the same sentiment with your comments.

https://youtu.be/R00g6E8Yet4

Dan has his heart and gut in the right place, comprehends the dangers of centralised blockchain systems and the overpowering dominance that can be exerted through them,

i am a big supporter of anything decentralised especially the most commonly used tool of the planet, ie money (medium of exchange), if people knew they are happy unwitting tax farm slaves we would not have privately owned central banks ruling planetary reality

Biggums

newbie

Activity: 45

Merit: 0

Quote from: RIGED on April 09, 2018, 08:48:03 AM

IM NOT MINING JUST FOR PROFITS

IM TRYING TO CHANGE THE WORLD

so if you are too stupid and/or psychologicaly/financialy complexed to comprehend that, stand out of the way and watch
(or go back to your amd sponsored cubicle)

Your comment made me think of this video. You can skip the first 5 minutes or so of this video... but as he nears the end he gets into a passionate talk that shares the same sentiment with your comments.

https://youtu.be/R00g6E8Yet4

RIGED

member

Activity: 182

Merit: 12

thnx guys, and please feel free to make suggestions

the programer tweaked QMS to keep the miner running even if a card drops offline, so you dont get continous restarts if a gpu has failed at a bios/driver/bus level

bandana121

newbie

Activity: 50

Merit: 0

Quote from: RIGED on April 10, 2018, 09:15:47 AM

im waiting for a moron to accuse me of photoshoping images next Tongue

Well wait no more.....just kidding!

Just to be clear though I do admire what your doing.

Keep up the good work.

bubaba

jr. member

Activity: 69

Merit: 2

i like what i read, gathering many tools in one nice software
keep it up and ignore the haters plz

looking forward to mine with QMS Wink

RIGED

member

Activity: 182

Merit: 12

4 new reg tables acting as default settings even on crash

XMR ECO preset, currently being developed, offers a minimum 25% wattage reduction to below 1000w for 6 Vega64 + peripherals

you can see the hash rate report
we are testing with cpu mining enabled for those of you that do, i personaly keep cpu resources only for system stability functions, so will be disabling it soon

we also added a driver version report (top right on QMS window)
look at lower state values on Ntool, these are default now via auto reg edit by clicking Activate on QMS
forgot to mention that QMS also disable/enable all gpu in event of crash/miner freeze to clear "stuck in thread" and keep on mining
crash minimum fan setting now 25% not 10rpm
setting up 0%fee pool with coin switch option, to auto connect with QMS pool button click
all this and much more with 50VGA satoshi participation that was $35...

https://imgur.com/N6Azt8g

im waiting for a moron to accuse me of photoshoping images next Tongue

to clarify again, im not a or the programer, VGA QMS v10.0.0 is coded by an expert hired by the VGA Community using VGA Community funds
all testing is conducted on my rigs so i take screen pics when he completes a task the VGA Dev team has set for this project

Biggums

newbie

Activity: 45

Merit: 0

Quote from: bandana121 on April 09, 2018, 03:06:43 PM

Quote from: RIGED on April 09, 2018, 02:22:31 PM

getting 7435h from 6x64 and 2815h from 2xFE
only 100w and 30w more

no oc yet just intensity tweaking

You'll have to excuse me but my knowlege of this subject is extemely limited.

Would I be correct in understanding that your working on getting the vega 64 running on 2 threads? If so, using this method what would be your 'ideal' outcome in terms of hashing power from a single card be? (potentially ~4000h/s working on the basis a single thread would be ~2000h/s?).

P.S. Please don't laugh too much if if I've completely misundertood the subject.

No the vega cards get hash around 2000 h/s on cryptonight when properly configured with 2 threads per card. In XMR stak one must go into the amd.txt file after running the software the first time and manually create a second thread for each card.

RIGED

member

Activity: 182

Merit: 12

Quote from: bandana121 on April 09, 2018, 03:06:43 PM

Quote from: RIGED on April 09, 2018, 02:22:31 PM

getting 7435h from 6x64 and 2815h from 2xFE
only 100w and 30w more

no oc yet just intensity tweaking

You'll have to excuse me but my knowlege of this subject is extemely limited.

Would I be correct in understanding that your working on getting the vega 64 running on 2 threads? If so, using this method what would be your 'ideal' outcome in terms of hashing power from a single card be? (potentially ~4000h/s working on the basis a single thread would be ~2000h/s?).

P.S. Please don't laugh too much if if I've completely misundertood the subject.

if you see the 2 amd.txt code boxes above you will notice the top has a single (GPU) index 0,1,2,3,4,5 entry at a high intensity whereas the bottom box has 2 entries per (GPU) index 0,1,2,3,4,5 at lower and scaled intensity

managed 1370h / gpu with the double thread per GPU (index) but have settled at 1230h with a 25% wattage reduction due to the QMS XMR ECO setting with core temps in the low 30s high 20s

we have reduced the values of the lower core states and their voltages by rewritting the powerplay table and creating 4 new registry variations so far that act as presets of user choice
Activation of a preset also means crossfire check and disable/enable refreshing the GPU after a possible automatic or manual restart

restarting the miner after crash has literaly been totaly automated with all Cryptonight Vega specific dual thread mining otherwise timeconsuming prerequisites satisfied

all by pushing 1 button,
i was able to do trial and error testing to find the sweet spot in hours instead of days because of this new software
and i still havent oc yet

bandana121

newbie

Activity: 50

Merit: 0

Quote from: RIGED on April 09, 2018, 02:22:31 PM

getting 7435h from 6x64 and 2815h from 2xFE
only 100w and 30w more

no oc yet just intensity tweaking

You'll have to excuse me but my knowlege of this subject is extemely limited.

Would I be correct in understanding that your working on getting the vega 64 running on 2 threads? If so, using this method what would be your 'ideal' outcome in terms of hashing power from a single card be? (potentially ~4000h/s working on the basis a single thread would be ~2000h/s?).

P.S. Please don't laugh too much if if I've completely misundertood the subject.

RIGED

member

Activity: 182

Merit: 12

getting 7435h from 6x64 and 2815h from 2xFE
only 100w and 30w more

no oc yet just intensity tweaking

RIGED

member

Activity: 182

Merit: 12

got the 2 FEs on 2 thread each, intensity 980 660 @2540H both,

ddu
install 2 drivers etc method to disable/enable

std settings as before

450W 2xFE, mobo, 2 fans, ssd, risers = 5.6H/W

I'm sure they all have more to give and I haven't even started overclocking....

RIGED

member

Activity: 182

Merit: 12

Quote from: Sgsg666 on April 09, 2018, 08:07:10 AM

Fan cooler ideas: https://www.youtube.com/watch?v=h6hvsqUEtZ4
https://www.youtube.com/watch?v=Iegpwo9SqSg
https://www.youtube.com/watch?v=-yQHAMz1v5g
Ball bearings: https://bitcointalksearch.org/topic/replacing-sleeve-bearings-with-ball-bearings-3267631

Optimized miner:
https://bitcointalksearch.org/topic/m.34028507 not sure if this will be updated)

thnx Sg

I'm going to be building a fortified, insulated, fanned shed to house the rigs (including 6 ASUS STRIX64's arrived this morning Grin

https://imgur.com/Akho3Cr )
had the idea of burying a long length of ventilation pipe in the ground and venting it out from a grill on the floor past the gpus and out through fans in the ceiling
obviously gpus have to be upright with blower facing up, like my current set up

i believe that the cooler soil about 1.5-2m down will cool the outside air intake on its way to the floor grill in the rig room, like a free airconditioner
well, im just waiting for the snow to melt so i can start, i will post pics of the build and report temp results during the summer

i have also been developing a Peltier system recycling heat to electricity in a feed back loop, but im far from being able to say more at this point

RIGED

member

Activity: 182

Merit: 12

preliminary results ECO mode:

6x64s 1 asus mining mobo, 4 fans, 1 ssd, 1 screen

mining Sumo heavy 835W@5200H approx

thats around 6.23H/W

used the above single thread amd.txt edit to achieve with standard cryptonight settings core 1212, 1408, and 1100Mem

try it out for yourselves and let me know

ps its nice to see print of temp, cpu load, cpu frq, mem frg, fan % (for now) while running xmr stak (or any miner),

https://imgur.com/6svEjln

Topic: VEGA 64 mining FREE WISDOM - page 4. (Read 10803 times)