Any word on amd vega hash rates?

[fr4nkthetank]

just got my 2 vega's in the mail, although im really busy these days, just had time to install them, blockchain drivers, win10, etc.  standard stuff.  but xmr-stak refuses to run.  sucks because I have no time for that shit   if anyone has an idea why, let me know.  so its dual mining now, about 190w each.  too high for me still but these need tinkering.

[DementedMindz]

just got my 2 vega's in the mail, although im really busy these days, just had time to install them, blockchain drivers, win10, etc.  standard stuff.  but xmr-stak refuses to run.  sucks because I have no time for that shit   if anyone has an idea why, let me know.  so its dual mining now, about 190w each.  too high for me still but these need tinkering.

I couldn't get it to work either so I'm using sgminer on Win10 64.

[roboshark]

Can someone tell me what's vega's 56/64 neoscrypt hashrate?

[rednoW]

Can someone tell me what's vega's 56/64 neoscrypt hashrate?

0 mhs

[rednoW]

err, I don't think that u understood what I mean -

Oldcomer's mod with blockchain is 1220 GPU clock, using around 1400W for me -> around 220W per card.
1000 GPU clock with 17.8.2  was using only 1100W. -> 170W.

The voltages that the cards were using with the 17.8.2 and underclocked were under 0.9. I think it was 0.875mV


But someone in the thread (if I am not mistaken it was you), said, that he can mine ETH with Oldcomer's mod at 150W per card. And this is the part that I seriously doubt, as my numbers are pretty different, and I single card, that its not modded the same way as in the mod cannot make 400W difference. If the 150W part was true, the whole system would draw only 1000W.


P.S. And I am still using the mod as we speak. 261 MHs ETH, very stable (as its night and its better for the cooling of the cards), they seem to lose a few MHs when temps hit 65.

I repeat again. When mining with Oldcomer's mod run gpu-z or hwinfo64 and check voltages for every card. It should be ~820mV. If not - you've messed something.
With Oldcomer's mod Vega should take ~170watt mining eth at 44mh/s

[DonBit]

...
Power tuning is private yet, sorry.

Not anymore

RX Vega 64 Undervolting Guide for Blockchain Drivers

If this guide was useful for you, then like or share or donate (check the video description for details).

Thanks for this. I've been playing with the power settings with a Sapphire RX Vega 64 Liquid Cooled and got 1990.7 H/s stable @ 0.875 VDDC. Couldn't get power draw at the wall because I need a kill-a-watt thing.


Vega is running thread 3 and 4 with 2600+1800 intensity (threads 0/1 is a RX 560 and thread 2 is a R9 Fury X).

[Skylake]

...
Power tuning is private yet, sorry.

Not anymore

RX Vega 64 Undervolting Guide for Blockchain Drivers

If this guide was useful for you, then like or share or donate (check the video description for details).

Thanks for this. I've been playing with the power settings with a Sapphire RX Vega 64 Liquid Cooled and got 1990.7 H/s stable @ 0.875 VDDC. Couldn't get power draw at the wall because I need a kill-a-watt thing.


Vega is running thread 3 and 4 with 2600+1800 intensity (threads 0/1 is a RX 560 and thread 2 is a R9 Fury X).

Wow first time i know it, so using vega is like using dual vga now?
i think7990 with dual R9 280x is better than vega?
but the power of vega and 7990 is high, curently im using only Rx 480/470 and not have any plan to using vega

[KaydenC]

...
Power tuning is private yet, sorry.

Not anymore

RX Vega 64 Undervolting Guide for Blockchain Drivers

If this guide was useful for you, then like or share or donate (check the video description for details).

The video guide hard bricked my system. It made my gpu fan 100% and stuck at windows loading screen. Had to reinstall windows.

[DonBit]

...
Power tuning is private yet, sorry.

Not anymore

RX Vega 64 Undervolting Guide for Blockchain Drivers

If this guide was useful for you, then like or share or donate (check the video description for details).

The video guide hard bricked my system. It made my gpu fan 100% and stuck at windows loading screen. Had to reinstall windows.

Perhaps you messed up some hex value when changing it. If this happens just unplug the card, Windows will boot. Then run DDU and reinstall the blockchain driver.

[l1xx]

...
Power tuning is private yet, sorry.

Not anymore

RX Vega 64 Undervolting Guide for Blockchain Drivers

If this guide was useful for you, then like or share or donate (check the video description for details).

The video guide hard bricked my system. It made my gpu fan 100% and stuck at windows loading screen. Had to reinstall windows.

Wrong! Its just a specific card that does not accept the values. All you had to do is to unplug the Vega from the system, reset it's value in the registry and start over.
I had same problem, and it was not the windows.

[l1xx]

err, I don't think that u understood what I mean -

Oldcomer's mod with blockchain is 1220 GPU clock, using around 1400W for me -> around 220W per card.
1000 GPU clock with 17.8.2  was using only 1100W. -> 170W.

The voltages that the cards were using with the 17.8.2 and underclocked were under 0.9. I think it was 0.875mV


But someone in the thread (if I am not mistaken it was you), said, that he can mine ETH with Oldcomer's mod at 150W per card. And this is the part that I seriously doubt, as my numbers are pretty different, and I single card, that its not modded the same way as in the mod cannot make 400W difference. If the 150W part was true, the whole system would draw only 1000W.


P.S. And I am still using the mod as we speak. 261 MHs ETH, very stable (as its night and its better for the cooling of the cards), they seem to lose a few MHs when temps hit 65.

I repeat again. When mining with Oldcomer's mod run gpu-z or hwinfo64 and check voltages for every card. It should be ~820mV. If not - you've messed something.
With Oldcomer's mod Vega should take ~170watt mining eth at 44mh/s

0.83-0.85V except one that needed a bit more to be stable, so I adjusted it manually and its around 1. But all 5 are running at 44 MHs. The 6th , where the mod was not applied is at 1.05, runs at 1100 Mem, but the gpu clock is not stable so its ~40.5.

[rednoW]

err, I don't think that u understood what I mean -

Oldcomer's mod with blockchain is 1220 GPU clock, using around 1400W for me -> around 220W per card.
1000 GPU clock with 17.8.2  was using only 1100W. -> 170W.

The voltages that the cards were using with the 17.8.2 and underclocked were under 0.9. I think it was 0.875mV


But someone in the thread (if I am not mistaken it was you), said, that he can mine ETH with Oldcomer's mod at 150W per card. And this is the part that I seriously doubt, as my numbers are pretty different, and I single card, that its not modded the same way as in the mod cannot make 400W difference. If the 150W part was true, the whole system would draw only 1000W.


P.S. And I am still using the mod as we speak. 261 MHs ETH, very stable (as its night and its better for the cooling of the cards), they seem to lose a few MHs when temps hit 65.

I repeat again. When mining with Oldcomer's mod run gpu-z or hwinfo64 and check voltages for every card. It should be ~820mV. If not - you've messed something.
With Oldcomer's mod Vega should take ~170watt mining eth at 44mh/s

0.83-0.85V except one that needed a bit more to be stable, so I adjusted it manually and its around 1. But all 5 are running at 44 MHs. The 6th , where the mod was not applied is at 1.05, runs at 1100 Mem, but the gpu clock is not stable so its ~40.5.

So you have 2 cards that are running > 300watt each ))
The speed on your 6th card is unstable because it overheats and throttles ugly.
Be careful, you can  burn your HBM2

[l1xx]

no, the speed of the card is stable. I could not make it run with the mod, as it was crashing. So with manual downclock via the wattman its not accepting the fixed values for the gpu clock and is running different MHz.
The manually adjusted 1mV card - how come 0.1mV make it draw double?

Whats the danger zone temp for the HBM2 ?

[rednoW]

no, the speed of the card is stable. I could not make it run with the mod, as it was crashing. So with manual downclock via the wattman its not accepting the fixed values for the gpu clock and is running different MHz.
The manually adjusted 1mV card - how come 0.1mV make it draw double?

Whats the danger zone temp for the HBM2 ?

Manual downclock in Wattman or OverdriveNtool makes nothing with blockchain drivers except for the case you hit some special ("bug") value.
Funny, all monitoring software will show 1000mhz clock (for example if you set 1000) but in fact it will be default clock.
Only trusted values are gpu voltage shown, mining speed and power-meter@the_wall readings.
And yes, difference between 1222@0.82v and 1500+@1.05+v is more then 130watt when ETH mining ))

I have no info about danger temp for HBM2 but I strongly recommend to keep it below 70 degrees.

[KL0nLutiy]

no, the speed of the card is stable. I could not make it run with the mod, as it was crashing. So with manual downclock via the wattman its not accepting the fixed values for the gpu clock and is running different MHz.
The manually adjusted 1mV card - how come 0.1mV make it draw double?

Whats the danger zone temp for the HBM2 ?

Manual downclock in Wattman or OverdriveNtool makes nothing with blockchain drivers except for the case you hit some special ("bug") value.
Funny, all monitoring software will show 1000mhz clock (for example if you set 1000) but in fact it will be default clock.
Only trusted values are gpu voltage shown, mining speed and power-meter@the_wall readings.
And yes, difference between 1222@0.82v and 1500+@1.05+v is more then 130watt when ETH mining ))

I have no info about danger temp for HBM2 but I strongly recommend to keep it below 70 degrees.

Did you write to amd support about this issue? They must to fix it.

[rednoW]

no, the speed of the card is stable. I could not make it run with the mod, as it was crashing. So with manual downclock via the wattman its not accepting the fixed values for the gpu clock and is running different MHz.
The manually adjusted 1mV card - how come 0.1mV make it draw double?

Whats the danger zone temp for the HBM2 ?

Manual downclock in Wattman or OverdriveNtool makes nothing with blockchain drivers except for the case you hit some special ("bug") value.
Funny, all monitoring software will show 1000mhz clock (for example if you set 1000) but in fact it will be default clock.
Only trusted values are gpu voltage shown, mining speed and power-meter@the_wall readings.
And yes, difference between 1222@0.82v and 1500+@1.05+v is more then 130watt when ETH mining ))

I have no info about danger temp for HBM2 but I strongly recommend to keep it below 70 degrees.

Did you write to amd support about this issue? They must to fix it.

I've read release notes for blockchain drivers and it said that this drivers are provided as is without support.
HWINFO64 author, Mumak, recently send to AMD bug report about error in power consumption readings with blockchain drivers - will see if it helps ))

From another side, think ... we will not see this cards in stock for reasonable prices if it becomes easy for slider-mover-tuners ))

[roboshark]

Can someone tell me what's vega's 56/64 neoscrypt hashrate?

0 mhs

Vega doesn't have a working miner for it?

[rednoW]

Can someone tell me what's vega's 56/64 neoscrypt hashrate?

0 mhs

Vega doesn't have a working miner for it?

at least I'm not aware of it ...

[gregfromo]

Can someone tell me what's vega's 56/64 neoscrypt hashrate?

0 mhs

Vega doesn't have a working miner for it?

sgminer has a neoscrypt.cl, ever tried it?

https://github.com/nicehash/sgminer-gm

[l1xx]

no, the speed of the card is stable. I could not make it run with the mod, as it was crashing. So with manual downclock via the wattman its not accepting the fixed values for the gpu clock and is running different MHz.
The manually adjusted 1mV card - how come 0.1mV make it draw double?

Whats the danger zone temp for the HBM2 ?

Manual downclock in Wattman or OverdriveNtool makes nothing with blockchain drivers except for the case you hit some special ("bug") value.
Funny, all monitoring software will show 1000mhz clock (for example if you set 1000) but in fact it will be default clock.
Only trusted values are gpu voltage shown, mining speed and power-meter@the_wall readings.
And yes, difference between 1222@0.82v and 1500+@1.05+v is more then 130watt when ETH mining ))

I have no info about danger temp for HBM2 but I strongly recommend to keep it below 70 degrees.

Uhm ... I was walking about the difference in consumption between 1222@0.82v and 1222@1.05v . As the other values are fixed by the mod. I adjusted the voltage manually with the Wattman, and it really accepts that setting, as I see no variations in GPU-Z.

About the 70C ... I cant imagine how this can be done to be honest. The cards running at 58C have HBM2 at ~75C. That at ~18C room temperature.
What exactly needs to be done to cool these?