R9 Nano Modded Rom

[Eliovp]

I will share my second rom as well. I will edit the op.

This one is even more energy efficient.
Although it is set at stock 800Mh Core clock. (you can up this to 1000 in afterburner if needed).

Greetings.

What is the difference between this one and the previous one? Did you change other voltages or something else?

Lower mvddc, better timings, fixed dpm values :-)

Some dwarfpool-side results, with a watt-meter reading, would be gladly appreciated.

Have you tried claymore's dual miner with such low voltages (eth+dcr)?

I suppose that it would be too unstable with blake256.

Keep up the good work ! thank you!

Hey Nikolaj.

First, i'm not mining on Dwarfpool. For several reasons.

They already have to much hash rate to start with.
I clearly get higher readings on other pools.

I'm currently mining over at ethermine.org

I have a picture here of my Amp Clamp hooked up on my testing rig.
This is with 6x nano all with my second Rom i shared at stock settings. Nothing extra changed.
16.1.1 drivers windows 10 64bit running Genoil Ethminer 1.0.6, doing 132Mh/s

https://drive.google.com/file/d/0B_cBBoZ_PTwOYTdtTGc1Um9MTW1ELTVuYWhiaDZoMEVFb1Vr
So 570watt


Funny thing, i got the same hash rate with 16.4.1 drivers but a little more energy usage
https://drive.google.com/file/d/0B_cBBoZ_PTwONTFKbURLUF82UE44S0dtdlNMOHhRcVJxMTFj
575watt


I have indeed tested Claymore's miner with these settings:
dcri 50, running both eth and dcr.

also everything at stock (from my second shared rom, so 800 core clock, 300 memclock).

This gave me 130Mh eth and i believe around 3,2Ghz on DCR.
Downside, it went to 5.1A, so 1120watt.  

I had to down clock my core clock to 650 to get the same wattage as before. Obviously this lead to a serious decrease in hashrate.

I also tested solo mining eth on all 3 of Claymores versions.

All of them obviously used more energy at only a little change in hash rate --> 135 Mh...


At this moment i'm doing a test on Windows 7, just for the fun of it.
it seems that it's not better at all. (hash rate -> energy).

Done a test with 15.12 Crimson drivers. 3.2A, also 132Mh.

Will be doing a lot more testing this week.

Will report back!

Greetings!

[asrilani]

I have a picture here of my Amp Clamp hooked up on my testing rig.
This is with 6x nano all with my second Rom i shared at stock settings. Nothing extra changed.
16.1.1 drivers windows 10 64bit running Genoil Ethminer 1.0.6, doing 132Mh/s

https://drive.google.com/file/d/0B_cBBoZ_PTwOYTdtTGc1Um9MTW1ELTVuYWhiaDZoMEVFb1Vr
So 570watt


Funny thing, i got the same hash rate with 16.4.1 drivers but a little more energy usage
https://drive.google.com/file/d/0B_cBBoZ_PTwONTFKbURLUF82UE44S0dtdlNMOHhRcVJxMTFj
575watt


I have indeed tested Claymore's miner with these settings:
dcri 50, running both eth and dcr.

also everything at stock (from my second shared rom, so 800 core clock, 300 memclock).

This gave me 130Mh eth and i believe around 3,2Ghz on DCR.
Downside, it went to 5.1A, so 1120watt.  

I had to down clock my core clock to 650 to get the same wattage as before. Obviously this lead to a serious decrease in hashrate.

I also tested solo mining eth on all 3 of Claymores versions.

All of them obviously used more energy at only a little change in hash rate --> 135 Mh...


At this moment i'm doing a test on Windows 7, just for the fun of it.
it seems that it's not better at all. (hash rate -> energy).

Done a test with 15.12 Crimson drivers. 3.2A, also 132Mh.

Will be doing a lot more testing this week.

Will report back!

Greetings!

How much more energy does the Claymore miner use? I tried on 7990, it seems the temperature is the same.

[Termie]

my MSI runs with 1100 MHz@-60mV nonstop since two weeks. Good compromise between power consumption and mining power.
Also overclocked HBM memory temporary to 525 MHz, but this seems to have no effect to MH/s, so I lowered it to 500 MHz again.
Anyway, R9 Nanos are amazing cards!

@Termie Can you say a little more about mining results using that settings? Power consumption, noise, hashpower, etc.

Sorry for the late answer, I´m not very often here. With my config (which contains a few more tweaks as only overclocking+undervolting) I get an average of ~32 MH/s, peaks up to 40+ MH/s (best as yet were 42.3 MH/s). Short-term drops are certainly also possible, depending on the "mood" ^^ of my pool (Coinotron).

Average 24h power consumption is ~148 watt (without mem), before with stock voltage is was ~168 watt. Noise isn´t very important for me, it runs at a separate room, typically rig-like without case. Most time with 58% fan speed and 63-65°. But best of all is that I managed all this via software. Because I guess if you mod&flash BIOS of your ridiculously-expensive Nano, warranty expires. Bad in case your card conks out, and loweres also the resale value.

[Zenixin]

my MSI runs with 1100 MHz@-60mV nonstop since two weeks. Good compromise between power consumption and mining power.
Also overclocked HBM memory temporary to 525 MHz, but this seems to have no effect to MH/s, so I lowered it to 500 MHz again.
Anyway, R9 Nanos are amazing cards!

@Termie Can you say a little more about mining results using that settings? Power consumption, noise, hashpower, etc.

Sorry for the late answer, I´m not very often here. With my config (which contains a few more tweaks as only overclocking+undervolting) I get an average of ~32 MH/s, peaks up to 40+ MH/s (best as yet were 42.3 MH/s). Short-term drops are certainly also possible, depending on the "mood" ^^ of my pool (Coinotron).

Average 24h power consumption is ~148 watt (without mem), before with stock voltage is was ~168 watt. Noise isn´t very important for me, it runs at a separate room, typically rig-like without case. Most time with 58% fan speed and 63-65°. But best of all is that I managed all this via software. Because I guess if you mod&flash BIOS of your ridiculously-expensive Nano, warranty expires. Bad in case your card conks out, and loweres also the resale value.

148W/32 = 4.65 J/MH. That is very good. For the R9 390 220W/30 = 7.3 J/MH.

For the OP's setting, 460W/132 = 3.5 J/MH.

[ldw-com]

my MSI runs with 1100 MHz@-60mV nonstop since two weeks. Good compromise between power consumption and mining power.
Also overclocked HBM memory temporary to 525 MHz, but this seems to have no effect to MH/s, so I lowered it to 500 MHz again.
Anyway, R9 Nanos are amazing cards!

@Termie Can you say a little more about mining results using that settings? Power consumption, noise, hashpower, etc.

Sorry for the late answer, I´m not very often here. With my config (which contains a few more tweaks as only overclocking+undervolting) I get an average of ~32 MH/s, peaks up to 40+ MH/s (best as yet were 42.3 MH/s). Short-term drops are certainly also possible, depending on the "mood" ^^ of my pool (Coinotron).

Average 24h power consumption is ~148 watt (without mem), before with stock voltage is was ~168 watt. Noise isn´t very important for me, it runs at a separate room, typically rig-like without case. Most time with 58% fan speed and 63-65°. But best of all is that I managed all this via software. Because I guess if you mod&flash BIOS of your ridiculously-expensive Nano, warranty expires. Bad in case your card conks out, and loweres also the resale value.

Had a dead Nano a few weeks ago, was flashed with my modified rom.
I switched the switch back to the original position.

Sent it out, received a new one without questions.


Further, Ridiculous expensive? Depends on what you call expensive.

I bought in bulk, got a very nice price (same as a 390x).
So if i ever resell, i could even make a profit out of that as well

 

Ps. I am Eliovp (this is my account i use at work..)

[Nikolaj]

Thank you Eliovp

The main issue is that the avg profitability with the current ETH and DCR USD value, too low to let the nano's be a good buy at the market prices. With a bulk order obviously the situation is a bit better, but right now here's the situation:

reference data: 6 NANO's in a single system, USD values, electricity cost 0.3$:

power consumption: 0.57KWh
24h gross profit ETH: 11.04
24h electricity cost: 4.1
24h gross NET profit ETH: 6.94

power consumption: 1.1KWh
24h gross profit ETH+DCR: 15.47
24h electricity cost: 7.92
24h gross NET profit ETH+DCR: 7.55

total cost of the platform in USD: 4397
linear break even NOT considering the electricity (ETH+DCR), in days: 284
linear break even considering the electricity, in days: 582

a "BIT" too much.. Even with an electricity cost of 0.10$/KWh we've got:

power consumption: 0.57KWh
24h gross profit ETH: 11.04
24h electricity cost: 1.368
24h gross NET profit ETH: 9.67

power consumption: 1.1KWh
24h gross profit ETH+DCR: 15.47
24h electricity cost: 2.64
24h gross NET profit ETH+DCR: 12.83

total cost of the platform in USD: 4397
linear break even NOT considering the electricity (ETH+DCR), in days: 284
linear break even considering the electricity, in days: 343

I live in Italy and to obtain 0.1$ I should spend at least 35k EUR to produce electricity with solar panels and wind turbines.

That's why I've stopped to go on with the nano's project, too un-optimized, too weak, too expensive, and finally too late considering the polaris and pascal architecture in a couple of months. What you could gain now it's a tiny fraction of the devaluation of the nano's, in the meantime.

To me, it's a failed project right now, with the current USD profit capabilities. Better to go on with nvidia ones in all honesty.

[Faxmate]

Thank you Eliovp

The main issue is that the avg profitability with the current ETH and DCR USD value, too low to let the nano's be a good buy at the market prices. With a bulk order obviously the situation is a bit better, but right now here's the situation:

reference data: 6 NANO's in a single system, USD values, electricity cost 0.3$:

power consumption: 0.57KWh
24h gross profit ETH: 11.04
24h electricity cost: 4.1
24h gross NET profit ETH: 6.94

power consumption: 1.1KWh
24h gross profit ETH+DCR: 15.47
24h electricity cost: 7.92
24h gross NET profit ETH+DCR: 7.55

total cost of the platform in USD: 4397
linear break even NOT considering the electricity (ETH+DCR), in days: 284
linear break even considering the electricity, in days: 582

a "BIT" too much.. Even with an electricity cost of 0.10$/KWh we've got:

power consumption: 0.57KWh
24h gross profit ETH: 11.04
24h electricity cost: 1.368
24h gross NET profit ETH: 9.67

power consumption: 1.1KWh
24h gross profit ETH+DCR: 15.47
24h electricity cost: 2.64
24h gross NET profit ETH+DCR: 12.83

total cost of the platform in USD: 4397
linear break even NOT considering the electricity (ETH+DCR), in days: 284
linear break even considering the electricity, in days: 343

I live in Italy and to obtain 0.1$ I should spend at least 35k EUR to produce electricity with solar panels and wind turbines.

The electricity price you use is  $0.3/kWh. I think the cost is mostly below $0.2. Maybe in Germany, it is that expensive.

[Nikolaj]

Just have a look at the numbers, it's not worth the time and the investments.

The maxwell architecture it's more balanced right now, I'm very curious about the gtx 1080 performance because given these behaviours we could have a game-changing model. Even if the 1080 will achieve the gtx 980Ti performance, consuming 35% less W with a similar price, it would be a better choice.

It's not relevant the electricity price, it's relevant the USD income of a 5k EUR platform. Not even 15$ net, I consider it a total failure, especially because of the situation will worsen over time.

If I need to invest such amounts, I would prefer right now to build 6x280x platforms, and invest in the meantime in renewable energy sources. The best way to go, especially considering the EUR purchase price of 280x models, and the general performance with the algos. The nano, right now, it's "too little, too late"

[Positid]

Just have a look at the numbers, it's not worth the time and the investments.

The maxwell architecture it's more balanced right now, I'm very curious about the gtx 1080 performance because given these behaviours we could have a game-changing model. Even if the 1080 will achieve the gtx 980Ti performance, consuming 35% less W with a similar price, it would be a better choice.

It's not relevant the electricity price, it's relevant the USD income of a 5k EUR platform. Not even 15$ net, I consider it a total failure, especially because of the situation will worsen over time.

If I need to invest such amounts, I would prefer right now to build 6x280x platforms, and invest in the meantime in renewable energy sources. The best way to go, especially considering the EUR purchase price of 280x models, and the general performance with the algos. The nano, right now, it's "too little, too late"

The 280x consumes more energy than the nano, ie less efficient. So it will make a loss earlier than nano.

[ldw-com]

Thank you Eliovp

The main issue is that the avg profitability with the current ETH and DCR USD value, too low to let the nano's be a good buy at the market prices. With a bulk order obviously the situation is a bit better, but right now here's the situation:

reference data: 6 NANO's in a single system, USD values, electricity cost 0.3$:

power consumption: 0.57KWh
24h gross profit ETH: 11.04
24h electricity cost: 4.1
24h gross NET profit ETH: 6.94

power consumption: 1.1KWh
24h gross profit ETH+DCR: 15.47
24h electricity cost: 7.92
24h gross NET profit ETH+DCR: 7.55

total cost of the platform in USD: 4397
linear break even NOT considering the electricity (ETH+DCR), in days: 284
linear break even considering the electricity, in days: 582

a "BIT" too much.. Even with an electricity cost of 0.10$/KWh we've got:

power consumption: 0.57KWh
24h gross profit ETH: 11.04
24h electricity cost: 1.368
24h gross NET profit ETH: 9.67

power consumption: 1.1KWh
24h gross profit ETH+DCR: 15.47
24h electricity cost: 2.64
24h gross NET profit ETH+DCR: 12.83

total cost of the platform in USD: 4397
linear break even NOT considering the electricity (ETH+DCR), in days: 284
linear break even considering the electricity, in days: 343

I live in Italy and to obtain 0.1$ I should spend at least 35k EUR to produce electricity with solar panels and wind turbines.

That's why I've stopped to go on with the nano's project, too un-optimized, too weak, too expensive, and finally too late considering the polaris and pascal architecture in a couple of months. What you could gain now it's a tiny fraction of the devaluation of the nano's, in the meantime.

To me, it's a failed project right now, with the current USD profit capabilities. Better to go on with nvidia ones in all honesty.

Nikolaj

First of all, thank you for your time and interest.

In your case, you are probably right.

In my case, it's different.
I do own solar pannels. So i tend to save a lot of energycosts.
Price here is also not that much, so again, a gain in cost.

Further, i bought my nano's almost 4 months ago. I was there when eth was SERIOUSLY profitable, so i made a very big profit in that time. And i still am..

Regarding polaris. I've read so many articles, even leaked examples. They will not (in the beginning) be a threat to fiji based cards.
Even more, you say in a few months? I don't believe that. The first cards (i think) will be available begin 2017, so that's still almost a year.
And what about the price?

True, a nano isn't cheap. But again, as mentioned several times. I did got a seriously good offer, even after months the prices are not even near what i've payed for them.

I do however will continue my project regarding power efficienty and will update this thread as much as i can.

Thx again for your interest!

Greetings!

[Marvell1]

my MSI runs with 1100 MHz@-60mV nonstop since two weeks. Good compromise between power consumption and mining power.
Also overclocked HBM memory temporary to 525 MHz, but this seems to have no effect to MH/s, so I lowered it to 500 MHz again.
Anyway, R9 Nanos are amazing cards!

@Termie Can you say a little more about mining results using that settings? Power consumption, noise, hashpower, etc.

Sorry for the late answer, I´m not very often here. With my config (which contains a few more tweaks as only overclocking+undervolting) I get an average of ~32 MH/s, peaks up to 40+ MH/s (best as yet were 42.3 MH/s). Short-term drops are certainly also possible, depending on the "mood" ^^ of my pool (Coinotron).

Average 24h power consumption is ~148 watt (without mem), before with stock voltage is was ~168 watt. Noise isn´t very important for me, it runs at a separate room, typically rig-like without case. Most time with 58% fan speed and 63-65°. But best of all is that I managed all this via software. Because I guess if you mod&flash BIOS of your ridiculously-expensive Nano, warranty expires. Bad in case your card conks out, and loweres also the resale value.

Had a dead Nano a few weeks ago, was flashed with my modified rom.
I switched the switch back to the original position.

Sent it out, received a new one without questions.


Further, Ridiculous expensive? Depends on what you call expensive.

I bought in bulk, got a very nice price (same as a 390x).
So if i ever resell, i could even make a profit out of that as well

 

Ps. I am Eliovp (this is my account i use at work..)

by in bulk how many did you buy to recive a voloume discount ?

[Marvell1]

Hey,

i use HxD and a calculator to modify the values.

After that i use Fiji Bios Editor to fix the checksum

Greetings

I would like your bios but its hash rate is a bit too low for me I have around 8 saphire nanos

I notice on afterburner you cant reduce the memory clock on any cards not even 390s  You can reduice voltage a max of 100mv of course and power tune and core clock.

would this tool work on the R9 Nanos and the 390s to reduce voltage and Memory clock ?

https://github.com/OneB1t/HawaiiBiosReader

or can you provide a new modifed bios that  hashes at around 150mh/s rather than 130ish leaving 30 mhs on the table seems kind of a waste.

Hey Marvell1,

The rom i shared has a mod on the core voltage (-100mV) and a mod on the mem voltage (also -100mV)

You can open afterburner and go even lower.

So if you set (-12) you will have -112mV

You are correct, you cannot reduce the memory voltage with afterburner or any other software.

This has to be done with a hardmod.

The Rom i shared can easily hit 150Mh because you can still set 1000 as core clock (it is stock set at 1000).
Thus giving you full power but with less energy usage.

I will share my second rom as well. I will edit the op.

This one is even more energy efficient.
Although it is set at stock 800Mh Core clock. (you can up this to 1000 in afterburner if needed).

Greetings.

Thanks for the response on this Eliovp , please do some of your brillaint work on the 390s they seem to be great cards not as good as the nano in saphire Trixx it seems you can lower thier
voltage all the way down to -200mv although it starts to climb back up.  Also with lower memory freqency and voltage like you did for the nanos they would be even better.

Ill try you new Rom on the six nanos i have but i have around 23 390s lol

[Eliovp]

Hey,

i use HxD and a calculator to modify the values.

After that i use Fiji Bios Editor to fix the checksum

Greetings

I would like your bios but its hash rate is a bit too low for me I have around 8 saphire nanos

I notice on afterburner you cant reduce the memory clock on any cards not even 390s  You can reduice voltage a max of 100mv of course and power tune and core clock.

would this tool work on the R9 Nanos and the 390s to reduce voltage and Memory clock ?

https://github.com/OneB1t/HawaiiBiosReader

or can you provide a new modifed bios that  hashes at around 150mh/s rather than 130ish leaving 30 mhs on the table seems kind of a waste.

Hey Marvell1,

The rom i shared has a mod on the core voltage (-100mV) and a mod on the mem voltage (also -100mV)

You can open afterburner and go even lower.

So if you set (-12) you will have -112mV

You are correct, you cannot reduce the memory voltage with afterburner or any other software.

This has to be done with a hardmod.

The Rom i shared can easily hit 150Mh because you can still set 1000 as core clock (it is stock set at 1000).
Thus giving you full power but with less energy usage.

I will share my second rom as well. I will edit the op.

This one is even more energy efficient.
Although it is set at stock 800Mh Core clock. (you can up this to 1000 in afterburner if needed).

Greetings.

Thanks for the response on this Eliovp , please do some of your brillaint work on the 390s they seem to be great cards not as good as the nano in saphire Trixx it seems you can lower thier
voltage all the way down to -200mv although it starts to climb back up.  Also with lower memory freqency and voltage like you did for the nanos they would be even better.

Ill try you new Rom on the six nanos i have but i have around 23 390s lol

Hey Marvell1,

Thx for the nice words.

I went ahead and did a small mod on one of my 390's (NON X) rom.

There are 2 roms in this pack.
one with -100mV on the memory voltage
and one with -150mV.

Core voltage is still stock (you can change this with trixx or afterburner.)
Core clock = 1040
Mem clock = 1500

Link

Make sure to use your backup rom position.

Good luck and have fun!

[Marvell1]

Hey,

i use HxD and a calculator to modify the values.

After that i use Fiji Bios Editor to fix the checksum

Greetings

I would like your bios but its hash rate is a bit too low for me I have around 8 saphire nanos

I notice on afterburner you cant reduce the memory clock on any cards not even 390s  You can reduice voltage a max of 100mv of course and power tune and core clock.

would this tool work on the R9 Nanos and the 390s to reduce voltage and Memory clock ?

https://github.com/OneB1t/HawaiiBiosReader

or can you provide a new modifed bios that  hashes at around 150mh/s rather than 130ish leaving 30 mhs on the table seems kind of a waste.

Hey Marvell1,

The rom i shared has a mod on the core voltage (-100mV) and a mod on the mem voltage (also -100mV)

You can open afterburner and go even lower.

So if you set (-12) you will have -112mV

You are correct, you cannot reduce the memory voltage with afterburner or any other software.

This has to be done with a hardmod.

The Rom i shared can easily hit 150Mh because you can still set 1000 as core clock (it is stock set at 1000).
Thus giving you full power but with less energy usage.

I will share my second rom as well. I will edit the op.

This one is even more energy efficient.
Although it is set at stock 800Mh Core clock. (you can up this to 1000 in afterburner if needed).

Greetings.

Thanks for the response on this Eliovp , please do some of your brillaint work on the 390s they seem to be great cards not as good as the nano in saphire Trixx it seems you can lower thier
voltage all the way down to -200mv although it starts to climb back up.  Also with lower memory freqency and voltage like you did for the nanos they would be even better.

Ill try you new Rom on the six nanos i have but i have around 23 390s lol

Hey Marvell1,

Thx for the nice words.

I went ahead and did a small mod on one of my 390's (NON X) rom.

There are 2 roms in this pack.
one with -100mV on the memory voltage
and one with -150mV.

Core voltage is still stock (you can change this with trixx or afterburner.)
Core clock = 1040
Mem clock = 1500

Link

Make sure to use your backup rom position.

Good luck and have fun!

looks great ill try these on my MSI 390s tonight, can you create one with a default core voltage reduction of 100 as well

thanks !

[Eliovp]

looks great ill try these on my MSI 390s tonight, can you create one with a default core voltage reduction of 100 as well

thanks !

Sure.

Try this one first.

-100mV both Core and Mem

Link

If it's stable, i'll lower it some more


Greetings

[Marvell1]

looks great ill try these on my MSI 390s tonight, can you create one with a default core voltage reduction of 100 as well

thanks !

Sure.

Try this one first.

-100mV both Core and Mem

Link

If it's stable, i'll lower it some more


Greetings

ok trying right now thanks will post update in a few

these 390s I have the MSI does not seem to have a toggle bios switch

[Marvell1]

looks great ill try these on my MSI 390s tonight, can you create one with a default core voltage reduction of 100 as well

thanks !

Sure.

Try this one first.

-100mV both Core and Mem

Link

If it's stable, i'll lower it some more


Greetings

** edit flashing from dos seems to be working now**

tried to load your bios to my MSI 390 using AtiWinFlash and I get a SYSTEM mistmatch error never seen that before on flashing

i made a backup of the bios from another MSI 390 and flashed it to the card that would not work with your 100 100 bios and it works any ideas ?

[Marvell1]

looks great ill try these on my MSI 390s tonight, can you create one with a default core voltage reduction of 100 as well

thanks !

Sure.

Try this one first.

-100mV both Core and Mem

Link

If it's stable, i'll lower it some more


Greetings

** edit flashing from dos seems to be working now**

tried to load your bios to my MSI 390 using AtiWinFlash and I get a SYSTEM mistmatch error never seen that before on flashing

i made a backup of the bios from another MSI 390 and flashed it to the card that would not work with your 100 100 bios and it works any ideas ?

Ok this rom did not work on the MSI 390s i have lead to the pc freezing and the card was having errors in dos

i created a boot disk and relashed it back to the default MSI ROM and it worked. 

It seems that its harder to flash ROMs from different card brands to other brands really sucks

[ldw-com]

Marvell1,

Well, i had a feeling that it wasn't going to work.

With those nano's, it's different, they're all reference designs, so my rom will work on any nano. I'm pretty sure of that.

I didn't know what kind of 390 you have so i just took one of mine.

Could you send me your rom, i'll mod that one and that will definitely work then

Greetings!

[Zitdadast]

Marvell1,

Well, i had a feeling that it wasn't going to work.

With those nano's, it's different, they're all reference designs, so my rom will work on any nano. I'm pretty sure of that.

I didn't know what kind of 390 you have so i just took one of mine.

Could you send me your rom, i'll mod that one and that will definitely work then

Greetings!

What is the brand of the 390 card you do the mod?

Edit: I just found, it is the sapphire as indicated in your file.