Question GIGABYTE 5700 XT Bios mod fails

AlleyCat

Veteran
Mitglied seit
Okt 9, 2020
Beiträge
290
Bewertungspunkte
17
Punkte
17
Hi,
I am failing to update Gigabyte 5700 XT. The same procedure updates the BIOS on MSI cards with no problem. I follow the instructions on Igors Lab.

The sign of having trouble with the card bios flash shows in GPU-Z. After flash using amdvbflash the values of GPU and memory frequencies is empty. With stock bios there are Mhz frequencies.

Is there any known problems with flashing Gigabyte cards?
Any suggestions on what other forums I may ask for assistance?
Is it possible that the OEM bios is signed, and any modified bios will be rejected?
If the bios is signed, any tools to resign, or would I need to buy a card from a different vendor?

Thanks,

Alley Cat
 
Zuletzt bearbeitet :
I am going to let the card run for a while at 760/760/1350.
If the GPU start reporting shares errors, which of the parameters I should try to increase to fix an Invalid shares error?

I run overnight the rig at 760/760/1350, and out of 710 accepted shares there was on invalid. The power was never above 104W.

I think what is important is knowing what needs to be increased to address any future errors in Invalid shares.
VDD is related the core clock
VDDCI and MVDD are related to the memory clock

Increase in Core Clock will need Increase in VDD, Decrease in Core Clock does not require decrease in VDD unless to reduce the power consumption and temperature.

Same for Memory Clock, however with memory clock here for mining using RX 5700 XT, it is better to fix it at 1350 mv and adjust the VDDCI.
 
The best voltages at 910 MHz memory clock is MVDD at 1350 mv, VDDCI at 765-770 mv and VDD at 765-770 mv,

And at 900 MHz memory clock is MVDD at 1350 MHz, VDDCI at 750 mv and VDD at 750 mv.
I will test both settings overnight. I will compare between the two setting with a whole day performance data, and monitoring for errors or crashes.
 
Can we start optimizing the Gigabyte cards?
Perhaps you could give me the steps with upper and lower limits. I can run the tests and periodically report the progress.
Certainly,
By the way the memory clock at 905 MHz is a fault spot in the Sapphire card with Samsung memory therefore, it is better to avoid the 905 MHz.

For Gigabyte test VDD and VDDCI from 750 mv to 850 mv at 940-950 MHz memory clock.
 
Certainly,
By the way the memory clock at 905 MHz is a fault spot in the Sapphire card with Samsung memory therefore, it is better to avoid the 905 MHz.

For Gigabyte test VDD and VDDCI from 750 mv to 850 mv at 940-950 MHz memory clock.
Well noted. I will avoid memory speed of 905 on the Sapphire card.
By the way, how can you tell that this frequency is a unstable?

Please find a summary report of the tests done one the Sapphire card with Samsung memory.
 

Anhänge

  • Igors Test Report.png
    Igors Test Report.png
    390,6 KB · Aufrufe : 130
Well noted. I will avoid memory speed of 905 on the Sapphire card.
By the way, how can you tell that this frequency is a unstable?

Please find a summary report of the tests done one the Sapphire card with Samsung memory.
Because it exceeded its power requirement,
By the way, please remember to increase the fan speed in the power table and make a new one with it as below:

For Gigabyte card:
Under Fan tab, set:
Fan Acoustic Limit 2400 RPM
Fan throttling 2400 RPM

For both Sapphire card with Micron and Samsung:
Under Fan tab, set:
Fan Acoustic Limit 1550 RPM
Fan throttling 3200 RPM
 
How do I chose the optimal Core Clock?
Depends on your goal, performance to power ratio, if power saving is preferred then decrease the core clock until you reach good MH/s like 55 with more power saved however less than 750 mv VDD or even 800 mv for some cards it yield to unstability of the system for the long run therefore, one can reduce the core clock until it match the 750 mv VDD.

if performance is preferred then increase the core clock until it match 800 mv and reach best MH/s with good power consumption.

The summary is you have 2 core clock to find for your crad, at 750 mv VDD and at 800 MHz VDD, after that compare the result for best performance to power ratio.
 
Depends on your goal, performance to power ratio, if power saving is preferred then decrease the core clock until you reach good MH/s like 55 with more power saved however less than 750 mv VDD or even 800 mv for some cards it yield to unstability of the system for the long run therefore, one can reduce the core clock until it match the 750 mv VDD.

if performance is preferred then increase the core clock until it match 800 mv and reach best MH/s with good power consumption.

The summary is you have 2 core clock to find for your crad, at 750 mv VDD and at 800 MHz VDD, after that compare the result for best performance to power ratio.
Is there a way to tell in HiveOS the cause of a crash? Was it under-volt of the core, or under-volt of the memory?
 
Depends on your goal, performance to power ratio, if power saving is preferred then decrease the core clock until you reach good MH/s like 55 with more power saved however less than 750 mv VDD or even 800 mv for some cards it yield to unstability of the system for the long run therefore, one can reduce the core clock until it match the 750 mv VDD.

if performance is preferred then increase the core clock until it match 800 mv and reach best MH/s with good power consumption.

The summary is you have 2 core clock to find for your crad, at 750 mv VDD and at 800 MHz VDD, after that compare the result for best performance to power ratio.
Based on the values for the extremes, here are the testing results for a Gigabyte with Micron memory.

I will test the card with Core 1430 750/750/1350. I am going to test for stability.
 

Anhänge

  • Igors Test Report.png
    Igors Test Report.png
    676 KB · Aufrufe : 161
  • Screen Shot 2020-10-21 at 12.49.54 PM.png
    Screen Shot 2020-10-21 at 12.49.54 PM.png
    255,4 KB · Aufrufe : 154
How do I chose the optimal Core Clock?
Depends on your goal, performance to power ratio, if power saving is preferred then decrease the core clock until you reach good MH/s like 55 with more power saved however less than 750 mv VDD or even 800 mv for some cards it yield to unstability of the system for the long run therefore, one can reduce the core clock until it match the 750 mv VDD.

if performance is preferred then increase the core clock until it match 800 mv and reach best MH/s with good power consumption.

The summary is you have 2 core clock to find for your crad, at 750 mv VDD and at 800 MHz VDD, after that compare the result for best performance to power ratio.
Finding the core clock is easier than finding the memory clock as there is only one variable related to it which is the VDD unlike the memory clock which has 2 variables the MVDD and VDDCI related to it,

Now we are going to do the same for the core clock as we did with memory clock, we fix all the values and we start changing the core clock and compare,
This is how we are going to find the best core clocks as below:

We are going to fix the values of MVDD at 1350 mv, VDDCI at 750 mv and memory clock at 950 MHz (for Micron and 900 MHz for Samsung),

After that we fix the VDD at 700 mv and we start changing the core clock values like below:

1250 MHz
1300 MHz
1350 MHz
1400 MHz
1450 MHz

After that we increase the VDD by 25 mv to 725 mv and repeat the clocks above until we reach 800 mv,

You can do the opposite as well, by fixing the core clock at a value like the lowest 1250 MHz and changing the VDD as below:

700 mv
725 mv
750 mv
775 mv
800 mv

after that increase the core clock by 50 MHz and repeat the VDD values above until we reach core clock at 1450 MHz,

Table is recommended to record and compare the results.
 
Zuletzt bearbeitet :
Finding the core clock is easier than finding the memory clock as there is only one variable related to it which is the VDD unlike the memory clock which has 2 variables the MVDD and VDDCI related to it,

Now we are going to do the same for the core clock as we did with memory clock, we fix all the values and we start changing the core clock and compare,
This is how we are going to find the best core clocks as below:

We are going to fix the values of MVDD at 1350 mv, VDDCI at 750 mv and memory clock at 950 MHz (for Micron and 900 MHz for Samsung),

After that we fix the VDD at 700 mv and we start changing the core clock values like below:

1250 MHz
1300 MHz
1350 MHz
1400 MHz
1450 MHz

After that we increase the VDD by 25 mv to 725 mv and repeat the clocks above until we reach 800 mv,

You can do the opposite as well, by fixing the core clock at a value like the lowest 1250 MHz and changing the VDD as below:

700 mv
725 mv
750 mv
775 mv
800 mv

after that increase the core clock by 50 MHz and repeat the VDD values above until we reach core clock at 1450 MHz,

Table is recommended to record and compare the results.
Please find is a report.
It looks like the efficient range is Core clock around 1420 to 1440, with core power of 725 - 750 mv.
I got a few crashes around VDD 725.
 

Anhänge

  • Igors Test Report.png
    Igors Test Report.png
    538,3 KB · Aufrufe : 162
Please find is a report.
It looks like the efficient range is Core clock around 1420 to 1440, with core power of 725 - 750 mv.
I got a few crashes around VDD 725.
You are right,
the first one, the core clock 1445 MHz, VDD 740 mv, is the memory temperature of it 84 at memory clock 960 MHz? this one is good,
And the core clock 1450 MHz at 750 mv is good as well,

If you like you can check the 1500 MHz as well for more information.
 
You are right,
the first one, the core clock 1445 MHz, VDD 740 mv, is the memory temperature of it 84 at memory clock 960 MHz? this one is good,
And the core clock 1450 MHz at 750 mv is good as well,

If you like you can check the 1500 MHz as well for more information.
I reached with one car 60.1MH/s 117W. Increased the core clock to 1500, and the volt down to 735mv, memory clock 960.
But it worked only for a short time before crashing. :)
 
I reached with one car 60.1MH/s 117W. Increased the core clock to 1500, and the volt down to 735mv, memory clock 960.
But it worked only for a short time before crashing. :)
Nice, now you need to find the right voltage for it, increase bit by bit above 735 mv to 850 mv until you reach stable one and check for the long run.
 
I reached with one car 60.1MH/s 117W. Increased the core clock to 1500, and the volt down to 735mv, memory clock 960.
But it worked only for a short time before crashing. :)
Nice, now you need to find the right voltage for it, increase bit by bit above 735 mv to 850 mv until you reach stable one and check for the long run.
Good morning,

I hope that you had a good sleep and enough rest,

Before trying to find the stable VDD for the 1500 MHz, check 1475 MHz as well to close the circle on the best core clock for performance to power ratio.

After that we will work on tuning the power for the memory of micron.

I am thinking about increasing the GFX Maximum Clock to 1700 MHz in the power table to require more information about the core clock impact on the performance to power ratio,
Would you like to do that? If you like to do that then let us use the temporary power table below,

- Under Features tab enable all the overdive features,
- Under Overdrive Limits tab, set:
GFX Maximum Clock (2000 Micron - 2250 Samsung) MHz
Memory Maximum Clock (1000 Micron - 1125 Samsung) MHz
Under Power and Voltages, set:
Maximum Voltage GFX 1250 mv
Maximum Voltage SoC 1250 mv
Minimum Voltage GFX 750 mv
Minimum Voltage SoC 750 mv
Power Limit GPU 180 W
TDC Limit GFX 166 A
TDC Limit SoC 14 A
- Under Frequency tap, set:
GFX Maximum 1400 MHz
Memory DPM 3 - (1000 Micron - 1125 Samsung) MHz
 
Zuletzt bearbeitet :
Good morning,

I hope that you had a good sleep and enough rest,

Before trying to find the stable VDD for the 1500 MHz, check 1475 MHz as well to close the circle on the best core clock for performance to power ratio.

After that we will work on tuning the power for the memory of micron.

I am thinking about increasing the GFX Maximum Clock to 1700 MHz in the power table to require more information about the core clock impact on the performance to power ratio,
Would you like to do that? If you like to do that then let us use the temporary power table below,

- Under Features tab enable all the overdive features,
- Under Overdrive Limits tab, set:
GFX Maximum Clock (2000 Micron - 2250 Samsung) MHz
Memory Maximum Clock (1000 Micron - 1125 Samsung) MHz
Under Power and Voltages, set:
Maximum Voltage GFX 1200 mv
Maximum Voltage SoC 1200 mv
Minimum Voltage GFX 750 mv
Minimum Voltage SoC 750 mv
Power Limit GPU 180 W
TDC Limit GFX 168 A
TDC Limit SoC (12 Micron - 14 Samsung) A
- Under Frequency tap, set:
GFX Maximum 1465 MHz
Memory DPM 3 - (1000 Micron - 1125 Samsung) MHz
Good morning,
The Power Table above breaks the card. I have to revert the bios settings.
 
Zuletzt bearbeitet :
Last night I had the production restarting a few times. Let's focus on stability and low power.

Here are a few ideas to consider:

1) Would it make sense to test the Micron memory speeds above 960?

2) Increasing the core speed makes a significant power increase, which reduces significantly the efficiency. Is there are a way to reduce the impact of core speed increase on power?
 
Oben Unten