Question GIGABYTE 5700 XT Bios mod fails

[AlleyCat]

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

 

[AlleyCat]

Greetings dear,

I would like to get a piece of your knowledge regarding when attaching the Gigabyte with the original bios as well as the reference card to the HiveOS, what are their default parameters regarding the MVDD and VDDCI and MEM clock?

I was trying to figure this our too. Take a look at my settings below. You will see that I let the driver/OS to select the power for the memory and the memory controller.

I am looking for a hardware monitor that could give us an insight to the power curve value at any given memory speed.

 

[Mini_Me]

I was trying to figure this our too. Take a look at my settings below. You will see that I let the driver/OS to select the power for the memory and the memory controller.

I am looking for a hardware monitor that could give us an insight to the power curve value at any given memory speed.

It is only a matter of time until we get our hands on the rest of the information for the power table,

For now the default ratio for the power and core clock is 1.7 and max safe ratio is 1.8, now I am trying to determine the minimum safe ratio if it is 1.6 or less in order to reduce the current (amperage) consumption and increase the voltage a little bit, this way the power at the wall may get reduced quite a bit.

 

[AlleyCat]

It is only a matter of time until we get our hands on the rest of the information for the power table,

For now the default ratio for the power and core clock is 1.7 and max safe ratio is 1.8, now I am trying to determine the minimum safe ratio if it is 1.6 or less in order to reduce the current (amperage) consumption and increase the voltage a little bit, this way the power at the wall may get reduced quite a bit.

I think if possible the end result should be a setting only the core and memory speed and the power table would set the amp/volt.

 

[Mini_Me]

I think if possible the end result should be a setting only the core and memory speed and the power table would set the amp/volt.

It is possible and it is possible to make a power table with the 1.8 ratio applied automatically without using navi curve helper as well.

 

[Pica]

I've got a question regarding the Gigabyte 5700xt Gaming OC Rev2.0. Does anyone know the difference between rev1.0 and rev2.0? I'm running 6 x rev 1.0 and 2 x rev 2.0. With the same OC settings and same apple straps, the rev2.0 cards run at about 20W higher. Note that this was the same with stock settings as well. Also, I'm having a hard time keeping my mem temps low. My core temps are fine, the rig is well ventilated and in an open air mining frame with additional CPU fans on each card. What would be a good mem temp to shoot for and what can I do to achieve that? I'm using auto-fan in HIveOS.

I have the same problem. Do you try to flash the lower consumption bios to the other GPU?
I haved try but the BIOS is signed and bricked the GPU. Any solution to that?

 

[Pica]

I will give it a try.

I have try to mod the BIOS Rev.02 but is signed, and when I flash the GPU bricked. Then I was mod the BIOS downloaded to the Gigabyte website, but this BIOS consumption +20w !!! Any solution?

 

[Calathia]

Hello,

My advice is to leave the soc clock at default as whether you set soc max at 950 MHz or at default 1267 MHz is the same, when you set the the memory clock at 910 MHz the soc will still run at 950 MHz and consume the same power however when the soc max set at default it will adapt properly to the power curve, if you are still want to set soc max then set it at most at 1093 MHz for mem at 950 MHz or at 1123 for Mem at 960 MHz.

Hmm, could you be more specific, I'm not sure if understood.

My SOC's are by default in range of 950-1267. By forcing all to 950 I save considerable amount of power consumption. However, system is not stable anymore and I cannot find a way how to research what GPU's causes reboots since miner logs get cleared on every reboot.

960 for mem seems quite hardcore overclocking? My max is at ~910-925

 

[Mini_Me]

Hmm, could you be more specific, I'm not sure if understood.

My SOC's are by default in range of 950-1267. By forcing all to 950 I save considerable amount of power consumption. However, system is not stable anymore and I cannot find a way how to research what GPU's causes reboots since miner logs get cleared on every reboot.

960 for mem seems quite hardcore overclocking? My max is at ~910-925

The power saving is done here not by locking the soc max value in vbios at 950 MHz however by setting the memory clock at (905 or 910) MHz, when you set the memory clock at (910 or 905) MHz the soc clock automatically will be set at 950 MHz,

Your rig is rebooting occasionally because your are setting your memory clock more than 910 MHz and forcing the soc clock at 950 MHz,

Memory clock at (1825 ÷ 2 = 912) MHz is the max memory clock for soc at 950 MHz, the recommended max memory clock for soc at 950 MHz is (1820 ÷ 2 = 910) MHz,

What was mentioned on other forums regarding locking the soc clock is useless and it was already verified and confirmed, unfortunately some is using this as a mean to collect money from those who do not have knowledge about this,

what really could change the outcome is the soc minimum clock, by increasing it it will improve the SMU performance however it was proven that might be fatal because of lack of knowledge regarding the limits.

Unlike locking the power limits which it has different purposes to it.

 

[Mini_Me]

I have try to mod the BIOS Rev.02 but is signed, and when I flash the GPU bricked. Then I was mod the BIOS downloaded to the Gigabyte website, but this BIOS consumption +20w !!! Any solution?

I will modify the bios file and upload it here whenever it is possible.

 

[Pica]

I will modify the bios file and upload it here whenever it is possible.

Thank you very much. I awaiting for this. Regards.

 

[Mini_Me]

Thank you very much. I awaiting for this. Regards.

You are very welcome.

 

[Pica]

I was trying to figure this our too. Take a look at my settings below. You will see that I let the driver/OS to select the power for the memory and the memory controller.

I am looking for a hardware monitor that could give us an insight to the power curve value at any given memory speed.

Do you send me the BIOS what do you use in your Gigabyte ?

 

[Sartori71]

I have the same problem. Do you try to flash the lower consumption bios to the other GPU?
I haved try but the BIOS is signed and bricked the GPU. Any solution to that?

I was able to “un brick” the two cards I bricked by using a different motherboard. I plugged the card into the x16 slot directly and was able to get the pc to post (previously the motherboard would not post if the card was plugged into any pcie slot, I would get a black screen with a white dot). Once i was able to get the pc to boot, I could see that the card was detected, an error was detected, but I then flashed it with the original bios And everything was fine. I then followed the flashing guide in this thread for the gigabyte cards. Note that I haven’t tried to mod the original bios for the rev 2.0 cards yet. I will try that when I’ve got some time this week. I’ve got 6 x rev 1 cards and 2 x rev 2 cards, all flashed with the rev 1 bios downloaded from gigabyte, and modified with the apple straps and micron settings from this thread. As noted the rev 2 cards run at about 20 watts higher.

 

[Calathia]

The power saving is done here not by locking the soc max value in morepowertool at 950 MHz however by setting the memory clock at (905 or 910) MHz, when you set the memory clock at (910 or 905) MHz the soc clock automatically will be set at 950 MHz,

Your rig is rebooting occasionally because your are setting your memory clock more than 910 MHz and forcing the soc clock at 950 MHz,

Memory clock at (1825 ÷ 2 = 912) MHz is the max memory clock for soc at 950 MHz, the recommended max memory clock for soc at 950 MHz is (1820 ÷ 2 = 910) MHz,

What was mentioned on other forums regarding locking the soc clock is useless and it was already verified and confirmed, unfortunately some is using this as a mean to collect money from those who do not have knowledge about this,

what really could change the outcome is the soc minimum clock, by increasing it it will improve the SMU performance however it was proven that might be fatal because of lack of knowledge regarding the limits.

Unlike locking the power limits which it has different purposes to it.

ahh, I was probably unclear on my first post... My mem speed is 900, which is default... I havent overclocked my mem speed at all, but I'm still getting reboots with SOC = 950.

And I measured power consumption from wall with default SOCS and 950 set... With 950, it lowers by ~2.7 watt per card (Which is ~40W with my system)! So thats why I would be happy to get this to work...If I could just somehow find the reason for reboots.

 

[Mini_Me]

ahh, I was probably unclear on my first post... My mem speed is 900, which is default... I havent overclocked my mem speed at all, but I'm still getting reboots with SOC = 950.

And I measured power consumption from wall with default SOCS and 950 set... With 950, it lowers by ~2.7 watt per card (Which is ~40W with my system)! So thats why I would be happy to get this to work...If I could just somehow find the reason for reboots.

What are the parameters that you set in the HiveOS regarding the core clock, VDD, Memory clock, VDDCI, MVDD and the DPM?

And if you believe that there is a 2.7 watt reduction in all cards and you believe it worths the risk then it is your decision.

 

[Calathia]

What are the parameters that you set in the HiveOS regarding the core clock, VDD, Memory clock, VDDCI, MVDD and the DPM?

And if you believe that there is a 2.7 watt reduction in all cards and you believe it worths the risk then it is your decision.

I dont trust power readings displayed by OS'ses. Thats why im using either ammeter or power consumption meter from the wall. And it really shows ~2.7 W reduction per card. This is average and reduction per card depends on what SOC value originally is.

what risk do you mean by capping soc to 950 ?

 

[Mini_Me]

Thank you very much. I awaiting for this. Regards.

[Deleted]

 

[Mini_Me]

Anhang anzeigen 7344

I dont trust power readings displayed by OS'ses. Thats why im using either ammeter or power consumption meter from the wall. And it really shows ~2.7 W reduction per card. This is average and reduction per card depends on what SOC value originally is.

what risk do you mean by capping soc to 950 ?

Regarding the settings, I believe it is the reason for your system reboot on the long run, use these settings below,

Core clock 1370 MHz, I believe more than this you will not see difference with memory clock 900 MHz,

VDD 760 mV less than this you will have system reboot on the long run, the ratio between core clock and VDD must not exceed 1.8, 1370 ÷ 760 = ~1.8

MVDD no less than 1310 mV for Micron at POD135 conditions or you put the RAM at risk of degradation and might die on the long run,

VDDCI 850 mV, no less than 845 mV for Micron at POD135 conditions you put the RAM at risk of degradation and might die on the long run,

Regarding the SoC, it is a very sensitive and important part of the gpu and you may be changing the power curve and thermal behavior by capping it at 950 MHz without any knowledge about it or the impacts on the long run which may lead to unpleasant outcome starting with odd behavior and on the long run might result in an unrecoverable and unrepairable card unless by the manufacturer,
If you know what you are doing then please do not mind me and proceed at your own discretion.

 

[Calathia]

MVDD no less than 1310 mV for Micron at POD135 conditions or you put the RAM at risk of degradation and might die on the long run,

VDDCI 850 mV, no less than 845 mV for Micron at POD135 conditions you put the RAM at risk of degradation and might die on the long run,

Regarding the SoC, it is a very sensitive and important part of the gpu and you are changing the power curve by capping it at 950 MHz without any knowledge about it or the impacts on the long run which may lead to unpleasant outcome starting with odd behavior and on the long run might result in an unrecoverable and unrepairable card unless by the manufacturer,
If you know what you are doing then please do not mind me and proceed at your own discretion.

Thank you very much for detailed description!

This is very interesting. Can you explain why too small voltages can degrade memory/GPU? One intuitively thinks that smaller voltages -> smaller stress to electronics -> longer lifetime.

Also practically all mining blogs / youtube videos undervolt GPUs to get best watts/hash ratio...everyone is doing it wrong? Should mining community be notified about this important manner about GPU degradation?

 

[Pica]

Alright, You can now download the modded rev. 2.0 vbios in the link below and modify it to your liking and reflash it,

And please do update whenever it is possible.

This bios is Rev.01 (FA0), the Rev.02 is (F30), right ?