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 :
Question regarding startup crashes in Phoenixminer. I'm using HiveOS which so far I'm enjoying. I've got everything dialed in, and previously it's been fairly stable.
Greetings,

First properly connect all power input to the power supply solution (ATX_12, ATX, Auxiliary and etc.) and make the power consumption in balance, also please note that it is important to provide sufficient power solution to the rig and it is highly recommended to provide higher power output around 250 watt more than the minimum power required for the rig,
After that clear the CMOS of your motherboard and update its bios to the latest version and please do not put PCI-e Configuration on Auto and set each slot to pcie 3.0 or 3.1 if supported, if not then set to pcie 2.0 or 2.1, and check,

If you still have same problems then please follow the guide in the kindly attached link below thoroughly,

 
Zuletzt bearbeitet :
Question regarding startup crashes in Phoenixminer. I'm using HiveOS which so far I'm enjoying. I've got everything dialed in, and previously it's been fairly stable. However since I've done the full bios mods with the apple straps Phoenixminer will crash and freeze during startup, usually 2 or 3 times before it finally starts. It always happens during the DAG loading part of the startup. Previously my bios mod was only with the 1500MHz straps copied to those below, however now I've done the MPT mod as well as the apple straps (with the settings suggested in this thread, max 140W). Currently here are my current miner config arguments: -coin %COIN% -amd -mode 1 -acm -altinit. Note that I did not have this issue previously on this same rig (8 x Gigabyte 5700xt Gaming OC), it's only after I've done the full bios mod and added the apple straps.

Also, I'm not a linux guru by any stretch, previously I've only mined in Windows. Where can I view the miner log files in HiveOS? I'm getting random GPU crashes where the Hashrate for one card will suddenly drop to 0, and then eventually the miner will crash or reboot, and of course when HiveOS reboots, I then most likely get my startup crash so I've got to physically go and power cycle the rig. Usually about 4 or 5 hours after mining starts.
Please do update whenever it is possible.
 
Question regarding startup crashes in Phoenixminer. I'm using HiveOS which so far I'm enjoying. I've got everything dialed in, and previously it's been fairly stable. However since I've done the full bios mods with the apple straps Phoenixminer will crash and freeze during startup, usually 2 or 3 times before it finally starts. It always happens during the DAG loading part of the startup. Previously my bios mod was only with the 1500MHz straps copied to those below, however now I've done the MPT mod as well as the apple straps (with the settings suggested in this thread, max 140W). Currently here are my current miner config arguments: -coin %COIN% -amd -mode 1 -acm -altinit. Note that I did not have this issue previously on this same rig (8 x Gigabyte 5700xt Gaming OC), it's only after I've done the full bios mod and added the apple straps.

Also, I'm not a linux guru by any stretch, previously I've only mined in Windows. Where can I view the miner log files in HiveOS? I'm getting random GPU crashes where the Hashrate for one card will suddenly drop to 0, and then eventually the miner will crash or reboot, and of course when HiveOS reboots, I then most likely get my startup crash so I've got to physically go and power cycle the rig. Usually about 4 or 5 hours after mining starts.

i was having same issue and thanks to @Mini_Me he helped me to make it stable watch this video and i hope it will help you
my GPUs

AORUS RX 5700 XT Bios
https://www.techpowerup.com/vgabios/217192/gigabyte-rx5700xt-8192-191105

Gigabyte Gaming OC RX 5700 8 GB BIOS
https://www.techpowerup.com/vgabios/213914/gigabyte-rx5700-8192-190812

I have mod my gigabyte 5700 non xt to XT with this bios
https://www.techpowerup.com/vgabios/214475/gigabyte- rx5700xt-8192-190909

this video to show u mod data
https://drive.google.com/file/d/1ywN-D_5XrZBzyr__IIsKCDLUy1hZjwUO/view?usp=sharing
 
I want to give a quick update of my Power Tale settings.

I made a change to the Power Table by reducing the SoC Max frequency to 950 and the results are a reduction of about 3-4W on my Gigabyte 5700 XT Gaming cards.

I got this idea from reading to comments provided by jgonzi at https://forum.hiveos.farm/t/reduce-watts-and-temp-memory-with-5700-and-5700xt/17078.

I am giving the credit for work done by jgonzi.

AlleyCat
Good day,

The SoC is really a sensitive thing, the outcome of altering it without proper knowledge and only by trial and error may lead to a severe and unrecoverable damage that will brick the card as this matter was already discussed and tested before and the result to the risk ratio made us not to modify what related to SoC that much, we even increased the minimum SoC clock from 507 MHz to 900 MHz and decreased the maximum from 1267 MHz to 950 MHz to improve the SMU performance and decrease power consumption however the outcome on the long run led to a fried SMU and the card died.

I hope that everything will be fine with you and I am still curious to see the outcome of jgonzi way on the long run,

Best Regards.
 
Zuletzt bearbeitet :
Good day,

The SoC is really a sensitive thing, the outcome of altering it without proper knowledge and only by trial and error may lead to a severe and unrecoverable damage that will brick the card as this matter was already discussed and tested before and the result to the risk ratio made us not to modify what related to SoC that much, we even increased the minimum SoC clock from 507 MHz to 900 and decreased the maximum from 1267 MHz to 950 MHz to improve the SMU performance and decrease power consumption however the outcome on the long run led to a fried SMU and the card died.

I hope that everything will be fine with you and I am still curious to see the outcome of jgonzi way on the long run,

Best Regards.
Thanks for your concern. Let me make sure that I understand; by reducing the SoC minimum frequency to 950 the long-term stability of the card could lead to a total breakdown?

Thanks.
AlleyCat
 
Thanks for your concern. Let me make sure that I understand; by reducing the SoC minimum frequency to 950 the long-term stability of the card could lead to a total breakdown?

Thanks.
AlleyCat
by increasing the soc minimum clock to 900 MHz to improve performance and decreasing the maximum soc clock to 950 MHz to reduce power consumption that led to a fried SMU on the long run.

It is all about the power curve, it is tricky,
What I found after your help and a friend help that the ratio between the core clock and core voltage should not exceed 1.8 as this verified it is the minimum for stable safe long run however with a 1% decrease in real performance of the core clock,
for an example core clock 1440 MHz at VDD 800 the ratio is
1440/800 = 1.8 more than this will lead to temperature increase and instability and the duration of system failure increases by this, for an example
1440/775 = 1.858 will run for a couple of hours
1440/750 = 1.92 will run for 1 hour,
1440/720 = 2 instant crash.

Now imagine everything have a ratio here, the memory clock to memory voltage and controller voltage, the soc clock to soc voltage as well, and their ratio are unknown and I am trying to figure it out with simplest way however it is really hard without knowing the detailed relationship between the components and their limits, therefore to change how these were designed for gaming to pure computing and mining purpose.
 
Zuletzt bearbeitet :
Very interesting. So you think that the ratio of 1.8 on memory clock to power is the maximum stable?
Would this 1.8 ration is applicable to any card of specifically to the Gigabyte cards, as referred to in your link above?
 
Very interesting. So you think that the ratio of 1.8 on memory clock to power is the maximum stable?
Would this 1.8 ration is applicable to any card of specifically to the Gigabyte cards, as referred to in your link above?
I believe it depends on the manufacturer restrictions as well,
and even this ratio it depends on the provided voltage and current, for an example in the post above in the link below,

there is 2 settings for Max gfx voltage and TDC gfx limit, one is for good temperature and power consumption by automatically charging the power curve that was set by the manufacturer and the other setting is to improve that, the ratio that I mentioned before for the second setting may be different from the first one which almost share the same ratio of the manufacturer,

Actually there is another setting regarding the memory that I believe no one try it yet, I am now looking for a volunteer who has 2 cards, one with Samsung memory and one with Micron memory other than the reference card with blower to help in verifying the changing of this setting in other brands.
 
Zuletzt bearbeitet :
i was having same issue and thanks to @Mini_Me he helped me to make it stable watch this video and i hope it will help you
my GPUs

AORUS RX 5700 XT Bios
https://www.techpowerup.com/vgabios/217192/gigabyte-rx5700xt-8192-191105

Gigabyte Gaming OC RX 5700 8 GB BIOS
https://www.techpowerup.com/vgabios/213914/gigabyte-rx5700-8192-190812

I have mod my gigabyte 5700 non xt to XT with this bios
https://www.techpowerup.com/vgabios/214475/gigabyte- rx5700xt-8192-190909

this video to show u mod data
https://drive.google.com/file/d/1ywN-D_5XrZBzyr__IIsKCDLUy1hZjwUO/view?usp=sharing
I am glad it was useful and helpful to you,

And no need to thank me at all for anything as it is not my intention at all to get credits or receive thanks or get benefits of any kind for sharing knowledge and easing things for others and anything regarding that, I am glad to share with others everything good and beneficial and if my sharing was useful and helpful then I am happy,

You may have a good day,

Best Regards.
 
Please do update whenever it is possible.
Thank you for the information. I will check the motherboard settings. I do have the latest Motherboard Bios, but I believe I only verified that the bios was set to Mining Mode (Asus B250 Mining Expert). The power connections are good, I've verified that all connections are good. I've got 1 x 1200W and 1 x 750 W for the entire rig (8x5700xt). Do you know the best optimal settings for this board specifically?

I believe my BIOS flash went well, I initially followed the process outlined in this thread with no issues. I've since done it again with the MPT settings and the Apple straps. The only real issue I'm having is that 2 of my cards are the Rev 2.0 of the Gigabyte 5700xt gaming OC, the other 6 are Rev 1.0. The Rev 2.0 run at about 20W higher than rev 1.0. This is with stock bios and settings as well as flashed bios and MPT settings in this thread. I haven't been able to find any difference in the cards that would cause this much difference in power draw. I've attached a screenshot showing the exact same settings, and I've verified that they are flashed with the exact same modified bios as the other 6 cards.

Currently my rig has been running stable for the last 17 hours, so I haven't made any adjustments yet, but this is by far the longest it's remained stable.
 

Anhänge

  • Screenshot 2020-11-07 104737.png
    Screenshot 2020-11-07 104737.png
    294,5 KB · Aufrufe : 88
i was having same issue and thanks to @Mini_Me he helped me to make it stable watch this video and i hope it will help you
my GPUs

AORUS RX 5700 XT Bios
https://www.techpowerup.com/vgabios/217192/gigabyte-rx5700xt-8192-191105

Gigabyte Gaming OC RX 5700 8 GB BIOS
https://www.techpowerup.com/vgabios/213914/gigabyte-rx5700-8192-190812

I have mod my gigabyte 5700 non xt to XT with this bios
https://www.techpowerup.com/vgabios/214475/gigabyte- rx5700xt-8192-190909

this video to show u mod data
https://drive.google.com/file/d/1ywN-D_5XrZBzyr__IIsKCDLUy1hZjwUO/view?usp=sharing
Thank you, I will go through the videos!
 
Update on Sappphire Pulse Micron Rx5700:
I haven't yet played with the CPU settings because I have the numbers of extensive testing and don't want to start all over, so that will come at the end.

But some interesting findings on core speed. The best memory profile for temps held in check, decent speed, good stability and a little bit of power saving is:

Memory Speed: 910 Mhz
MVDD: 1305 mV
VDDCI: 775 mV

This allows the card to mine at stable 57 MH/s.


Interestingly, core speeds have been very important for both speed, stability, and of course, power usage. While impact on power is significant, we're talking about 3W per card difference, so not a back breaker.

The best for power consumption vs speed is :
Core Speed: 1350 MHz
VDD: 740 mV
Card is stable and producing around 57 MH/s, HOWEVER, and this is a big one, of the 13 cards in one rig and 5 in the second, there is always MULTIPLE cards being weak-mining around 55 MH/s , and multiple cards going so-so (56.0, 56.5, 56.6).
The WEIRD thing is-this happens to RANDOM cards at every startup! So one card will go at 57.0 MH/s, then you restart and it goes at 55.1 MH/s.
And the total speed remains pretty similar every time for the entire rig-so different cards get faster while different others go slower

So I tried to get them all stable at 57.0 MH/s and it is possible at:

Core Speed: 1380 MHz
VDD: 760 mV

At this setting, all cards are 57.0 MH/s flat.

Question: is it possible that the lower clock/VDD are not only producing less MH/s, but also making the cards unstable over longer period of time, it seems like the point of break for them because they behave like they did-losing mh/s for no reason. Could it also be damaging the card over longer periods of time?

Weird thing: At Core speed 1350 MHz and VDD at 750 mV the rig produces LESS Mh/s than at 1350 MHz/740 mV.


1340 MHz is no good, crashes it.
1390 MHz and over gains nothing.
 
Very interesting. So you think that the ratio of 1.8 on memory clock to power is the maximum stable?
Would this 1.8 ration is applicable to any card of specifically to the Gigabyte cards, as referred to in your link above?
Good morning,

If you are going with jgonzi method then I advise not to let the value of SoC Maximum less than 1090 MHz and preferably 1093 MHz in the power table, and to use a script or MorePowerTool from inside the system to set it at 950 MHz which is the safer way, and it is entirely at your own discretion.

Please stay safe and healthy,

Best Regards.
 
Zuletzt bearbeitet :
Update on Sappphire Pulse Micron Rx5700:
I haven't yet played with the CPU settings because I have the numbers of extensive testing and don't want to start all over, so that will come at the end.

But some interesting findings on core speed. The best memory profile for temps held in check, decent speed, good stability and a little bit of power saving is:

Memory Speed: 910 Mhz
MVDD: 1305 mV
VDDCI: 775 mV

This allows the card to mine at stable 57 MH/s.


Interestingly, core speeds have been very important for both speed, stability, and of course, power usage. While impact on power is significant, we're talking about 3W per card difference, so not a back breaker.

The best for power consumption vs speed is :
Core Speed: 1350 MHz
VDD: 740 mV
Card is stable and producing around 57 MH/s, HOWEVER, and this is a big one, of the 13 cards in one rig and 5 in the second, there is always MULTIPLE cards being weak-mining around 55 MH/s , and multiple cards going so-so (56.0, 56.5, 56.6).
The WEIRD thing is-this happens to RANDOM cards at every startup! So one card will go at 57.0 MH/s, then you restart and it goes at 55.1 MH/s.
And the total speed remains pretty similar every time for the entire rig-so different cards get faster while different others go slower

So I tried to get them all stable at 57.0 MH/s and it is possible at:

Core Speed: 1380 MHz
VDD: 760 mV

At this setting, all cards are 57.0 MH/s flat.

Question: is it possible that the lower clock/VDD are not only producing less MH/s, but also making the cards unstable over longer period of time, it seems like the point of break for them because they behave like they did-losing mh/s for no reason. Could it also be damaging the card over longer periods of time?

Weird thing: At Core speed 1350 MHz and VDD at 750 mV the rig produces LESS Mh/s than at 1350 MHz/740 mV.


1340 MHz is no good, crashes it.
1390 MHz and over gains nothing.
Great work you have done there,

I am going in the future to make a new thread and summarize everything,

I recommend to use the power table at the tope of this page and to study the change of Memory Timings Control 1 instead of 2 for cards with both Samsung and Micron memory that you may find an interesting outcome from it.

You may have a good day,

Best Regards.
 
I am having very good results with Red Team Miner. The miner reports a lower hash rate, BUT the pool reports more shares. Also the reported hash rate on the pool is very close to the reported rate, which suggests the correct amount paid for dev fees.
And finally, at the same OC settings the rig runs are 10 to 15 w lower. Must be due to the optimization algorithm.
I left the default miner settings and I’m only using the miner with Navi cards.
I would love to hear confirmation to my observations.
AlleyCat
 
I am having very good results with Red Team Miner. The miner reports a lower hash rate, BUT the pool reports more shares. Also the reported hash rate on the pool is very close to the reported rate, which suggests the correct amount paid for dev fees.
And finally, at the same OC settings the rig runs are 10 to 15 w lower. Must be due to the optimization algorithm.
I left the default miner settings and I’m only using the miner with Navi cards.
I would love to hear confirmation to my observations.
AlleyCat
didn't noticed any change except -1MH/gpu on the software. But my test is slightly different as i'm getting better performance with flash to 5700.
 
didn't noticed any change except -1MH/gpu on the software. But my test is slightly different as i'm getting better performance with flash to 5700.
As it is already mentioned before,

Software optimization is important likewise the hardware.

What is needed is that first to optimize the software at fixed settings until reaching a state when there is no more changes in the outcome or finding a good optimized one, after that optimizing the hardware comes and likewise the software optimization.
 
I am having very good results with Red Team Miner. The miner reports a lower hash rate, BUT the pool reports more shares. Also the reported hash rate on the pool is very close to the reported rate, which suggests the correct amount paid for dev fees.
And finally, at the same OC settings the rig runs are 10 to 15 w lower. Must be due to the optimization algorithm.
I left the default miner settings and I’m only using the miner with Navi cards.
I would love to hear confirmation to my observations.
AlleyCat
Are you saying that with Team Red Miner, your "mean" (in hiveOS: workers -> stats) is close what hive is diplaying as "reported" ?

Because for some reason I always have mean to be ~5% lower than reported. Even though I live relatively near of one worker API's and I have good and stable internet :/

1604910250186.png

Do I misunderstand something here? Its not the ping to your worker API thats important but ping to miner? How do I ping to my miner (phoenixmoner) or other available miners?

Sorry for offtopic!
 
Zuletzt bearbeitet :
The ping is important to the pool server, but what's more important is the actual shares, valid vs stale and invalid which is the final product of mining, connection quality, card stability etc. I see you have around 5% stale shares, me too, it is around 4%. Hiveon pool just isn't working for me. I have 0.8% stale shares on ethermine. So I'll pay the $3 to HiveOs and improve actual paid mining by 3.2%.
 
Oben Unten