Overclock.net banner

41 - 60 of 178 Posts

·
Registered
Joined
·
373 Posts
Thank you for the informative reply..

If AMD didnt see this overvolting in testing, how can they be certain that inconsistent components in the card are not causing it?

It seems like Amd will likely have to down clock the card, whats that going to do to components and the longevity of the card?

I still think my 480 is going to be returned or sold on ebay, and dont particularly want to stick someone else with it under the circumstances.... I feel amd just rushed another untested product to market, pushing the testing to the consumer to save money and its biting them in the rear... I just dont trust them now..
 

·
Premium Member
Joined
·
2,753 Posts
Discussion Starter · #43 ·
Quote:
Originally Posted by NomNomNom View Post

@The Stilt
Does this fix work for multiple card rigs? Lets say for 6 gpus.
Try following (with at least two cards installed):

"MSIAfterburner /sg0 /wi06,08,1E,BB"
"MSIAfterburner /sg0 /wi06,08,1F,B0"

Then

"MSIAfterburner /sg1 /ri06,08,1E"

If the response is not ": BB", then give the just repeat the commands while switching the /sgx parameter.

"MSIAfterburner /sg0 /wi06,08,1E,BB"
"MSIAfterburner /sg0 /wi06,08,1F,B0"

"MSIAfterburner /sg1 /wi06,08,1E,BB"
"MSIAfterburner /sg1 /wi06,08,1F,B0"

"MSIAfterburner /sg2 /wi06,08,1E,BB"
"MSIAfterburner /sg2 /wi06,08,1F,B0"

"MSIAfterburner /sg3 /wi06,08,1E,BB"
"MSIAfterburner /sg3 /wi06,08,1F,B0"

"MSIAfterburner /sg4 /wi06,08,1E,BB"
"MSIAfterburner /sg4 /wi06,08,1F,B0"

"MSIAfterburner /sg5 /wi06,08,1E,BB"
"MSIAfterburner /sg5 /wi06,08,1F,B0"

That would cover six different cards. However you better be sure as hell that you have a extra power plug attached to the motherboard, since otherwise the ATX connector is still likely to fry.
 

·
Registered
Joined
·
7 Posts
@The Stilt

As always, you are very knowledgeable. Thanks a lot for the effort!

On other topic about the IR3567B,
Since the programming for the IR3567B is confidential, do you know some values we can write via afterburner to the VRM controller to 'fool' the VRM into thinking that it detects lower power draw? i.e actual GPU power draw is 150W but the VRM reads 75W or something like that (sorry for bad English).

I've been able to send offset volts to the GPU to get more than 1.15v, but even at 1400Mhz @ 1.18V (after vdroop) it throttles a lot due to power limit. Any command that we could send to the VRM controller as to elevate the power limit ?
 

·
Registered
Joined
·
715 Posts
as its now possible to reflash polaris bios you can make yourself modified bios with higher limits
wink.gif
or modify VRM table inside bios with higher values (check hawaii bios editing thread it will be same)
 

·
Premium Member
Joined
·
2,753 Posts
Discussion Starter · #46 ·
Quote:
Originally Posted by AlvaJonathan View Post

@The Stilt

As always, you are very knowledgeable. Thanks a lot for the effort!

On other topic about the IR3567B,
Since the programming for the IR3567B is confidential, do you know some values we can write via afterburner to the VRM controller to 'fool' the VRM into thinking that it detects lower power draw? i.e actual GPU power draw is 150W but the VRM reads 75W or something like that (sorry for bad English).

I've been able to send offset volts to the GPU to get more than 1.15v, but even at 1400Mhz @ 1.18V (after vdroop) it throttles a lot due to power limit. Any command that we could send to the VRM controller as to elevate the power limit ?
IR or CHIL based controllers don't limit the power. They have per phase and total current limits, which are usually high enough. On RX 480 the limit is 36A per phase, i.e 216A in total. So unless your card crashes or the system shuts down the problem has nothing to do with the VRM controller.

The software power management is most likely the reason. RX 480 has default TDP of 110W. The driver allows it to be raised by 50% (165W). Most likely the power limit is simply not enough when the card is overclocked. You need to increase the allowed power limit by either increasing the default TDP (110W) and TDC (107A) or by increasing the PowerTune margin (from ±50%).
 

·
Registered
Joined
·
7 Posts
Quote:
Originally Posted by The Stilt View Post

IR or CHIL based controllers don't limit the power. They have per phase and total current limits, which are usually high enough. On RX 480 the limit is 36A per phase, i.e 216A in total. So unless your card crashes or the system shuts down the problem has nothing to do with the VRM controller.

The software power management is most likely the reason. RX 480 has default TDP of 110W. The driver allows it to be raised by 50% (165W). Most likely the power limit is simply not enough when the card is overclocked. You need to increase the allowed power limit by either increasing the default TDP (110W) and TDC (107A) or by increasing the PowerTune margin (from ±50%).
I see, I already maxed out the PowerTune margin to 50% but the card still throttles when GPU Volts is set to above 1.17v @ 1400Mhz (after vdroop), 1.15-1.16v is the maximum I can set with +50% PT before it starts downclocking. I'll try to see If I can mod something else then. Thanks
wink.gif
 

·
Premium Member
Joined
·
9,327 Posts
Quote:
Originally Posted by AlvaJonathan View Post

I see, I already maxed out the PowerTune margin to 50% but the card still throttles when GPU Volts is set to above 1.17v @ 1400Mhz (after vdroop), 1.15-1.16v is the maximum I can set with +50% PT before it starts downclocking. I'll try to see If I can mod something else then. Thanks
wink.gif
What temps? Both core and vrm
 

·
Meddling user
Joined
·
7,423 Posts
@The Stilt

The question AlvaJonathon asking is something similar I asked der8auer after viewing his video (waiting on him to reveal register),
.

This
him changing this value which is supposed half what the IR3567B sees as GPU draw from what der8auer states earlier.
 

·
Premium Member
Joined
·
2,753 Posts
Discussion Starter · #51 ·
Bios files were updated with proper roms, anyone who downloaded the previous package (128KB flash files), please discard it immediately.

See OP.
 

·
Premium Member
Joined
·
9,327 Posts
Quote:
Originally Posted by AlvaJonathan View Post

fan at 100%, room temp 22c, GPU max temp around 68C in Fire Strike. VRM I didn't monitor yet
Glofo loves low temps with higher volts. And from the reviews most vrms temps is around 90C-98C so you lose a bit of efficiency there. I am not so sure if you hit a wall because of the temps or the power limit at the moment
 

·
Registered
Joined
·
189 Posts
Dusted off my OCN account to say thank you, OP. Really appreciate the temp fix and it demonstrates that the issue can be resolved without lowering performance through software/BIOS updates, as opposed to being a hardware flaw.
 

·
Registered
Joined
·
270 Posts
First I want to say EXCELLENT work to the OP!
Quote:
Originally Posted by I-o-I View Post

thanks for the reply
thumb.gif

Yes, I understand those basics.
However, Igor from Tom's Hardware deduced that actually 4 phases are connected to the slot and only 2(+1) to the 6-pin connector.
Toms was speculating when they made that pic. Here's the layout via direct measurement (I followed from the PWM output all the way to the FETs of each power phase):
 

·
Premium Member
Joined
·
2,753 Posts
Discussion Starter · #55 ·
Quote:
Originally Posted by malventano View Post

First I want to say EXCELLENT work to the OP!
Toms was speculating when they made that pic. Here's the layout via direct measurement (I followed from the PWM output all the way to the FETs of each power phase):
Good job
wink.gif
 

·
PC Evangelist
Joined
·
47,497 Posts
Quote:
Originally Posted by xzamples View Post

AMD was suppose to give an official update on this situation today.... where is it at? it's 10PM EST
I am sure they have their reasons. They might have found a fix but not implemented in drivers.

Also Stilt have you contacted PCPer. You be cool if they tested it.
 

·
Registered
Joined
·
35 Posts
Hey guys, Allyn ran it through some quick testing and it appears to be doing some good. I didn't make it back in the office to put testing together, validate performance, etc.

I am hearing that AMD will have a response tonight, but not a fix. Likely tomorrow.

I would GUESS it will be very similar to what is being done with this temp mod.
 

·
PC Evangelist
Joined
·
47,497 Posts
Quote:
Originally Posted by RyanShrout View Post

Hey guys, Allyn ran it through some quick testing and it appears to be doing some good. I didn't make it back in the office to put testing together, validate performance, etc.

I am hearing that AMD will have a response tonight, but not a fix. Likely tomorrow.

I would GUESS it will be very similar to what is being done with this temp mod.
You guys Rock. Thanks for testing so fast. Keep up the good work.
 
41 - 60 of 178 Posts
Top