Overclock.net banner

21 - 40 of 178 Posts

·
Meddling user
Joined
·
7,423 Posts
Quote:
Originally Posted by PCIEgate View Post

Thanks a lot for your effort.
Can you please explain how applying this fix directly on the BIOS would work?
Maybe we can include this function on the Polaris BIOS Editor: http://forums.guru3d.com/showthread.php?t=408469
This fix via bios would mean editing VoltageObjectInfo. There is discussion on how to edit VoltageObjectInfo in hawaii bios mod thread.
i) add appropriate registers/data values to VoltageObjectInfo in i2c programming section.
ii) correct table length for VoltageObjectInfo and pointer to i2c programming within VoltageObjectInfo as/if required.
iii) delete original VoltageObjectInfo in ROM, insert new VoltageObjectInfo in ROM, due to length of new VoltageObjectInfo the UEFI module will shift offset location, delete padding to bring it back to correct offset location.
iv) update data/command table pointers in upper section of ROM as required.
v) fix checksum on legacy section of ROM.
vi) flash.
Quote:
Originally Posted by Visceral View Post

Most important question: Does it effect performance at stock or when overclocked?
This will not effect GPU performance, it has only shifted how VRM phase loading is.
 

·
Premium Member
Joined
·
2,753 Posts
Discussion Starter · #23 ·
Quote:
Originally Posted by Visceral View Post

Most important question: Does it effect performance at stock or when overclocked?
Performance, no. However you might hit the overcurrent protection at some point, if you increase the core voltage heavily from the default.
 

·
Premium Member
Joined
·
2,753 Posts
Discussion Starter · #24 ·
Quote:
Originally Posted by rjc862003 View Post

what do the hex values correspond to
AMPS ? Percent ?
It is a fractional number, representing percentages. However I cannot disclose it any futher for the reasons stated before (confidential).
 

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

Performance, no. However you might hit the overcurrent protection at some point, if you increase the core voltage heavily from the default.
Since your loading half the phases with more current, it *could* have an impact on stability especially if your under-volting/over-volting. The higher loaded phases will not respond as well to transients and ripple currents, but again like everyone mentioned this DC-DC design is way overbuilt so I doubt youd see any impact unless your going for extreme overclocks.
 

·
Meddling user
Joined
·
7,423 Posts
+rep for share
smile.gif
.

I'm assuming from what I've noted on Hawaii/Fiji bios mod these ROMs will not boot when mobo is in "pure UEFI" mode (ie CSM=Off, SB=On, FB=On) even though they have UEFI/GOP module? This is due to the modification to legacy section for fix and signature protecting legacy section not being updatable by us, which the UEFI refs?
 

·
Premium Member
Joined
·
2,753 Posts
Discussion Starter · #28 ·
I've never have any such issues before, however our tools may slightly differ. Obviously I have no tools which work with Polaris based cards, which means manual labour (and lot of it).
 

·
Registered
Joined
·
43 Posts
if anything shifting the load toward the 6 pin should allow for some more headroom
currently VDROP due to insufficient current available is what is limiting vcore,but thats a separate issue
 

·
Registered
Joined
·
1,627 Posts
Quote:
Originally Posted by SanguineDrone View Post

Fixes something that isn't even a problem, hmmm.
The only real problem is the media blowing this way out of proportion.
While I agree with you on that, this isn't the thread for it, the OP is offering assistance for the problem and should be thanked.
My opinion atleast.

+rep OP
 

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

As I haven't been taking apart RX 480 ROM much due to lack of flasher, which is solved now
smile.gif
. I thought I'd see the VOI you've done and noted stock ROM had no VDDC limit, is that correct?
 

·
Premium Member
Joined
·
2,753 Posts
Discussion Starter · #34 ·
That's correct...
A mistake IMO, since I've killed some CPUs accidentally when the limit has not been set.

1.36875V was set for enhanced safety and should be more than enough, even on custom water cooling.
 

·
Meddling user
Joined
·
7,423 Posts
I'm amazed stock ROM/MTP had none and yeah agree good addition.

Only today I was viewing EVC tool by Elmor on HWBot, as
had it, on stock ROM coupled with say that tool or i2c programming of VDDC offset via MSI AB could equal dead GPU.
 

·
Registered
Joined
·
373 Posts
so moving more of the power load to the 6pin, whats that going to do to your psu?

A good psu costs half of what a 4gig 480 costs and a good motherboard cost the same..

What about the longevity of the product, if amd didnt catch this in testing, just shows someone or some dept completely dropped the ball?

My sapphire rx480 8gig is to be delivered tomorrow. I spoke with the place I ordered it from and they will accept a return and full refund being its unopened.

I'm doubting it and its build quality and I no longer trust amd or their architecture or test procedures..
 

·
Registered
Joined
·
43 Posts
nothing is gonna happen to the psu
most power supply's the 6/8 pin's share the same rail/physical connection anyway
the way AMD has the 6 pin wired is,its capable of well over 150W
 

·
Premium Member
Joined
·
2,753 Posts
Discussion Starter · #38 ·
Quote:
Originally Posted by danjal View Post

so moving more of the power load to the 6pin, whats that going to do to your psu?

A good psu costs half of what a 4gig 480 costs and a good motherboard cost the same..

What about the longevity of the product, if amd didnt catch this in testing, just shows someone or some dept completely dropped the ball?

My sapphire rx480 8gig is to be delivered tomorrow. I spoke with the place I ordered it from and they will accept a return and full refund being its unopened.

I'm doubting it and its build quality and I no longer trust amd or their architecture or test procedures..
Even in the worst case scenario, the PCI-E 6-pin connector can handle 177 - 207W of power draw. The worst case scenario being: 2/3 VDD and VSS pins connected on the PCI-E power connector and low grade connectors used in the power cable or in the cable header. Low grade connectors are rated for 8A per contact and the high grade ones are rated for 11A. The ATX specification recommends to use 16AWG wires (12A rated each) for these connectors but allows 18AWG (8.6A rated each) to be used. So at minimum you are looking at 16A at 11.04V - 12.96V.

And when a fully connected high quality connector, paired with 16AWG wires is used you can draw safely at least 265 - 311W from a single 6-pin PCI-E connector. If your PSU can deliver, that is.
 

·
Registered
Joined
·
374 Posts
Quote:
Originally Posted by iTurn View Post

That's just it though, the OP basically just called you and asked if your refrigerator is running, then offered you a net to catch it with.
While I agree with you on that, this isn't the thread for it, the OP is offering assistance for the problem and should be thanked.
My opinion atleast.

+rep OP
 
21 - 40 of 178 Posts
Top