Overclock.net › Forums › Graphics Cards › AMD/ATI › A temporary fix for the excess PCI-E slot power draw for the reference RX 480 cards
New Posts  All Forums:Forum Nav:

A temporary fix for the excess PCI-E slot power draw for the reference RX 480 cards - Page 8

post #71 of 178
Quote:
Originally Posted by rjc862003 View Post

3% lol
what
thats within the margin of error for any benchmark of the planet
do they mean 30% ?

No, they mean 3%. Because that's probably around how much you expect to lose when you gimp the power draw to 150W exactly. The improvements were likely already in the source tree, so they simply fast-tracked the RX 480 fixes and pushed a release forward into their internal testing system.
post #72 of 178
I am assuming they are setting the IC2 register at boot via the driver
which makes sense as it saves needless end-user bios update risk
post #73 of 178
Quote:
Originally Posted by looncraz View Post

No, they mean 3%. Because that's probably around how much you expect to lose when you gimp the power draw to 150W exactly. The improvements were likely already in the source tree, so they simply fast-tracked the RX 480 fixes and pushed a release forward into their internal testing system.

They mean 3% performance improvement from driver improvement as is normal with new releases and will be 3% higher than current performance. they've also likely implemented a similar redistribution as the stilt has for power management, reducing the draw on the PCI-e slot and increasing it from the 6 pin, not actually lowering it unless you use the compatibility toggle. I wonder if they undervolted the GPU at all though, seems conservatively high from reports.

Seems like a timely (ie very well done to get it out so quickly) "fix" to allay the fears of the masses that were whipped up by the media about a non issue.
Edited by LiamG6 - 7/5/16 at 10:03pm
post #74 of 178
I doubt they will touch the vcore via the driver thats a bit risky for the end-user (from AMD's stand point anyway)
I am very interested to see how much of the load they shift around using the driver (potential driver mod?)
post #75 of 178
Quote:
Originally Posted by rjc862003 View Post

I doubt they will touch the vcore via the driver thats a bit risky for the end-user (from AMD's stand point anyway)
I am very interested to see how much of the load they shift around using the driver (potential driver mod?)

I believe it changes from the stock 50/50 from PCI-e slot and 6 pin to 30/70 from PCI-e slot and 6 pin. ie PCI-e slot now only handles 30% of total power and 6 pin handles 70%. The "compatibility" mode is probably to bring the 6 pin power draw within spec, even though it is capable of far far higher, at least there is an option to use very cheap PSU's without exceeding 6 pin specs. For the rest of us, have at it and ignore the compatibility mode

from reports I think they could easily lower voltage 100mv for stock core speed.
Edited by LiamG6 - 7/5/16 at 10:12pm
post #76 of 178
see my data collection thread here for number I am well aware they could easily knock 0.100 - 0.200mv off the v core across the board and nobody with blink
https://www.reddit.com/r/Amd/comments/4qupw4/super_psa_all_rx480_owners_please_attempt_to/
but I doubt AMD is going to
post #77 of 178
perhaps the compatibility mode does just that. or lower total power limit.

I think I read that 20% voltage drop ended up with like 4% lower performance, 10% lower was virtually the same performance as stock
post #78 of 178
why even have a compatibility mode (its just going to confuse people)
its not a issue not on any power supply on the planet
not even on crappy HEC built units
assuming you have the worse PSU on earth and it uses crappy 22AWG wire and sub-standard connectors

you can still safely pull 150W out of it assuming the psu can output that much (and thats being extremely extremely conservative
post #79 of 178
neither was the issue of drawing too much power over the pci-e slot, but here we are. they most likely did it to stop the whole flame war of ermahhgerdd it is using more than 75w over 6 pin. it is a preemptive fix to the logical next thing that everyone will start complaining about, to which they can now respond, simply apply the compatibility toggle if you don;t want to exceed 6 pin specs, even though every 6 pin on every PSU is capable of it.
post #80 of 178
unfortunately we live in a world where somebody opens there mouth takes a little bit of fact and suddenly they are a expert on the subject

and people are lazy and stupid enough to believe anybody that sounds like they might, maby possibly, probably know a little bit more then they do instead of taking the time to research and learn and most of ALL QUESTION

I may not be 1337 at bios modding like the slit but god knows I would like to be but I am competent enough to understand topic at hand and craft my own assumptions and test them which is more I can say then 90% of the userbase here or any other techforum for that matter
Edited by rjc862003 - 7/5/16 at 10:42pm
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD/ATI
Overclock.net › Forums › Graphics Cards › AMD/ATI › A temporary fix for the excess PCI-E slot power draw for the reference RX 480 cards