AMD update to correct power consumption on RX480.
-
It seems that between today and next Friday they will release a new driver for the RX480 (Crimson Edition 16.7.1) adding an option in the global settings that allows the user to enable/disable a "compatibility" mode (disabled by default) to reduce excess current on the PCIe bus. As this will surely reduce performance a bit, they are covering their bases by saying that they have applied optimizations to the driver to increase performance by approx. 3%, so the reduction in current applied is offset.
To be honest, I find it incredible that a newly released product needs such a major fix (in my opinion). How did they not detect this problem in their test benches?
On the other hand, their quick response is appreciated and it's fortunate that it can be fixed via software. There would be many unhappy customers if this had to be corrected with a new PCB revision of the card.I quote their press release:
@AMD:We promised an update today (July 5, 2016) following concerns around the Radeon RX 480 drawing excess current from the PCIe bus. Although we are confident that the levels of reported power draws by the Radeon RX 480 do not pose a risk of damage to motherboards or other PC components based on expected usage, we are serious about addressing this topic and allaying outstanding concerns. Towards that end, we assembled a worldwide team this past weekend to investigate and develop a driver update to improve the power draw. We're pleased to report that this driver-Radeon Software 16.7.1-is now undergoing final testing and will be released to the public in the next 48 hours.
In this driver we've implemented a change to address power distribution on the Radeon RX 480 - this change will lower current drawn from the PCIe bus.
Separately, we've also included an option to reduce total power with minimal performance impact. Users will find this as the "compatibility" UI toggle in the Global Settings menu of Radeon Settings. This toggle is "off" by default.
Finally, we've implemented a collection of performance improvements for the Polaris architecture that yield performance uplifts in popular game titles of up to 3%. These optimizations are designed to improve the performance of the Radeon RX 480, and should substantially offset the performance impact for users who choose to activate the "compatibility" toggle.
AMD is committed to delivering high quality and high performance products, and we'll continue to provide users with more control over their product's performance and efficiency. We appreciate all the feedback so far, and we'll continue to bring further performance and performance/W optimizations to the Radeon RX 480.
-
Unfortunately, AMD's performance doesn't surprise me much... that card is very capped and with only one 6-pin connector, it's normal for a card with a TDP of 150W to have problems because the consumption, due to efficiency issues, depends on the temperature.The VRM the hotter they get, the more energy they need to power the GPU and GDDR5... this causes, among other things, the input current to be higher.
It also doesn't help that the silicon gets close to 80-90º... the higher the temperature, the higher the leakage current in the chip and therefore the greater demand for the VRM.
Unfortunately, the only trick manufacturers use is to lower the TDP and therefore the card performs in a jerky way because it lowers the Mhz to reduce average consumption... unfortunately it's a shoddy solution and makes each card perform differently depending on the ASIC and ambient temperature.
The only solution is usually to edit the BIOS, unlock the TDP and put in better cooling than the one that comes with the card, but if they keep patching the drivers and putting limitations through software, I don't know.
Also, I don't understand how the card is going to choose through software where to get or not the energy that powers the VRM... that's what happens when they don't put an 8-pin connector to cap the card at 150W and limit the performance of that beast just so that the high-end silicons perform much better when this mid-range silicon can do much more.
And mind you, nvidia is not safe either because the mid-to-high range of their 10 series could also be much better but they have limited them... even the custom models with more power connectors go up the same way unless they are Founders Edition to not harm the high-end Pascal that haven't come out yet.
-
In any case, it seems that there is an xfx or a sapphire (I can't remember now) that has an 8-pin connector, which, as with the 6+8 ones that many assemblers mounted an 8+8. I think that these same manufacturers possibly saw that the ones with the 6-pin pulled more from the pcie, which is the one that messes up the board.
But I have also seen that with powerbank the oc is greater in proportion than the 1080, and that the reference ones are mounting 8GB capped at 4GB. That, I think, is a great success for those who got one these days
I don't think it was a mess up by AMD. Let's remember the 970 and its 3.5+0.5GB, or something that happened with the 1080 that I can't remember now. Moreover, many of those whose board blew up had 3 mining like crazy, and although I read that one had had three 280s before, from what I saw the connector of the 280s only pulled 35w from the pci, and these were pulling 225w. No big deal. But I don't excuse AMD. This was a failure and they should have done all the possible crap to the cards before releasing them to the market, but this is something that is happening too much to manufacturers. And it's not new.
-
I had no idea that Polaris had already been screwed up by mining; I thought that Toms Hardware, with their review, had found the flaw when measuring power consumption with specific equipment, since they are the only ones who perform this type of measurement.
Anyway, if this really screws up the cards, I think it's a bit disastrous and I don't understand how they could have made such a mess... you see, it would have cost them nothing to put in a bigger connector and avoid this kind of problem... in any case, I suppose that the next revisions will bring it, as well as the custom models, but well... they are talking about an RX 485 instead of a 480 REV 2.
It is also true that the production process is new and they will be refining it, but that is no excuse.
-
They have already posted results of the new driver:
-
You can see that drop. As long as it doesn't affect performance, we're doing well
