AMD promises code fix for power-hungry Radeon RX 480 GPU AMD says it will soon release a software update for its Radeon RX 480 graphics card to stop it slurping unexpectedly large amounts of power. The world's "other" purveyor of x86 and GPU chips says it will distribute the update later this week, but in the meantime it does not believe any motherboards are at risk for damage from …
Brand new hardware so it's bound to need a little initial tweaking. All I can say is... I WANT ONE! Been waiting for the shrink to ~14Nm from 28Nm for too long. Sadly, Nvidia has been batshit crazy on pricing for years now so thankfully AMD seems to be in a position to take advantage of Jen-Hsun Huang's greed with their new RX 480. I believe this gutsy move will be great for AMD.
Go AMD go!
Yeah didn't mean to start a holy war and considering I have maybe bought 15 or 20 graphics cards total in my somewhat old lifetime it's hardly much of representative sample. I just know I have had at least two or three ATI/AMD (that merger went well) cards go bad including an 1900XTX (that was branded ATI, so not 3rd party maker's fault) that immediately comes to mind. I have never had an NVIDIA go bad and have bought from some pretty no name Taiwan companies with NVIDIA chipsets and been lucky I guess.
Never seen a dead Nvidia card? Well for certain some brands/rebadgers/manufactures do make mistakes. One was giving instant swaps/replacements as one particular model fried it's power line after 20 months. :P
It was a 570, and they replaced it with a 660 for me. :)
"never seen a dead Nvidia card."
I've had several. Including a GTX 280 which was overheating like crazy.. On inspection I discovered the heatspreader on the chip (not the heatsink) was badly fitted. After much careful work to cut through the silicone holding it in place I managed to remove it, apply some decent thermal grease and stick it back together. Temps went from 110C forcing thermal shutdown to about 65C under load.
You may want to hold off for the AIB cards that have more power pins, better cooling, and might be able to compete with the GTX980 when OC'd.
Of course, if you can wait until early next year, Vega will be along to *really* compete with Nvidia.
Oh, and 'high end' of the market? It's a mainstream card, and deliberately so.
With both xbox and PS using AMD chips more software will be optimised for AMD rather than nvidia in the future, meaning nvidia's iron grip on the market will slip. They are getting by on brute force power at the moment, whereas AMD seem to be taking a more measured approach.
AMD haven't won the crown yet, but they are putting in the required training :)
I had a fun Nvidia card once. Think it was an MSI one, bought from Scan (or Dabs - I can't remember), but for some reason it was deemed ok for it to have the "Pre-release firmware - not for resale" stamped in its BIOS. End result: No full screen acceleration. Software or full-screen window was fine. Outright full screen - it just died and crashed the machine.
I had a Gigabyte Nvidia GT9800 1Gb silent with a massive heatsink that failed twice. Once under one year and I returned it to Gigabyte Service and it was repaired and was OK for another six months then it failed again in the same way. Digital noise and ghosting. The card was used in a huge Lian Li Server case which has 4 case fans and was not overheating. Not even a gamer other than casual stuff like Bejeweled 3, Luxor and Mahjong. It was replaced by Gigabyte with an equivalent Radeon card. Gigabyte have 3 year warranties. My other Gigabyte GT9800 1Gb Silent is chugging along in a Dell T3400 Core 2 Quad and has been no problem. I have had GPU's from both the Red and Green teams fail and have no loyalty to either side apart from I would never buy an Intel CPU new. Only used when they have dropped in price to less than $100 which is why I have AMD CPU's in my newer machines I build. When I am given a good Core 2 Duo machine (T3400) I upgrade it to Quad core, SSD, USB3 for not much cash. I am happy to be several generations behind as on a fixed income on disability I can't buy new and shinny very often but can get 90% of the performance with my old bangers and my 2 x FX 8350 and FX 8320. It is possible to be really Geeky and Nerdy on a budget and I always buy and support shareware authors such as D Opus, XYPlorer, Xplorer2, Bitsum Process Lasso, Goldwave, Terracopy and WinAmp 5. With Windows 10 sucking the big one I am still on 7 and I am using Linux Mint so that is where I'll be headed. It'll be cheaper and less hassle than the Microsoft route.
Before you go all bonkers about this video card you might want to look at the real competition, which is the GeForce GTX 970. You can see the performance comparison at the PassMark webpage. the GTX 970 has a score of 8658 and the Radeon RX 480 has a score of 7686. The GTX 970 costs $40 more, and for my money the lousy $40 buys you a significant difference in performance for the whole time you use the card. The GTX 970 has a great reputation and has stood the test of time with gamers. The Radeon RX 480 lacks performance, and has not yet endured the gamer tests. Look at the PassMark web page to see the results of their tests: http://www.videocardbenchmark.net/high_end_gpus.html
Also see their base videocard benchmarks page at: http://www.videocardbenchmark.net/
Although in games the 480 tends to win out and with DX12 I suspect that the 480 will be the better longer term card:
http://www.pcgamer.com/radeon-rx-480-review/
Personally I am waiting for either the 480 with a dual fan and an 8 pin power plug or the 1060.
It's using the latest Vulkan, OpenGL and DX12 libraries where the AMD card really shines because AMD have introduced asynchronous compute on the graphics chip.
You'll find a reasonable overview here, I have no doubt that nvidia will eventually catch up so AMD have only a small window to prise the market open.
NV employee??
Plenty of tests on games have been done, the RX lagged slightly on games that are known to be optimised for the NV cards, but trounced the 970 in nearly all others, even beating AMDs own 390 in some.
The tests were done with a stock RX480 against O/C variants of a range of cards.
I would post a link, but cant be arsed.
I really like AMD hardware, but had to jump to NVidia, because working around poorly implemented soft PCIe reset (a necessity when doing PCIe passthrough) on newer AMD cards was driving me crazy. NVidia hardware was expensive like hell and I would really like AMD to be viable competition, also in "my" market segment. I will be waiting for newer generation of Polaris-based FirePro cards with VDI support, perhaps this will do the trick.
So you want to go back to the old argumente eh?
It was a 32bit card that saved space space and power by dithering to 16 bit and then upscaling.. they claimed 22 bit and was almost 24bit quality.
That said, the quality difference on "normal" quality CRTs of the time was very similar between 16 bit and 24bit. Yoo needed an Eizo or a Trinitron to see the difference.
No software patching crap can fix a bad hardware design.
The RX 480 is not PCI-Express compliant !
It isn't a safe product to use either because it can destroy the motherboard.
So a software bug in any new driver release and the fix couldn't work either... so safe.. uh?
AMD must not be allowed to get away with this !
They should replace all the defective cards. Re-design the RX 480 and stop selling the defective designs. Period.
Well I've got mine and I couldn't be happier.
Actually I could, HDMI audio on Linux needs the not-yet-mainlined DAL drivers, which I haven't had any luck with, but that's not a massive deal.
I'm not overly concerned about this issue. Hell, if my motherboard did go pop, they'd probably be doing me a favour as it's 6 years old. I have heard that post 1.0 versions of PCI-Express should allow more power and even my board is 2.0. This is further backed up by the reported failure having been from a cheap 2008 Abit motherboard.
Of course software can fix it. The power limits for all of the internal supply rails inside the card are controlled by firmware and the drivers on existing cards; there's no reason to think the 480 is any different.
One of my PCs has a Nvidia Kepler GPU, and it resisted my attempts to overclock it using tools like MSI Afterburner. I increased the overall TDP limit with Afterburner, but it was still scaling back its boost clock under heavy loads, with the power-limit (TDP) bit set.
Once I edited the card's firmware in Kepler BIOS editor or whatever it was called, I could see that I could adjust the power limits for each of the card's three power sources (each its own internal power rail)-- the PCIE slot, the 6-pin power connector, and the other 6-pin power connector.
By tweaking the individual power limits that controlled the draw from the 6-pin connectors and increasing the range of the overall TDP target (the max it can be set to in Afterburner) from 110% to 120%, I was then able to maintain about one step higher in boost clock under heavy loads than I had ever seen before under any amount of load (and never under heavy load; it would always throttle back about 2 steps from its maximum when under heavy load), and by making a slight tweak to the boost clock in Afterburner, I got it to go one more step up in boost clock under load, for a total of some 4 steps under heavy load.
All of that was done by altering the BIOS.
One thing I could not do is increase the voltage to allow even higher overclocks. The peak overvoltage on my card was all of 10mv-- trivially tiny, and because it was set in the driver, there was no way to use the BIOS tweaker to raise the limit. Not effectively, anyway; the BIOS has its own internal limit, but no matter what was done with it, the driver would prevent the overvoltage from exceeding 10mv.
BIOS would seem to be the proper place for AMD to fix the 480, but there is really no way for me to know whether it uses a system like my Kepler card or something quite different. Anything the BIOS can do can also be done by the driver, and the enforcement of power limits most certainly is done in BIOS in some (perhaps all) modern GPUs.
Now, it is possible that there is some engineering defect in the power control hardware on the card, causing it to disregard power limitations set by the driver or firmware, but that has not been established. Given that some media outlets have reported performance increases by undervolting the 480, it is quite possible that this is nothing more than a driver or firmware defect, and that the card will emerge from all of this as a solid product.
Or it could fall flat on its face.
We won't know until we see what happens. I'm inclined to believe it's the former; this just has the feel of a firmware bug, and while the driver fix could be a useful Band-Aid to prevent any more motherboards from burning up, it will probably take a firmware revision to fix it permanently. I don't really know, though; it's possible that all of this is functionality built into the driver too.
I think that firmware of a modern GPU is usually loaded by the driver at the moment of OS startup, and anything that the GPU can do before then is only dumb VGA BIOS (or UEFI) support. Meaning that yes, an updated driver can fix a lot of things.
So rebelling against the status quo by purchasing from the status quo... It hardly qualifies as a Rebellion when they are number two in their market with number three a mere speck compared to them in market share. I'd expect a 'rebellion' to come form one of the little guys like Matrox or VIA.
I love how both sides are so blinded by brand loyalty that they don't notice that they stink just as bad as the other side. I've used cards from both ATI and nVidia, and probably an equal number of Intel GPUs with a smattering of Matrox cards and some other smaller manufacturers.
Given their performance and failure rates, they all seem to suck just as hard as the others. They all seem to love driver packages weighing in 200+ MB, and have pointless selection criteria to download them (why do I have to find mine in a list of hundreds when I'll just end up with the same driver no matter what I select?). There also the fact that they both try to install to C:\AMD or C:\nvidia (Seriously guys, either use the Program Files folder or the temp folder like proper software). The both install 'update' services that don't actually update the drivers. Both bundle stupid, useless programs with the drivers (that are quite annoying to try to deselect since neither company uses a standard UI for their installer and tries to make their own but makes it unusable in the process). Even when the driver package is uninstalled, it still leaves nice little surprises behind that will completely screw things up to where a fresh OS install is needed.
A few months ago, I upgraded my monitors to a 2x3 grid, so I swapped out the nVidia 560Ti for a Radeon HD 9700 series card since it had 6x mDP connectors where the nvidia card only had a pair of DVI connectors and a mini-HDMI port (which couldn't be used simultaneously with both DVI ports). Spent a week trying to get the new card to work with my system but random cruft left over from the nVidia card kept messing with the AMD drivers. There was a slot the nice surprise that by deleting the c:\nvidia folder on my system, Windows stopped booting. Plus I would constantly get messages about nvidia services failing to start because they didn't detect nvidia hardware (they were still hanging around even after I removed all of nvidia's stuff through Programs and Features). Eventually, I found it easier to just perform a fresh install of Windows and start over from scratch.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
