AMD Crimson 16.7.1 driver puts Radeon RX 480 PCIe power draw in check
As a result of this rejiggering, the power draw in the 6-pin PCIe connector goes over spec. That shouldn't be much of an issue—it's better equipped to handle the added power draw than a motherboard's PCIe slot—but if that's still a concern, there's an optional "compatibility" toggle in Crimson's Global Settings that brings down the total power consumption of the card. Enabling compatibility mode means everything is under spec. There's a performance hit for doing so, though AMD claims it's "minimal" and offset by performance tweaks introduced by Crimson 16.7.1, which bumps performance in popular titles by up to 3 percent. Note that the toggle is turned off by default.
http://www.pcgamer.com/amd-crimson-1671-driver-puts-radeon-rx-480-pcie-power-draw-in-check/
>it wouldn't hurt performance they said
>AMD doesn't gimp their cards they said
>AMD will die in your lifetime
Was it ever explained how a non-compliant card was certified?
>>344310187
>buying AMD ever
>>344310536
Was it ever certified?
I smell a class action lawsuit.
Isn't this why people buy non-garbage power supplies? So if some hardware ends up performing out of spec (design flaw or malfunction) there is no issue because the PSU is built to cover it
POO
IN
>>344310723
Yeah, it's the hidden cost of AMD cards, along with requiring much more CPU cycles and needing enough power on the 12v rail in order to power two 480s.
LOO
>Have to underpower your card to run it safely
Nvidia killer.
>>344310692
yes it was certified.
if it wasn't it wouldn't need to apply to any PEG standards, all they would have to do is remove "PCI-Express" label from packaging.
>>344310187
DELETE THIS
>>344310723
if your motherboard draws too much power for what it's rated, no PSU will cover it. it will fry itself. same with PCIe lanes.
It's not "buy 1200W Platinum PSU, be set for life". that's not how it works. that's not how anything works.
you can't just fill up oil in a car. you have to check your lower control arm, raqu below boots, inner tie rods, radiator, axle, outer tie rods, bushings, and so on. if your drive axle falls off or is otherwise broken, you literally will not go anywhere.
>>344310187
Just wait for 1060 vs 480 benchmarks
the perform actually went up overall after the update though
>>344311172
Reminds me of the cucks that go online and lie about jelqing working.
>>344310187
But it doesnt hurt performance
>>344310187
>it wouldn't hurt performance they said
>AMD doesn't gimp their cards they said
http://www.pcper.com/reviews/Graphics-Cards/AMD-Radeon-RX-480-Power-Consumption-Concerns-Fixed-1671-Driver/Performance-Va
http://www.tomshardware.com/reviews/amd-radeon-rx-480-polaris-power-fix,4668.html
http://www.anandtech.com/show/10477/amd-posts-radeon-1671-drivers-fixes-power-consumption-issues
ur an moron
>>344310886
This.
People still regurgitate the same "AMD is cheap" argument, when in reality you need better equipment to compensate. It is like that Mr. Plinkett joke with the Kodak Printer, you are not saving shit.
You would need at least a silver standard PSU for AMD when any cheap built using a low end Nvidia budget card runs perfectly fine with a cheap bronze standard.
In addition, better cooling too because AMD shitty architecture is not energy efficient.
>>344310459
They won't die, they'll go bankrupt and sell off a lot of divisions/sell the whole company.
I know it's probably very unlikely, but it'd be funny if Nvidia bought AMD's processor department and Intel bought AMD's Radeon department so that now Nvidia and Intel were directly competing in GPUs and CPUs.
>>344311370
>>344311393
>clocked lower to compensate
>it didn't hurt performance because the driver had some optimizations which improved performance
Fantastic mental gymnastics from the AMD fanboys.
>>344311393
>AMD really delivered on a driver update that overrides the BIOS defaults to relieve the motherboard slot without a performance loss
this is concurrences in all of your reviews.
>>344312074
>Clocked lower
Except it isn't.
my R9 380 hasnt been working since last august
I kept thinking they would fix it with the next driver update so I held onto it....
....but it never came.....
IT NEVER CAME
>>344310459
You're a fucking retard if you geniunely want AMD to get out of the competition. A NVIDIA monopoly will ensure you'll pay premium for shit-tier cards since they'll have no reason to keep innovating
>>344312295
>Finally, when it comes to power consumption, measurements at the wall back up our earlier findings. In standard mode, power consumption at the wall bobbles by a couple of watts compared to the 16.6.2 drivers. With compatibility mode on, we see power consumption drop by 18W under Crysis 3, and 13W under FurMark. The average GPU clockspeed in both cases is similarly reduced, with Crysis 3 shaving off around 50MHz.
>hurr I can't read because I'm a retarded AMD drone.
>>344310187
>>344312074
Fuck off shill. Compatibility mode is not the default mode. With the new driver the default mode actually clocks slightly higher despite drawing 10W less than with the old driver.
AMD fixed the power issue and improved boost clocks by 10Mhz or so.
>>344310723
Are you retarded?
The issue is not that AMD requires a larger or better PSU is that the fucking card is drawing more power from the motherboard than it should, and the shitty card burns the lane in the process.
I.E even if you had a 1200W supply the damn card is still a piece of shit because it is drawing power from a shitty 6-pin connector and the PC-E slot. It should had been 8-pin.
>>344312613
>A NVIDIA monopoly will ensure you'll pay premium for shit-tier cards since they'll have no reason to keep innovating
You're half right.
They could and would charge a premium, but they WOULD still have reason to innovate.
>>344312807
Because there's such a huge leap from 2xxx to 6xxx series Intel CPUs
>>344312678
>50Mhz
Is it lower clock or is it thermal throttling?
>>344312735
proof that clock rates are actually improved despite the 10W reduction in PCI slot power draw
>optional "compatibility" toggle
>Note that the toggle is turned off by default
Nvidiots are literally retarded.
>>344312032
Fed would hit them with anti-trust hard.
>>344312735
As the article says, in default mode you're still out of spec, you're just drawing power from another source.
If you enable compatability mode, aka stay within spec, you lose 3-5% performance.
The driver itself has optimizations that increase performance by the same amount, so if you want to stay in spec, you're still losing clockspeed independently of the optimizations that would create a net zero change.
It's typical smoke and mirrors with AMD.
>June
>"This baby can compete with the 970, and overcloacked it can compete with the 980, the 980ti EVEN under bullshit circumstances! Look at Ashes of Singularity"
>Today
>"AMD releases a driver so its ORC-ish architecture does not burn down motherboards"
Oh red team, you never fail to disappoint.
>>344312996
>The within-spec mode isn't enabled by default.
Can AMD do anything right? Well, I guess they can trick their customers.
>>344310187
>As a result of this rejiggering, the power draw in the 6-pin PCIe connector goes over spec. That shouldn't be much of an issue
Wow it's literally nothing.
>>344312613
>A NVIDIA monopoly will ensure you'll pay premium for shit-tier cards since they'll have no reason to keep innovating
You mean like right now because AMD cannot compete at all?
Under a different market, Nvidia newer cards would be awful. A $799+ card that cannot do 4k or 1440p on DX12 games without a severe loss in performance? LOL.
Current Nvidia cards are not suitable for DX12, but AMD is so bad that even a Nvidia card would outperform it ion its own territory.
Nvidia is sitting pretty on many technologies, the GTX 1060 was simply an after thought it case AMD would released a non shit card for once.
That future you have described has been going on for a while now.
So your options are
>A performance increase from the previous driver, but you burn alive
>A performance decrease from the previous driver, and you get the privilege of not being flambé'd
And they try to hide this by advertising half and half, so they mention the napalm free gaming and the performance increase without clarifying that they are mutually exclusive.
Wow.
>>344313275
literally nobody with any technical knowledge cares about how much you draw from a 6 pin. Even high end PSUs don't have true 8 pin connector these days. It's all 6+2pins where the +2 are just dangling off the same ground leads as the 6 pin.
>>344313664
>I defend shoddy business practices because I'm a fanboy
Truly pathetic.
>>344314240
>I have no idea how 6-pin PCIe power works
Good goy.
>>344314174
>literally nobody cares if your motherboard burns
>literally nobody cares if your PSU burns
AMD drones, ladies and gents.
>>344311318
it does work jew cuck shill look my picture of my horribly disfigured penis
t. /v/ poster
>>344314442
>I have no idea what I'm talking about
kill yourself Rajesh. You will never get a real tech job
>>344314543
Thank you Pajeet.
₹25 has been deposited to WCCFTech Indian Bank इंडियन बैंक #138702940
>>344312116
I was quoting OP's retarded babbling and demonstrating that he was wrong.
>>344310187
>>it wouldn't hurt performance they said
>>AMD doesn't gimp their cards they said
But they are right.
http://www.pcworld.com/article/3093406/components-graphics/tested-amds-new-driver-fixes-the-radeon-rx-480s-power-draw-and-boosts-performance.html
The thing about AMD stressing the price/perf of the rx480 is that obviously Nvidia is going to release a card that reasonably matches it. However AMD's marketing is probably a resounding success.
The 1060 will probably have the performance of a rx480/970/980 for ~$250. This is just ass-pull speculation though.
>>344314745
From AMD:
>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.
How much of a fanboy do you have to be to ignore something negative the company acknowledges themselves?
>>344314442
>2-3 reports from random nobodies online of motherboards breaking
>no single group has been able reproduce any of the reported cases.
Nvidia shills, ladies and gents.
>>344313275
>driver optimizations - that were announced in the press release and specifically noted as being able to offset the slight performance decrease imposed by compatibility mode - are smoke and mirrors
>>344315025
Yeah, AMD acknowledged and attempted to fix this issue because of trolls :^)
My god you kids are retarded.
Mfw I still have to wait for the 1060 to decide whether the 480 is for me or not
>>344314745
>5.5% performance increase
>just from the first driver update
Nvidiots on suicide watch. Custom RX 480 with good clocks will compete with a 1070 at this rate.
>>344314745
So the safe mode doesn't decrease more performance than what is gained with the driver update? Nice to hear that if it ends up being true obviously. However, I'm happier with the fact that said "fix" is optional since it will not gimp the aftermarket models
>>344315021
Well the end result is that the card performs better than the benchmarks showed at launch.
The issue was unfortunate but I'd say they resolved it just fine.
>>344315097
They have the mode off by default in a sad attempt to keep benchmarks high.
>>344315025
Two reports I've seen were both by idiots, one with a disgusting computer with shit all over the motherboard and the other was some retarded buttcoin miner.
>>344315303
The end result is no net gain in performance unless you want to operate out of spec at stock speeds.
>>344314174
For something like the RX 480 shouldn't they had designed it with two 6-pin connectors like other power consuming GPUS if it is true that 8-pin does nothing at all? and I only follow with this because I need to read more about power supplies because I am reasonably skeptic about your statement.
What is is funny about this is that I brought energy consumption as point back when it was announced and I was only called Nvidiadrone for asking. So I take /g/ is worthless outside phone threads and basic language threads. It is pretty much highschool entry computing back there.
>>344315465
I think you've misread the graphs; the top bar is the no-fix launch performance while the bottom bar is compatibility mode post-fix. So the performance was improved.
How can a Software, Driver fix a design flaw like energy consumption? Even if it only tells the clock to run a bit slower isn't the flaw still there?
>>344315308
They have compatibility off because offloading excess power onto the PCI-E lead where it belongs solves the issue. Compatibility mode exists as a security blanket for paranoid worrywarts who need that extra placebo to keep their anus from prolapsing and isn't actually necessary other than to stop people from bitching about something that isn't an issue any longer.
>>344315792
Lowering the clock decreases the power consumption.
>>344310187
Even with the fix the power draw on the PCIe slot is off spec. So forget about overcloking. As a result the card is also warmer and loses about four frames average on games.
>>344316106
What I am getting at is that the card has an inherent design flaw.
From the article:
"That thrusts the 6-pin power connector over its theoretical 75 watt rating, but that shouldn’t be an issue; the connector can withstand the higher power"
>Withstand.
So is overclocking out of the question for now? Will after market cards fix this design flaw, and the shitty stock fan?
>>344314083
Actually, the options are free performance increase and securing the safety of your PCI-E slot, or maintaining the same performance from before the update and lowering overall power draw.
Even with the garbage reference cooler with a tiny heatsink, the housefire meme was never going to make a comeback with the RX 480.
>>344314442
PSU will not burn from drawing a lot from the 6 pin. Each pin is rated for like 8-9A at 12V. That's like 300W. The only way something could go wrong is if you're using one of those molex to 6 pin things.
>>344316641
You talk of it. as if you had won anything by saying that your little AMD shit card was any better for under-performing bellow anyone's expectations; without it being a safety hazard.
Not only did at release, was its performance while overclocking was disappointing, now overclocking is out of the question entirely. So you effectively got a GTX 970 that can run DX12 games without a penalization. Congratulations, but realize that this card cannot be paired with other budget components anymore, and runs even hotter. Well played AMD.
Can someone competent please join the GPU market? I honestly hope Samsung or HTC do.
>>344312613
AMD CAN'T win against Intel or Nvidia though. They keep shooting themselves in the foot. I want competition, but that doesn't mean I'll support an underdog if they can't support themselves. AMD had better hope there are a lot of poorfags considering they've virtually abandoned the high-end market. Their CPU's are absolute trash compared to Intel's, and only the most optimistic /g/tard thinks that Zen will be worth waiting for.
>>344317479
>but realize that this card cannot be paired with other budget components anymore, and runs even hotter.
I think this is a good point, the 480 isn't really a "budget" card considering you have to piar it with a non-shit PSU/mobo. Compare to the GTX 950, fuck they just came out with one that runs completely on the PCI-e. There are a lot of cards that you can just throw into a shitty OEM computer and not worry about it, this doesn't seem like one of them. I'd say you have to wait for the 470 for a "real" budget card.
>>344317479
This is a fix for anyone who was stupid/impatient enough to get the reference card. None of it was ever going to be an issue for custom designs with 8-pin connectors. Since the review embargo lifted all signs have been pointing to the 480's weak OC capability being due primarily to being limited to a 6-pin, and talk has been constant about partner boards easily hitting 1400MHz.
>realize that this card cannot be paired with other budget components anymore
But it can. Anyone with a shitty PSU will probably want compatibility mode enable at all times, but those PSUs are going to be a constant liability even if they were using integrated graphics because their poor quality is the most pressing concern in those cases. Everything else will be 100% a-okay.
Nice. Now you can't even OC your card one bit otherwise the issue will return since it will draw more power.
Sasuga, AMD.
>please literally fell for the "w-wait f-for the n-next AMD's s-shit!" again
lmao
>>344314747
1060 already confirmed for 300-350$
>>344320848
The 1050 will be the card that competes with the rx480 then.
Pretty sad for AMD.
>>344320917
1060 is theorized to have specs similar to the 480 though
>>344320848
Maybe the reference card, pretty much every rumor has it at $250.
>>344321264
>reference card
Founders Edition, anon.
>1070 and 1080 are limited to two-way SLI unless you unlock it manually
>1060 may not even support SLI because Nvidia admits budget SLI is a waste of money
AMD shilling Crossfire harder than ever
Well this is strange. ATI only developed Crossfire as a reaction to SLI, now Nvidia seems to be abandoning SLI. Seems they probably saw DX12 multi-GPU scaling and didn't think it was much of an improvement to warrant investing in.
>>344321264
>rumor
Nvidia confirmed launch price at $300 reference, $250 custom. But we're using our brains and observing recent history here. Custom 1070 and 1080 cards have all launched at FE prices, not the lower MSRP they were supposedly going to have. And that's on sites like Newegg. Price gouging everywhere else like eBay and resales on Amazon have taken prices even higher, and they've even hit the 480 to a lesser degree. 1060 FE will hit $350-$400 within 48 hours, and there is absolutely no way partner boards actually launch below $300.
>>344321714
Nvidia isn't abandoning it, they are just focusing on optimizing the most common setups
Very few people have more than two graphics cards, and the scaling and optimization beyond that are garbage.
Plus, with how fast gpus are these days they can start to bottleneck even high end cpus
As for the 1060, yeah no SLI is shitty, and the only reason why they did it is to not cannibalize 1080 or 1070 sales since two 1060 would perform on par with them for less money
Why is Nvidia so stingy with VRAM? It wouldn't be a big deal but it seems like all new games have bloated textures or something, I have to turn textures to low just to prevent stuttering/pop-in on some games despite running everything else on high. 1060 only has 6GB compared to the 480 8GB and even the 1080 only has 8GB as well.
>any mobo about 40 bucks and a bad quality power phase gets burned by AMD's power draw
>any decent motherboards 70+ bucks dont
>this is now a meme of AMD KILLS MOBOS LOL
niggers on this forum are so fucking retarded.
>>344322897
Because 192 bit bus can't do 8GB of memory unless you use asymmetric config. Nvidia has gotten shit for that before so I guess they aren't looking to try again.
>AMD fixes the 480'scritical issue AND gives it a performance boost in a single update
>People still insist that AMD is an unredeemable piece of shit
This board is undeniably infested by Nvidia shills and 3dfx ex-employees.
>>344322228
That's how it's always been. Few cards are actually MSRP, most cost more with better cooling. Even the AIB partner rx 480s so far cost more than $200/$249. And there is a $400 gtx 1070 (which is the historic price of the x70 cards before the 970) on Newegg ($20 over MSRP), and it's not even a blower cooler. The cheapest 1080 I could find is $620 ($20 over MSRP). If the 1060 is $20 over MSRP minimum, that still puts it close to the rx 480 in pricing. The price of the sapphire rx 480 (250£) converts to $330, which subtracting the VAT and overpricing euros have to deal with, should put it at around $260-280 US AKA on par with the 1060s with similar coolers.
>>344322897
The 1070 has 8gb vram. They need the 1060 to have lower specs as to not cannibalize sales from retards who think there's no reason to buy a 1070 if they have the same vram. Also 6gb is enough for 1080p, this cards target audience, so it's probably a cost cutting decision too.
>>344322956
>advertise it as a budget card
>be surprised when people use it in shitty toasters
I actually put my 960 in an OEM for my first "gaming" PC, could barely fit it in but it worked. Yeah the CPU and PSU were garbage but it let me play Skyrim with mods.
>>344311121
any non shit motherboard would shut down before their was damage
>>344321470
What's this founder's edition scheme?
>>344312032
>Nvidia bought AMD's processor department
There's no point, the x86 license is not transferable and x86's days are numbered anyway.
Nvidia can't into CPUs anyway, which is why their shit gets wrecked by Qualcomm and Apple.
>>344323285
More like
>AMD provides two options for faulty hardware: Gimp performance or use the card with a less severe out of spec problem
>Do what's expected out of drivers normally
LOL
>AMD KEKS TRYING TO DAMAGE CONTROL WHEN LITERAL CARD BURNS YOUR FUCKING COMPUTER UP
LOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOL
>>344324728
But doesn't Intel have to engage in good faith negotiations about re-licensing the x86 licence if AMD is sold? I'm sure Intel would agree to letting someone buy it to avoid the FTC breaking Intel up.
this is literally only for reference cards
why are people pretending like it's an issue for the consumer?
>>344312613
>an NVIDIA monopoly
this wouldn't happen, someone in the US would buy AMD and make sure they get back on their feet, or fund some other GPU business. Having monopolies over here is illegal
Also, why do you faggots want AMD to die? Less competition means you don't need to try as hard. They'll release shittier GPUs for higher and higher prices.
inb4 poorfag
>>344325015
because /v/ is full of shitposters and retards who get their news from those shitposters, anon
>>344325003
Probably, that's why they REALLY don't want Microsoft to buy AMD. If there's even a chance that the feds would force Intel to license x86 to Microsoft things would get pretty crazy in the desktop market.
>>344324215
Yea you're not supposed to rely on that. That's a really bad idea.
>>344320306
>review just came out with a 480 overclocked to 1500 MHz
>>344325993
delete this please
>>344325529
>Microsoft to buy AMD
Why would they want to buy AMD?
I think Samsung, Apple, or Qualcomm would make more sense.
>no else realizes but me that this generation is a money trap
>bith both sides will push DX12, and Volta will outperform Vega in async computing
>the 480 is already too weak for future proofing and 1000 series is obsolete in one year
Gg guys.
>>344326568
It's well known that Microsoft has shown interest in buying AMD before, Intel has basically bribed them before to keep them away. The obvious benefit would be they could manufacture their own chips for Xbox and OEMs, which would reduce costs. A more cynical answer is that they could optimize the chips for newer versions of Windows, which would encourage people to upgrade. However they don't need their own chips to do this apparently, it's already confirmed that new Intel chips will no longer support Windows 7 or 8.1.
>>344327157
>manufacture their own chips for Xbox and OEMs
They haven't even made it clear how interested they are in the future of Xbox, and AMD divested itself of all its manufacturing capabilities years ago.
As to the forced upgrade shit, yeah I hate what Intel has done with their processors. If it weren't for the compatibility issues I'd gladly see x86 die.
