http://videocardz.com/61783/amd-testing-driver-fix-for-rx-480-pci-express-overcurrent-issue
>As you know, we continuously tune our GPUs in order to maximize their performance within their given power envelopes and the speed of the memory interface, which in this case is an unprecedented 8 Gbps for GDDR5. Recently, we identified select scenarios where the tuning of some RX 480 boards was not optimal. Fortunately, we can adjust the GPU’s tuning via software in order to resolve this issue. We are already testing a driver that implements a fix, and we will provide an update to the community on our progress on Tuesday (July 5, 2016).
>undervolt
>underclock
Enjoy your even slower poo in loo
>>55375874
it's going to need a hardware fix, and AMD should recall the card, and replace motherboards, for users who've had boards damaged by the card.
as the longer it goes the harder it will be to do anything.
>>55375874
>all those amd faggots beta testing for amd
Will these faggots ever learn? u never ever buy the reference design
>>55376410
>>55375874
nvidia shills
>>55376423
House fires
>>55375874
>"""""tuning"""""
>wait until reviews are in
>lower performance with driver to make sure people don't fry their mobo's
AMD is just as bad as Nvidia
>>55376423
better being nvidia shill, then amd cuck taking it in the ass are you enjoy the crispy smell of your pcie slot?
>>55376471
>this
>amd will underclock and undervolt there cards. >There is nothing else they can do.
>This might have been the reason they actually inluded voltage regulation in the actual drivers.
>Where were you when AMD cucked 1000's of loyal cucks?
>>55376530
Who are you quoting, retard?
>>55375874
How many times have i already wrote this but il have to write it again:
BAIT AND FUCKING SWITCH
>>55376440
>being a newfag
>>55376682
wut
>>55376423
>AMD testing driver
>nvidia shill
You are worst than nvidiot. Get over it, Polaris is a big fail.
>buying mid/mid-high performance GPUs
>using trash tier motherboard that can't handle a little bit of overvoltage
>"b-b-but /g/ said I shouldn't worry about fight end mommy boards bc they're unnecessary :^("
>>55376931
>be a poorfag
>buy a card FOR poorfags
>impying poorfags use +$600 momboard
>>55376931
It's not overvoltage. It's overcurrent. If it was overvoltage it wouldn't be caused by the graphics card.
>>55376999
can modern games even use 2 cpus? they have trouble using all the cores on single cpu..
>>55376931
>using trash tier motherboard that can't handle a little bit of overvoltage
Are you stupid? You think people use high end motherboard for a $199 card?
>>55377085
It's overkill for only gamers, you will get like 5fps. It's more like for render, compilation ect.
>>55376254
this...
https://www.youtube.com/watch?v=eVTXPUF4Oz4
>"Drawing too much over the PCIe bus is totally going to blow boards!!1"
>People show the 750 Ti and 960 doing the same thing, except rather than just an average overdraw, there's also peaks up to 250W but they still aren't blowing boards under normal use
>"Hey, we're now going to ignore and the fact that bad OCs also blow board by using an example of someone badly overclocking a 480 and blowing his motherboard as a result"
I see summer is in full swing again...
>>55377860
Anything is a bad overclock when you go above 3% and suddenly you're drawing 100w from the pci-e
>>55377860
>People show the 750 Ti and 960 doing the same thing
What?
>>55377860
>drawing
>microsecond power spikes the same as sustained over the rating power draw
>Those cards literally draw less power on average
>Thinking it's power spikes that destroy hardware not heat
FUCKING 6-9A OF CURRENT RUNNING THROUGH YOUR SLOT IS GOING TO MELT IT NOT MICROFUCKINGSECOND VOLTAGE SPIKES
>>55377860
Citation pls?
>>55377860
>drawing too much
Like OP says, the correct term for this is OVERCURRENT.
Here is how you calculate power.
Voltage multiplied by the Current gives you the wattage.
Voltage spikes are common because not all the power you get is clean.
The problem is when your card is drawing 6-9A over the 5.5A average.
Current is what destroys hardware. Voltage spikes are literally nothing compared to the heat that will be generated by the current running through the slot.
You can harp on and on about how there are voltage spikes in Nvidia cards. The issue is still and always has been overcurrent.
>>55377860
>card are shit
>poor oc capable
>amd fail pci limit spec
>blame nvidia
>>55377924
the rated 5.5A for the PCI-E slot*
>>55377871
We're talking about someone who forced his card to 1350 MHz, which is more than 7% over the standard boost clock and 21% over the base clock. May not be something you'd expect to get out of a high end CPU with a good cooler, but we're talking about a reference card here.
>>55376832
Why is it sold out everywhere?
>>55377978
Because it's dirt cheap and new
>>55377978
AMD overhype
http://seekingalpha.com/article/3985508-amds-polaris-revealed-overhyped-disaster
>>55377978
They talk about this '14nm' trash card since January
>>55377880
>>55377893
Picture related
>>55377883
>>55377924
>Implying it's just a few peaks and not consistently drawing over 75W
>>55377860
>ASUS Strix
ASUS Strix
>ASUS Strix
You try so hard...
>>55378116
>strix
replicate this same issue with the other 960s including the reference please
protip: you can't
>Get the performance of a midrange card from 2 years ago
>For the same price of said midrange card
How does the RX480 even make sense?
>>55378116
Asus Strix
>>55378141
>>55378116
Microsecond spikes
You see the blue line there that says average and how that average is below 75 watts?
Well for the 480 that average is 80+
>>55377759
https://www.youtube.com/watch?v=R9HKxMgekCo
>>55378116
>some asus custom card = amd reference card
Nigger stop
>>55378147
>>55378141
>Implying it matters when the Strix was nether fixed or recalled, but still didn't blow any boards under normal circumstances
Even if the problem is limited to the Strix model, it proves is that you can draw too much on the PCIe bus and still not blow boards because of something called tolerances.
Sorry to break this to you summerfags, but you didn't get your "gotcha"
>>55378013
> And the power consumption under typical gaming loads didn't show anywhere near the factor of 2.8x improvement in efficiency AMD has been claiming.
Why are they lying? AMD never claimed 2.8x power efficiency for the 480. Pic related.
>>55378182
Asus is Nvidia?
>>55376624
haha, wtf. Kid, quit being a faggot, I know that is likely impossible for you, but really, try.
>they are highlighting truths
>not a game was mentioned
>AMD is marketed for poor children, so they can play games..
>You are a poor faggot and while one day you can change the poor status perhaps, the faggot status you will carry to the grave with you.
>>55378172
>Non-reference that consistently drew too much, but still didn't blow any boards boards
Sorry to break this to you, but it being a reference board or not is irrelevant. If it didn't blow boards, which it didn't, then nether will the reference 480 before they lower the voltages (which third parties have successfully done without affecting performance)
>>55378160
The microsecond spikes are to 225W and beyond, it however consistently draws over 75W
>RX 480 will get gimped before Nvidia will start to gimp GTX 9XX
All this shitposting aside, would it still be reasonable to upgrade from R9 390 8GB to R9 480 8GB AFTERMARKET? Like pick up an MSi or Sapphire when this shit gets worked out?
>>55378199
Asus makes graphics cards and with the Strix 960 they accidentally proved that you can draw more than the PCIe spec allows and still not break motherboards.
>>55378272
>would it still be reasonable to upgrade from
are you stupid?? the 390 costs more than 480 for a reason.
>>55378272
Why?????????????? Your 390 is fatser...
>>55378264
See >>55378116
>Other card had an even worse problem with spikes going as high as 250W over PCIe
>Still didn't blow boards
Stop trying to act like the RX 480 is going to blow boards in default configuration. If the Strix 960 didn't blow boards, nether is the RX 480.
>>55378310
my board costed 120 jewro mate.. I'm not gonna risk it family.
>>55378310
read this...
>>55378160
>>55378310
It won't blow up a motherboard but it will do damage. It's the heat generated from being constantly over the 75w limit which will do some serious damage. It can fry circuitry like the on board sound before the failsafe kicks and the motherboard shuts itself off.
>>55378310
continuously drawing over the spec is what causes problems you idiot, not spikes
>buy GTX970 performance 2 years after the 970
>Now you cant even reach 970 performance due to AMD gimping the card so it wont fry your mobo
Another
Massive
Dissappointment
>>55378372
more accurately it's the heat generated by the constant draw over the spec.
The spikes are caused by voltage spikes which are normal as fuck. The constant power draw is caused by the current.
The current is what produces most of the heat as it runs through the slot and weakens the solder/fries other components nearby because the current is more than the thing can handle. It's like plugging 5 toasters into one wall socket. Your wiring will catch fire but your socket should remain undamaged until the flames get it.
CUSTOMERS CHOICE
>>55378449
wut?
>>55376624
>>>/neo_g/
>>55378449
>9000h of paint
Not bad
>>55378449
The card is actually so fucking ugly in general and whoever thought up this car crash design needs to be thrown off a cliff.
>>55378476
>he has that picture in his device
>>55378508
the guy who put a 6-pin on it should jump first family
>>55378508
>that card
>ugly
back to /v/ with your edgy founders edition eyesore
>>55378711
Not him but the 480 look so cheap
>>55378760
you have an incredibly poor eye for design, i would recommend you stay away from that career path.
>>55378711
>>55378795
Blind people need to leave.
![razer-espark-scooter-sparks[1].jpg](https://i.imgur.com/VPh6bxw.jpg "razer-espark-scooter-sparks[1].jpg")
>>55378844
can you children who have an interest in these dank gamer designs go back to /v/
>>55378711
The 480 is ugly...It look like nothing
>>55378876
>thinks a black rectangle looks better than a similarly basic but more out of the box design which is satisfying for the eyes
![16537110-Emoticon-showing-thumb-up-Stock-Vector-smiley-face-smile[1].jpg](https://i.imgur.com/Smrs8O6.jpg "16537110-Emoticon-showing-thumb-up-Stock-Vector-smiley-face-smile[1].jpg")
>>55378891
minimalism at its finest.
let me guess you like the cancer that was web2.0 skeuomorphism like pic related and thought windows xp through 7 actually looked visually appealing? fuck outta here.
>>55378932
>sent from my edgy razer keyboard
>AutisticMotherboardDestroyers couldn't contain their autism.
>Muh OC 480 will be able to reach 980 Ti performance
>Meanwhile in reality AMD has to underclock it to don't fry poorfag PCI-E slots.
I fell for AMD meme once and thank god I remembered that lesson.
>>55378953
>listen /g/ and not your common sense
Amd releases a card with shitty ppwer management
>Releases driver to fix it
Nvidea releases a faulty card with 3.5 gb of ram
>Its a feature
>>55379079
>fix it
*gimp the card
at least the gtx 1060 launches next week, and its faster than a 480
>>55377860
> I know nothing about electronics
Mobos are made for spikes. Spikes in electronics mean NOTHING. The power specification is for average/constant draw it's not built for that which is why plenty of people are having problems with it. Seriously don't talk about electronics if you don't understand them.
>>55379079
>fix it
Yes so now it will only be 20% under the performance of a 970 instead of only 10
>>55379110
And then nvidia will stop releasing useful drivers for it after 5 months. And then, the RX 480 will slowly get more and more optimized, and 2 years after its release, it will be 20%+ faster than the 1060.
>>55379202
>10%
>>55379271
>5 shitty dx12 shovelware games included
wew lad.
>>55379261
And by then nobody will care as there will be new cards out, and the cycle of nvidia always being faster will continue.
>>55379289
>cuckvidia damage control
i r8 3.5/4
>>55377924
>Current is what destroys hardware
It can be but voltage can too. Too much voltage will blow MOSFET gates apart and permanently damage them. Considering GPUs are full of 14nm FETs you probably do not want to exceed their voltage rating.
Also, power is not as straightforward as P = VI. In reality there are all sorts of impedances that complicate power draw and while that only holds true for AC you have to figure there are square wave oscillators generating clock signals and other digital logic signals. This is either AC or pulsed DC and calculating power by P = VI because of phase shift introduced by capacitors and inductors which ideally dissipate no power.
>>55379289
http://www.gamespot.com/articles/amd-radeon-rx-480-review/1100-6441354/#1080p_Benchmarks
Would you like me to kwwp going until you find benchmarks you agree with?
To recap:
1. Every single review site praises the RX480 for best price/performance card on the market
2. Few review sites point out that the Rx480 draws more than standardized through the pci-e slot.
3. Nvidiots still butthurt about their 3.5gb cripple card try to force a powergate meme
4. Despite what nvidiots say nothing happens. Not a single mobo explodes
5. AMD releases a patch which undervolts the card slightly so the RX480 squeezes through the 75watt pcie spec
6. All is well. Nvidiots on suicide watch because there is no software patch to fix their 970 christopher reeves edition
MAXIMUS KEKKUS. NVIDIOTS BLOWN THE FUCK OUT.
>>55379331
>too much voltage
Yeah the chances of that happening are slim to none.
>Not as simple as P= VI
>PCI-E slot specifications are 75 w
>12 v x 5.5A = 66w
>3.3v x 3A = 10w
>= 76 w
>not as simple as that
lol no
>>55379373
this needs to be switched now, it's AMD blowing up and nvidia BSODing
>>55379390
Maybe these guys mught be able to fix the 3.5 issue
http://www.downloadmoreram.com/download.html
>nvidia
top laugh
>>55379331
No you can't reliably calculate card power draw by P = VI because it has so many capacitive and inductive elements on it which have a reactance dependent on frequency. You have oscillators outputting different frequencies to different parts of the circuit, the duty cycle might be varying depending on the task being performed, etc. This will all affect total power consumption.
Listen if you're confused look up apparent power and reactive power.
>>55379461
meant for
>>55379395
Come watch this guy burn his house down
buildzoid (twitch)
>>55379390
>Slower than the old generation (390)
>Noisy
>Card runs into temperature and power limit, resulting in reduced clocks
>Limited GPU overclocking potential
>Power draw exceeds PCI-Express specification
>High temperatures
>Multi-monitor and Blu-ray power draw still terrible
>DVI output removed
>Poor Crossfire driver support
>Low performance per watt even for a 14nm technology
http://www.anandtech.com/show/10446/the-amd-radeon-rx-480-preview
https://www.techpowerup.com/reviews/AMD/RX_480/
http://www.tomshardware.com/reviews/amd-radeon-rx-480-polaris-10,4616-11.html
It's the classic
>it's all nvidias fault
>tip kok nvidiots btfo by amd fucking up
Denial and asshurt at its finest.
Why can't the card just draw more from the 12V supply directly? Are the third parties going to use two 12V connectors?
>>55379261
>And then nvidia will stop releasing useful drivers for it after 5 months.
What? My 960 has much better performance today than when it was released. My 560 Ti also still works with current drivers. Nvidia isn't Apple.
>>55379610
>Noisy
>Card runs into temperature and power limit, resulting in reduced clocks
>Limited GPU overclocking potential
>High temperatures
why are you talking about pascal now?
>>55379390
actually ppl already posted about pci-e slots dieing. Most reviewer sites are literally retard status, or paid shill status. And AMD wont undervolt there cards, why? cuz it might make them unstable on these clocks, what they will do is alter power target by something like -10% and move some of the W to the 6pin slot. But guess what? reducing power target reduces performance as well. AS example my card gives me 117fps in furmark in 100% power target, if i reduce power target by 10 % it instatly drops to 108%. So its almost 1 to 1 actual performance to actual power target %.
>>55379722
>Limited GPU overclocking potential
>pascal
wew lad
>>55379752
>nobody can hit 2ghz clock other than reviewers with 'press' drivers
>fan issues
>pixel clock issue on dvi
wew lad paxwell is amayyzin
>>55379783
so what if nobody can hit 2k ghz, ocing by over 200mhz is already good enough. 480 cant even oc 50mhz in most cases.
>>55379806
>most people hitting 1350mhz
>"cant even oc 50mhz"
you retarded shills need to at least learn a thing or two
>>55379610
390 is $300 though lol
>>55379783
>1.5Ghz to 2Ghz
>+500Mhz of OC
>nobody can hit 2ghz clock
kek
>>55379826
who are these people show me.
>>55379783
>1500 to 1900 mhz in avegae
>Limited GPU overclocking potential
Stop your stupid damage control Roy.
>>55379783
>10/15% oc on shrunk 2014 28nm optimized architecture
Vs
>4% oc on architecture "optimized for 14nm" as claimed by amd
>4% is somehow better than 15%.
>10% performance gains on poolaris oc vs 15% performance gains on based pascal
Stay mad son, stay mad.
>>55379839
you made the first claim with burden of proof on you yet you expect others to cite their claims.
pathetic.
https://www.guru3d.com/articles-pages/amd-radeon-r9-rx-480-8gb-review,35.html
there's a 1375 for you
>>55379833
>hurdur base clocks
lol 1080 can barely hit 1800. rx 480 can get 250mhz overclocks just like it, poor argument
>>55379870
>literally admits paxwell is nothing more than an overpriced die shrink
kek, get cucked by your beloved some more
>>55376999
this is the first time hardware has given me an actual erection.
Thank you.
>>55379783
>nobody can hit 2ghz clock other than reviewers with 'press' drivers
[Citation]
>>55379867
>1500 to 1900 mhz in avegae
>>55379911
>prove to me something that doesn't exist doesn't exist
?? do you want me to prove to you god doesn't exist too?
>>55379900
you fucking shill, you fucking fucking shill you fucking contradicting yourself claiming only nvidia card reviewers are getting good overclock and then fucking posting a fucking reviewer for your proofs, fucking dirty shill exposed.
>>55379900
>a revision of the most efficient architecture ever is a bad thing
Amdrone logic everyone. Everyone tale a good look and laugh at this guy.
>>55379783
>nobody can hit 2ghz clock other than reviewers with 'press' drivers
BUUUUUUULSHIT
>>55379934
Take*
>>55379927
>aib cards will come with those clocks stock
>>55379934
yeah they are pretty amazingly efficient spaceheaters, got that 100C on the vram, thanks titan x
even the 1080 is now an efficient housefire, how nvidia can get so much heat output out of such a 'low' power draw is incredible
>>55376254
>undervolt
>underclock
you think this is an nvidia card?
7950 was perfectly fine on the same clocks while undervolted from 1.25 to 1.1
paid shills are on overdrive 2 days now i guess nvidia has too much money to burn on the to shill instead of creating a card that can actually do async
>>55379922
I still wait your cittation anon
>nobody can hit 2ghz clock >>55379783
>>55376423
Well, it's possible he is a Nvidia shill, since /g/ is basically a playground for Rajesh and Pajeet to flood AMD/Nvidia threads, but he's also right; NEVER, EVER, EVER, EVER by a reference board for, or from ANY card or device.
>>55376832
>asrock motherboard
>not knowning that pcie errors are code 99
>literally a case without fans
>nvidia paid shills dont have any grasp of technology
>>55379783
https://youtu.be/ng-pmHV-RMM
>literally the first video on youtube "gtx 1080 overclock" that isn't a reviewer
>>55379972
show me somebody who has using release drivers
>>55375874
>"""Fix"""
Pajeet is on damage control
>>55379989
>the only person in the world to somehow hit 2.1ghz and have the gpu running at 60C
kek, yeah thermal throttling was something everyone else just made up
>>55379965
You know that async is marketing bullshit right?
Just like 8gb ram, 390/x never ever and never will even more then 5gb ram.
>>55379965
Hiring a poo in the loo shill costs one 100th of what it costs to hire an electrical engineer
>>55379992
>shitposting and unsourced claim
['''''''nobody can hit 2ghz clock''''''''] Where is the link anon?
>>55380043
it's okay when nvidia shills make claims without citation right?
>>55379732
It's literally just one guy who overclocked his card to 1350 Mhz (7% higher than the default boost clock and 21% than the standard clock) and a motherboard failure is something that can actually happen when overclocks go bad. As for the undervolt, third party sites have already found that you can do that without having to lower the clocks so the situation is a bit like the original Nahalem chips (which could be overclocked and undervolted at the same time).
However I guess it's just the usual summer circlejerk in action again...
>>55379732
Doesn't matter since the benchmarks are already out :^)
>all the amdfags ITT who blames nvidia
>>55380020
sure thing faggot he was only one in the world, keep making shit up it will make u feel better about your crappy rx 480 purchase
>>55380084
you're so fucking mad rn
>>55380020
>this retard doesn't know what a custom fan curve is
Kekkkk
https://youtu.be/ZgLPwbiXtDw
This guy gets above 2ghz with a stock fan curve
question
out of all the people who reported broken motherboard so far, did any of them report problems with intel boards? I've seen like 3 threads and they all used either Phenom 2 or an FX-8350.
>>55380064
undervolting your card might make it unstable, @make it crash under load, even if its like 5% of the market, amd wont risk it, they will move the power target down and it will effect actual card performance, like it or not.
>>55380072
>N/v/idiots go on about how it's going to fail and perform worse than a 970
>Actually performs pretty much how people expected it, between a 970 and a 980 on average and above or tied with a 980 in DX12
>N/v/idiots try to defect this by using the power draw over PCIe issue, but get BTFO'd by facts when people point out that there's been loads of cards that have done this in the past, including Nvidia ones, but none of them have blown boards under normal circumstances
>"all the amdfags ITT who blames nvidia"
Yup... Summer is definitely in full swing kids.
>>55375874
>select scenarios
AAAAAAAAAAAHAHAHAHAHAHAHAHAHAHAHAHHAHAHA
>>55380132
This
>>55380116
Im might be slighty drunk and mad, but you are much more mad than me, you stupid amdrone
>>55376832
What in the name of fuck did that paste eating retard do to his case?
>>55380021
>async is a lie
>390 never goes up of 8gb
4chan the place that you learn something new from paid shills
>>55380152
>It's not an issue but AMD are trying to fix this thing that isn't an issue
kys pajeet
>>55380145
As I said, third parties have already discovered that it's completely stable when undervolted and other cards, including ones from Nvidia, have proven that drawing too much over the PCIe bus is not an issue.
Yes, I know it's annoying when you think you have a "gotcha" on your hands, but then find out that you don't.
>>55380152
And people who have a 390 are not disappointed? Why blame nvidia is Polaris is DOA?
>>55380152
>that pathetic damage control
minimum wage shill
>>55380197
>Not being an issue as proven by cards doing exactly the same thing in the past
>AMD still doing driver tweaks to shut people up
Not exactly that good from a PR perspective, but knowing murrican laws I have a feeling the reason why they're doing the driver tweaks is to avoid a class action lawsuit rather than anything technical.
>>55380152
>N/v/idiots go on about how it's going to fail and perform worse than a 970
>Actually performs pretty much how people expected it, between a 970 and a 980 on average and above or tied with a 980 in DX12
on average 480 perform 10% below 980
>N/v/idiots try to defect this by using the power draw over PCIe issue, but get BTFO'd by facts when people point out that there's been loads of cards that have done this in the past, including Nvidia ones, but none of them have blown boards under normal circumstances
>"all the amdfags ITT who blames nvidia"
expect not, 950/960/750ti, might have the power peaks, but constant draw is well below the limit of 66W. The specification is for a reason and if your card is constant drawing 80W and you OC it to draw 100W, your pcie will fucking die.
>>55380185
He's right. No game as of yet uses more than 4/5gb vram on 8gb video cards other than mirrors edge catalyst which has a specific setting which was made for the 8gb vram in the 1070 and 1080. You can even check the report they gave out conceding hyper settings and it was literally made to tax 8gb 1070/80.
Devs won't suddenly start making games which need more than 4gb at default settings because there is way more of a user base of people who have 4gb vram and under than there is with 8gb vram. If you check steam hardware survey all the cards with 6-8gb vram are like 1% of all users.
>>55380202
>other cards, including ones from Nvidia, have proven that drawing too much over the PCIe bus is not an issue.
None of those use as much overcurrent for as long as the rx480. They only use more current sporadically (if very often), while the rx480 takes more current CONSTANTLY. The average power draw is higher on average.
>>55380202
> including ones from Nvidia, have proven that drawing too much over the PCIe bus is not an issue.
Why do you keep making shit up, nvidia cards never ever fucking reacjed or when over specified 66W pcie constant draw. while stock RX 480 is have constant 80W...
>>55380253
yes steam survey that the most popular amd card is 360........
literally its broken they know it and they havent fixed it..
also som is another game that reaches 7gb.. just saying
>>55380152
>Actually performs pretty much how people expected it, between a 970 and a 980
100% agree but the shit overclocking, the shit power consumption, the pcie issue, the temperature. Polaris is literally a 28nm card...
>>55380293
>also som is another game that reaches 7gb.. just saying
Show me fucking proof of RX480 or 390 ever fucking reaching 7gb of ram, these cards dont even have the processing power to reach that.
>>55380322
Not him but 8GB of RAM is useful for Crossfire
This is such a fucking stupid thing, if they included a 8-pin connector instead then the rx480 would have been rock stable AND reach way higher clocks very easily.
The GPU itself is okay but this is such a fucking big design failure.
>>55379289
People like you need to cut off your penis, and shove it up your ass as you drink bleach.
>>55380322
>this cards dont even have the power to reach that
THE FUCK
>day1
>4chan starts to vomit random things
>memory needs precessing power to fill it up
>i think i accidentally switched on the iq vaccum machine
>>55380208
You do realize that the RX 480 uses a 232mm^2 chip? The 390 used a chip almost twice the size with 500 million more transistors, so the RX 480 wasn't going to beat it no matter what AMD did.
>>55380209
>Pointing out facts
>"Damage control"¨
Oh well... Fanboys will be fanboys...
>>55380249
>on average 480 perform 10% below 980
At what point did I say that RX 480 beats the 980? In DX12 it's better or on par, but on average the 980 is faster, but also about twice the price.
>WAH WAH!!!!1 LIES!!!1
Except the Strix 960 was consistently drawing well over 75W with peaks to 250W if that didn't blow boards, then nether will the RX 480.
>>55380343
>The GPU itself is okay
nope
>overwatch was lagging at 1440 144hz
>i want to buy new pc when vegas and zen comes, see if they are good
i bought this card untill than, and got lucky on gpu lottery
it can run stock speeds at very low voltage (1.15v stock)
shall i return it, wait for aftermarket cooler?
wont get as good asic again for sure.
>>55380352
they do you fucking shill, same reason 960 with 4gb of ram have literally same performance as 960 with 2gb ram.
>>55380367
because surely the die size on a node shrink with new tech boasting the chip is a perfect way to compare them...
>>55380257
>Drawing power like >>55378116
>Only "sporadically" drawing over 75W
If that doesn't kill boards, then nether will the 480...
>>55380287
See >>55378116
>>55380371
performance is good enough and only costs $230. It's only opposition is a card that literally takes all its users as idiots.
>>55380389
please stop...dont post anymore
>>55380343
then they couldnt sell it to oems, it was calculated move my amd, overclock the card on release for favorable reviews. Then release driver fix, which fucks its performance. They need the 6pin for oem's.
>>55380352
>I am a colossal and uneducated retard
http://www.techspot.com/review/1114-vram-comparison-test/page5.html
>As for the 4GB R9 290 vs. 8GB R9 390 comparison, there was nothing to see here. Most games didn’t consume more than 4GBs of VRAM, but there were cases such as in Assassin's Creed Syndicate where with almost 5GBs of memory usage, there was still not a single frame difference. Moreover, in this Assassin's Creed Syndicate example the 290 and 390 were good for just 26fps on average, so clearly the quality settings will need to be reduced for achieving playable performance and consequently memory use will drop.
>The 390 and 390X are really graphics cards we never wanted. At the time of their release the Radeon R9 290 and 290X were exceptional buys. The 290X cost just $330, while today the 390X costs around $100 more for no additional performance and it is no different with the 290 and 390.
>We see plenty of gamers claiming that the 390 and 390X are excellent buys due to their 8GB frame buffer ensuring that they are "future proofed," and well, that simply isn’t the case, as neither GPU has the horsepower to efficiently crunch that much data. Perhaps the only valid argument here is that the larger frame buffer could support Crossfire better, but we haven’t seen any concrete evidence of this yet.
>>55380343
>AND reach way higher clocks very easily
if that was the case I'm pretty sure that the card isn't a good overclocker
I think even the poo in the loo said something about the card being on the sweet spot of performance x watt x temperature. I guess that is the reason that price of the cards range from 100 to 300, pay 300 for 3 coolers and 10% extra performance and 350 watts card!
>>55380403
What is the average watt load on the slot? Random spikes aren't as likely to kill a board if the average is still within spec. I mean, they can, just not as likely to do as constant, non-stop overcurrent.
>>55375874
they didn't even test the cards or what lol
AMDead fucking sucks ass
>>55380422
because he's right and you want to live in denial?
>>55380437
and yet here we are
shadow of mordor uses 7+gb on 390
catalyst also uses 7+
but you know
>they dont have the horsepower for it
clearly what we saw its fake
because muh nvidia
>>55380434
In that case it was a well executed, but really, they should've released it undervolted so the card doesn't end up killing boards during normal function. Then they can just blame overclocking morons on killing their gear.
>>55376254
I think some German magazine has better performance after undervolting
>>55380448
>if that was the case I'm pretty sure that the card isn't a good overclocker
It would be if it had enough power. AIBs already reported 1600MHz overclocks, and they use 8-pin or dual 6-pin power.
>>55380479
do you actually read your graphs?
>>55380401
>because surely the die size on a node shrink with new tech boasting the chip is a perfect way to compare them...
A die shrink really does fuck all to performance on it's own. What actually improves performance is adding more hardware onto the die and running something at higher clocks. The 480 is not a successor to the 390. which is a considerably more expensive card, it's a successor to the 380 and it beats it. The real successor to the 390 is coming with vega which is going to be bigger chips about the same size as the GTX 1080 and 1070, both of which IIRC are based on a 320mm2 chip. That should give you an idea of how much more hardware is going to be on the silicon for the 390 successor coming later this year.
>>55380499
you're still falling for this shit after seeing what the card is capable of?
>>55380453
Just by looking at the graph you can see that while the spikes above 225W are sporadic, going above 75W is not sporadic, it's being done constantly meaning that the average draw is going to be above 75W.
>>55379390
Noice.
>all this shilling and damage control
kill yourselves AMD losers
>>55380473
Shadow of mordor runs perfectly fine at 4gb at the maximum setting. You are too retarded to understand what a cache is.
https://youtu.be/COEwgUI6D3w
And catalyst hyper settings was made exactly for the reason to use 8gb vram. Ultra settings doesn't use more than 3gb.
>Mirror's Edge™ Catalyst from EA DICE is now available, offering gamers hours of fun in a massive city tailor-built for fast, fluid movement in first-person. For this latest release, DICE has taken PC visual fidelity to a new level, taking full advantage of the power of our new GeForce GTX 1080 and 1070 graphics cards.
>Making full use of their blistering performance, and 8GB framebuffer, DICE is able to dramatically increase environmental detail and shadow quality on the new 10 Series graphics cards, and further improve the quality of reflections, visual effects, environmental maps, image based lighting, motion blur, and Resolution Scale’s downsampling. Classed as “Hyper” settings, these enhancements enable users to explore a richer, more detailed city on the very best PCs and graphics cards, delivering the definitive Mirror's Edge™ Catalyst experience.
Once again, you have just proven how much of a retard you are. If all games are using more than 4gb like you are claiming then how are all the 99% of people who have 4gb vram and less playing them? Off yourself.
>>55380529
>it's being done constantly meaning that the average draw is going to be above 75W.
I think you don't know the meaning of the word "constantly". It happens often, but it is not a constant over-75W power draw. It stays under 75W almost as often as it goes over it, meaning it is less taxing than the RX480 which does not spend a single second under 75W on average.
>>55380512
oh i know that but its natural to be right near or on par with 390
all the mid tier cards once "evolved" goes on par with the previous high end cards
when i asked robert about their little power gating patent is on 480 or it will come later with vega he said "you will see soon enough"
i guess that 480 was really a test lab for it thats why it has only 2 HWS
>>55380566
are you actually bringing a 970 into the discussion being used by a gameworks game?
that was the previous point in case you missed it
>nvidia kinda fucks up maxwell 970 memory
>mass refunds and class action lawsuits, reputation ruined for years
>AMD 480 blows up motherboards
>FUCK OFF NVIDIOTS AMD ROCKS DRIVERS WILL FIX IT JUST YOU WAIT SOME MORE IT'S A FEATURE SUCK UP I DON'T WANNA HEAR IT AAAAHHH
>>55380499
>reported
I also feel for the 1500mhz meme and was disappointed on 29, but that doesn't mean I can't accept that the card isn't that great. It isn't shit, but I think for the price that the AIB will come it will probably be a shit buy. Even thought you might get a bit more of performance it will probably be have less performance x price than a good AIB 970. I guess the 8gb and dx12 support are a good thing, but I'm not sure if I want to spend that kind of money on that level of performance.
We are just masturbating here, we have to wait to see if the power throttle was really a problem.
Can't someone just hack a 8 pin on a 480 to give more energy?
>>55375874
>>55380520
>you're still falling for this shit after seeing what the card is capable of?
I did not "fall" for any of that shit, I make my own opinion based on what I see on multiple reviews and technical breakdowns. And what I see is a card that is heavily gimped because it only uses a 6-pin power connector.
Simply put it cannot be overclocked at all as long as the power issues remain, so until we see AIB cards with 8-pin connector, we won't know exactly what the card is capable of.
>>55380667
switching pins isn't going to magically give the card a 300% oc increase, this is just impossible
use your head
>>55380584
>oh i know that but its natural to be right near or on par with 390
When you have half a billion less transistors, you really shouldn't expect 390 levels of performance unless there's some massive architectural improvements.
>>55380583
The graph spends more time above the 75W line than under it making it clear that we are in fact talking about an average power draw in excess of 75W. The fact that the line occasionally goes under the 75W mark does not get you out of trouble when you have an average above it and spikes up to 250W.
>>55380618
>960 Strix drew similar average wattages with spikes up to 250W, but still didn't blow motherboards
>Delusional N/v/idiots somehow think that this is going to blow motherboards
Fanboys everyone...
>>55380667
actually it can we have seen alot of people oc it already problem is obviously it has a lot of headroom that cant reach it with the 6 pin only...
it is literally 7xxx series all over again...
amd downclocked them so they can all have a nice headroom...
>>55380599
It's got nothing to do with Gameworks retard. We're not comparing performance between amd and nvidia here. We're looking at vram utilisation in general and it's clear from the video that even a card like the 970 which has had all that 3.5gb issue is running perfectly fine at the maximum settings and since it's not buckling under the decreased bandwidth of the last 0.5gb we can safely assume it's just using the last .5gb segment as a cache, which would be the same in the 390s case.
>>55380703
>Constrained by power draw and cooling
>Thinks that adding better cooling and changing a power connector from one that can provide 75W to 150W isn't going to change anything
>>55380703
The card throttles itself or crashes because of insufficient power draw.
If it had proper power draw then it would not do that and actual OC limits would be whatever the silicon can actually do.
That combined with a cooler that isn't a complete fucking joke will increase OC potential.
>>55380722
there is well 2 things mostly
1)the primitive discard acc on the HWS they literally created a conservative rasterazation chip that is programmable i bet that whole tessellation gimp on amd is finished once more people test the card on games that had heavy use of it
2) the HWS themselfs..literally they are improved ACE engines with "hyperthreading tech"
>>55380629
A guy is voltmodding onw on twitch right now. Buildozer
He was explaining the weird desing of the board then he soldered two things together i dunno
>>55380758
>>55380765
I don't doubt it will improve it, but if you're thinking that the 1600mhz FUD isn't setting you up for disappointment you'd be wrong
Set reasonable expectations
>>55380667
>Performance already being gimped with drivers
>Massively innefficient
>Can't be overclocked at all because of the power gate
Beyond that who knows what issues exist.
Adding an extra 2 pins worth of power is not going to undo what is clearly a colossal fuck up created by panic to rush a card out the door in the AMD camp.
>>55378272
It's as fast or slower than a 390
>>55380667
It struggles to even maintain stock clocks with the normal power target. You have to max out the power rather slider at even stock clocks but then that increases heat and thus makes the card sound like a leaf blower.
>>55380722
>The graph spends more time above the 75W line than under it making it clear that we are in fact talking about an average power draw in excess of 75W.
No.
>>55380744
so additional pp filters on the game that on nvidia pass on their dedicated physics chip has nothing to do with the impact of them on memory
right.......
on other news the sun comes from the north
>>55380839
Those are total draws so roughly about a half of the values are drawn from pci-e slot.
I've been an AMD/ATi guy since 2005 when I replaced my Geforce 4 Ti4600. 9600 PRO > x850 > x1950 PRO > 5770 > 270X > 380.
This fuckup has me leaning back towards Nvidia for my next card. I know the PCI-E slot power draw issue is probably getting overblown, but this kind of oversight is just disgusting and reeks of poor engineering or poor decision making in the name of marketing.
They wanted that 6 pin connector so badly instead of just modestly accepting the fact that Polaris 10 would need an 8 pin on the reference PCB. All of the AIBs will be 8 pin after this clusterfuck anyway (or they should be) so it all just backfired in their face.
Instead of talking about the good performance/price the 480 delivers at $199 we're debating a critical design flaw. I didn't give 3.5 a pass and I won't give this a pass either. RIP AMD.
>>55380722
>The graph spends more time above the 75W line than under it
No it fucking doesn't. Those are spikes. The measured average is slightly below 50W, while the RX480 is over 80W. The asus 960 strix isn't healthy with all those fucking spikes but it doesn't outright kill the card because the average is still well below the limit.
>>55380876
Also EVGA version has an 8 pin connector if anyone's wondering.
>>55380881
and here's the RX480, in comparison.
>>55380876
>Those are total draws so roughly about a half of the values are drawn from pci-e slot.
Then you are proving him right; if the card takes half of that 100W from the pcie slot then it only takes 50W on average.
>>55380853
Are you legit retarded or just trolling?
>gets blown the fuck out with hard facts and sources
>b-but muh nvidia conspiracy
If shadow or mordor doesn't use less than 7gb vram like you're saying then why does it work fine at max textures on my friends 290x oc?
You haven't been able to provide any proof of your claims yet I have. Me > you now neck yourself.
>>55380877
They got blindsided by Pascal and completely shit the bed with what is a clearly massively inferior architecture.
Inefficient garbage targetting a part of the market that doesn't know any better.
From my cursory glances around the place I've seen a few reports of PCI slots being burnt out by the things, it's fucking horrendous.
I'm so glad I bought my 1080 FE. I can keep it at 70c stable crushing 1850mhz stable at only 80% fans (which is inaudible behind my headphones).
Waiting for Polaris was a pretty retarded meme.
But back to your post, buy a 1070 or wait for the 1060, you won't regret it. AMD completely screwed the pooch with this one.
>>55380803
>Adding an extra 2 pins worth of power is not going to undo what is clearly a colossal fuck up created by panic to rush a card out the door in the AMD camp.
No, but it might finally give us a sub-$300 card with good performance.
>>55380796
>I don't doubt it will improve it, but if you're thinking that the 1600mhz FUD isn't setting you up for disappointment you'd be wrong
Hey, everyone said that Ivy Bridge can overclock to 4.6ghz easily, but the one I got shit itself at 4.3 unless I turn the voltage up to 1.4.
Overclocking is a lottery like that, high binned chips may still possibly hit 1600mhz, but obviously not all will.
People using the 480 to mine memecoins are already seeing their hardware fail after only a few days.
https://bitcointalk.org/index.php?topic=1433925.msg15438155#msg15438155
>>55380995
>I'm so glad I bought my 1080 FE.
>Waiting for Polaris was a pretty retarded meme.
You are glad you bought a $650 card because you were disappointed with the performance of the $230 one? How fucking stupid is that?
>>55380558
>>55381008
>No, but it might finally give us a sub-$300 card with good performance.
Unlikely. When the reference design is a complete and utter shambles, with the first set of post release drivers being a straight up gimp then you know the card is doomed. You can't fix an inherent issue with the architecture and design with another 2 pins worth of power.
Thing is horrendously thirsty for such an underwhelming card.
>>55380964
shalalala
https://www.youtube.com/watch?v=8i0docmTrGc
>>55380995
Yeah, I'm debating stepping up to a 1070 for 1080p/144 Hz soon. I did spy a very cheap ~$300 980 non-Ti on Newegg that has me intrigued as a holdover until HBM2 stuff shows up. I think the 1070 at $379 would be a great deal but they don't actually exist at that price yet. I think they're a little overpriced at $470.
>>55381018
IDE connector
IDE FUCKING CONNECTOR
yeah mine too will explode if i use such an old motherboard.......
>>55380877
its not an issue unless you bought a shit motherboard and psu
and if you did those things you only have yourself to blame when they fail under load
>>55381067
>290x
>no memory compression
>4gb vram
>ultra settings
>still runs perfectly fine
https://youtu.be/KSj2xIHlbTI
I'm having fun blowing you out. You literally have no argument left. You can't even refute the 2 videos I've provided showing max textures runs fine on 4gb vram and anything above is just a cache.
>>55381074
It's all relative, I paid nigh on £640 (about $900 or so) for my 1080 and it was worth every penny because I had that kind of money to spend on a card that'll stay at the top of benchmarks for quite a while.
I mean I'm not making an argument that the 480 and the 1080 are comparable, I'm simply saying that the 480 isn't worth the $300 that it costs where the 1080 this far has been worth every penny.
>>55381059
>You can't fix an inherent issue with the architecture and design with another 2 pins worth of power.
But the problem isn't the fucking architecture, but the fact that the reference design is a piece of shit. It had to be, for them to be able to sell the card to OEMs which need a 6-pin connector.
That 2 pins worth of power is a +75W which gives the card enough headroom so it can work properly. Same for the improved coolers.
I agree that the card is underwhelming, I expected it to use way less power. But the performance is decent for the price, and if AIBs can get it to run safe and actually OC (it currently cannot do that at all), then it will still be a good choice.
Though honestly I'm more interested in the 470 at this point; if it is 15% slower than it is still fast enough for me, while being way cheaper and three times faster than my current card.
>>55381101
>only have yourself to blame if your out of spec GPU blows up your PCI-E slot
Yeah, fuck no brah. If it's out of spec, it's out of spec. Can't ever blame the motherboard in that case for failing to do something it isn't designed to do.
I'm not going to defend AMD's shitty design just because I've owned AMD/ATi for the past 10 years. Wake the fuck up. Side note, my motherboard is not cheap as it's X99, but I'd still never use a reference 480. Why fucking chance it after knowing what we know about it?
>>55381085
IDE connectors are still on AM3 and AM3+ motherboards. The system was less than 5 years old.
>>55381085
>yeah mine too will explode if i use such an old motherboard.......
And he was using THREE CARDS, not one.
>>55381101
>What is a specification?
Are you serious? The specification is there for a reason, when something is marked as PCI-E and you plug it into any PCI-E slot you do so with the understanding that both parts have been certified to not blow up in your face.
>>55381141
>Why fucking chance it after knowing what we know about it?
AMD sold it to the low end of the market for a reason, they simply don't know any better most of the time.
>>55381162
nope the motherboard was old as shit
1 x PCIe 2.0 x16
1 x PCIe 2.0 x16 (x4 mode, 2.5 GT/s)
2 x PCIe 2.0 x1 (2.5 GT/s)
3 x PCI
he literally used it on a fucking pcie 2.0
>>55375874
so they are going to cripple performance in favor of reduced power draw and call it success?
duh
>>55381211
PCI-E is PCI-E. There's no fucking excuse for a card to blow up a slot it's designed to plug in to. Just stop posting.
lol AMD is like a backyard mechanic
over spec but she can handle it
>>55381211
pcie 2.0 was the standard up until around 2013, kike. nothing old about it nor would it be killed by newer cards.
the blames lies solely on AMD for designing a product that damages motherboards.
>>55381226
yeah, pcie is pcie
>>55381211
is retarded, the issue is not that he used pcie 2.0, rather that he used a riser to put a x16 card on a x4 slot
>>55376254
>or microcode update pushed to make it draw more from 6-pin
Seriously, these chips probably aren't hardwired into specific phases except for the memory chips which draw less than 60W.
>>55381226
it wasn't the pcie that blew but the atx24 connector. And he was using three cards.
>>55381264
>it wasn't the pcie that blew but the atx24 connector.
where exactly do you think the current that goes through the pcie slot comes from?
Reminder that the only way the RX 480 blows Nvidia away is if you install it on an old nForce motherboard.
What if the 1060 is a rebranded 970, with some overclock from factory and 6gb? I bet nvidia could do it.
why did AMD do this retarded 6 pin thing?
Like whats so wrong with giving it extra and proper power. Even poorfag psus should have enough connectors
>>55381336
They'll do it and sell it for $350 ($400 for founders edition).
And there is nothing you can do about it because they have monopoly on the performance market.
>>55381336
Nah, the new die was already leaked/confirmed. Besides, the 970 chip has been confirmed out of production for almost a year like all maxwell silicon.
>>55381336
Bloody unlikely and downright not economical. It's an even lower binned gp104.
My crystal ball says 10-15% above RX 480 and roughly same price just to make the butthurt even worse.
>>55381368
>why did AMD do this retarded 6 pin thing?Like whats so wrong with giving it extra and proper power. Even poorfag psus should have enough connectors
OEM sales.
>>55381370
>>55381376
they can sell their last 970 stock as 1060, then release 1060Ti to counter Vega.
