AMD Radeon RX 480 8 GB

[Bluescreendeath]

With 1060 coming, nVidia has effectively killed the hope of AMD's Polaris GPUs.

The 1060 likely won't be close to the price range of Polaris. (I've read that it will probably cost ~$300). Nvidia is still keeping its lower tiered Maxwells. The Polaris would still be perfectly competitive at the $200-$250 range unless the 1060 is somehow priced at $250 or lower.

 

[ikeke]

Before the launch date....buzz is:

"Insane overclocking!"
"390X performance for $200 at a fraction of the power draw!!!"
"It'll be a Crossfire monster!!"

AMD launches 480...

"Terrible overclocking"
"Sub 390 performance"
"Crossfire doesn't work with really anything."

Moral of the story is...whenever AMD pimps a performance increase based on power draw, you know when it finally releases it's going to disappoint. They did it with Fury and they did it with Nano. With 1060 coming, nVidia has effectively killed the hope of AMD's Polaris GPUs.

Moral of the story - dont buy into hype from Videocardz/wccftech et al.

AMD said to expect R9 290/GTX970 level of performance for $200. Delivered.

 

[john_]

@john_ go ahead and ignore the fact that 290 series was still new and current then. They did that to compete. Nvidia did this with old stock because it is competing with its own overpriced product. Not even close to the same situation. If it were, 1080 and 1070 prices would have been slashed.

Don't act like I'm the AMD hater here. Read the thread. I'm one of the only ones that hasn't said how much the 480 sucks compared to the expectations they had for it.

I'm out of this conversation. It's extremely hard to have a logical conversation with extreme belief people and not worth getting worked up over.

290 series was one year old when Maxwell came out. And yes they did that to compete. I wrote that, I guess you had to wrote it too. On the other hand you keep talking about Nvidia like they don't ave any competition. Like they are alone in the market. You make it look like if Maxwell cards where no where to be found, people would be spending at least $450 for a 1070. They would completely ignore everything AMD had in the market. Nvidia's only competition is Nvidia itself.

That's your "logic" here. You are not an AMD hater. You are an Nvidia worshiper, at least in this case. And thank you for restraining yourself for saying how much 480 sucks

I was thinking to say "I stop here" in my previous post, but I thought you would say that I try to escape from the conversation. And yes, it is extremely hard to have a conversation with you. Because you are trying to convince me that, that photo in your avatar showing a Roman soldier using his smartphone, is an authentic 2000 years old photo.

 

[Assimilator]

The 1060 likely won't be close to the price range of Polaris. (I've read that it will probably cost ~$300).

$250 for 3GB and $300 for 6GB is what I've heard. Personally I feel NVIDIA could shave another $20 - $30 off those prices and still come away with a nice profit, but they're obviously worried about cannibalizing their existing Maxwell stocks.

 

[BiggieShady]

AMD said to expect R9 290/GTX970 level of performance for $200. Delivered.

With some DX12 sugar on top

 

[john_]

Those announcements came after they found out what they had.

They had a 232mm2 chip. I am pretty sure they had enough experience in the market to know what to expect from a 232mm2 chip against a 314mm2 chip.

Moral of the story - dont buy into hype from Videocardz/wccftech et al.

AMD said to expect R9 290/GTX970 level of performance for $200. Delivered.

I guess all the hype about insane overclocking and 390X performance could happen with custom cards. But not at $200, except if Nvidia decides to be very aggressive with 1060's pricing.

AMD wasn't only trying to give R9 290/GTX 970 performance, but also to restrict what someone could archive with the reference card. That's where the 6pin connector, instead of an 8pin, comes in the equation. RX 480 looks to have been developed as a low cost to manufacture card. AMD could probably sell it at a 4GB R7 370's price and still make a profit out of it. Other than the cost of Polaris GPU over Pitcairn GPU, I can't really find anything else to make the RX480 board more expensive than a R7 370 board.

$250 for 3GB and $300 for 6GB is what I've heard. Personally I feel NVIDIA could shave another $20 - $30 off those prices and still come away with a nice profit, but they're obviously worried about cannibalizing their existing Maxwell stocks.

Nvidia's problem is that GTX 1070 and GTX 1080 prices are still too high. A GTX 1060 with 6GB and a price at $250 is very much possible and much more logical, because of RX480, but then the gap between GTX 1060 and GTX 1070 would be, in reality, $200. GTX 980 Ti can fill temporarily that gap with a price at $350-$400. If GTX 1060 is slower than a good GTX 980, maybe GTX 980 can also become a good option at $300. But Nvidia will need to lower the prices of GTX 1070 and GTX 1080 soon. The rumors where saying that GTX 1060 was scheduled for August, which would have gave NVidia more time to bring prices of it's GP104 cards closer to their original MSRP. Now they will have to move faster with their plans.

With some DX12 sugar on top

The sugar is necessary only when comparing with an overclocked GTX 970.

 

[BiggieShady]

The sugar is necessary only when comparing with an overclocked GTX 970.

Even if you compare 4th generation GCN to previous generations, there are now dual hardware schedulers active for async compute ... interestingly there were two of them since Hawaii but only one was active

 

[nexus_a]

How come no one mentioned that RX480 fails the PCI-E specification??
If you don't want your motherboard fried, don't buy this card.

 

[bug]

How come no one mentioned that RX480 fails the PCI-E specification??
If you don't want your motherboard fried, don't buy this card.

Apparently the card has passed compliance tests and going over 150W is unexpected behaviour (AMD is investigating). Still, something to keep in mind when making a buying decision.

 

[john_]

How come no one mentioned that RX480 fails the PCI-E specification??
If you don't want your motherboard fried, don't buy this card.

I think it was already mentioned somewhere in here.

It seems to affect only really old motherboards, like 10 years old motherboards. Newer motherboards, even the ultra cheap ones, don't seem to have any problem, giving more power through the pci bus.

Two more thing to consider.

Many graphics cards at their defaults are very close to the limits of what the motherboard can give them, for example the GTX 950 with NO extra power connector. Overclock it and you have gone over 75W. And all those watts will have to come from the pcie bus.

Second, ASUS, MSI, Gigabyte for example, not only sell RX 480 cards, but also ultra cheap $40 motherboards. You think they will have start selling RX 480 cards if there was any possibility the graphics cards to burn their own motherboards?

And one last thing. It doesn't kill the motherboard. The system shuts down, meaning you can't really play any games.

 

[64K]

How come no one mentioned that RX480 fails the PCI-E specification??
If you don't want your motherboard fried, don't buy this card.

I'm not sure how rigid the PCIe spec is. I remember the R9 295x2 had two 8 Pins and counting the slot power that is specced for 375 watts but according to the tests here it averaged 430 watts in gaming, peaked at 500 watts and running Furmark went all the way to 646 watts.

https://www.techpowerup.com/reviews/AMD/R9_295_X2/22.html

 

[nexus_a]

I think it was already mentioned somewhere in here.

It seems to affect only really old motherboards, like 10 years old motherboards. Newer motherboards, even the ultra cheap ones, don't seem to have any problem, giving more power through the pci bus.

Two more thing to consider.

Many graphics cards at their defaults are very close to the limits of what the motherboard can give them, for example the GTX 950 with NO extra power connector. Overclock it and you have gone over 75W. And all those watts will have to come from the pcie bus.

Second, ASUS, MSI, Gigabyte for example, not only sell RX 480 cards, but also ultra cheap $40 motherboards. You think they will have start selling RX 480 cards if there was any possibility the graphics cards to burn their own motherboards?

And one last thing. It doesn't kill the motherboard. The system shuts down, meaning you can't really play any games.

There are few reports of motherboard components dying right now.
https://community.amd.com/thread/202410
http://www.overclock.net/t/1604421/various-amd-rx-480-review-thread/1800_100#post_25309056
Doesn't prove that the RX 480 causes the damage, but all buyers should be aware of the potential risk.

I'm not sure how rigid the PCIe spec is. I remember the R9 295x2 had two 8 Pins and counting the slot power that is specced for 375 watts but according to the tests here it averaged 430 watts in gaming, peaked at 500 watts and running Furmark went all the way to 646 watts.

They could beef up the amperage of two 8-pins, but not one 6-pin. Just a guess.

 

[john_]

There are few reports of motherboard components dying right now.
https://community.amd.com/thread/202410
http://www.overclock.net/t/1604421/various-amd-rx-480-review-thread/1800_100#post_25309056
Doesn't prove that the RX 480 causes the damage, but all buyers should be aware of the potential risk.

There where negative and positive reviews of RX480 on amazon before the card become available. Internet is not a place to believe anything. The first link could be from a troll/nvidia fanboy, the second from someone who burned his audio card long ago and now found a chance to throw the blame on RX480 and demand an RMA.

We will have to wait and see. A strong indication their is a problem will come from the motherboard monufacturers. If they start posting on their motherboard pages importand notifications that RX 480 is incompatible with some motherboards, then yes. AMD F UP in a way that almost comes even with Nvidia's old bumpgate fiasco.

 

[bug]

I'm not sure how rigid the PCIe spec is. I remember the R9 295x2 had two 8 Pins and counting the slot power that is specced for 375 watts but according to the tests here it averaged 430 watts in gaming, peaked at 500 watts and running Furmark went all the way to 646 watts.

https://www.techpowerup.com/reviews/AMD/R9_295_X2/22.html

2 8pin connector is already outside the spec. Any video card configured like that is not PCI-SIG certified. At the same time, it was built from the start to work outside specs.
The 480 is, but that makes it even more dangerous, because it means it's working in a way it's not supposed to.

 

[KainXS]

How come no one mentioned that RX480 fails the PCI-E specification??
If you don't want your motherboard fried, don't buy this card.

I did a few pages ago

Sorry I missed this but I say there is no OC headroom on this reference board because at stock the card is already at or near it max TDP it seems. Not only is it at its max TDP but multiple reviewers had reported(If they had the equipment to test it) that their review cards and some of them have retail cards also, all of which go way above PCI-E specification which is 5.5A from the motherboards to about 6.5A of draw and in some games that gets as high as about 8A which is worrying for a cheap motherboard.(many people buying budget cards buy budget boards) An expensive board might handle that but a cheaper one could be ruined or have interference from that over time. You can slap whatever heatsink you want on it and overclock but its going to be out of spec and the more you OC this thing the farther it will be out of spec it seems. This could just mean a little interference in your speakers though(and thats probably all you will get) but I'm not taking that chance because I know I will OC if I get one.

With AIB's though this would not be the case more than likely as this seems like a PCB issue and I will probably buy an AIB card with a better heatsink and 8pin though to OC it.

edit: those dead board and component failure reports look fishy though, if something would fail it would take a while.

 

[64K]

They could beef up the amperage of two 8-pins, but not one 6-pin. Just a guess.

I'm way out of my depth speculating on power delivery capacities but the only difference between a 6 Pin and an 8 Pin is that the 8 Pin has 4 ground wires and 4 power delivery wires and the 6 Pin has 2 ground wires and the same number of power delivery wires. Since the issue has been brought up as a concern about the 480 it would be cool if an Electronics Engineer would chime in on this subject and make it clearer how this works and if it matters.

 

[TRWOV]

Supposely the PCIe power overload is an issue with some cards and AMD is looking into is as it shouldn't be happening (according to them).

https://www.reddit.com/r/Amd/comments/4qfwd4/rx480_fails_pcie_specification/d4sy0c3

I might wait a little and see if this gets solved with a bios update or something.

 

[Dethroy]

it's worth reminding people that only a very small number of hundreds of RX 480 reviews worldwide enountered this issue.

Yeah, because only a handful of reviewers have the equipment to measure PCIe power draw. Way to cloud an issue ...

 

[BiggieShady]

Yeah, because only a handful of reviewers have the equipment to measure PCIe power draw. Way to cloud an issue ...

Pcper has detailed measurements http://www.pcper.com/reviews/Graphics-Cards/Power-Consumption-Concerns-Radeon-RX-480
TL;DR it slightly exceeds pcie spec at stock, it draws up to 100W from pcie when overclocked

 

[bug]

View attachment 75943
Pcper has detailed measurements http://www.pcper.com/reviews/Graphics-Cards/Power-Consumption-Concerns-Radeon-RX-480
TL;DR it slightly exceeds pcie spec at stock, it draws up to 100W from pcie when overclocked

100W is 33% more than the standard 75W. If you want to call that slight, I'm not going to stop you.

 

[rhythmeister]

100W is 33% more than the standard 75W. If you want to call that slight, I'm not going to stop you.

It's significant but even my budget Biostar A880GZ allows me to increase the amount of power available to the PCI-E x16 slot!

 

[R-T-B]

It's significant but even my budget Biostar A880GZ allows me to increase the amount of power available to the PCI-E x16 slot!

So does my board. But specs exist for a reason, and this should be mentioned in reviews.

 

[BiggieShady]

100W is 33% more than the standard 75W. If you want to call that slight, I'm not going to stop you.

Slight was meant for stock clocks where it peaks at 80 W

 

[KainXS]

100W is 33% more than the standard 75W. If you want to call that slight, I'm not going to stop you.

its not 33% because of the tolerance to be applied to the voltage, but its still over the limit a good bit

teclab reported 126W at the slot but they are nvidia biased(to me at least)

Science Studio are the only ones reporting real problems that I remotely believe myself

 

[nexus_a]

Slight was meant for stock clocks where it peaks at 80 W

"The PCI Express specification language is more specific on currents than wattage limits, calling for a maximum of 5.5A over the +12V line and 3A over the +3.3V."
"The highest power draw I measured with the RX 480 at stock settings showed 80-85 watts of power draw at over 7A on the +12V line and 4.5-5.0 watts of power draw on the 3.3V line. These were consistent power draw numbers, not intermittent spikes, and users have a right to know how it works."

80W actually doesn't tell you much because you need to look at both voltage and amperage. In the test the RX480 draws 12V @ 7V = 84W on average. 7A is 27% over the absolute maximum current limit of 5.5A.