AMD Hints that rx 480 has gtx 980/r9 nano Level Performance

[xAcid9]

26 minutes ago, ONOTech said:

An Asian site, Chiphell, leaked benchmarks of the RX 480. When underclocked to 1080MHz, it beat the 980, and at 1266 it scored 3500, placing it right under the Fury.

Here's the reddit thread: https://www.reddit.com/r/Amd/comments/4noslj/rx480_leaksrumors_timeline_megathread/

 

The YouTube Video w/ "proof": 

Screenshot of settings

 

  Reveal hidden contents

 

 

 

 

 

 

More video note-worthiness (shows the system specs after testing)

Comparitive chart with the 480 at 1080 MHz.

 

In the last BIOS and driver video he get Valid 3DMark.. i wonder what driver he used.

[Humbug]

1 hour ago, Rodinski said:

What's with the "VR" talk?  Isn't it no different in rendering than any other way?

ya it's similar to normal high resolution rendering. Framerate drops are more noticeable though and can sometimes induce nausea. But rendering is similar, if the GPU is powerful it will be good at both. Most of the rendering differences are at the game engine level not at the GPU hardware level. It's just that both HTC vive and Oculus Rift have mentioned R9 290 / GTX 970 and up as the recommended spec. With VR currently being a hot topic in the news AMD can market the new polaris architecture by claiming to surpass the last gen VR-ready spec for only $200. Even if many buyers have no intention of purchasing an HMD it's good marketing as it demonstrates affordable performance of the new architecture. It's much better than trying to compare it to a GTX 1080 since that is irrelevant in the target market. E.g. Nobody who bought an R9 285 cares how the GTX 980 performed as it's irrelevant and out of budget. I think the crossfire AOTS benchmark was stupid, they could have just compared single GPU performance with current gen cards like GTX960 and R9 380 to show the performance jump. They don't need to talk about beating GTX 1080 until they launch vega.

[Jahramika]

On 6/10/2016 at 0:13 PM, Enderman said:

i have literally no idea what he means by " incorporated DX12 support only to  their engine midway during development and aren't using the API like AotS is."

 

like, what the heck, if they're using DX12 for the engine that means its using the DX12 API...

AOTS is just biased to AMD, like heaven is biased to nvidia

No AOTS is the only Dx12 gamestop the ground up the rest are ported dx11 titles running on Dx12. Only 1 Dx12 titles pairs a  Nvidia and Amd card together for a reason.

[Trixanity]

1 minute ago, Jahramika said:

No AOTS is the only Dx12 gamestop the ground up the rest are ported dx11 titles running on Dx12. Only 1 Dx12 titles pairs a  Nvidia and Amd card together for a reason.

Supporting explicit multiadapter isn't a requirement for a ground-up DX12 title. Mostly will probably use implicit because it's easier and it off-loads the work to the driver like it is with previous APIs.

[Jahramika]

12 minutes ago, Trixanity said:

Supporting explicit multiadapter isn't a requirement for a ground-up DX12 title. Mostly will probably use implicit because it's easier and it off-loads the work to the driver like it is with previous APIs.

Okay if you say so. If another current Dx12 was capable of achieving it then it would be know. Their are northern for good obvious reason. If the game is not coded correctly from the start it will not have access to all of the Dx12 functions, like a dx11 port.

[Enderman]

25 minutes ago, Jahramika said:

No AOTS is the only Dx12 gamestop the ground up the rest are ported dx11 titles running on Dx12. Only 1 Dx12 titles pairs a  Nvidia and Amd card together for a reason.

http://wccftech.com/hitman-feature-implementation-dx12-async-compute-amd/

[Trixanity]

12 minutes ago, Jahramika said:

Okay if you say so. If another current Dx12 was capable of achieving it then it would be know. Their are northern for good obvious reason. If the game is not coded correctly from the start it will not have access to all of the Dx12 functions, like a dx11 port.

The whole multiadapter thing requires a lot of work. Most developers will avoid it, I'm sure. The explicit modes (linked and unlinked) will give developers the most control and open up new possibilities of combinations and performance but implicit is easier, so we all know what that means.

 

So if explicit mode is the requirement for a DX12 title, then we probably won't see many. And some will probably not use any multiadapter mode at all because while implicit is easier, it still requires work, so not all developers will be up to the task. Just like today where many DX11 titles don't support SLI/Crossfire.

 

DX12 gives developers tons of possibilities and new stuff to play with but at least 50% won't have the skill or time to use everything DX12 have to offer and use it well.

[Big-G]

18 hours ago, Delicieuxz said:

How much does 3DMark benefit from async? I'd like to know the Rx 480's straight DX11 performance.

At the moment, no benefit. It's based on DX11.

 

2 hours ago, Jahramika said:

Okay if you say so. If another current Dx12 was capable of achieving it then it would be know. Their are northern for good obvious reason. If the game is not coded correctly from the start it will not have access to all of the Dx12 functions, like a dx11 port.

Another one who thinks DX12 = Async. There's a difference between minimum required functionality and feature sets. mGPU and Async are optional features and not a requirement for DX12.

[ivan134]

9 hours ago, GidonsClaw said:

At the moment, no benefit. It's based on DX11.

 

Another one who thinks DX12 = Async. There's a difference between minimum required functionality and feature sets. mGPU and Async are optional features and not a requirement for DX12.

You can go bury your head in the sand all you want. You people would not be saying this if Nvidia had better async compute performance. PC gamers were waiting for dx 12 for a long time because it would help with the inefficiencies of all the previous APIs we've had, and asynchronous compute is the main driving factor for that. If you know another dx 12 functionality that does that please let us know. Since the company of choice of so many fanboys sucks at said functionality, they now turn to downplaying. When Maxwell came out, power consumption now suddenly became the most important thing and the amount of stupid things we've been hearing in regards to power consumption has been fucking amazing. That's apparently how it works: whatever Nvidia is good at suddenly becomes the most important thing, regardless of the fact that their "efficiency" comes from not having the extra hardware on their chips needed for things like asynchronous compute.

[Khvarrioiren]

8 minutes ago, ivan134 said:

whatever Nvidia is good at suddenly becomes the most important thing

This has been the case since the Fermi revision days, and sadly will not go away until AMD steps up their game.

[Jahramika]

On 6/12/2016 at 6:31 PM, Enderman said:

http://wccftech.com/hitman-feature-implementation-dx12-async-compute-amd/

Does not do what AotS can do and is a dx11 port.

[Enderman]

1 minute ago, Jahramika said:

Does not do what AotS can do

yeah, and what it can do is use a huge amount of resources to cripple nvidia performance

which regular triple A games do not do since they dont need to have thousands of AIs running at the same time

 

and no, makes no difference if its DX11 port or not, it still coded for DX12 and utilizes DX12 featuers

[SteveGrabowski0]

[SteveGrabowski0]

On 6/10/2016 at 6:55 AM, SteveGrabowski0 said:

It sounds like BS. If it really performed at the level of the Nano why did we get no single card benchmarks at Computex? Why no benchmarks still on the day of the 1070's launch? R9 Nano performance for $200 would make a ton of people reconsider their desire to buy a 1070 at launch.

 

[DocSwag]

4 hours ago, SteveGrabowski0 said:

 

Well it was wrong bro I guess...

Then again, basically every rx 480 rumor was wrong . They were all like saying "GTX 980 performance and it can overclock to over 1.5 ghz!" Instead it's 970 performance that barely ocs to 1.4 ghz

[khothmod]

5 minutes ago, DocSwag said:

Well it was wrong bro I guess...

Then again, basically every rx 480 rumor was wrong . They were all like saying "GTX 980 performance and it can overclock to over 1.5 ghz!" Instead it's 970 performance that barely ocs to 1.4 ghz

With a reference card...

[DocSwag]

1 minute ago, khothmod said:

With a reference card...

True. I guess....

Stilll though. I'm not expecting it to hit 1.6 ghz, which some sites said it would hit *cough* wccftech *cough*

[awesomeness10120]

I'm just glad it's decent for the money.

[khothmod]

1 minute ago, DocSwag said:

True. I guess....

Stilll though. I'm not expecting it to hit 1.6 ghz, which some sites said it would hit *cough* wccftech *cough*

It looks like it's stuck behind power limits, and an 8-pin connector from an after-market card will probably fix that.

 

From the sounds of it, the reviewer cards were a bit messed up, with huge power draw issue, and some people are pointing to it being something to do with how the reviewers could freely switch between 4GB and 8GB.

[Kobathor]

7 minutes ago, awesomeness10120 said:

I'm just glad it's decent for the money.

Unless the 1060 is better for the same price, the 480 is perfect for my budget. It has more features than the R9 390 and GTX 970, which is really the only reason to consider it.

[DocSwag]

5 minutes ago, Kobathor said:

Unless the 1060 is better for the same price, the 480 is perfect for my budget. It has more features than the R9 390 and GTX 970, which is really the only reason to consider it.

Not to mention that it probably has a lot of performance that can be unlocked through better drivers which would put it right on par with a 390 in the future.

 

[awesomeness10120]

8 minutes ago, Kobathor said:

Unless the 1060 is better for the same price, the 480 is perfect for my budget. It has more features than the R9 390 and GTX 970, which is really the only reason to consider it.

Knowing Nvidia, the 1060 will probably be more expensive than the RX 480. 

[MoonSpot]

17 minutes ago, khothmod said:

It looks like it's stuck behind power limits, and an 8-pin connector from an after-market card will probably fix that.

 

From the sounds of it, the reviewer cards were a bit messed up, with huge power draw issue, and some people are pointing to it being something to do with how the reviewers could freely switch between 4GB and 8GB.

I doubt it.  The thermals go too far out of control with minor OC'ing, no cooler is going to be good enough to rein it in and get more than a superficial perf boost.  Still, would be nice to drop the stock operating temps down a bit more.

 

As for if it has anything to do with the reviewers cards and their bios, we've no way to know; and that's just reaching for hopes. 

Still an awesome card IMO, particularly if one is expecting to upgrade their monitor quite a bit later to a beefier resolution.

1080p tits completely out ready to rock everything

1440 kill some/most filters and carry on,

4k do the setting dancing we've all done for years and years to make shite work/fumble

[Kobathor]

1 minute ago, awesomeness10120 said:

Knowing Nvidia, the 1060 will probably be more expensive than the RX 480. 

Well duh. But, if it's close in price and performs better, I might consider it.

[DocSwag]

Lol why is this 3 week old thread suddenly alive again