• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

have AMD stopped competing

Your own chart shows the GTX 480 7% faster than HD 5870 @1600p. @1080p it was 10% faster.

perfrel_1920.gif

What about the HD 5970? Non existent? When it is 30-40% faster? :D
 
AMD's AF problems have been fixed for ages now - not sure exactly but its possible 5+ years.

Isn't there still a big performance issue? When AMD released their own benchmarks of Fuji beating the 980ti the settings we're all weird but the standout was AF was always switched off or low. Some subsequent testing by a reviewer showed how several AMD cards took a much bigger performance hit
 
Dual GPU card?

Nothing wrong with it. If you remember, starting with HD 3870 / 3870 X2; HD 4870 / 4870 X2, HD 5870 / HD 5970, it was the AMD strategy to build one sweet spot, relatively small chip, and based on it, dual-GPU card every generation.
And it worked perfectly fine.
 
Nothing wrong with it. If you remember, starting with HD 3870 / 3870 X2; HD 4870 / 4870 X2, HD 5870 / HD 5970, it was the AMD strategy to build one sweet spot, relatively small chip, and based on it, dual-GPU card every generation.
And it worked perfectly fine.

Well given multi-GPU, to say it worked perfectly fine is a bit much.
I owned a 4870x2 and at one point owned a 5970.
 
This one post proves you have zero idea what you are talking about

Except that I never say anything without some solid facts behind it :D

Things get more complicated with Explicit Multiadapter, which is new in DX12. Explicit Multiadapter will have two distinct API patterns: Linked GPUs and Unlinked GPUs. Linked GPUs refer to the special pairing of specific hardware, similar to what we're familiar with via SLI and Crossfire. DirectX 12 will view linked GPUs as a single GPU, allowing them to collaborate more closely and share resources in each other's rendering pipeline.

https://www.pcgamer.com/directx-12-will-be-able-to-use-your-integrated-gpu-to-improve-performance/
 
AMD should build Radeon RX 590 based on two Polaris 10 chips.
Rely on DirectX 12 to see them as one GPU.
You mean so they can finally try to keep up with a gtx980 from 3-4 years ago by fudging it with crossfire because they can’t build a proper chip? Lol

What about the HD 5970? Non existent? When it is 30-40% faster? :D

It’s a pretty poor show that card... 2 gpu and they still only make a 30% gain over the previous nvidia flagship.
 
It’s a pretty poor show that card... 2 gpu and they still only make a 30% gain over the previous nvidia flagship.

HD 5970 was already released prior to the GTX 480. So, GTX 480 couldn't touch it.

These 30-40% are on average, including cases where CrossFire scales 60-70% and beyond, where CrossFire doesn't scale, and where it scales poorly.

Modern gual-GPU solutions like the R9 2950X2 scale 90-100% in some cases.
 
Lets look at there current stack of products,

Low end, sub 570 area... ok - all competitive with there nvidia counterparts. - bigge
mid range - 570/580 both very competative with nvidias offerings however they are outsold hugely.
high end - vega 56 does very well against the 1070(ti) and 1080, it is very competitive there. V64 is not quite as good vfm but still is able to go toe to toe with the 1080 in most titles.
Enthusiast Vega water is not as good as the 1080ti or various Titans. This is really the only part of the stack they struggle in.
They also have intel hammered with the APUs - at least until intel drops the vega based one they are working on ... but its still vega lol.
Finally of course the big consoles are AMD and likely to continue to be next round.

So... really only the very highest end of the market are they struggling with competing - however this has a big knock on effect to the rest of there product stack for some reason... It shouldnt but... big egos and short memories run rampant in the PC games sector so thats what you get.

True but they have nowhere to go to make up the gap, Vega is bigger die than the GP102 (1080ti let alone the GP104 (1070/1080) and there already losing money per chip, as well as generating to much heat and power while being bandwidth starved.

So they can't make a bigger die for more performance, dual GPU is pretty dead, they need a new Architecture, remaking Bulldozer didn't work for them in the CPU space, it took a Ryzen redesign to get them back in the game , they just need to do the same with in the GPU space.

Scrap what you had and start again/

Intel had to do the same after PressHOT
Nvidia had to do the same after Rankine (NV30) and Fermi
AMD/ATI had to as well after the R600 and Bulldozer
 
HD 5970 was already released prior to the GTX 480. So, GTX 480 couldn't touch it.

These 30-40% are on average, including cases where CrossFire scales 60-70% and beyond, where CrossFire doesn't scale, and where it scales poorly.

Modern gual-GPU solutions like the R9 2950X2 scale 90-100% in some cases.
Congrats... I'll take a single GPU with the same horsepower as 2 ati gpus... kthx
 
Except that I never say anything without some solid facts behind it :D

Things get more complicated with Explicit Multiadapter, which is new in DX12. Explicit Multiadapter will have two distinct API patterns: Linked GPUs and Unlinked GPUs. Linked GPUs refer to the special pairing of specific hardware, similar to what we're familiar with via SLI and Crossfire. DirectX 12 will view linked GPUs as a single GPU, allowing them to collaborate more closely and share resources in each other's rendering pipeline.

https://www.pcgamer.com/directx-12-will-be-able-to-use-your-integrated-gpu-to-improve-performance/


You have this the wrong way round.
DX12 Linked GPUs is more or less exactly what we have with DX11. From the developers perspective there is a single GPU, with the driver doing a lot of magic to balance workloads. in DX12, just like DX11, this uses Alternate Frame4 Rendering (AFR). In this case the GPUs need to be the same, same performance, brand, architecture, GPU chip-set. memory amounts can be different, but the card with more memory will be capped to the amount of memory of the smallest card. DX12 unlinked GPUs is where DX12 differs form DX11, in this mode the developer see multiple different GPUs and can directly control and communicate rendering to each independently. This is not possible in DX11. In this mode the developer has more control,. but has to put in much ore development work. In this mode the GPUS can be different brands and chip-sets.



So if you consider that the DX11 SLI/XF performance and limitations is acceptable then you can do the exact same thing in DX12.

DX12 does not magcially make 2 GPus perform like one, it just allows the standard SLi/XF AFR rendering.
 
Explicit MGPU is not in anyway the same as AFR, keep the thread on topic.

DX12 linked GPU is exactly AFR.
DX12 unlinked GPU is whatever the developers want it to be. Unlinked is refereed to as explicit MGPU, because the develop has explicit control over multiple independent GPUS.
these post are on top[c, refuting 4K8KW10 inaccurate and ridiculous comments.
 
DX12 linked GPU is exactly AFR.
DX12 unlinked GPU is whatever the developers want it to be. Unlinked is refereed to as explicit MGPU, because the develop has explicit control over multiple independent GPUS.
these post are on top[c, refuting 4K8KW10 inaccurate and ridiculous comments.

WTH? :D :confused: You've just repeated what I said and what I linked.
 
Back
Top Bottom