Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
AMD Confirms RX 480 At $199 USD, Other APU & Polaris Announcements
- Thread starter lowendlinux
- Start date
- Sort by reaction score
But we don't see it mentioned much do we? It's odd.
I also don't give too much credit to Chinese rumors. But I do believe it will be close enough, NVidia will not loose the opportunity to brake the 480.
I also don't give too much credit to Chinese rumors. But I do believe it will be close enough, NVidia will not loose the opportunity to brake the 480.
The 480 will still be a nice 1080p card even if the 1060 comes out on top. It all depends on what price will the 1060 be sold.
That's the MSRP, not the actual price that you'll get in store... Supply and demand at play...
Tested the Doom 3 Vulkan update on 1070
1200P
All settings ultra
Anti aliasing maxed out
Cinematic filter
Nightmare stats overlay
OpenGL average : 155FPS
Vulkan average : 175FPS
It can go up to 200FPS in a view with less FX. The frame rate counter maxes at 200 so it could be higher.
150FPS when viewing particles and fog
120-160FPS when action gets really heavy
1200P
All settings ultra
Anti aliasing maxed out
Cinematic filter
Nightmare stats overlay
OpenGL average : 155FPS
Vulkan average : 175FPS
It can go up to 200FPS in a view with less FX. The frame rate counter maxes at 200 so it could be higher.
150FPS when viewing particles and fog
120-160FPS when action gets really heavy
Last edited:
That's pretty awesome! Any chance you can test at 1440p?
My poor old 680 actually performs better with OpenGL. Under Vulkan I lose about 20% FPS :/
OCN?
[doublepost=1468392770][/doublepost]
My monitor is 1920x1200
To be frank the game runs so well even on a 2GB 680 with OpenGL that it doesn't feel different to play on a 1070 with Vulkan. Just the filters make it look grainier and sharper.
Last edited:
Overclock.net
Most of the info you see in this thread is pulled directly from there.
==edit==
My that Sapphire RX 480 is sexy!
Last edited:
Because it's the first AAA title with Vulkan support that highlights the importance of how a GPU should function without all that overhead and also by using Async, which is the GPU equivalent of hyper threading. This and OpenCL are the strengths of the Radeon brand. Nvidia could have chosen to have better Async, but they chose high clock speeds. They could have chosen better OpenCL and Double Precision, but they want people to support CUDA. This weakening of their architecture allows them to reduce power consumption and increase clock speeds, but it also means Nvidia isn't the greatest choice for workstation applications.MThis is all elementary.
I still think the 1070 is the best all rounder now, but I think the Nano offers tremendous value these days if you need a productivity card. The 480 also, just fantastic.
I still think the 1070 is the best all rounder now, but I think the Nano offers tremendous value these days if you need a productivity card. The 480 also, just fantastic.
I agree, the R9 Nano and the 480 certainly do look promising, but try and frame this within the Mac Pro sphere.
Right now, I could get an Nvidia 980, install drivers provided and worked on fairly regularly by Nvidia, and use it in both Windows and OS X. Can we do that with the R9 Nano or the 480? No.
Sierra looks interesting though. Better built-in Nvidia drivers (though still behind), elements of Polaris support. But again, the Nvidia stuff just works, or has available drivers for it, which we can reasonably confidently expect to continue. AMD GPUs, not so much.
AMD stuff may well perform better with Vulkan / async workloads. But if we're discussing Macs (and on MR, I assume we are!) then isn't this a bit academic? If there isn't any support, why argue that it's a better solution?
Now, of course, from reading netkas' forums and MVC efforts (and it's obviously early days yet), it looks like there could be a way to relatively simply run the R9 Nano or the 480 with minimal tweaks to the OS. But that's no guarantee these will work in future.
Nvidia on the other hand look fair more promising from this perspective. So, all things considered, doesn't that matter more?
Thanks for testing.
How is this implemented in the application? Is there a configuration setting for choosing a rendering method, or was Vulkan simply enabled by an update and now it is automatically enabled? (Or something else?)
Because we will discover in the coming months how Metal will run on equivalent AMD and Nvidia hardware. Until now we have been using that **** version of OpenGL and mostly crap old drivers. El Capitan has been a disappointing waste of time and Sierra looks like the first time in a long time that there is a proper refresh of drivers and APIs at the same time.
[doublepost=1468445557][/doublepost]
You just choose which API in the video settings. You can also use GeForce Experience to crank everything up to max.
[doublepost=1468446064][/doublepost]These are the settings I used. In GFE, click on DOOM Vulkan from the games list. Then in the settings click on the spanner and slide the quality to maximum. With the latest GPUs you can crank that slider to max in any game. No sweat.
Attachments
http://videocardz.com/62250/amd-vega10-and-vega11-gpus-spotted-in-opencl-driver
Like I have said before, Greenland is different chip than Vega 10. It shares similar architecture however.
That SOC 15 is about Memory subsystem integrated into the GPU(HBM).
Also timing for Vega is between 2016 and 2017. What is however strange is that Raven1X codename.
Like I have said before, Greenland is different chip than Vega 10. It shares similar architecture however.
That SOC 15 is about Memory subsystem integrated into the GPU(HBM).
Also timing for Vega is between 2016 and 2017. What is however strange is that Raven1X codename.
I don't know if I would read too much into it. Did "Iceland" ever turn in to anything? "Greenland" could be nothing.
That groups are names for the architectures, or families of architectures. Vega and Greenland are the same family. That is all.
GPU core counts can be different. For Example Vega 10 can be 3072 GCN core chip with HBM1, and Vega 11 can be 5120 GCN core Chip with HBM2. Greenland while being in the same family can have 4096 GCN cores.
This is only theorising, about how can architecture family look like.
GPU core counts can be different. For Example Vega 10 can be 3072 GCN core chip with HBM1, and Vega 11 can be 5120 GCN core Chip with HBM2. Greenland while being in the same family can have 4096 GCN cores.
This is only theorising, about how can architecture family look like.
Id explain why Async and compute is so important
http://wccftech.com/doom-dev-going-...vely-idtech-6-developers-will-take-advantage/
http://wccftech.com/doom-dev-going-...vely-idtech-6-developers-will-take-advantage/
Futuremark released a directx12 benchmark including support for async compute and explicit multi adapter. Interestingly both the RX 480 and GTX 1070 see similar performance gains when async compute is enabled.
http://www.anandtech.com/show/10486/futuremark-releases-3dmark-time-spy-directx12-benchmark
http://www.anandtech.com/show/10486/futuremark-releases-3dmark-time-spy-directx12-benchmark
Sure, but I was especially thinking about the relative gains. Something like this is advertised as "50% gain for Radeon graphics, little gain for Team Green":
I don't see anyone mention that AMDs high-end Fury X got outperformed by a GTX 970 in OpenGL.
Those Vulkan scores are definitely nice and show what async compute can do (RX 480 being just 20% behind a GTX 1070 is totally awesome!), but the relative gain is just so enormous because AMD's OpenGL drivers were obviously crap.
The reduced CPU overhead is also nice to have, but it doesn't seem to have a notable real world benefit, unless you're trying to play Doom on an Intel Atom.
Btw, I'm seeing very different benchmark scores around the net. Seems to be difficult to benchmark 'Vulkan' at the moment, with Fraps and similar tools lacking proper support.
Sure, but I was especially thinking about the relative gains. Something like this is advertised as "50% gain for Radeon graphics, little gain for Team Green":
View attachment 640248
I don't see anyone mention that AMDs high-end Fury X got outperformed by a GTX 970 in OpenGL.
Because it's a 2 frames difference (inside margin of error), doesn't reflect OGL performance in all apps and games, and regardless of that Vulkan is far ahead. Why continue to develop for a slower and inconsistent performing API? It's time to move on from OpenGL.
The other thing Vulkan can do is bring Linux gaming up to par with PC gaming. It can probably allow easier ports to Metal too.