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
Manuel, there is Nvidia Pascal thread in this very forum. Can you post things related to Nvidia there? I already did this in the past. Lets not mix threads together, shall we? 
Direct link: https://forums.macrumors.com/threads/so-is-anyone-actually-running-a-gtx-1070-80.1981630/
Direct link: https://forums.macrumors.com/threads/so-is-anyone-actually-running-a-gtx-1070-80.1981630/
Yes, it goes upward of 150W but does NOT draw more than 75 watts from the slot. So yes, they fixed it.
They fixed it by offering a reduced performance "PCIe compliant" mode. A Band-Aid for a failed design.
...
It is actually quite funny that nobody have spotted that there is so far no mobile GPUs based on Polaris architecture, available anywhere
It is just packaged differently.
"Interestingly, RX 460 does not ship with a fully enabled Polaris 11 GPU. AMD previously disclosed that Polaris 11 was a 16 CU part, whereas RX 460 only ships with 14 CUs enabled. ... "
http://www.anandtech.com/show/10530...x-460-specifications-shipping-in-early-august
The 480 shipped before the 470 did. It is likely AMD makes more money selling the lower yielding stuff at higher prices as a entry desktop solution.
When they have a 16 CU part then can take it mobile ( wider cores at slower speeds which is less of a performance hit than fewer cores at slower speeds. Same thing Nvidia did with their mobile. more cores at slower clocks with confusingly the same model number. ). They probably aren't selling them because they don't have them in bulk yet.
Make a sizable number of wafers, tweak the process, and yields should go up. Release when have enough product volume. Scarcity might work in the "tech porn" marketplace (vendor get to slap peak surcharge on the Add-In cards) , but as a system component vendor if you screw up the overall product's production logistics, people don't pick you again. You are screwing with other people's money.
Last edited:
Definitely...
Big win for AMD i feel for the Polaris 10 chips.
In DaVinci resolve 3x Rx 480's equal 2x Titan X's (pascal) . Thats 750usd vs 2400usd...
xsmi, PCIe spec isn't for power but current. It's still out of spec as far as I know, less so now, but still out.
The problem is put off because if a board is well designed, as most of late are, it's not a problem.
Could try and find it but the spec calls for 66W or so, not 75W. Google it if you want to know for sure.
The problem is put off because if a board is well designed, as most of late are, it's not a problem.
Could try and find it but the spec calls for 66W or so, not 75W. Google it if you want to know for sure.
Interesting. I hadn't seen any compute benchmarks for pascal this comprehensive. This is a good comparison to gauge efficiency since the GTX 1070 and RX 480 use roughly the same amount of power. It looks like Nvidia holds a small but still significant edge across the board. This mostly mirrors the theoretical results that are up on anandtech.
It would be fun to see how much better the GTX 1080 and Titan X do.
Correction to the PCIe 4 power delivery:
http://www.tomshardware.com/news/pcie-4.0-power-speed-express,32525.html
[doublepost=1472135331][/doublepost]http://www.fudzilla.com/news/graphics/41440-new-alienware-machines-powered-by-Polaris
Polaris seems to be a hit.
[doublepost=1472135649][/doublepost]No $h|t?!
https://www.techpowerup.com/225287/nvidia-ansel-post-fx-based-on-stolen-mastereffect-reshade-fx-code
https://www.techpowerup.com/225297/nvidia-to-credit-marty-mcfly-modding-for-leftover-ansel-code
http://www.tomshardware.com/news/pcie-4.0-power-speed-express,32525.html
[doublepost=1472135331][/doublepost]http://www.fudzilla.com/news/graphics/41440-new-alienware-machines-powered-by-Polaris
Polaris seems to be a hit.
[doublepost=1472135649][/doublepost]No $h|t?!
https://www.techpowerup.com/225287/nvidia-ansel-post-fx-based-on-stolen-mastereffect-reshade-fx-code
https://www.techpowerup.com/225297/nvidia-to-credit-marty-mcfly-modding-for-leftover-ansel-code
Negligible. It was never going to be good enough for people who want nvidia cards anyway. The issue is moot as far as I'm concerned and it seems to be with the buying public because they're flying off the shelves.
Thanks for the kind words.
Yes I'm trying to get a hold of a Titan X (Pascal) although its troublesome as Nvidia won't ship to me.
eBay currently has some at 400-700us above the rrp.
xsmi, don't take this wrong, at least not from me. I was really hoping that AMD would rise again.
I believe they weren't at their best on this but still they have a formidable product. And to back that up their share is going up.
It's just I believe their come back could have been smoother, and it wouldn't have been that difficult.
Still, it's good to see them back up.
[doublepost=1472220792][/doublepost]CrossfireX between 480 and 470 seems to work well.
Same Polaris 10 core so no sweat.
I believe they weren't at their best on this but still they have a formidable product. And to back that up their share is going up.
It's just I believe their come back could have been smoother, and it wouldn't have been that difficult.
Still, it's good to see them back up.
[doublepost=1472220792][/doublepost]CrossfireX between 480 and 470 seems to work well.
Same Polaris 10 core so no sweat.
This thread is 32 pages long and at least 20 pgs of it are dedicated to why there should be nvdia boards in mac pros. Yes, I love to see apple give us the opportunity to chose what boards we want in our machines. But to call the RX 480 an out and out failed attempt? That's crap.
They've shown time and time again that when looking at next level API's Polaris is the best bang for the buck. Why is none on this board not taking nvidia to task about Vulcan and DX12 performance improvements? Because of the bias to that side.
Polaris is a forward looking tech. It was meant as a low to moderate cost solution. When you look at it in that vein, you realize why AMD is taking marketshare. As soon as Newegg gets the Sapphire nitro back in stock, it's going in the kid's gaming rig.
They've shown time and time again that when looking at next level API's Polaris is the best bang for the buck. Why is none on this board not taking nvidia to task about Vulcan and DX12 performance improvements? Because of the bias to that side.
Polaris is a forward looking tech. It was meant as a low to moderate cost solution. When you look at it in that vein, you realize why AMD is taking marketshare. As soon as Newegg gets the Sapphire nitro back in stock, it's going in the kid's gaming rig.
AMD gets a large improvement going from DX11 to DX12 because their DX11 driver is pretty crap. AMD gets a large improvement going from OpenGL to Vulkan because their OpenGL driver is pretty crap.
NVIDIA doesn't get much of an improvement going from DX11 to DX12 because their DX11 driver is very good (e.g. beats AMD's Mantle driver in CPU overhead tests). NVIDIA doesn't get much of an improvement going from OpenGL to Vulkan because their OpenGL driver is very good.
What's the issue exactly?
Now that is pure speculative nonsense with nothing to back it up.
The only way to quantify the above hypothesis that the AMD DX11 driver is crap or that the Nvidia DX11 driver is great would be to compare Maxwell vs Fiji or Polaris vs Pascal at exactly the same clockspeed.
And then you will just how far ahead AMD is. Their only issue is power consumption, not performance per clock.
But then...we don't need to take advice from you about drivers when you had been promoting Nvidia cards and web drivers full of serious bugs that were affecting pro users.
No, it's also ASYNC compute. Pascal is STILL not optimized as it should be for that. Yes, it's better than Maxwell, but NOWHERE where you would expect them to be.
http://www.pcper.com/reviews/Graphics-Cards/Whats-Asynchronous-Compute-3DMark-Time-Spy-Controversy
Last edited:
No, it's really easy -- look at lower resolution (i.e. CPU limited) performance. That's where driver overhead will actually show up. AMD gets a large gain going from older APIs like DX11 and OpenGL to newer APIs like DX12 and Vulkan because the newer APIs are low level and thus low overhead by design. NVIDIA doesn't see the same level of percentage increase on their side because their starting point (the older APIs) is so much faster to begin with.
I've also never told people to go buy NVIDIA GPUs, I merely provide assistance when they do since that's what I use. I also don't use pro apps and thus don't make recommendations regarding those as well. Perhaps you have me confused with someone else?
What's the issue here? Pascal sees improvements with async compute. Are you complaining that they don't see the same percentage increase as AMD? Perhaps their architecture just has fewer execution gaps that they can fill with async compute work, which one could argue means their architecture just has a better balance to begin with (i.e. for the same workload, NVIDIA can saturate their GPU while AMD cannot).
So going forward, it doesn't bother you that a $280 mainstream card gets pretty close in performance in forward looking API's to a $450-$650 card? Those are BEASTLY cards. But for my money, on the API's that are coming out, they have to have a better showing.
I play games, so no?
http://www.anandtech.com/show/10540/the-geforce-gtx-1060-founders-edition-asus-strix-review/5
http://www.anandtech.com/show/10540/the-geforce-gtx-1060-founders-edition-asus-strix-review/6
http://www.anandtech.com/show/10540/the-geforce-gtx-1060-founders-edition-asus-strix-review/7
http://www.anandtech.com/show/10540/the-geforce-gtx-1060-founders-edition-asus-strix-review/8
I'm not seeing the RX 480 come anywhere near a 1070 in those tests, including DX12 games. On paper, the RX 480 should destroy the 1060, yet it's behind the 1060 in all those games. If you care about an app where the RX 480 is close to a 1070, then by all means buy the RX 480 instead. I'm happy with my 1080 because it performs well in the games I play.
AMD's cards do get a significant boost on some DX12 games. AMD cards tend to be very wide architectures, making it tough to fill the pipe with the single threaded limitations of DX11. Thats why you see such boosts with DX12, AMD can finally utilize their entire GPUs for games that are programmed to take advantage of it. However, since games have been using DX11 for years, most new games are still DX11, and not every DX12 is well optimized, Nvidia tends to beat AMD with its narrow, higher clocked architecture in games.
I think the 1060 is a good chip for Nvidia. The RX 480 will probably sell more than it because of its slightly lower price but the 1060 usually beats it out in most gaming scenarios. Add to that the 1060 is a smaller die and uses less power and its pretty clear Nvidia is making more profit on it.
In related news it looks like Vega is coming out in the first half of 2017. I hope this means Apple won't wait for it and release a new mac pro this fall. I think they will hold some sort of press event in october where they refresh the macbook pro, mac pro and release a new thunderbolt 3 display.
https://www.computerbase.de/2016-07/doom-vulkan-benchmarks-amd-nvidia/
10% difference between RX 480 and GTX 1070. There are DX12 benchmarks at Golem.de that compared the performance difference between RX 480 and GTX 1060.
10% difference between RX 480 and GTX 1070. There are DX12 benchmarks at Golem.de that compared the performance difference between RX 480 and GTX 1060.