Sidebar Sidebar
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

I was contemplating a dual 480 setup but against the 1070 I just don't think it's worth it. And that was before all these issues with power draw.

I'm sure AMD can correct it all with a driver update, but what kind of reckless company puts out a product in this state? Don't get me wrong, I'm no massive Nvidia fan; I've seen enough dead 2008 MacBook Pros with failed 8600m GTs to last a lifetime, but I think this round Nvidia might just have clinched it.

I was thinking of getting an R9 Nano, but they were never prices right, and now seem to be unavailable (just when MVC/netkas/Flor!an look like they've figured out how to get it working!)

I guess it could be prudent to wait for Vega, but then there's no guarantee it'll work either. And if Apple drop support for it, you're stuffed. The Nvidia web driver is at least a much more reliable source of comfort for those of us with an Nvidia card.. but then who knows how much longer that will hang about for either.
 
Draeconis said:
I was contemplating a dual 480 setup but against the 1070 I just don't think it's worth it. And that was before all these issues with power draw.

I'm sure AMD can correct it all with a driver update, but what kind of reckless company puts out a product in this state? Don't get me wrong, I'm no massive Nvidia fan; I've seen enough dead 2008 MacBook Pros with failed 8600m GTs to last a lifetime, but I think this round Nvidia might just have clinched it.

I was thinking of getting an R9 Nano, but they were never prices right, and now seem to be unavailable (just when MVC/netkas/Flor!an look like they've figured out how to get it working!)

I guess it could be prudent to wait for Vega, but then there's no guarantee it'll work either. And if Apple drop support for it, you're stuffed. The Nvidia web driver is at least a much more reliable source of comfort for those of us with an Nvidia card.. but then who knows how much longer that will hang about for either.
Click to expand...

It really is a waiting game at the moment. 1070 web drivers would do it for most though some might prefer the AMD cards. Either way I'm watching closely and sticking with what I got.
 
My idea is GeForce are better for gaming and all round use, Radeon better for compute. So my 1070 went in my Skylake PC and I'm waiting for a Radeon solution for the cMP, which will probably be the last card it ever sees because the cMP will be truly a vintage relic after 10.13.
 
SoyCapitanSoyCapitan said:
My idea is GeForce are better for gaming and all round use, Radeon better for compute. So my 1070 went in my Skylake PC and I'm waiting for a Radeon solution for the cMP, which will probably be the last card it ever sees because the cMP will be truly a vintage relic after 10.13.
Click to expand...

I hear you. The 1070 in a PC sounds like a good choice particularly for gaming and VR too.
I have held back from buying or building a PC, well for now at least. Whenever I need a gaming fix I bootcamp into Windows 10.
If the cMP does get a look in with the 1070 drivers I might live with that and see.
 
  • Like
Reactions: SoyCapitanSoyCapitan
I ended up buying one on launch day as an upgrade from a gtx760.

I am happy and mine has no power draw problems. Bang for buck it's great, runs everything I throw at it on ultra, 60+ FPS at 2560x1080. Doom runs at 75-115 FPS like that with everything maxed. I'm happy. In a machine with i5-4430 and 16GB RAM.

Considering another one in crossfire, not because I need it but because I've not played with multiple GPUs yet and it is pretty cheap.
 
Pending said:
I hear you. The 1070 in a PC sounds like a good choice particularly for gaming and VR too.
I have held back from buying or building a PC, well for now at least. Whenever I need a gaming fix I bootcamp into Windows 10.
If the cMP does get a look in with the 1070 drivers I might live with that and see.
Click to expand...

I'm ready and in position to test the 1070 in cMP for everyone's benefit. If Nvidia has properly ported the 368+ driver code from Windows then it should be excellent for Sierra's Metal. Even Maxwell should finally get most of the optimisations. But will Nvidia fix OpenCL bugs? We know they don't like to support that API because they have their own.
 
Definitely interested in a 480 to replace my 7950 as a final card for my cMP too. We'll see how it shakes out with support and power draw (I doubt it'll be an issue with 3rd party versions, especially with 8-pin options already available, so plenty of time to wait it out.)
 
  • Like
Reactions: h9826790
fuchsdh said:
Definitely interested in a 480 to replace my 7950 as a final card for my cMP too. We'll see how it shakes out with support and power draw (I doubt it'll be an issue with 3rd party versions, especially with 8-pin options already available, so plenty of time to wait it out.)
Click to expand...

I am still looking for dual RX480 setup. I usually lower the voltage as much as possible on the graphic card to lower the temperature and reduce noise. It won't affect the performance (confirmed by benchmark). If that also works on the 480, then I think I can still safely install 2x 480 in my cMP.

I did try the single card OC route, but don't like that high temperature, high power draw, and ineffective diminish return (especially true on the AMD 79xx series). I am more happy to go for dual card but lower voltage (or even slightly down clock if required) route. Of course, I pay more for one extra card, but the return is worth. My current dual 7950 is quiet and high performance. I am looking for the similar setup but with the RX480.
 
  • Like
Reactions: orph
SoyCapitanSoyCapitan said:
I'm ready and in position to test the 1070 in cMP for everyone's benefit. If Nvidia has properly ported the 368+ driver code from Windows then it should be excellent for Sierra's Metal. Even Maxwell should finally get most of the optimisations. But will Nvidia fix OpenCL bugs? We know they don't like to support that API because they have their own.
Click to expand...

I'd be very interested to see what results you get.

I'm currently having no issue running Doom 4 at 1080p with my current setup. In the future, if Apple make my Mac unsupported, I'll just turn it into a Windows 10 box and keep it until it dies and I can't economically repair it I think. The 1070 would be the last upgrade I'd consider, can always take that across to the new machine as and when I build one. Done with Mac desktops I think.

Only tricky thing is my storage setup. I use OpenZFS and replicate to ZFS on a FreeNAS box. I'll probably convert that to ReFS and change my backup box to run Windows 10 too. Shame though.
 
NVidia is really hard at it:


http://wccftech.com/nvidia-pascal-gp100-geforce-gtx-titan-launch/
[doublepost=1467735148][/doublepost]And no word from AMD so far, right?

I know this is nothing to go by but still:
http://www.fudzilla.com/news/graphics/41034-vega-10-amd-hbm-2-can-launch-in-1h-2017
If it's not October but only March 17 they're done...
[doublepost=1467737467][/doublepost]Tom's second take on the subject:
http://www.tomshardware.com/reviews/amd-radeon-rx-480-power-measurements,4622.html
 
ManuelGomes said:
NVidia is really hard at it:


http://wccftech.com/nvidia-pascal-gp100-geforce-gtx-titan-launch/
[doublepost=1467735148][/doublepost]And no word from AMD so far, right?

I know this is nothing to go by but still:
http://www.fudzilla.com/news/graphics/41034-vega-10-amd-hbm-2-can-launch-in-1h-2017
If it's not October but only March 17 they're done...
[doublepost=1467737467][/doublepost]Tom's second take on the subject:
http://www.tomshardware.com/reviews/amd-radeon-rx-480-power-measurements,4622.html
Click to expand...

AMD isn't going out of business. Sony, Microsoft and Apple do not want a Nvidia Monopoly.
 
Soy, I was saying that the leverage they might have if releasing Vega in October as rumored earlier, and beating big Pascal to the punch in the high end, might be gone if this is true.
NVidia is really keen on having the full lineup out even if just to beat the hell out of AMD and don't let them take back market share, by any means.
If this happens, AMD is again an underdog and fall back to where it was (has been).
Too bad.
They're not helping themselves either.
 
Nvidia is really firing on all cylinders. (G)P100, GP104 and GP106 are all landing within a few months of each other. Thats pretty impressive. It will be interesting to see if the Pascal Titan keeps the same DP ratio as P100 or if its GP102 that is more focused on SP compute.

ManuelGomes said:
Soy, I was saying that the leverage they might have if releasing Vega in October as rumored earlier, and beating big Pascal to the punch in the high end, might be gone if this is true.
Click to expand...

I would bet Vega is coming next year. There doesn't seem to be much substance to the Vega in 2016 rumors.
 
ManuelGomes said:
Soy, I was saying that the leverage they might have if releasing Vega in October as rumored earlier, and beating big Pascal to the punch in the high end, might be gone if this is true.
NVidia is really keen on having the full lineup out even if just to beat the hell out of AMD and don't let them take back market share, by any means.
If this happens, AMD is again an underdog and fall back to where it was (has been).
Too bad.
They're not helping themselves either.
Click to expand...

Where AMD makes most of their revenue they have no competition. Where Nvidia makes most of their revenue they have some competition. Where they leapfrog each other for a few months every summer it's for a niche sector where only big spending consumers and enthusiasts are the target. It's going to stay like that for the foreseeable future because it works out best for both of them.
 
ManuelGomes said:
Soy, I was saying that the leverage they might have if releasing Vega in October as rumored earlier, and beating big Pascal to the punch in the high end, might be gone if this is true.
NVidia is really keen on having the full lineup out even if just to beat the hell out of AMD and don't let them take back market share, by any means.
If this happens, AMD is again an underdog and fall back to where it was (has been).
Too bad.
They're not helping themselves either.
Click to expand...
About Vega, from what I understand, porting Fiji to 14 nm, would bring the GPU into 150W range of TDP with 1050 MHz core clocks. What would make the GPU faster is solving the problems with scheduler. That alone at 1050MHz would bring the GPU into GTX 1080 ranges in graphics performance. Will it happen?

Im not so sure... But who knows.

Also: http://www.tomshardware.co.uk/amd-radeon-rx-480-power-measurements,review-33595.html

It should not disappear in wall of information.
 
Last edited:
koyoot said:
About Vega, from what I understand, porting Fiji to 14 nm, would bring the GPU into 150W range of TDP with 1050 MHz core clocks. What would make the GPU faster is solving the problems with scheduler. That alone at 1050MHz would bring the GPU into GTX 1080 ranges in graphics performance. Will it happen?
Click to expand...

At this point I don't believe it. Polaris is a much smaller chip than vega and its currently at 150-165 W. I don't see how AMD is going to increase performance and efficiency using a bigger chip while maintaining the same TDP since Vega is likely just Polaris scaled up with HBM.
 
  • Like
Reactions: tuxon86
Stacc said:
At this point I don't believe it. Polaris is a much smaller chip than vega and its currently at 150-165 W. I don't see how AMD is going to increase performance and efficiency using a bigger chip while maintaining the same TDP since Vega is likely just Polaris scaled up with HBM.
Click to expand...

I'm not sure that Vega is scaled Polaris I don't think AMD would bet it's future on that
 
GPU's scale much better with compute work than games.
that might get fixed with volcan/DX12 maybe :E
 
Stacc said:
At this point I don't believe it. Polaris is a much smaller chip than vega and its currently at 150-165 W. I don't see how AMD is going to increase performance and efficiency using a bigger chip while maintaining the same TDP since Vega is likely just Polaris scaled up with HBM.
Click to expand...
The GPU die is 110W. Board consumes over 50W. 37W GDDR5 consumes, alone. HBM consumes 8W.

I also talk about scaling down the GPU. 14nm Finfet brings 1.9 times better efficiency compared to 28 nm.

Fiji shrinked down to 14 nm 1:1 would be around 150W board power, at 1050 MHz core clock.

P.S. Vega is not the same as Polaris. Polaris from what I understand was designed for 14nm LPP process, to maintain very low cost of the GPUs, and bring the R9 390 levels of performance to masses.

At this point only AMD knows what Vega is, however few things are apparent. If it is 4096 GCN4 chip with HBM2, we can also expect redesigned scheduler. I have heard 90 GFLOPs/watt some time ago as rumored efficiency, but I think much more believable value at this point is 60 GFLOPs/watt.
 
Last edited:
Looks like a good card from what I've seen so far it will be interesting to see if anyone can merge EFI to the rom film at 11 hehe
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back