I don't understand how anyone hasn't heard of the news, but the UK voted to leave the EU.
This is off-topic, so I'll just leave it at that.
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
I don't understand how anyone hasn't heard of the news, but the UK voted to leave the EU.
This is off-topic, so I'll just leave it at that.
seems RX480 is shipping..
seems Fury support in Sierra can be enabled:
citing netkas (http://netkas.org/?p=1456):
"Just add device id (0x73001002) into Baffins section (AMDBaffinGraphicsAccelerator) in /System/Library/Extensions/AMDRadeonX4000.kext/Contents/Info.plist"
can you test adding Polaris device id to X4000.kext in similar to see if you get full OpenGL/OpenCL acceleration?
seems Fury support in Sierra can be enabled:
citing netkas (http://netkas.org/?p=1456):
"Just add device id (0x73001002) into Baffins section (AMDBaffinGraphicsAccelerator) in /System/Library/Extensions/AMDRadeonX4000.kext/Contents/Info.plist"
can you test adding Polaris device id to X4000.kext in similar to see if you get full OpenGL/OpenCL acceleration?
Hey guys ! I have a RX 480 in hand, tried on my Mac Pro 4,1 (patched 5,1), and the good news is that is starts and don’t give KP, it’s a good start. But bad news is that the fan were spinning super fast and 3D acceleration wasn’t working.
Card was recognized at “AMD Radeon R9 xxx 8192 MB”.
Maybe that’s where the problem is ? Didn’t recognize it was RX but recognized R9 ?
Unless AMD decided to change the name of the Polaris 10 cards, because I saw some places were also thinking the name would be R9 something
Anyway, 10.12 is in early beta still, and card isn’t released yet, so that’s why I said it’s a good sign… we might have working driver when 10.12 comes !
NB : photo is of the card is the one I shot
It's pseudo Fury support. The test scores are very very slow compared to true support.
It's running now in cMP with improved performance: http://forum.netkas.org/index.php/topic,11566.msg33282.html#msg33282
I think the performance isn't too bad. PCs with a R9 Nano running Valley @ Extreme HD achieve approx. 50% more FPS:
http://extreme.pcgameshardware.de/a...ranking-unigine-valley-valleybench1.0_new.png
http://extreme.pcgameshardware.de/a...738-ranking-unigine-valley-valleybench1.0.png
Considering the old CPUs (and the GPU running @ x4) this is okay, I guess. Would like to see a benchmark from a more modern Hack. My Skylake rig with GTX 780 isn't too far away from those PC R9 Nano benchmarks:
Yeah it scores about 4000 on a PC but I think it's with DX.It's running now in cMP with improved performance: http://forum.netkas.org/index.php/topic,11566.msg33282.html#msg33282
I think the performance isn't too bad. PCs with a R9 Nano running Valley @ Extreme HD achieve approx. 50% more FPS:
http://extreme.pcgameshardware.de/a...ranking-unigine-valley-valleybench1.0_new.png
http://extreme.pcgameshardware.de/a...738-ranking-unigine-valley-valleybench1.0.png
Considering the old CPUs (and the GPU running @ x4) this is okay, I guess. Would like to see a benchmark from a more modern Hack. My Skylake rig with GTX 780 isn't too far away from those PC R9 Nano benchmarks:
The benchmark you cited in the other thread (where it achieved ~4000) was without AA, if I remember it correctly.
Looks like AMD is working the weekend on a RX 480 fix.
http://www.anandtech.com/show/10465/amd-releases-statement-on-radeon-rx-480-power-consumption
http://www.anandtech.com/show/10465/amd-releases-statement-on-radeon-rx-480-power-consumption
Wouldn't affect third party cards with better designs. They should have launched with 8 pin.
[doublepost=1467456221][/doublepost]
Can't remember but
Can't remember but I think since we are talking about Sierra's better Metal update then we should finally come to the point when we never have to talk about macOS's **** OpenGL. Unigene won't update their tests so it's best to wait for apps that directly compare Metal to DX12.
Maybe they can fix it with a driver update.
I wonder though why they released the card with only a 6pin connector. If they knew, and they must have, that the card is already borderline on the max 150W combined PCIe+6pin, and they tout the card as highly OC'able (which is not the case with that config), this just leads to discredit.
Which is a shame since it appears that the power delivery system is quite beefy, much better than 1080 even. So why this? Maybe to let the OC capabilities to the AIB partners with different coolers but it also hurts AMD's image in the process.
It's also funny how the GloFo process seems (I'm not saying it is) inferior to TSMC's in terms of OC potential, even being shorter (14 vs 16nm). 10x0 can OC like hell and start off in much higher freq. It could the architectural differences that make AMD's cards slower.
On another note, I'm puzzled with the GDDR5X issue, nowhere to be seen. Will it be in 490 or at all? Does Polaris even support GDDR5X?
I also can't see any concrete info of DP hardware. HP support is there, AMD stated so. But DP is not mentioned at all, is there dedicated hardware and how much of it? Maybe 490 will be 480 better binned to achieve higher clocks, have GDDR5X at 10Gbps+ and have higher DP throughput enabled (in the drivers maybe). Or that will be left for FirePro only.
Just a few thoughts...
I wonder though why they released the card with only a 6pin connector. If they knew, and they must have, that the card is already borderline on the max 150W combined PCIe+6pin, and they tout the card as highly OC'able (which is not the case with that config), this just leads to discredit.
Which is a shame since it appears that the power delivery system is quite beefy, much better than 1080 even. So why this? Maybe to let the OC capabilities to the AIB partners with different coolers but it also hurts AMD's image in the process.
It's also funny how the GloFo process seems (I'm not saying it is) inferior to TSMC's in terms of OC potential, even being shorter (14 vs 16nm). 10x0 can OC like hell and start off in much higher freq. It could the architectural differences that make AMD's cards slower.
On another note, I'm puzzled with the GDDR5X issue, nowhere to be seen. Will it be in 490 or at all? Does Polaris even support GDDR5X?
I also can't see any concrete info of DP hardware. HP support is there, AMD stated so. But DP is not mentioned at all, is there dedicated hardware and how much of it? Maybe 490 will be 480 better binned to achieve higher clocks, have GDDR5X at 10Gbps+ and have higher DP throughput enabled (in the drivers maybe). Or that will be left for FirePro only.
Just a few thoughts...
Answer is simple. For OCability, of the GPU you need additional power connectors.
Yeah, but using an 8pin or dual 6pin OOB wouldn't go with the image of a power efficient cards, would it? I believe that was the reason AMD used the 6pin instead, for marketing reasons alone. And that can now get them in trouble. If they had used an 8pin no issues whatsoever.
But they wanted to market it as a low power or efficient card.
And after all this time the DVD and multi-monitor power draw are still horrible. They need to learn fom NVidia how to manage those clocks.
But they wanted to market it as a low power or efficient card.
And after all this time the DVD and multi-monitor power draw are still horrible. They need to learn fom NVidia how to manage those clocks.
If this was just a marketing stunt then they screwed themselves. I find it hard to believe they want to ruin their credibility.
I hope to use one of these in the fall for a build so I have time to wait but people that have them right now are probably a little leary if their MB is going to burn.
I'm still fond of the days when 3DfX add in card cost 100 bucks and made a world of difference
)
And I still remember times where 199$(2008 year to be precise) GPU cost in my country 800 PLN with all the taxes, whereas today 229$ GPU costs 1399 PLN with taxes.I'm still fond of the days when 3DfX add in card cost 100 bucks and made a world of difference
So yeah, times change
.
Once the NDA was lifted, it's pretty obvious why an nnMP with Polaris didn't debut at Bill Graham early in June. "Hot dog" can be either a sausage in a bun, or a GPU that uses a lot of power for lackluster performance.
And I'm sure that the cute guy in the video didn't mean to warn you about "older PCIe 2.0" boards - the warning is about any board that doesn't handle a GPU drawing more than the maximum spec'd PCIe power - regardless of age or PCIe version. (To be honest, I would pay extra for a mobo that shutdown if a card misbehaved.)
And, within a week or so, we should expect to see drivers and/or firmware that throttles the GPU to keep it within the PCIe power spec (although what that does to FPS (at least two interpretations of the TLA are OK) benchmarks is unknown).
GPU that did not had any other connection to power other than PCIex. 141W maximum power draw from PCIex. And it is reference model.
Stop making fuss about no-problem just because it is AMD branded.
Average power draw for the Nvidia 750ti was 64 watts - what's your point?
Spikes are normal and expected. What a card draws for 1 msec is mostly irrelevant to what it draws for 500 msec.
That's why we've more or less moved away from fuses (that tend to trip on spikes) to circuit breakers (that tend to trip on average draw).
The 750ti didn't generate a bunch of web reports of system shutdowns and burned out motherboards. Polaris 10 is generating those reports. HUGE DIFFERENCE. Nobody worried about a Maxwell frying their motherboard. A lot of people are worried that a Polaris might fry their motherboard.
AMD has a real problem to address here - which most likely will be done by software changes to lower the performance of the Polaris cards.
AMD will fix it - but the real point of my post was to highlight why we didn't see Apple announce an nnMP with Polaris a month ago. It's a hot dog.
Last edited:
2 reports out of 38 english reviews.
PCPer did 30 minute film on the topic.
And yes, spikes are normal.