AMD RX 400 series (Polaris) MegaThread! FAQ & Resources

[IAEInferno]

I see thanks, I mainly do gaming so I had no Idea that it works like that.

 

[littleleo]

What would be the use of 32gb of Vram for?

For anything you want to use it for.

 

[Martell1977]

Lol i also see opposite things happen. Remember how people say fermi is hot and going to increased your power bill? Back then many said efficiency is vert important. But when the tables got turned around we see people claiming that power consumption does not matter only real performance does. In short it happens on both sides actually.

That kind of thing is expected as things change. What I was talking about is listing certain specs and delivering a different product or naming 2 different parts essentially the same thing. One is false advertisement and the other is just shady and makes the less informed think they are getting the same GPU but with less memory, as has always been done in the past. Maybe they should have done it as 1060 and 1060 ti, that would have been more honest IMO.

 

[Martell1977]

What would be the use of 32gb of Vram for?

I was thinking that if it's indeed a gaming card, maybe AMD believes that VR will take off and high resolution VR experiences will require a large VRAM pool.

 

[manleysteele]

What would be the use of 32gb of Vram for?

I was thinking that if it's indeed a gaming card, maybe AMD believes that VR will take off and high resolution VR experiences will require a large VRAM pool.

Makes sense to me.

 

[renz496]

Lol i also see opposite things happen. Remember how people say fermi is hot and going to increased your power bill? Back then many said efficiency is vert important. But when the tables got turned around we see people claiming that power consumption does not matter only real performance does. In short it happens on both sides actually.

That kind of thing is expected as things change. What I was talking about is listing certain specs and delivering a different product or naming 2 different parts essentially the same thing. One is false advertisement and the other is just shady and makes the less informed think they are getting the same GPU but with less memory, as has always been done in the past. Maybe they should have done it as 1060 and 1060 ti, that would have been more honest IMO.

personally i do think 1060 3GB should have different naming. but if that "3GB" also included in the name instead just spec then they still can say the name is different than "GTX1060 6GB" because the other was "GTX1060 3GB". this remind me of GTX460 256bit 1GB and GTX460 192bit 768MB. both card still GTX460. but back then people simple refer to GTX460 256bit or GTX460 192bit to differentiate the two. and no people complaining about it at all. with 1060 the memory interface still the same (192 bit) but the core config is different. rather than calling them "GTX1060 1280 Core Edition" and "GTX1060 1152 Core Edition" it is much simpler to call them as 6GB version and 3GB version?

 

[renz496]

I was thinking that if it's indeed a gaming card, maybe AMD believes that VR will take off and high resolution VR experiences will require a large VRAM pool.

so that's why Fury X isn't doing so well in VR vs 980Ti? well in any case AMD need to double or triple their effort on VR because they talk a lot about VR.

 

[Sakkura]

I was thinking that if it's indeed a gaming card, maybe AMD believes that VR will take off and high resolution VR experiences will require a large VRAM pool.

so that's why Fury X isn't doing so well in VR vs 980Ti? well in any case AMD need to double or triple their effort on VR because they talk a lot about VR.

It's definitely not the reason with current VR headsets, as their VRAM requirements are low. But if new headsets come out with much higher resolution, that could change.

In any case, there's no reliable benchmark data for VR yet, so I don't know on what basis you're saying the Fury X is doing poorly in VR compared to the 980 Ti.

 

[renz496]

I was thinking that if it's indeed a gaming card, maybe AMD believes that VR will take off and high resolution VR experiences will require a large VRAM pool.

so that's why Fury X isn't doing so well in VR vs 980Ti? well in any case AMD need to double or triple their effort on VR because they talk a lot about VR.

It's definitely not the reason with current VR headsets, as their VRAM requirements are low. But if new headsets come out with much higher resolution, that could change.

In any case, there's no reliable benchmark data for VR yet, so I don't know on what basis you're saying the Fury X is doing poorly in VR compared to the 980 Ti.

hey i never say Fury X perform "poorly" vs 980ti. just said it is not doing as well as 980ti and pointing that 980ti advantage probably lies on that extra VRAM. hardocp made several test on VR lately.

 

[Martell1977]

hey i never say Fury X perform "poorly" vs 980ti. just said it is not doing as well as 980ti and pointing that 980ti advantage probably lies on that extra VRAM. hardocp made several test on VR lately.

Considering the possible time frame for the "Vega20", I was just speculating that besides being for workstations, AMD might be trying to use some "Forward thinking" and/or think that a truly fluid and steady VR experience will require a LOT of VRAM.

The Fury X does pretty good, the main complaint I've seen in the Fury X vs. 980ti debate has been the lack meaningful overclocking on the Fury X. I haven't seen the review you speak of, but will look it up.

 

[Sakkura]

HardOCP is using SteamVR to report performance. While it's probably one of the best options right now, I would not put much faith in it. SteamVR in general is not the most amazing piece of software, and performance measurement doesn't seem to be a strong point of Valve's right now.

 

[renz496]

Just like how game being review tool will improve over time. In the past most review onlyclook at frame rates now we also have frame time. Hardocp has been long talking about in game smoothness in their review long before TR come up with their 'inside the seconds' concept. And jjst back then with this VR test they also explain how it really feels when playing with VR with both of hardware. Looking at the results so far i don't think there is anything wrong with AMD hardware. but as usual amd need to focus more on their drivers.

 

[Sakkura]

Meh, Nvidia has had more driver problems with VR than AMD.

 

[renz496]

Meh, Nvidia has had more driver problems with VR than AMD.

problem in what manner? if you were thinking about those issue when nvidia cards cannot detect VR headset it's already been fixed. and so far the test shown by hardocp it is amd that need to work on their driver more.

 

[TehPenguin]

Maybe it's fixed now but it was an issue. Let's not get too fanboy-ish in here and admit errors on both sides.

 

[Sakkura]

Meh, Nvidia has had more driver problems with VR than AMD.

problem in what manner? if you were thinking about those issue when nvidia cards cannot detect VR headset it's already been fixed. and so far the test shown by hardocp it is amd that need to work on their driver more.

Those problems are fact. The performance aspect is highly speculative as it's not based on good data.

 

[renz496]

Meh, Nvidia has had more driver problems with VR than AMD.

problem in what manner? if you were thinking about those issue when nvidia cards cannot detect VR headset it's already been fixed. and so far the test shown by hardocp it is amd that need to work on their driver more.

Those problems are fact. The performance aspect is highly speculative as it's not based on good data.

it is not speculative when some of the problem even visible to the naked eye. the data was just there to explain why such issue happen. anyway i will stop here. i agree with tehpenguin that there are issues on both side. but is seems to you readily accept that there is issue on nvidia side but take it that the issues on AMD just as "speculative" by people that actually experienced using the device themselves.

 

[BurgerandChips66]

https://youtu.be/wstG14DmZVk

Interesting example of what water-cooling can do with the reference RX 480.

It managed 1390MHz without any additional voltage.

With extra voltage it passed Firestrike at 1460MHz.

All the while temps were not going over 39/40 degrees and power about 150 watts or so.

Using a custom bios, I'd expect increasing TDP + more voltage would get it over 1500MHz. Maybe an 8-pin card could do more.

 

[Sakkura]

it is not speculative when some of the problem even visible to the naked eye. the data was just there to explain why such issue happen. anyway i will stop here. i agree with tehpenguin that there are issues on both side. but is seems to you readily accept that there is issue on nvidia side but take it that the issues on AMD just as "speculative" by people that actually experienced using the device themselves.

Because it is speculative and subjective stuff that might very well not be real. There's no reliable info one way or the other. Only a fanboy would equate that with solidly established facts.

 

[renz496]

https://youtu.be/wstG14DmZVk

Interesting example of what water-cooling can do with the reference RX 480.

It managed 1390MHz without any additional voltage.

With extra voltage it passed Firestrike at 1460MHz.

All the while temps were not going over 39/40 degrees and power about 150 watts or so.

Using a custom bios, I'd expect increasing TDP + more voltage would get it over 1500MHz. Maybe an 8-pin card could do more.

this is probably how it is goes with GCN in general. if you can cool down the card properly you can also reduce the power leakage and in turn will make the card more power efficient. this is also one of the reason why AMD equipping Fury X with AIO.

One other power optimization in the Fury X really isn't a GCN improvement, but it helps explain why Fiji is able to run at ~1GHz with less board power than the 290X. It has to do with that liquid cooler. AMD has cited operating temperatures around 52C for the GPU on this card, and operating at such low temperatures tamps down on leakage power in pretty dramatic fashion. The transistors on a warmer chip will leak more and thus require more power. By cooling Fiji aggressively, the Fury X likely saves a non-trivial amount of wattage that would otherwise be wasted. This fact is noteworthy in this context because it suggests the power-oriented improvements in Fury X aren't all related to more efficient GPU architecture per se.

http://techreport.com/review/28499/amd-radeon-fury-x-architecture-revealed/2

 

[BraveVN]

I'm so confuse about buying those 2 cards: RX 480 & GTX 1060.
Should I get GTX 1060 now and upgrade to another card after 2 years of using ?
Or should I buy RX 480 for better price but less performance & get another RX 480 for Crossfire in next 2 years ?

 

[cdrkf]

I'd say hey the 480, performance is looking stronger than the 1060 in Vulcan and dx12, and it's fast enough in dx11.

Crossfire is an option as well long term, although multi gpu setups can be a pain if nothing else the fact it's crossfire compatible should help it's resale value imo.

 

[-Fran-]

I will check it first hand today when my monitor arrives to see if its really a thing to consider, but...

I'd also say Freesync is a nice thing to have with AMD cards. The premium for monitors with GSync is *at least* USD$60 for the same monitor equipped with it. So, you might get more in terms of brute FPS'es going with a 1060, but if you intend to pair it with a GSync monitor, the price premium goes even higher. That being said, there is no "value" with any Freesync or GSync monitor and they are well above the cheapest monitors you can get anyway (I agree with that sentiment from Tom's reviews), but like I said, if you plan on getting a better monitor down the road it is something to consider.

And since I love being redundant, I'll check if Freesync is something to consider tonight (I hope) and post back my totally subjective impressions on it.

Cheers!

 

[BraveVN]

thanks guys, I will think about it

 

[renz496]

I'm so confuse about buying those 2 cards: RX 480 & GTX 1060.
Should I get GTX 1060 now and upgrade to another card after 2 years of using ?
Or should I buy RX 480 for better price but less performance & get another RX 480 for Crossfire in next 2 years ?

look at the game you will be playing and decide. and i dare to bet even if you end up with RX480 you will not going to get another RX480 two years from now. it is a different story if you intended to get that second RX480 within 3-6 months from now.