• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Official AMD Polaris Review Thread: Radeon RX 480, RX 470, and RX 460

Page 56 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
USER8000 said:
First review for the Sapphire RX480 Nitro+ 4GB:

http://www.pcworld.com/article/3098...laris-rethought-and-refined.html?page=1#toc-1
Click to expand...

Good find. Kinda crazy how much worse the 480 does in VR despite all the boasting about it. This is also interesting:

nitro-480-time-spy-100673126-large.png
 
Shades of 290X all over again. Looks like the RX 480 will get a respectable boost just from not throttling in addition to a small core clockspeed bump.

I imagine if you OC the memory to 2200MHz or higher (should be easy, all 3 of mine do) then you will see another 3-4% uplift.
 
new review "Sapphire Nitro+ RX 480" from ITworld

http://www.itworld.com/article/3098...iew-polaris-rethought-and-refined.html?page=9

Up to 1405Mhz.

Now for the bad news: That epic (for Polaris) bump in clock speed still didn’t result in massive performance increases, though there were slight improvements in the games and benchmarks we tested. Seeing that, Sapphire’s decision to keep the clock speeds lower (and thus, also keep the card cooler, quieter, and drawing less power) seems rational.
Click to expand...
 
My takeaway on the 480 is that the Polaris chipset, at this level, is locked-in. With a few more driver tweaks, it might eak out a 3-5% gain but the ceiling is definitely unmovable or unavoidable. I think the Sapphire team saw this in their own testing when they were looking at how they'd position their card in the market. That's why they went with the Dual-X instead of the Tri-X system for cooling. Any other attempts at more efficient cooler would just be wasted.

So do you get this card, expecting to play more and more DX12 or Vulkan-enabled games in the next two to four years? You don't get 60fps on all games with all the bells and whistles enabled, so do you buy instead the 1060 to get all that on DX 11 games? Or, do you take a pass and hug your 7950 a little bit more and wait for Navi to potentially blow everyone's socks off?

Don't get me wrong, the idea of the 480 @ $200 made nVidia grit their teeth and push out the 1060 at a much lower price point they they had planned, that I'm pretty sure of. And for people with the old Ti's or 6 or 7-series Radeons, it's a big bump in performance and energy savings, so the card is very much welcomed. It's just now becomming apparent that this card will fill a specific niche without any room for growth.
 
boozzer said:
doesn't do much at all. doubt anyone can live with the increase noise.
Click to expand...
He didn't mention thermal throttling temp.. if it's the stock 76C then there is definitely room for tweaking.. like 84C target at 2500rpm instead of 3000rpm.

In either case it's still a pretty sweet card.. for $220.

3 fan cards like PowerColor Devil could be another option if noise is an issue. I have an Extreme Accelero on my rx480 reference.. and it's dead silent.. I also get 1385 Mhz on stock voltage and the card stays around 60C.
 
sirmo said:
He didn't mention thermal throttling temp.. if it's the stock 76C then there is definitely room for tweaking.. like 84C target at 2500rpm instead of 3000rpm.

In either case it's still a pretty sweet card.. for $220.

3 fan cards like PowerColor Devil could be another option if noise is an issue. I have an Extreme Accelero on my rx480 reference.. and it's dead silent.. I also get 1385 Mhz on stock voltage and the card stays around 60C.
Click to expand...
just saying the oc doesn't matter at all. just stick with the sotck nitro. 220$ is pretty sweet for everything he listed in the video.

220$ is super sweet for the performance. man, think about comparing this to the 299$ turd fe.
 
boozzer said:
just saying the oc doesn't matter at all. just stick with the sotck nitro. 220$ is pretty sweet for everything he listed in the video.

220$ is super sweet for the performance. man, think about comparing this to the 299$ turd fe.
Click to expand...
For 1080p the 4gb is great. With driver updates it should do well in the future. I'm undecided if I want to stick with 1080p or move up. But if I do stay with 1080 then I'll get a 4gb model.

Unless of course I find a nice 1440p monitor before I can buy this card.
 
Really glad to see the 4GB AIB card, although the slightly lower OC is a little sad. Still that kind of performance with a solid cooler at $220 is a killer deal.
 
AMD Radeon RX 460 release date on July 28
13781779_1340899105924568_5009479007525825382_n.jpg

The RX 460 has already been tested against the R7 260X on 1080p and it was found that the former performed 1.7x faster than the latter in both MOBA games
 
It is expected to be 70% faster than R7 260X acc to leaked benchamarks. This puts it as fast as GTX760 which is the same perf. as GTX950. I find this a bit unrealistic considering the specs of RX460 but hopefully its true.GTX950 perf. at $99 sounds appealing.
perfrel.gif
 
Silverforce11 said:
70% faster than 260X is above 7950 or 960/280/380 level, previously $199.

Probably in selected games only, because yes, the specs look really low power.
Click to expand...
Great to some of our predictions coming true, though perf/W is still a (minor) concern but 470/460 should be more efficient than the 480.
 
behrouz said:
new review "Sapphire Nitro+ RX 480" from ITworld

http://www.itworld.com/article/3098...iew-polaris-rethought-and-refined.html?page=9

Up to 1405Mhz.

Now for the bad news: That epic (for Polaris) bump in clock speed still didn’t result in massive performance increases, though there were slight improvements in the games and benchmarks we tested. Seeing that, Sapphire’s decision to keep the clock speeds lower (and thus, also keep the card cooler, quieter, and drawing less power) seems rational.
Click to expand...
Click to expand...

I strongly suspect a memory bottleneck. They tested the 4GB version of the Nitro card, and that has GDDR5 running at only 7Gbps, compared to 8Gbps for the 8GB version. This was easy to fix on the reference 4GB cards, which were just 8GB cards with a different BIOS, but it's likely to be an actual physical change on AIB models. Even during overclocking, they said they were only able to push the memory clock an additional 100 MHz (which would equate to 7.4Gbps effective RAM clock). We could see a few percent extra on the 8GB Nitro, especially if the RAM is also pushed to 9Gbps (which the reference cards can usually do).

I think AMD needs to consider a respin/new stepping of Polaris 10, to increase clocks and bring down power consumption.
 
JDG1980 said:
I strongly suspect a memory bottleneck. They tested the 4GB version of the Nitro card, and that has GDDR5 running at only 7Gbps, compared to 8Gbps for the 8GB version. This was easy to fix on the reference 4GB cards, which were just 8GB cards with a different BIOS, but it's likely to be an actual physical change on AIB models. Even during overclocking, they said they were only able to push the memory clock an additional 100 MHz (which would equate to 7.4Gbps effective RAM clock). We could see a few percent extra on the 8GB Nitro, especially if the RAM is also pushed to 9Gbps (which the reference cards can usually do).
Click to expand...

AMD's bandwidth efficiency is a major problem. Nvidia is far more efficient and can provide similar or better perf with a smaller die and much lesser bandwidth. I think Polaris would benefit significantly from a move to G5X but I don't see that happening for atleast a year.

I think AMD needs to consider a respin/new stepping of Polaris 10, to increase clocks and bring down power consumption.
Click to expand...

A second revision is planned for according to the Rx 400 series naming convention. But the GF 14LPP process needs to improve and that might take atleast till Q1 2017.

http://videocardz.com/61721/amd-radeon-rx-400-series-naming-scheme-explained

If GF can improve the process then AMD can get a Rx 485 with 1400 Mhz clocks with slightly reduced power consumption. I think we might see the second revision some time in H1 2017.
 
raghu78 said:
AMD's bandwidth efficiency is a major problem. Nvidia is far more efficient and can provide similar or better perf with a smaller die and much lesser bandwidth. I think Polaris would benefit significantly from a move to G5X but I don't see that happening for atleast a year.
Click to expand...

You really need to bare in mind that it's a $200 GPU before you start saying it needs better anything. There is a whole series of compromises needed to get it to that price point.
 
3DVagabond said:
You really need to bare in mind that it's a $200 GPU before you start saying it needs better anything. There is a whole series of compromises needed to get it to that price point.
Click to expand...

I said that at any price or performance point AMD needs a lot more bandwidth to compete with Nvidia GPUs. Eg: Rx 480 with 256 GB/s bandwidth is still losing to GTX 1060 with 192 GB/s bandwidth which has a smaller die size and 1.3 billion transistors lesser. Nvidia can easily bring other GP106 variants with even lesser bandwidth (say 168 GB/s using 7 Gbps chips) and still beat AMD GPUs like Rx 470 which will have 224 GB/s . AMD's Bill of Materials is always going to be higher than Nvidia as they are far more efficient in terms of perf/transistor, perf/sq mm and perf per GB of bandwidth.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top