AMD Polaris Thread: Radeon RX 480, RX 470 & RX 460 launching June 29th

[Shivansps]

A 480 would be almost 4 times as many shaders, presumably improved in terms of efficiency/performance and running at a higher clock speed. No idea where the 470 will come in at, but I suspect it will have a similar price (~$160) as the 7770 at launch.

I don't think I've had that big of a performance jump since I got an 8800 back in the day and it was staggering, like going from dial-up to broadband.

mid-range cards are stuck in performance for like 4 years now, for example R7 360 is a rebranded HD7790... R7 370 rebranded HD7850...

So yeah, its kinda the first jump in performance in a long time.

 

[Silverforce11]

mid-range cards are stuck in performance for like 4 years now, for example R7 360 is a rebranded HD7790... R7 370 rebranded HD7850...

So yeah, its kinda the first jump in performance in a long time.

I keep seeing people confuse mainstream with mid-range.

There are distinct tiers of pricing for GPUs as well as chip size within a generation.

Typically;

$89 to $149 = Entry (ie. 750Ti, 950, 360, 370)
$149 to $249 = Mainstream (ie. 960, 380/X)
$299 to $399 = Mid-range (ie. 970, 390/X)
$400 to $499 = High-end (980, Fury, Nano)
>$500 = Enthusiast

The price for these tiers are creeping up, in particular the high-end and enthusiast tier.

Chipwise:

GM200 = Enthusiast & High-end (cut variant)
GM204 = High-end (980) & Mid-range (970)
GM206 = Mainstream
GM207 = Entry

AMD likes to label their tiers as Entry -> Mainstream -> Performance -> Enthusiast.

Cards around $200 are certainly not mid-range anymore but squarely mainstream. They used to be mid-range, like the 560Ti and 660Ti, where they used x06 Fermi/Kepler chips but sold at a much lower price.

NV has changed their positioning of cards, with mid-range chips being sold at high-end prices so it's confusing people.

In terms of Polaris 10, it's a mainstream chip, 232mm2, low power. GP104 is a traditional mid-range chip in size & power.

Polaris 10 should not reach GP104 performance, if it does, it means there's either a failure on Pascal or an excellent architecture for AMD.

 

[Despoiler]

This is enthusiasm from ebay:
https://youtu.be/H6Ni2gPo5Us

That is a thing of beauty!

 

[Mopetar]

mid-range cards are stuck in performance for like 4 years now, for example R7 360 is a rebranded HD7790... R7 370 rebranded HD7850...

So yeah, its kinda the first jump in performance in a long time.

Both AMD and NV being stuck on 28 nm for so long didn't leave a lot of room for improvement in the middle. However, the 370 has 60% more shaders and double the memory of a 7770, so it's not as though there was no improvement at all.

Pitcairn was also a solid chip, so it's not too surprising that AMD got a lot of mileage out of it. It aged surprisingly well and has had rather good driver support.

 

[Shivansps]

Both AMD and NV being stuck on 28 nm for so long didn't leave a lot of room for improvement in the middle. However, the 370 has 60% more shaders and double the memory of a 7770, so it's not as though there was no improvement at all.

Pitcairn was also a solid chip, so it's not too surprising that AMD got a lot of mileage out of it. It aged surprisingly well and has had rather good driver support.

The 360 and 370 are still the exact same cards of the HD7000 series you could have brought back then, althought if i remember well the HD7790 was the last card to be launched. Serving in the same sector at almost the same price (HD7850 was $250 back then, the R7 370 is about $170 now, for a +4 year old card), so we where litterally stuck on performance for several years now.

And Nvidia did almost nothing on that sector as well.

 

[Mopetar]

The 360 and 370 are still the exact same cards of the HD7000 series you could have brought back then

The 7770 wasn't Pitcairn though, only the 7850 and 7870 were. Going from Pitcairn to Polaris is still going to be a considerable upgrade in terms of performance, but replacing a 7770 with a 470 (480 is really in the class above so it's more fair to use a 470 for comparison) is going to be a fairly staggering performance increase. Conservatively it's 200% more shaders operating at 15% clock rates with a 10% performance gain.

If it comes out better than that it's not much of a stretch to see almost quadruple the performance relative to a 7770, so it's going to be a big step up. Some of that does have to do with the fact that it's a 4 year old card being replaced, but it really doesn't matter because the improvement is going to be very noticeable just like anyone who's going from a 680 to a 1080.

 

[monstercameron]

Hype train is now blasting off, we're going to space bois!

 

[IllogicalGlory]

The RX 480 will have the same performance an OC headroom as the 390 and 80% of Pascal's performance/watt. It'll be a good product, not incredible, but good (TPU gives it an 8.9), but a major engineering loss for AMD. They'll have to work overtime on Vega to even compete with GP104. Maybe they'll pull off another Hawaii - losing in performance/mm^2 with Tahiti, then pulling ahead decisively, but that's unlikely to happen.

I'll be very unhappy on the 29th when when we see this and the AMD bashers are vindicated. Unfortunately, that's been the trend, and the leaks aren't giving me too much hope.

But hey, at least my expectations are low, and there's plenty of room for me to be surprised.

 

[Yakk]

A 480 or 470 will be a blast then. Something to look forward to.
Until then enjoy tf2 or bf3.
Enthusiasm is not about the speed of the gfx.
... But oc is ! Lol.

This is enthusiasm from ebay:
https://youtu.be/H6Ni2gPo5Us

Awesome, and strangely appropriate. LoooooL...

 

[R0H1T]

The RX 480 will have the same performance an OC headroom as the 390 and 80% of Pascal's performance/watt. It'll be a good product, not incredible, but good (TPU gives it an 8.9), but a major engineering loss for AMD. They'll have to work overtime on Vega to even compete with GP104. Maybe they'll pull off another Hawaii - losing in performance/mm^2 with Tahiti, then pulling ahead decisively, but that's unlikely to happen.

I'll be very unhappy on the 29th when when we see this and the AMD bashers are vindicated. Unfortunately, that's been the trend, and the leaks aren't giving me too much hope.

But hey, at least my expectations are low, and there's plenty of room for me to be surprised.

Citation needed :hmm:

Personally I think GCN4 will be within 5~10% of Pascal's efficiency in gaming perhaps even beating it, yet blowing them away in terms of compute efficiency. Also Toms had an efficiency test with the most efficient AMD at the time, Nano, & it was around that range. I think the TDP numbers from Nvidia make it seem like they're untouchable, so far as efficiency is concerned, but that's not all there's to that story!

 

[Vesku]

The RX 480 will have the same performance an OC headroom as the 390 and 80% of Pascal's performance/watt. It'll be a good product, not incredible, but good (TPU gives it an 8.9), but a major engineering loss for AMD. They'll have to work overtime on Vega to even compete with GP104.

TPU 480 review is out?

A staffer with OCUK, which has received its 480 launch day product, did say 480 is a nice card but don't expect it to match 980 ti. They didn't clarify if they meant OC vs stock. If so it fits with 390ish at stock clocks.

 

[dark zero]

The RX 480 will have the same performance an OC headroom as the 390 and 80% of Pascal's performance/watt. It'll be a good product, not incredible, but good (TPU gives it an 8.9), but a major engineering loss for AMD. They'll have to work overtime on Vega to even compete with GP104. Maybe they'll pull off another Hawaii - losing in performance/mm^2 with Tahiti, then pulling ahead decisively, but that's unlikely to happen.

I'll be very unhappy on the 29th when when we see this and the AMD bashers are vindicated. Unfortunately, that's been the trend, and the leaks aren't giving me too much hope.

But hey, at least my expectations are low, and there's plenty of room for me to be surprised.

Major engineering loss from AMD to be NEAR GTX 1070? considering that the numering tier is more going to be 2 tiers lower than the competition?
I mean... the 480 was intended to fight toe by toe against the 1060... but is going NEAR the 1070.

Is no way to be a defeat. Not a major win, but not at defeat after all.

 

[trane]

TPU 480 review is out?

A staffer with OCUK, which has received its 480 launch day product, did say 480 is a nice card but don't expect it to match 980 ti. They didn't clarify if they meant OC vs stock. If so it fits with 390ish at stock clocks.

He later posted in the thread that he hasn't run any benchmarks. Just plugged it in to see how loud it was, and presumably going by industry sources for performance. Pretty sure he meant stock vs stock in that case.

 

[JDG1980]

The RX 480 will have the same performance an OC headroom as the 390 and 80% of Pascal's performance/watt.

I'm expecting closer to R9 390X performance at stock, though admittedly that's a fairly small difference. At the default 1266 MHz boost clock, it should have almost as much raw power (TFlops) as a 390X, so even discounting any major architectural improvements since GCN 1.1, it should be as good as 390X at a minimum.

Overclocking headroom should be better, since the 390X is basically a factory OC'd card to begin with. Hawaii usually couldn't overclock higher than 1100-1150 MHz (less than 10% over 390X), while we've already seen leaks indicating that P10 should be able to hit 1500 MHz without much trouble (~18% over stock RX 480). In practice, I expect that to translate to about 10%-12% additional performance.

Also, a lot of people are placing too much stock in 3DMark scores. While it's not insanely vendor-biased (nothing like, say, Project CARS in that regard), it does tend to lean a few percentage points towards Nvidia. R9 390X beats GTX 980 at 1080p in TPU's combined real-world gaming benchmarks, while it falls behind by a couple percent in 3DMark. Perhaps more importantly, this is just a single number, while a major real-world consideration is minimum FPS. Based on the Hitman demo and AMD's mention of a "primitive discard accelerator", there's some reason to hope that minimum FPS might see improvements on Polaris. That means a better gaming experience at given settings even if average FPS is similar to older products. I'm interested to see what TechReport has to say about frame latencies on this card (once they work through their review backlog).

As for perf/watt, I think it will come reasonably close to GP104 (GTX 1070/1080) and beat GP106 when that comes out. GTX 1080 peaks at roughly 185W during gaming. R9 390X is about 60% of 1080's performance, so if we assume RX 480 also lands there, then it would need peak gaming power of 111W or lower to match 1080's perf/watt. Reference R9 270X only peaked at 122W in gaming, and with this chip being similarly sized and having reasonably conservative stock clocks compared to what the silicon is capable of, I would not be surprised to see similar results. This would mean probably about 90% of GTX 1080's perf/watt.

It'll be a good product, not incredible, but good (TPU gives it an 8.9), but a major engineering loss for AMD. They'll have to work overtime on Vega to even compete with GP104. Maybe they'll pull off another Hawaii - losing in performance/mm^2 with Tahiti, then pulling ahead decisively, but that's unlikely to happen.

That's an oddly specific prediction about the TPU score. Just to be sure I double-checked their site and found nothing. Even tried manually formulating a URL to see if they maybe had it posted but just not linked, but got a 404.

I don't see how P10 is an engineering failure. Similar die size and power consumption to Pitcairn, with roughly 2x the performance in both compute and gaming, plus a full complement of modern video processing features. What's the problem? If GP106 is half a GP104 (the most likely scenario, IMO) then P10 should beat GP106 in both performance and perf/watt.

I'll be very unhappy on the 29th when when we see this and the AMD bashers are vindicated. Unfortunately, that's been the trend, and the leaks aren't giving me too much hope.

Expecting that P10 would match 1070 was always a stretch. At this point it's fairly clear that isn't happening, at least not at this time. AMD never promised any such thing.

 

[JDG1980]

TPU 480 review is out?

A staffer with OCUK, which has received its 480 launch day product, did say 480 is a nice card but don't expect it to match 980 ti. They didn't clarify if they meant OC vs stock. If so it fits with 390ish at stock clocks.

GTX 980 Ti beats R9 390X by about 25% at 1080p. If RX 480 matches 390X performance at stock, then that's a lot of ground to make up. There are apparently going to be 1500 MHz AIB cards, but that's only an 18% overclock, which could be as little as 10-12% in actual performance improvement.

That said, all the above numbers (from TPU) are based on average FPS. It's possible that the Polaris architecture will do better on minimum FPS, as the Hitman demo seemed to indicate. In other words, the actual gaming experience might be closer to GTX 980 Ti than the numbers alone would indicate.

 

[Xellos2099]

How much better would a 480 be compare to 7870 XT?

 

[StrangerGuy]

AMD Poland also just confirmed that reference RX 480 8GB has a $229 USD MSRP.

My local retailer is already accepting preorders for the Sapphire RX 480 8GB at exactly half the going price of a 1070 FE.

 

[sirmo]

How much better would a 480 be compare to 7870 XT?

rx480 should be twice as powerful at least.

 

[Xellos2099]

rx480 should be twice as powerful at least.

That's nice. Now need to look into real life bench to see if 480 can handle 1080p 60fps.

 

[trane]

That's nice. Now need to look into real life bench to see if 480 can handle 1080p 60fps.

That's a given (barring outliers). A better question is if 470 can handle 1080/60.

 

[Armsdealer]

It'll be a good product, not incredible, but good

Probably a good time to blow out of the stock. The hype train is indeed in overdrive relative to what you're describing

 

[redzo]

... I am one of those who think the cards get to much vram. One can always lower texture a bit and it will solve the perf issue.
...

Quite the opposite if you ask me. Texture quality is the first setting that I want maxed out. It impacts image quality too much. I'd rather lower something else. I can't stand blurry textures.

 

[antihelten]

Probably a good time to blow out of the stock. The hype train is indeed in overdrive relative to what you're describing

I don't really get all these claims of the hype being in overdrive.

Right now the rumours seem to be solidifying around 390X performance at stock, 110W power usage in games and decent overclocking (20-25% to 1500-1600 MHz from 1266 MHz stock).

All of this is about as modest as you can get. With 2304 and a 1266 MHz clock (5.833 TFLOPS) and zero architectural improvements, performance should be equal to a 390X (2816 shaders at 1050 MHz, 5.913 TFLOPS). 110W power usage puts it at about 90% the efficiency of GP104 (RX 480 would have 60% the performance of GP104, with 66% of the power usage). 20-25% overclocking is decent but not amazing (certainly not GM200 level, which could do 30-35%).

 

[Silverforce11]

I don't really get all these claims of the hype being in overdrive.

Right now the rumours seem to be solidifying around 390X performance at stock, 110W power usage in games and decent overclocking (20-25% to 1500-1600 MHz from 1266 MHz stock).

All of this is about as modest as you can get. With 2304 and a 1266 MHz clock (5.833 TFLOPS) and zero architectural improvements, performance should be equal to a 390X (2816 shaders at 1050 MHz, 5.913 TFLOPS). 110W power usage puts it at about 90% the efficiency of GP104 (RX 480 would have 60% the performance of GP104, with 66% of the power usage). 20-25% overclocking is decent but not amazing (certainly not GM200 level, which could do 30-35%).

Correction.

GM200 also did 20-25%.

35% is a ridiculous level that people who don't understand how boost clocks and OC % work.

Example.

GTX 980, in-game boost clocks on stock = 1,265mhz.

http://www.anandtech.com/show/8526/nvidia-geforce-gtx-980-review/22

If you have a good sample and get 1,500mhz, what's the OC %? ~19%, not this mythical 30-35%.

The same applies to the 980Ti, which even reference cards on stock often boost to 1,200mhz or above. A good OC to 1,500mhz is a 25% OC.

Maxwell is a great overclocker, 20-25% is really good for GPUs.

Both Pitcairn and Tahiti are also excellent over-clockers, with 25-50% (yes, 7950 and 7850 had a 50% OC potential, 800mhz to 1200mhz!)...

Even Hawaii isn't so shabby.

R290 947mhz base, can get to 1.2ghz. Likewise, R290X 1ghz base can reach 1.2ghz.

Only Fiji really sucked, with OC potential that's similar to GP104 actually! Peak of around 10% performance gained.

 

[coercitiv]

Right now the rumours seem to be solidifying around 390X performance at stock, 110W power usage in games and decent overclocking (20-25% to 1500-1600 MHz from 1266 MHz stock).

The unhype train wants 150W typical board power, 390 performance, less than 10% overclock: mostly a mediocre Hawaii die shrink on a weak 14nm process from GF. Looking at the hype train from this perspective, it does seem supersonic.