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

[RussianSensation]

For the sake of argument. Let's assume the RX480 ~ 390x /GTX 980

Based on the data above, if the RX480 is able to clock upward of 1500mhz, that would make it roughly ~15% faster if the scaling is similar to previous GCN. If you add 15% to the 60% from the overclock, that would give it ~69%. . Compared to the 73% performance index of the GTX980TI, that would make the RX480 @1500mhz ~6% slower than a stock GTX980TI; or 94% of the performance of the GTX980TI at 1440p. Not bad if at all if it can actually overclock to 1500mhz.

That's a very reasonable observation. Also, with what seems like AMD now moving to an NV-supplied blower fan, the reference card should be much quieter than the older generation HD7970->R9 290X cards that not only used an inferior fan but also used much more power than the RX 480 will.

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

"Translation:
Suggested (retail) price is 229 USD (for 8GB model, 199 for 4GB). The price in Polish zloty will be determined closer to the launch, to avoid differences in exchange rates."

http://videocardz.com/61262/amd-radeon-rx-480-8gb-to-cost-229-usd

The absolute cheapest GTX 1070 in the U.S. is $419.99.
http://www.nowinstock.net/computers/videocards/nvidia/gtx1070/

Even if RX 480 were to retail for $249, instead of $229.99, GTX1070 would need to be 68% faster to have a similar price/performance ratio. People continue using paper spec $379.99 MSRP for 1070 that has nothing to do with reality right now.

Using the targeted 1080p 60Hz gaming resolution that RX 480 is aimed at, GTX1070 is 43% faster than R9 390X. The 1070 is nowhere near RX 480's price/performance.

That means GTX1070 already lost the key price/performance metric that is so important for mainstream/performance gamers.

A lot of gamers who have a backlog of DX11 games may also consider just buying an RX 480 as a stop-gap and once they upgrade their monitor to 1440p 60-144Hz or 4K or start playing more demanding DX12 games in 2017, they can just buy Big Vega/Pascal. That's another reason why the stop-gap RX 480 is also attractive. Another option for mainstream/performance gamers is just to buy a $230-250 RX 480 and sell it in 2 years and just purchase another $250 GPU in 2018. As I stated before, GTX1070 is a great 1440p 60Hz card but as soon as RX 480 8GB can be purchased for $229-249, a $420 GTX 1070 is grossly overpriced for 1080p 60Hz gaming.

 

[Silverforce11]

For the sake of argument. Let's assume the RX480 ~ 390x /GTX 980

Based on the data above, if the RX480 is able to clock upward of 1500mhz, that would make it roughly ~15% faster if the scaling is similar to previous GCN. If you add 15% to the 60% from the overclock, that would give it ~69%. . Compared to the 73% performance index of the GTX980TI, that would make the RX480 @1500mhz ~6% slower than a stock GTX980TI; or 94% of the performance of the GTX980TI at 1440p. Not bad if at all if it can actually overclock to 1500mhz.

~1.5ghz is close to 20% OC, though performance never scales perfectly. So 15% more is about right.

As I said, if RX 480 matches a 390X, it's still a great product because of price, power and heat profile. If it's fast than that, it's a bonus icing on the cake.

Imagine starting with 390X + 10%, or Nano/Fury... that would make it one heck of a disruptive GPU.

 

[3DVagabond]

I still don't get why any sites test ROTTR in DX12. Its slower than DX11 and AMD especially has huge performance hit. They don't use any advanced DX12 techniques, they even removed async compute used on the XB1 version.

Hmmm? What would motivate the sites to use RotTR? :sneaky:

People do things for a reason. Look at who it benefits and you'll have your reason.

 

[Silverforce11]

That means GTX1070 already lost the key price/performance metric that is so important for mainstream/performance gamers.

A lot of gamers who have a backlog of DX11 games may also consider just buying an RX 480 as a stop-gap and once they upgrade their monitor to 1440p 60-144Hz or 4K or start playing more demanding DX12 games in 2017, they can just buy Big Vega/Pascal. That's another reason why the stop-gap RX 480 is also attractive. Another option for mainstream/performance gamers is just to buy a $230-250 RX 480 and sell it in 2 years and just purchase another $250 GPU in 2018. As I stated before, GTX1070 is a great 1440p 60Hz card but as soon as RX 480 8GB can be purchased for $229-249, a $420 GTX 1070 is grossly overpriced for 1080p 60Hz gaming.

This is why NV have MSRP and "Early Access" pricing schemes.

In the month before Polaris is launched, their cards are sold at inflated prices due to zero competition.

After Polaris launch, expect them to drop down to the MSRP prices naturally.

At $379, even if the $299 RX 480 matches the 1070 (maybe, maybe not), the $79 extra is justified as a premium NV tax and it will still sell well.

At similar performance and other metrics, NV GPUs can be priced higher, that's just market reality. If NV is slightly faster, they can charge massively higher.

 

[lukart]

Sorry if it was to sarcastic.
As i actually brought the issue of availability prior to the table pics I will be more specific:
20 cards on a table doesnt indicate anything about stock
...even if its 5 pictures of the same cards taken from 5 different angles or stacked on top of each other.
I think we can all agree on that yet we need 5 pictures of 20 similar cards.
Missing cards in stock is a serious issue if you want a card or the resulting skyrocketing prices with very low stock. The vendors can take some flak for that or they can set expectations right from day one eg saying when there will be stock or what the real price is.
Imo.

There will be plenty of 8GB cards selling, but not the 4GB card.
The 4GB is like 1/10th of the amount of 8GB cards that will be available on the first day.

Will be funny to see everyone thinking the 4GB is sold out because its the most popular (and probably will be) but its not because it sold more. :sneaky:

 

[DiogoDX]

Hmmm? What would motivate the sites to use RotTR? :sneaky:

People do things for a reason. Look at who it benefits and you'll have your reason.

TPU added Tomb Raider dx12 in their test suit but still bench hitman in dx11.:thumbsup:

 

[krumme]

There will be plenty of 8GB cards selling, but not the 4GB card.
The 4GB is like 1/10th of the amount of 8GB cards that will be available on the first day.

Will be funny to see everyone thinking the 4GB is sold out because its the most popular (and probably will be) but its not because it sold more. :sneaky:

I once bought a 1 gig 7850 for one of the kids. Its long gone but it proved to be fantastic value. 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.
Now i have no doubt the 8 gig 480 will sell like hot cakes and far more than the 4 gig. Solely on the reason of the ram amount and the good marketing specs it gets for that. If you keep it for 2 plus years its probably far safer but 4 gig is ... well like a 980. Its imo damn solid for 200 usd.

 

[Silverforce11]

TPU added Tomb Raider dx12 in their test suit but still bench hitman in dx11.:thumbsup:

They also do open bench testing, no long warm up period. Best case scenario for NV's boost and their reference 1070 and 1080 have inflated numbers compared to sites that do proper testing.

The gap between custom model 1070/1080 vs reference is bigger than TPU's results, because the custom models hold their boost clocks in real gaming (in a case, longer duration).

 

[3DVagabond]

TPU added Tomb Raider dx12 in their test suit but still bench hitman in dx11.:thumbsup:

Thanks. I haven't actually read their reviews in earnest for a while now.

After many years of reading it became painfully obvious they selected their bench suite to skew the results. Not to give us accurate results.

 

[3DVagabond]

They also do open bench testing, no long warm up period. Best case scenario for NV's boost and their reference 1070 and 1080 have inflated numbers compared to sites that do proper testing.

The gap between custom model 1070/1080 vs reference is bigger than TPU's results, because the custom models hold their boost clocks in real gaming (in a case, longer duration).

It's shocking how much the 1080 throttles. After ~15 minutes it's at it's base clock. TPU runs ~2 minute benches though.

 

[parvadomus]

I cant wait for the 480s reviews to see what AMD did with memory compression. Im afraid 480s might be bottlenecked with that 256-bit bus if they didnt deliver with this.
I am almost 100% sure we might see a 485 or something with GDDR5X later that truly enters the fury x/980ti/1070 territory.

 

[sze5003]

I'm wondering how much better AIB versions will be and when they will release. Almost makes me want to try and grab an 8gb 480 on the 29th just because I can. But at the same time I don't want to miss out on something better.

Then again if I get a 1440p monitor I'm going to worry about how it will handle it. I've been spoiled with my current card just cranking highest settings an go.

 

[antihelten]

It seems that Hitman shows the potential more than RotTR. We'll see. I just think it's a better bet that GCN will gain more performance than Pascal will in DX12. YMMV.

Besides, all I was showing is that claiming a 40% difference by comparing to the 390X the way the post did wasn't necessarily very accurate.

Or maybe it's a question of Hitman simply being broken for Nvidia, rather than showing the potential of AMD, who knows.

Right now these are the DX12 games out there:

AotS - slightly broken on Nvidia (drivers not rendering certain shaders properly).
Hitman - Episode 1 is fine, Episode 2 is broken (not rendering textures properly).
RotTR - Broken on both AMD and Nvidia (worse performance than DX11).
Quantum Break - Broken on Nvidia (they haven't fixed this yet, have they?)
GoW - was originally broken on AMD, but later fixed.

Basically DX12 has been a mess until now, so I would be very careful about taking any one game and proclaiming it to be a better model for future DX12 games over any of the other DX12 currently available.

I still don't get why any sites test ROTTR in DX12. Its slower than DX11 and AMD especially has huge performance hit. They don't use any advanced DX12 techniques, they even removed async compute used on the XB1 version.

Yeah it's a mess, I wonder if they (crystal dynamics) even intend to fix it at this point

 

[monstercameron]

Hmmm? What would motivate the sites to use RotTR? :sneaky:

People do things for a reason. Look at who it benefits and you'll have your reason.

Hanlons razor?

 

[96Firebird]

So should I pencil in TPU as an Nvidia-biased review site yet, or hold off until minds change after the RX 480 review?

 

[Silverforce11]

It's shocking how much the 1080 throttles. After ~15 minutes it's at it's base clock. TPU runs ~2 minute benches though.

Yeah but at this point it's not really of concern because custom 1070 and 1080 are available.

The real comparison for RX480 will be against custom 1070s, and the performance gap is a big one.

So $229 8GB vs $379 to $399 8GB 1070s (currently ~$449) will be just fine for NV.

I don't believe TPU is bias, it just so happens the way the test with open bench and short runs gives an advantage over GPUs that throttle clocks over time due to temps/TDP.

 

[Sonikku]

I once bought a 1 gig 7850 for one of the kids. Its long gone but it proved to be fantastic value.

Man, I feel like such a scrub still using a 7770. :'( I just keep telling myself, just one more week...

 

[Silverforce11]

Right now these are the DX12 games out there:

AotS - slightly broken on Nvidia (drivers not rendering certain shaders properly).
Hitman - Episode 1 is fine, Episode 2 is broken (not rendering textures properly).
RotTR - Broken on both AMD and Nvidia (worse performance than DX11).
Quantum Break - Broken on Nvidia (they haven't fixed this yet, have they?)
GoW - was originally broken on AMD, but later fixed.

Basically DX12 has been a mess until now, so I would be very careful about taking any one game and proclaiming it to be a better model for future DX12 games over any of the other DX12 currently available.

Yeah it's a mess, I wonder if they (crystal dynamics) even intend to fix it at this point

Some of those games had issues early on but in the patches that came, they are all fine except for RotTR DX12 which is still worse than DX11. Crystal Dynamics has no intention of fixing it, they've probably already moved onto the next game project.

Hitman, ep 3 now, they've improved it's performance with each episode. It's not actually a finished game, more like an early access thing, despite that, it runs quite well today.

Ashes runs fine, the shader bug is only on the NV press drivers. Public drivers work. Performance is also very good for the insane unit counts in multiplayer, as an example of DX12 advantage for RTS games.

QB had issues with vsync, patched now. You could argue the performance is low, but their dynamic lighting engine is the most impressive. Clearly they built their game engine around GCN, even though this was a neutral non-sponsored title.

GOW runs fine these days.

Forza DX12 also runs fine on both vendors, great performance all round and very optimized.

It's more a reflection of games these days, they need patches post release.

I am looking forward to the DX12 TW: Warhammer update, I bought the game already and it's been great fun but in really big battles, the CPU is still bottlenecked.

The next big titles are Deus Ex MD, Watch Dogs 2 and Battlefield 1.

There's only a handful of major AAA titles each year that are cutting edge on graphics and these titles are the ones that make our expensive GPUs shine. Most other titles are small indie ones where GPU performance is almost a non-factor. The trend is a shift to DX12/Vulkan for big AAA titles lately and it should increase in the future.

 

[DarkKnightDude]

These guys got a few to play with

Is that 17 way crossfire? :sneaky:

Maybe those are 470s on the left side? They look different from each other unless its just lighting.

 

[3DVagabond]

So should I pencil in TPU as an Nvidia-biased review site yet, or hold off until minds change after the RX 480 review?

Do whatever you'd like. Isn't that what you usually do?

 

[3DVagabond]

Yeah but at this point it's not really of concern because custom 1070 and 1080 are available.

Well not on pcpartpicker or Newegg they aren't. Newegg has one 1070 and it's a founders edition ($449+shipping)

 

[krumme]

Man, I feel like such a scrub still using a 7770. :'( I just keep telling myself, just one more week...

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

 

[guskline]

Nine days until the release. I'm interested in how the RX480 will actually perform once the tests results are up.

 

[Shivansps]

Ill still have my hand on the eyection seat just in case, the hype train is going supersonic by now.

 

[Mopetar]

A 480 or 470 will be a blast then. Something to look forward to.

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.