Radeon RX 480 "Polaris" Launched at $199

Mauller · 9 Jun 2016 at 12:30

People also have to take into consideration that the polaris series cards could have greater performance than the shader improvements alone in some situations. But we need to find out what the 'Primitive discard accelerator' does first to determine that.

If the 'primitive discard accelerator' is a form of culling for non visible vertices before they hit the geometry engine then it could improve performance in general. If it only effects tesselation then it will be tesselated games that see a bigger improvement.

Lokken86 · 9 Jun 2016 at 12:31

Mauller said:
In the original quote he was talking about a theoretical 480X with 40CU's at 1.4ghz. which is what the Fury-x part is in reference to.

the 480 bit underneath is just me explaining the increased per shader performance giving the part better performance than the 390x, even though its theoretical performance is lower than the 390x.

hence a larger part with higher clocks and more shaders should match the fury x or exceed it, even if its theoretical performance is lower.

That was the answer I was after. I guess clock speed would have more impact on performance since shaders cannot always be utilised fully.

LoadsaMoney · 9 Jun 2016 at 13:28

humbug said:
The 980TI is barley more than 20% faster than the 390X

Lokken86 · 9 Jun 2016 at 14:07

What's funny? He's right.

humbug · 9 Jun 2016 at 14:09

Ignore him he's trolling ^^^

Lokken86 said:
A 2560 shader part@ 1400mhz would still only be 7.1Tflops. Still not even close to a FuryX.

The Fury-X is massively bottlenecked by its front end.

A 390X has 6 TFlops (Even that is somewhat bottlenecked)
A Fury-X has 8.6 TFlops, thats 43% more power, its 21% faster than the 390X.

Polaris 10 is meant to have a much better front end.

Tute · 9 Jun 2016 at 14:15

I actually thought "nah, that's BS, gotta be more than that" but at 4K and 1440p it's actually just under 20%.

[Edit] Sauce: http://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_1080/26.html

Lokken86 · 9 Jun 2016 at 14:18

The 390x is no mug.

bru · 9 Jun 2016 at 14:19

Lokken86 said:
What's funny? He's right.

Silly thing is, using the same website but just a different reviews shows he is completely wrong and that is the MSI version of the 390X against a stock 980ti.

1080p 29% faster
1440p 27% faster
4K 24% faster

But of course we know that we can all find benchmarks to support a position if we look for them hard enough.

https://www.techpowerup.com/reviews/MSI/R9_390X_Gaming/30.html

Tute · 9 Jun 2016 at 14:21

To be fair bru I went for the 1080 review as it's the best chance of getting upto date drivers. I believe at 390X launch the gap was indeed bigger but has closed up a fair bit in the last twelve months.

humbug · 9 Jun 2016 at 14:22

bru said:
Silly thing is, using the same website but just a different reviews shows he is completely wrong and that is the MSI version of the 390X against a stock 980ti.

1080p 29% faster
1440p 27% faster
4K 24% faster

But of course we know that we can all find benchmarks to support a position if we look for them hard enough.

https://www.techpowerup.com/reviews/MSI/R9_390X_Gaming/30.html

Drivers my friend

i used the latest review deliberately, unlike most reviewers TPU retest with each new Driver and AMD's continue to improve.

Lokken86 · 9 Jun 2016 at 14:27

bru said:
Silly thing is, using the same website but just a different reviews shows he is completely wrong and that is the MSI version of the 390X against a stock 980ti.

1080p 29% faster
1440p 27% faster
4K 24% faster

But of course we know that we can all find benchmarks to support a position if we look for them hard enough.

https://www.techpowerup.com/reviews/MSI/R9_390X_Gaming/30.html

Iv said before I'm not a fan of techpowerups performance summarys but if i were to use them it'd be the newest version. They did updated chat with the driver boost AMD got a while back so whatever the difference is its not as big as it was then.

tommybhoy · 9 Jun 2016 at 14:28

bru said:
Silly thing is

Your trying to pass off year old performance as todays.

bru · 9 Jun 2016 at 14:34

Drivers, ok Ill agree that things will have changed since last year.

I was going to use this review.

July 3rd 2015
https://www.techpowerup.com/reviews/MSI/GTX_980_Ti_Gaming/30.html

1440p 980ti 100% 390X 68% somebody else can do the maths but it is definitely a good bit more than 20% faster. but as you say drivers and that review is also quite old.

So here is a more modern one.

Mar 31st 2016
https://www.techpowerup.com/reviews/Gigabyte/GTX_980_Ti_XtremeGaming/23.html

1440p 980ti 100% 390X 66% again not doing the maths but definitely way more than 20%.

It just goes to show that you can find benchmarks to support just about any position.

Mauller · 9 Jun 2016 at 14:37

The difference becomes smaller when gameworks heavy games are not taken into account.

humbug · 9 Jun 2016 at 14:38

bru said:
Drivers, ok Ill agree that things will have changed since last year.

I was going to use this review.

July 3rd 2015
https://www.techpowerup.com/reviews/MSI/GTX_980_Ti_Gaming/30.html

1440p 980ti 100% 390X 68% somebody else can do the maths but it is definitely a good bit more than 20% faster. but as you say drivers and that review is also quite old.

So here is a more modern one.

Mar 31st 2016
https://www.techpowerup.com/reviews/Gigabyte/GTX_980_Ti_XtremeGaming/23.html

1440p 980ti 100% 390X 66% again not doing the maths but definitely way more than 20%.

It just goes to show that you can find benchmarks to support just about any position.

Still older drivers than i used.

16.1 vs 16.4.2 in the ones i used.

All we can take from that is AMD got another performance boost from 16.1 to 16.4.

The performance difference is what it is today.

Nexus18 · 9 Jun 2016 at 14:39

Hasn't he also posted numerous times before about a 980 matching or/and beating a fury x???

humbug · 9 Jun 2016 at 14:45

Nexus18 said:
Hasn't he also posted numerous times before about a 980 matching or/and beating a fury x???

Me? probably.

The thing is the 390/X Fury/X have more power than their drivers imply.
There is more performance in them to be unlocked, AMD have been working on unlocking that performance for 2 years, and it does keep creeping up, relentlessly.

Nexus18 · 9 Jun 2016 at 14:45

humbug said:
Me? probably.

The thing is the 390/X Fury/X have more power than their drivers imply.
There is more performance in them to be unlocked, AMD have been working on unlocking that performance for 2 years, and it does keep creeping up, relentlessly.

No, loadofsmoney.

LoadsaMoney · 9 Jun 2016 at 15:00

Nexus18 said:
Hasn't he also posted numerous times before about a 980 matching or/and beating a fury x???

Yes i have, as it was when the Fury X was released, its not now though.

Lokken86 · 9 Jun 2016 at 15:01

bru said:
Drivers, ok Ill agree that things will have changed since last year.

I was going to use this review.

July 3rd 2015
https://www.techpowerup.com/reviews/MSI/GTX_980_Ti_Gaming/30.html

1440p 980ti 100% 390X 68% somebody else can do the maths but it is definitely a good bit more than 20% faster. but as you say drivers and that review is also quite old.

So here is a more modern one.

Mar 31st 2016
https://www.techpowerup.com/reviews/Gigabyte/GTX_980_Ti_XtremeGaming/23.html

1440p 980ti 100% 390X 66% again not doing the maths but definitely way more than 20%.

It just goes to show that you can find benchmarks to support just about any position. [/

Funny I would never have guessed the difference grows when you use the extreme gaming oc version.

Normal 980ti is <20% there.