Radeon RX 480 "Polaris" Launched at $199

[TheRealDeal]

Agred But now Kaapstad is saying it needs a 40% overlock to match a 980ti and hence a80% one to match a 1070

On that basis even two 480s wouldnt beat a 1070.

Yea Kaap is overestimating the gap between a 390x and the 980ti at stock speeds.

 

[Kaapstad]

How? In thosescharts a 390x is only 18% slower than a 980ti. I know you dont get linear gains but I would have thought a 20% or more overclock would put it in the ball park.

A 980 Ti is also about 35% to 40% faster than a 980.

You can use the heaven bench table I posted earlier to check.

 

[BluD]

Yeah but Kaapstad is hardcore Nvidia

lol Kaapstad hardcore nvidia?, hes hardcore everything. Hes owned every card on the planet 5 times over.

 

[Jono8]

How? In thosescharts a 390x is only 18% slower than a 980ti. I know you dont get linear gains but I would have thought a 20% or more overclock would put it in the ball park.

Not this again. We did this weeks ago and took many averages from three separate reviews including DX12 games etc and the 980Ti was still 30% faster across the board compared with a 390x.


https://forums.overclockers.co.uk/showpost.php?p=29602465&postcount=1741

https://forums.overclockers.co.uk/showpost.php?p=29599918&postcount=1671

 

[Mauller]

something else is that techpowerups charts are also distorted by some gameworks heavy games.

 

[Kaapstad]

Yea Kaap is overestimating the gap between a 390x and the 980ti at stock speeds.

What are stock speeds for either of them, most people use non reference for both.

 

[TheRealDeal]

A 980 Ti is also about 35% to 40% faster than a 980.

You can use the heaven bench table I posted earlier to check.

I would rather not use that as the benchmark as it's mainly down to tessellation performance. Why not use the tpu chart i put up as it's game testing. The gap in games is no where near as big especially in dx11.

 

[TheRealDeal]

What are stock speeds for either of them, most people use non reference for both.

Stock speed is the reference model that was reviewed at launch. That's what people are referring to. Nobody in there right mind was expecting the RX480 to reach 980ti oc models and the likes.

 

[Kaapstad]

Yeah but Kaapstad is hardcore Nvidia, literally any sniff of news which swings that way he'd use the extreme end of any figure to bend the argument his way.

To be honest mate i'd just ignore it and wait, we'll all find out soon what the true performance is. Lets all hope AMD can pull it out of the bag and add a bit of competition to the discrete GPU space. It sorely needs it.

Think you need to look at that Heaven 4 bench table again, I have a number of AMD entries on it including the highest score for the red team lol.

 

[D.P.]

something else is that techpowerups charts are also distorted by some gameworks heavy games.

They are also distorted by things like AotS and Hitman.A game will always favor a particular architecture over another.

 

[Kaapstad]

Stock speed is the reference model that was reviewed at launch. That's what people are referring to. Nobody in there right mind was expecting the RX480 to reach 980ti oc models and the likes.

Could you post a link to a reference 390X please.

 

[sprite-]

Think you need to look at that Heaven 4 bench table again, I have a number of AMD entries on it including the highest score for the red team lol.

I buy Coca Cola often, doesn't mean i don't prefer Pepsi

 

[TheRealDeal]

Think you need to look at that Heaven 4 bench table again, I have a number of AMD entries on it including the highest score for the red team lol.

Kaap why are you going on about Heaven. It's one of AMD's weakest bench marks and it makes the 980ti/Titan X look stronger than they are against the lower model 9 series cards.

 

[Mauller]

They are also distorted by things like AotS and Hitman.A game will always favor a particular architecture over another.

far more game-works games in comparison.

 

[Kaapstad]

I would rather not use that as the benchmark as it's mainly down to tessellation performance. Why not use the tpu chart i put up as it's game testing. The gap in games is no where near as big especially in dx11.

You can use it as Tessellation does not matter when comparing a 980 and 980 Ti.

 

[TheRealDeal]

Could you post a link to a reference 390X please.

I can post plenty running stock speeds.

 

[TheRealDeal]

You can use it as Tessellation does not matter when comparing a 980 and 980 Ti.

It clearly does and i would rather base performance on loads of game benchmarks than one that's not really comparitive to games.

 

[Jono8]

something else is that techpowerups charts are also distorted by some gameworks heavy games.

Hahahahaha

Sorry this made me laugh because I had this debate with Humbug ages ago and his argument was the exact opposite - he was using techpowerup because it was the only graph that came close to matching his claim that a 980ti was only 20% faster than a 390x, and criticising others for being gameworks biased).

I then ran the numbers on tomshardware and guru 3d's latest benchmarks featuring both cards and the average of both of them put the 980Ti as 33.5% faster across the board.

Just going to post them again -

https://forums.overclockers.co.uk/showpost.php?p=29602465&postcount=1741

https://forums.overclockers.co.uk/showpost.php?p=29599918&postcount=1671

 

[Greebo]

You can use it as Tessellation does not matter when comparing a 980 and 980 Ti.

Well i for one never knew there was a 40% performance game between a 980 and a 980ti.

Those 980 were massively overpriced and surprised anybody bought one.

 

[Kaapstad]

Kaap why are you going on about Heaven. It's one of AMD's weakest bench marks and it makes the 980ti/Titan X look stronger than they are against the lower model 9 series cards.

Because people were going on about the performance of NVidia cards. To show the difference between NV cards it is the best bench to use as it is totally about GPU grunt and type of CPU hardly matters.