• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD Polaris architecture – GCN 4.0

small pcb huge cooler beats me, i hated that on fury pro, i can't believe they did again with the 480, i honestly hope that they do like 300series, and let only AIBs custom to retail or review samples, because i dont trust a cooler designed for 300 series to be any good for polaris, might get the job done at some extent, but will probably be as bad cooler as the 1080's or 290's

If AMD produced a blower cooler which was any good,I think that would be a bigger revelation than the performance!! :p:D
 
Not only Apple but possibly Sony and Microsoft as well; although Apple had sickeningly massive discounts when it came to those D700 FirePro cards on their 2013 Mac Pro line.

I do hope AMD saying they're launching cards from $100-300 means we get a 480X as well that's got a healthy increase in Stream processors, and possibly clocks as well.

Even so the 480 does look like a great performance to cost card, but AMD does need more.

So y and Microsoft are not using a discrete GoU but a custom APU again.
 
Ok, i updated it, i didn't think it would make a noticeable difference, seems there is a small difference.

So again...

Gigabyte WF OC GTX 970

Stock = 1178/1750
Overclock @ 1500/1950 = 27% Overclock

RX 480
Stock = 1266/2000



RX 480


GTX 970: stock


GTX 970: 1500/1950
 
So y and Microsoft are not using a discrete GoU but a custom APU again.

I am actually wondering about that.


I am pretty sure I read Sony is sticking with 28nm jaguar cores albeit at an increased clock rate.

It might be that they get separated and a seperate GPU / CPU is used?, I have no idea however. Cost wise it might make more sense for now due to the 28 - 14nm shrink?

I am probably completely wrong though.
 
So y and Microsoft are not using a discrete GoU but a custom APU again.

It seems to be that the PS4 Neo has 2304 shaders too. It makes me wonder how much of the console development and Polaris development overlapped.

There is also a hint by Raja Koduri about the PS4 Neo. He said during his speech,that the RX480 would be the standard for VR experiences for the next three to four years.

This makes me wonder whether he is hinting at the PS4K having a Polaris 10 derived GPU.
 
The great sales of the gtx970 were nothing to do with the gtx980. It was more what people were getting for the money.

It's purely because most people don't research at all and simply buy "the latest 2nd best nvidia". If it's also popular on forums among more educated buyers that's another matter.

They used to buy the 60 card but now NV has herded them away from that, made it low-end and taught them to buy the 70 instead.
 
I am actually wondering about that.


I am pretty sure I read Sony is sticking with 28nm jaguar cores albeit at an increased clock rate.

It might be that they get separated and a seperate GPU / CPU is used?, I have no idea however. Cost wise it might make more sense for now due to the 28 - 14nm shrink?

i think AMD will do what they are planning for Navi, put multiple chips on interposer for sony's custom soc.
 
I am actually wondering about that.


I am pretty sure I read Sony is sticking with 28nm jaguar cores albeit at an increased clock rate.

It might be that they get separated and a seperate GPU / CPU is used?, I have no idea however. Cost wise it might make more sense for now due to the 28 - 14nm shrink?

I am probably completely wrong though.

There are two SKUs which look to be launched:
1.)PS4 Slim
2.)PS4 Neo

PS4 Neo would be a shrink to 14NM as the CPU cores are clocked higher and the GPU is more powerful.

The PS4 Slim might be a shrink too,or maybe a refined 28NM SOC.
 
Ok, i updated it, i didn't think it would make a noticeable difference, seems there is a small difference.

So again...

Gigabyte WF OC GTX 970

Stock = 1178/1750
Overclock @ 1500/1950 = 27% Overclock

RX 480
Stock = 1266/2000



RX 480


GTX 970: stock


GTX 970: 1500/1950

I was just looking at fury scores for the same API version and settings, they range from 46 - 60fps (ignoring the 41fps using an 8320 cpu), probably nano - fury-x since it doesn't seem to differentiate which score is for which card. Nor the clocks of the card, the top one is more than likely an overclocked furyx.

But the RX480 scores you were comparing against are also a month old, using an older version of the game and more than likely on less mature drivers to what they used for the event.
 
I was just looking at fury scores for the same API version and settings, they range from 46 - 60fps (ignoring the 41fps using an 8320 cpu), probably nano - fury-x since it doesn't seem to differentiate which score is for which card. Nor the clocks of the card, the top one is more than likely an overclocked furyx.

But the RX480 scores you were comparing against are also a month old, using an older version of the game and more than likely on less mature drivers to what they used for the event.

Keep's bench thread is different settings, i can run the 970 stock at those settings as a scaling measure if you like?

Also, don't forget, as this is now its a $200 390X/980.
 
Keep's bench thread is different settings, i can run the 970 stock at those settings as a scaling measure if you like?

Also, don't forget, as this is now its a $200 390X/980.

I was making my comparison to benches on the AoTS benchmark site, not this forum.
 
Its been pick out of AoTS submission servers. ^^^ thats not thin air.

I was making my comparison to benches on the AoTS benchmark site, not this forum.

Ok, well i have run it now....

The RX 480 is 20% faster so in Kaaps thread it would be 29.3.

 
We dont know, because Humbug's 40fps figure has been plucked out of thin air

His 40 fps figure comes from this older benchmark of the part. the ones used in the demo do not appear on the benchmark archive as they were more than likely using a special offline version.

Note that this is also on an older version of the benchmark compared to what Humbug run his 970 on.

gfhxdfh_zps1frw5lak.jpg
 
Back
Top Bottom