AMD Polaris architecture – GCN 4.0

[Davedree]

no they couldn't match with PB; but zen with 40% ipc over Ex; it will put it at least Haswell if not possibly a little higher.

Unified platform also will help with FX chips using the same mb as Zen APUs....

that's the platform I'll upgrade to

You gotta take the Amd hype with a pinch of salt, afterall the 40% higher ipc rumour is based on mobile excavator which doesn't scale in performance after 3.6ghz and neither does it run at it's intended clocks due to oems hampering it with a 15tdp (<2.1 ghz).
Piledriver and Steamroller provides more throughput after 3.6 to 5ghz but of course is inefficient compared to intel.

Putting the cpu extensions aside Llano k10.5 is still the highest Ipc cpu/apu Amd have made to date.

 

[CAT-THE-FIFTH]

You gotta take the Amd hype with a pinch of salt, afterall the 40% higher ipc rumour is based on mobile excavator which doesn't scale in performance after 3.6ghz and neither does it run at it's intended clocks due to oems hampering it with a 15tdp (<2.1 ghz).
Piledriver and Steamroller provides more throughput after 3.6 to 5ghz but of course is inefficient compared to intel.

Putting the cpu extensions aside Llano k10.5 is still the highest Ipc cpu/apu Amd have made to date.

IPC has nothing to do with clockspeed. Plus Bristol Ridge is being released for desktop with Excavator cores. Plus the 40% IPC statement is from an AMD slide. There is a good cumulative 25% to 35% IPC increase from Bulldozer to Excavator,so it probably isn't far off K10.

 

[flopper]

So why did Carrizo have the updated Uvd engine and Vce and hdmi2.0 support?

they had to redesign fury to make that happen so they didnt.
simple really as they already had HBM world leading technology memory on the fury. Nano world leading small card power envelope performance,small is bigger and better.

a brighter future simply

 

[Freddie1980]

I do now AMD has made RTG it's own division that can choose it's own design philosophy they will now give us video card that will challenge the top end products that have been occupied by Nvidia since the 8800 GTX. Ever since AMD bought ATI there top end single GPU card has always been a fair way off Nvidia top offering, which I is now about to change.

 

[Orangey]

Ever since AMD bought ATI there top end single GPU card has always been a fair way off Nvidia top offering

This is a myth.

 

[Poneros]

What do you expect, people have short memories.

 

[humbug]

I thought the 290X was what forced Nvidia to bring out the 780TI after the GTX Titan, and even that was only marginally faster.

The Fury-X is just as fast as the 980Ti @ 1440P, faster at 4K

 

[CAT-THE-FIFTH]

This is a myth.

Yep. He seems to have forgotten the HD5870 and the HD7970 and HD7970GE.

 

[artur_k]

I thought the 290X was what forced Nvidia to bring out the 780TI after the GTX Titan, and even that was only marginally faster.

The Fury-X is just as fast as the 980Ti @ 1440P, faster at 4K

And by now 290x left 780ti in the dust.

The same will be with FuryX/980ti come DX12.

 

[P.B]

But even if it matches Haswell, by the time they release them they will be 2 or 3 gens behind ? And that's if it is as they say it is.

 

[humbug]

But even if it matches Haswell, by the time they release them they will be 2 or 3 gens behind ? And that's if it is as they say it is.

Wrong room, Haswell is a CPU, this is GPU's

 

[Davedree]

IPC has nothing to do with clockspeed. Plus Bristol Ridge is being released for desktop with Excavator cores. Plus the 40% IPC statement is from an AMD slide. There is a good cumulative 25% to 35% IPC increase from Bulldozer to Excavator,so it probably isn't far off K10.

Lol, seriously you are telling me like I have not the foggiest. Carrizo on it's current low performance process is tuned for mobile, Ipc and clockspeed and headroom are very relevant if you wish to port this design onto desktop 65-95w (Bristol ridge). Also if benchmarks are going to be made upon performance of a throttling Excavator Apu, then that affects the comparsion being made to their ipc gains.
They cannot simply use the gf28lp for desktop, if they port it to 14nm then awesome, but I imagine they'll use the 28shp which to be honest is pointless.

The Stilt (a well know guru) proved that Carrizo loses it scaling and efficiency at higher frequencies.

'' Like I said before, these numbers are not perfectly accurate. Still, based on the numbers it seems that Carrizo cannot maintain it´s power efficiency at higher frequencies. Most likely the reason lies in the differences between the manufacturing process versions used for these designs. At low frequencies Carrizo is definitely more power efficient than Steamroller designs, however at ~2600MHz the two designs are already even. At frequencies higher than that Steamroller designs are more power efficient rolleyes.gif''


There is no magic in Carrizo: The design itself isn´t more power efficient at hardware level than the previous generations were. Carrizo simply has much more advanced, effective and most importantly finally functional power management. In fact at hardware level Carrizo is less power efficient than the previous generations due the changes in the design and the manufacturing process. At higher frequencies (>3200MHz) Kaveri / Godavari can easily outperform Carrizo in power efficiency.

http://www.overclock.net/t/1560230/jagatreview-hands-on-amd-fx-8800p-carrizo/520

 

[FoxEye]

a brighter future simply

Ugh, every time some regurgitates marketing spiel, I die a little inside.

Can we dispense with the "brighter future" nonsense, please It literally means nothing at all (like 99% of marketing utterances), and nobody was using that phrase before AMD's Polaris reveal.

 

[Davedree]

they had to redesign fury to make that happen so they didnt.
simple really as they already had HBM world leading technology memory on the fury. Nano world leading small card power envelope performance,small is bigger and better.

a brighter future simply

Traditionally Amd Apu's used an older gen gpu core from their latest , until the stagnation in their cpu department and the delay in process problems, Carrizo has a more advanced Gpu core than Tonga/Fury. So if like you said Amd held back the Display I/o in the fury as their next gen wasn't far away, it only shows how late and how short lived the Fury will be.

 

[humbug]

Davedree, utterly irrelevant, you can't compare Excavator to Zen, they are not the same CPU.

Carrizo has nothing what so ever to do with Zen.

Also THIS IS THE GPU ROOM!!!!!!!

 

[Davedree]

I thought the 290X was what forced Nvidia to bring out the 780TI after the GTX Titan, and even that was only marginally faster.

The Fury-X is just as fast as the 980Ti @ 1440P, faster at 4K

The Fury-x is also marginally faster than a 290x/390x at 1080p


You say direct x 11 overhead is the cause, I say front end scheduling problem like I predicted

Afterall the Gcn 4.0 slides mentions improvements with
Hardware Scheduler
Improved shader efficiency
Instrucion prefetch

 

[Davedree]

Davedree, utterly irrelevant, you can't compare Excavator to Zen, they are not the same CPU.

Carrizo has nothing what so ever to do with Zen.

Also THIS IS THE GPU ROOM!!!!!!!

Yes it has if you are basing 40% ipc gain of Zen vs Carrizo.
Carrizo is Excavator.
Don't blame me I didn't bring up zen in the thread, I'm just stating facts and stopping people getting amd hyped.

 

[humbug]

Perf comparisons only, has nothing in conmen in clocks and scaling, nothing. ^^^^

The Fury-x is also marginally faster than a 290x/390x at 1080p


You say direct x 11 overhead is the cause, I say front end scheduling problem like I predicted

Afterall the Gcn 4.0 slides mentions improvements with
Hardware Scheduler
Improved shader efficiency
Instrucion prefetch

Front end scheduling doesn't explain the huge performance scaling difference between 1080P to 1440P to 4K.

DX11 Overheads do, perfectly.

 

[Davedree]

Front end scheduling doesn't explain the huge performance scaling difference between 1080P to 1440P to 4K.

DX11 Overheads do, perfectly.

So why does 3584 4096 cores barely provide gains in comparsion to a 2816 Hawaii Grenada at 1080p?

 

[humbug]

So why does 3584 4096 cores barely provide gains in comparsion to a 2816 Hawaii Grenada at 1080p?

Draw Call Thoughput bottlenecking, at 1080P the CPU is doing more of the work, the higher the res goes the more of that work is off loaded to the GPU, thats when the true power of the GPU architecture comes in to play, where its not bottlenecked by a combination of a crap API and its own Drivers.