Overclock.net › Forums › Industry News › Rumors and Unconfirmed Articles › [WCCF] AMD RX 480 Can Hit 1.5Ghz+, New Overclocking Tool With Voltage Control Coming
New Posts  All Forums:Forum Nav:

[WCCF] AMD RX 480 Can Hit 1.5Ghz+, New Overclocking Tool With Voltage Control Coming - Page 217  

post #2161 of 2870
Quote:
Originally Posted by lolerk52 View Post

And that is exactly my point.

NVIDIA has already shown their engineering grunt with Maxwell, meanwhile AMD was stuck with a chip designed to compete with Kepler. Now NVIDIA are staying put in uArch this gen, while AMD has a chance to catch up... And they achieve basically nothing on the uArch front.
In shader efficiency perhaps not, but they seem to have done quite a bit with power consumption. Still, I get what you're saying.
post #2162 of 2870
Quote:
Originally Posted by lolerk52 View Post

And that is exactly my point.

NVIDIA has already shown their engineering grunt with Maxwell, meanwhile AMD was stuck with a chip designed to compete with Kepler. Now NVIDIA are staying put in uArch this gen, while AMD has a chance to catch up... And they achieve basically nothing on the uArch front.

Soo..... since when has it been set in stone that rx-480 is their flagship for this arch? Naming scheme alone says otherwise.

Edit:
Quote:
Originally Posted by Dudewitbow View Post

If anyone is bashing AMD for releasing midrange polaris as its first die into something new, I want to know your opinions when the first Maxwell die was a 750ti.

Go search this very site, the card was 'amazing' and a 'miracle of Nvidia engineering'. Seriously. I even bought one for my spare pc. Yet here comes AMD with a card that follows suit and look at the reaction differences.
The Struggle (4k)
(20 items)
 
File Server
(12 items)
 
Lenovo G50-45
(6 items)
 
CPUGraphicsRAMOS
Athlon II X2 250u Nvidia 6150SE  2gb DDR3 1066mhz Windows 10 Home 64 bit 
CPUGraphicsRAMHard Drive
AMD A8-6410 AMD R5 Crucial Ballistix 8GB DDR3L 1866 CAS10 Crucial BX100 250gb 
Optical DriveOS
DVD Windows 10 Home 
  hide details  
The Struggle (4k)
(20 items)
 
File Server
(12 items)
 
Lenovo G50-45
(6 items)
 
CPUGraphicsRAMOS
Athlon II X2 250u Nvidia 6150SE  2gb DDR3 1066mhz Windows 10 Home 64 bit 
CPUGraphicsRAMHard Drive
AMD A8-6410 AMD R5 Crucial Ballistix 8GB DDR3L 1866 CAS10 Crucial BX100 250gb 
Optical DriveOS
DVD Windows 10 Home 
  hide details  
post #2163 of 2870
Quote:
Originally Posted by lolerk52 View Post

If you look at perf/mm^2, it should have been in Fury X territory in comparison to Pascal. So saying "small die" doesn't save it from an engineering stand point.

The chip actually could - yet. It won't get there on the reference card though. The AIB cards will push higher clocks. That's higher clocks in the same perf / mm2 ratio you mention - same GPU, just not power limited with better cooling. So unlike Pascal where the AIBs apart from reducing noise have been absolute fails at improving clocks - it will actually help the RX480.

There's quite a few leaks that indicate 1500mhz + level of performance in the OC department, couple that with a driver update or two and bingo you are near Fury X performance.
post #2164 of 2870
Quote:
Originally Posted by lolerk52 View Post

And that is exactly my point.

NVIDIA has already shown their engineering grunt with Maxwell, meanwhile AMD was stuck with a chip designed to compete with Kepler. Now NVIDIA are staying put in uArch this gen, while AMD has a chance to catch up... And they achieve basically nothing on the uArch front.

See this post:
Quote:
Originally Posted by magnek View Post

Actually thinking deeper, I'm not sure if these comparisons could capture the full IPC improvements.

980 had 29% less shaders than 780 Ti (2048 vs 2880) but its boost clock was also 31% faster (1216 vs 928). After drivers matured it ended up being ~10% faster than 780 Ti. So you'd think it was only a 10% IPC improvement, but apparently it's supposed to be 35%:



(officially, 1 SMM with 128 shaders is supposed to have 90% the performance of 1 SMX with 192 shaders, assuming equal clockspeeds)

If we take this at face value, it means P10 already has about 25% "silent" IPC improvement from Hawaii. Catch my drift?
post #2165 of 2870
Quote:
Originally Posted by lolerk52 View Post

And that is exactly my point.

NVIDIA has already shown their engineering grunt with Maxwell, meanwhile AMD was stuck with a chip designed to compete with Kepler. Now NVIDIA are staying put in uArch this gen, while AMD has a chance to catch up... And they achieve basically nothing on the uArch front.

You just don't get it. They have yet to release the cards that are going to compete with Nvidia's upper range. This card was never meant to be compared to those flagship cards. You can argue with their decision to release cards from the bottom up this generation if you want to but you can't argue with the fact that the 480 is a bombshell of a card at the $200 price point that has literally no competition at this time. And that's just a fact.
post #2166 of 2870
Quote:
Originally Posted by Majin SSJ Eric View Post

Even though it's a much smaller chip with less shaders and half the memory bandwidth? And ignoring price point and market position completely? OK then... rolleyes.gif
It's a smaller chip because we got a 2x density increase from node. That's not AMD's engineering. And it's less shaders but higher clockspeed, having effectively the same raw performance if not looking at uArch improvements.

As for memory bandwidth, that's an improvement made with Tonga, not a Polaris advancement.
post #2167 of 2870
Quote:
Originally Posted by tkenietz View Post

That's pretty much exactly what I said. Lol

Take or pay means you use it or dont, you pay either way (usually a reduced rate but not much).

That was for 28nm tho
The terms of the "Take or pay" agreement have been renegotiated a couple of times since it was a major problem for AMD.
AMD hasn't been in a disadvantage for years regarding the wafer contracts they have with GloFo.
post #2168 of 2870
Quote:
Originally Posted by blue1512 View Post

It doesn't consume as much power as 390x for the same performance, mate.

If AMD had threw the power efficiency away and fed the card with 390x's power consumption, it would have had 1070 performance. Yeah, it is inferior to Pascal, that is why it is cheap rolleyes.gif
They could have shrunk Hawaii and had a similar chip in perf/mm^2 and perf/watt to the 1070.
post #2169 of 2870
Quote:
Originally Posted by lolerk52 View Post

And that is exactly my point.

NVIDIA has already shown their engineering grunt with Maxwell, meanwhile AMD was stuck with a chip designed to compete with Kepler. Now NVIDIA are staying put in uArch this gen, while AMD has a chance to catch up... And they achieve basically nothing on the uArch front.

It is a marketing strategy to target a different segment. Can't just based it on engineering strategy.
Second Intel Rig
(16 items)
 
  
CPUMotherboardGraphicsRAM
2700 4.5/ 1.28 77 1050 16 / 1866 
Hard DriveCoolingOSMonitor
1000 212 10 64 32 1080 
PowerCase
700 912 
  hide details  
Second Intel Rig
(16 items)
 
  
CPUMotherboardGraphicsRAM
2700 4.5/ 1.28 77 1050 16 / 1866 
Hard DriveCoolingOSMonitor
1000 212 10 64 32 1080 
PowerCase
700 912 
  hide details  
post #2170 of 2870
Quote:
Originally Posted by Dudewitbow View Post

If anyone is bashing AMD for releasing midrange polaris as its first die into something new, I want to know your opinions when the first Maxwell die was a 750ti.

Exactly. By these guys reckoning maxwell should have been declared an immediate fail as soon as the 750 TI released because it didn't beat a 780 TI in performance. I literally can't roll my eyes any harder right now.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Rumors and Unconfirmed Articles
This thread is locked  
Overclock.net › Forums › Industry News › Rumors and Unconfirmed Articles › [WCCF] AMD RX 480 Can Hit 1.5Ghz+, New Overclocking Tool With Voltage Control Coming