SemiAccurate Forums  

 
Go Back   SemiAccurate Forums > Main Category > GPUs

GPUs Talk about graphics, cards, chips and technologies

Reply
 
Thread Tools Display Modes
  #391  
Old 05-01-2012, 06:18 PM
distinctively's Avatar
distinctively distinctively is offline
2^11
 
Join Date: Sep 2009
Location: Waterloo, Ontario
Posts: 2,517
distinctively will become famous soon enoughdistinctively will become famous soon enough
Default

Quote:
Originally Posted by Thrillseeker View Post
Ok, I thought leakage was more a function of process than of design. What can they change on the design that would significantly affect leakage, and yet have about the same chip performance?
Leakage can be from both process and design. As far as what they can change? That's what R&D is all about at this stage. There's no magic procedure rather than study the problems and find solutions.
__________________
There are two kinds of people in the world: Those who like bacon and those who will be used as fodder in the case of a zombie apocalypse.
Reply With Quote
  #392  
Old 05-01-2012, 06:51 PM
Alexko Alexko is online now
2^11
 
Join Date: Jun 2009
Posts: 2,298
Alexko will become famous soon enough
Send a message via MSN to Alexko
Default

Quote:
Originally Posted by charlie View Post
If they respin it, it will be for die area, not for speed. I would guess they can lop 50mm^2 off of the total.

-Charlie
But it will be for power, and power is speed.

Quote:
Originally Posted by Thrillseeker View Post
Ok, I thought leakage was more a function of process than of design. What can they change on the design that would significantly affect leakage, and yet have about the same chip performance?
As distinctively said, it's both. Exactly how you can reduce leakage through better design, I have no idea, but AMD claims to have done it for Pitcairn & CV.
__________________
My (currently dormant) blog: Teχlog

Last edited by Alexko; 05-01-2012 at 06:53 PM.
Reply With Quote
  #393  
Old 05-01-2012, 07:43 PM
Thrillseeker Thrillseeker is offline
2^10
 
Join Date: Aug 2009
Location: USA
Posts: 1,872
Thrillseeker will become famous soon enough
Default

Quote:
Originally Posted by Alexko View Post

As distinctively said, it's both. Exactly how you can reduce leakage through better design, I have no idea, but AMD claims to have done it for Pitcairn & CV.
Well, I guess I'm assuming that AMD is using some kind of standard FET cell that TSMC provided. If TSMC has a new tweaked cell, or if AMD is sufficiently advanced enough to work with TSMC to customize their own cell I guess I can see it.
__________________
"Why waste time learning when ignorance is instantaneous?" - Hobbes
Reply With Quote
  #394  
Old 05-02-2012, 11:00 AM
charlie charlie is offline
Administrator
 
Join Date: Feb 2009
Posts: 7,326
charlie has disabled reputation
Default

Quote:
Originally Posted by Thrillseeker View Post
Ok, I thought leakage was more a function of process than of design. What can they change on the design that would significantly affect leakage, and yet have about the same chip performance?

Are there open spaces on the current layout? Are parts of the chip not being used that can be removed?
No, think of it more as incorporating learning about the 28nm process, and where they can do things better. Tahiti was the first 28nm chip, and a lot was learned from it.

-Charlie
Reply With Quote
  #395  
Old 05-02-2012, 11:01 AM
charlie charlie is offline
Administrator
 
Join Date: Feb 2009
Posts: 7,326
charlie has disabled reputation
Default

Quote:
Originally Posted by Alexko View Post
But it will be for power, and power is speed.

As distinctively said, it's both. Exactly how you can reduce leakage through better design, I have no idea, but AMD claims to have done it for Pitcairn & CV.
Power can be speed, and speed can be power. Both are non-linear, and some times non-interchangeable. In this case, it wouldn't surprise me if they gained both, but that isn't the main goal.

-Charlie
Reply With Quote
  #396  
Old 05-02-2012, 12:06 PM
265586888's Avatar
265586888 265586888 is offline
>intel 4004
 
Join Date: Jun 2009
Posts: 11,272
265586888 has a spectacular aura about265586888 has a spectacular aura about
Default

AMD always does something on the GPU when it comes to smaller nodes.
If anyone could spot this on the RV770 story article from AnandTech...
Reply With Quote
  #397  
Old 05-02-2012, 01:04 PM
Guild Guild is offline
2^10
 
Join Date: Mar 2010
Posts: 1,944
Guild is on a distinguished road
Default

AMD has beat Nvidia to a new process by ~6 months in the last 6-7 years at least. In fact this generation is the closest Nvidia has come, probably due to the whole 32nm mess.
Reply With Quote
  #398  
Old 05-03-2012, 02:04 AM
zomg zomg is offline
itanic
 
Join Date: Sep 2010
Posts: 89
zomg is on a distinguished road
Default

Possible HD7790 or HD7830 chip found - Pitcairn with 768 stream processors in AFOX Radeon :



thg via tpu

Last edited by zomg; 05-03-2012 at 02:09 AM.
Reply With Quote
  #399  
Old 05-03-2012, 05:31 AM
Guild Guild is offline
2^10
 
Join Date: Mar 2010
Posts: 1,944
Guild is on a distinguished road
Default

Makes sense... huge gap between 7770 $139 and 7850 $250 is begging to be plugged. With 768SP it looks more like a 7790 than a 7830, but on the other hand it has a bizarrely high memory bandwidth and 32 rop.

Last edited by Guild; 05-03-2012 at 05:32 AM.
Reply With Quote
  #400  
Old 05-03-2012, 05:50 AM
kalelovil kalelovil is offline
640k who needs more?
 
Join Date: Apr 2011
Location: Auckland, New Zealand
Posts: 738
kalelovil is on a distinguished road
Default

Quote:
Originally Posted by Guild View Post
Makes sense... huge gap between 7770 $139 and 7850 $250 is begging to be plugged. With 768SP it looks more like a 7790 than a 7830, but on the other hand it has a bizarrely high memory bandwidth and 32 rop.
With that clock rate its going to lose to the 7770 in a number of situations.

Presumably the clock rate is incorrect (presumably set as it is for the same reason the gpu has ended up in a 7850 branded product) and the true retail clock is around that of the 7870.

Not that the retail clock matters for these kind of salvage cards which often end up in all kinds of vender specific versions with different clocks, outputs and cooling systems. Based on the overclocking ability of the 7850 there should be plenty of leeway for vendors.

Based on past trends there is also a chance it will never be sold outside of China, but considering the large gap between the 7770 and 7850 it seems an obvious product to produce.
Unless AMD's yields on Pitcairn are so good there are practically no sub-7850 Pitcairn dies to sell.
__________________
"Both Intel & AMD Are Breaking Thru Ice Cave into Completely new era"
Reply With Quote
Reply

Tags
amd

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off


Forum Jump


All times are GMT -5. The time now is 04:39 PM.


Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2014, vBulletin Solutions, Inc.
SemiAccurate is a division of Stone Arch Networking Services, Inc. Copyright 2009 Stone Arch Networking Services, Inc, all rights reserved.