Go Back   Hardware Canucks > HARDWARE > Video Cards

    
Reply
 
LinkBack Thread Tools Display Modes
  #11 (permalink)  
Old March 4, 2014, 09:40 AM
SKYMTL's Avatar
HardwareCanuck Review Editor
 
Join Date: Feb 2007
Location: Montreal
Posts: 11,670
Default

Quote:
Originally Posted by great_big_abyss View Post
Is there ANY completely updated architecture and a die-shrink in view for AMD? How about a die-shrink for Nvidia? As I understand it, all Maxwell's will be on the 28nm node...
AMD is very much a generation behind on the high end and risks going behind two generations in the mid tier markets.

Here's the thing:

Kepler was designed to go head to head against Hawaii (or whatever NVIDIA predicted it to be), but ended up going up against Tahiti. For evidence of this, look no further than what NVIDIA needed to go toe to toe against the HD 7970: a GK104 part, the GTX 680.

In order to keep things fresh, NVIDIA kept cascading first generation 600-series parts into the pipeline while AMD very much held fast with HD 7000 parts. Bonaire was a change but only a minor one and certainly not a volume seller.

Then NVIDIA started releasing the second generation Kepler cards. The GTX 760, GTX 770 and GTX 780 were born. AMD on the other hand could do nothing other than lower prices and launch Boost / GHz Edition alternatives.

Now the situation is even worse for AMD. While they have Hawaii on the high end (the part that was ORIGINALLY supposed to trade blows with NVIDIA's 600-series), it's more that obvious the architecture has been pushed to its limit to just compete on a level footing against GK110. This has a trickle-down effect in a way too: it becomes extremely hard to "scale down" such an architecture to lower end parts.

So now comes Maxwell. At first glance this isn't an issue for AMD but its actually a huge threat. NVIDIA will continue launching Maxwell parts this year and, as evidenced by the R9 280 launch, they'll be competing against an AMD lineup that's not one but two generations old.
__________________
Reply With Quote
  #12 (permalink)  
Old March 4, 2014, 10:01 AM
Phill's Avatar
Allstar
 
Join Date: Mar 2007
Location: Morden MB
Posts: 592

My System Specs

Default

Quote:
AMD is very much a generation behind on the high end and risks going behind two generations in the mid tier markets... So now comes Maxwell. At first glance this isn't an issue for AMD but its actually a huge threat. NVIDIA will continue launching Maxwell parts this year and, as evidenced by the R9 280 launch, they'll be competing against an AMD lineup that's not one but two generations old.
This is not news to me but clearly it is Not good. I am upset that AMD is going to ruin things for all of us. Capitalism only works reasonably well *if* there is competition.

How is it that since the 5XXX series and the success for AMD thereafter (GPU wise) did they run out of a road map? Perhaps my memory fails me but they had a number of generations in a row that yielded (5,6 & 7 series) well and sold well. Why did that not pave the way for more success?
Reply With Quote
  #13 (permalink)  
Old March 4, 2014, 10:13 AM
sswilson's Avatar
Moderator
F@H
 
Join Date: Dec 2006
Location: Moncton NB
Posts: 14,535

My System Specs

Default

How closely does that tie into the time frame that AMD bought ATI?????

edit... maybe just the rebranding as opposed to the purchase? ;)
__________________
MSI Z87I Gaming AC / i5 4670K / 2X 4G Gskill 1866 DDR3 / XFX XTR 750 / EVGA GTX 680 SC+ 2GB / Intel DC S3700 200G / random 160G Sata HDD
Inwin 904 / Swiftech MCP655-b / Alphacool NexXxos XT45 120 Rad / 2X Scythe GT AP-15 / EK Supreme HF / Dell UltraSharp U2412M

Asrock AM1H-ITX / AM1 Athlon 5350 / 2X4G Gskill PC3-14900 / Intel 6235 Wi-Fi / 90W Targus Power Brick / 320G Seagate Momentus / Mini-Box M350 / 1X 22" Dell IPS / 1X 22" HP
Reply With Quote
  #14 (permalink)  
Old March 4, 2014, 12:25 PM
MVP
 
Join Date: Mar 2010
Location: Ottawa
Posts: 447
Default

I don't see an issue with rebranding. It's just a product refresh. It's logical to incorporate a consistent naming scheme across a product line. It's more expensive than its predecessor because GPU prices are high.
Reply With Quote
  #15 (permalink)  
Old March 4, 2014, 01:35 PM
Vittra's Avatar
Top Prospect
 
Join Date: Oct 2009
Location: Ontario
Posts: 168

My System Specs

Default

Both sides have re-brands in their current lineups, but AMD is certainly more firmly entrenched in that camp currently.

Quote:
Originally Posted by SKYMTL View Post
AMD is very much a generation behind on the high end and risks going behind two generations in the mid tier markets.

Here's the thing:

Kepler was designed to go head to head against Hawaii (or whatever NVIDIA predicted it to be), but ended up going up against Tahiti. For evidence of this, look no further than what NVIDIA needed to go toe to toe against the HD 7970: a GK104 part, the GTX 680.

In order to keep things fresh, NVIDIA kept cascading first generation 600-series parts into the pipeline while AMD very much held fast with HD 7000 parts. Bonaire was a change but only a minor one and certainly not a volume seller.

Then NVIDIA started releasing the second generation Kepler cards. The GTX 760, GTX 770 and GTX 780 were born. AMD on the other hand could do nothing other than lower prices and launch Boost / GHz Edition alternatives.

Now the situation is even worse for AMD. While they have Hawaii on the high end (the part that was ORIGINALLY supposed to trade blows with NVIDIA's 600-series), it's more that obvious the architecture has been pushed to its limit to just compete on a level footing against GK110. This has a trickle-down effect in a way too: it becomes extremely hard to "scale down" such an architecture to lower end parts.

So now comes Maxwell. At first glance this isn't an issue for AMD but its actually a huge threat. NVIDIA will continue launching Maxwell parts this year and, as evidenced by the R9 280 launch, they'll be competing against an AMD lineup that's not one but two generations old.
Was the R9 originally designed for 22nm and scaled back to 28nm, or was that always pure speculation?

I know that the 6xxx series was originally designed for 32nm but scaled back to 40nm due to fab concerns. I ask because that would then indicate within the span of three generations, twice AMD has had to fall back to a larger process. Would the blame of this be squarely on AMD's shoulders for jumping the gun? Or the Fabs for not being able to keep up? nVidia seems to be playing a much smarter (and safer) game by giving processes time to mature before jumping to them.
__________________

Reply With Quote
  #16 (permalink)  
Old March 4, 2014, 02:22 PM
Rookie
 
Join Date: Mar 2010
Posts: 11
Default

Quote:
Originally Posted by SKYMTL View Post
AMD is very much a generation behind on the high end and risks going behind two generations in the mid tier markets.

Here's the thing:

Kepler was designed to go head to head against Hawaii (or whatever NVIDIA predicted it to be), but ended up going up against Tahiti. For evidence of this, look no further than what NVIDIA needed to go toe to toe against the HD 7970: a GK104 part, the GTX 680.

In order to keep things fresh, NVIDIA kept cascading first generation 600-series parts into the pipeline while AMD very much held fast with HD 7000 parts. Bonaire was a change but only a minor one and certainly not a volume seller.

Then NVIDIA started releasing the second generation Kepler cards. The GTX 760, GTX 770 and GTX 780 were born. AMD on the other hand could do nothing other than lower prices and launch Boost / GHz Edition alternatives.

Now the situation is even worse for AMD. While they have Hawaii on the high end (the part that was ORIGINALLY supposed to trade blows with NVIDIA's 600-series), it's more that obvious the architecture has been pushed to its limit to just compete on a level footing against GK110. This has a trickle-down effect in a way too: it becomes extremely hard to "scale down" such an architecture to lower end parts.

So now comes Maxwell. At first glance this isn't an issue for AMD but its actually a huge threat. NVIDIA will continue launching Maxwell parts this year and, as evidenced by the R9 280 launch, they'll be competing against an AMD lineup that's not one but two generations old.
I not sure where to start with this revisionist history!

Well there's the other side… there was a GK100 part that couldn’t be made on 28Nm, and Nvidia it went back to redesigned as GK110. AMD saw it as to big, expensive chip and power hungry to it actually being a threat to a $500 gaming enthusiast market at the end of 2011. Using what was optimized around complex work of super computers for gaming was a far-fetch dream. That GK110 finally got shown to professional markets in May 17, 2012, it took a year after Tahiti to released Titan February 19, 2013. There no knowledge that Hawaii was even on any road-map as a design to lead off the 7000 series debut of January 2012, you believe that but that’s just an option. Hawaii got commissioned right after the Titan released and AMD saw that yes there is gamers needing and paying more, and 20Nm is was too far off plus Nvidia still could move GK110 chip in other gaming product and not be upside down money-wise.

There was no “second generation” Keplers… the 760,770 were rebrands, the 780’s are just Titan’s they rebrand as they could no longer fool folks at the price for that gelding, once Hawaii brought some sense back to the market. If Bonaire R7 260X wasn't getting under Nvidia skin why release Maxwell? The GK106 was a dud and Nvidia wasn't seeing a good return keeping it to via in that mainstream market.

Frankly Nvidia would be in a worse spot if "miners" weren't causing the shortages and stupid crazy pricing. If this was a normal gaming market AMD would be pricing below MSRP and they wouldn’t need to be considering moving what little binned (geldings) like this and the R7 265 to feed the demand. None of these SKU’s should be here right now, but the market is burning through everything, and AIB are clamoring to get anything.

Just because you have a pulpit don't go abusing it.
Reply With Quote
  #17 (permalink)  
Old March 4, 2014, 04:16 PM
SKYMTL's Avatar
HardwareCanuck Review Editor
 
Join Date: Feb 2007
Location: Montreal
Posts: 11,670
Default

Quote:
Originally Posted by Casecutter View Post
I not sure where to start with this revisionist history!

Well there's the other side… there was a GK100 part that couldn’t be made on 28Nm, and Nvidia it went back to redesigned as GK110. AMD saw it as to big, expensive chip and power hungry to it actually being a threat to a $500 gaming enthusiast market at the end of 2011. Using what was optimized around complex work of super computers for gaming was a far-fetch dream. That GK110 finally got shown to professional markets in May 17, 2012, it took a year after Tahiti to released Titan February 19, 2013. There no knowledge that Hawaii was even on any road-map as a design to lead off the 7000 series debut of January 2012, you believe that but that’s just an option. Hawaii got commissioned right after the Titan released and AMD saw that yes there is gamers needing and paying more, and 20Nm is was too far off plus Nvidia still could move GK110 chip in other gaming product and not be upside down money-wise.

There was no “second generation” Keplers… the 760,770 were rebrands, the 780’s are just Titan’s they rebrand as they could no longer fool folks at the price for that gelding, once Hawaii brought some sense back to the market. If Bonaire R7 260X wasn't getting under Nvidia skin why release Maxwell? The GK106 was a dud and Nvidia wasn't seeing a good return keeping it to via in that mainstream market.

Frankly Nvidia would be in a worse spot if "miners" weren't causing the shortages and stupid crazy pricing. If this was a normal gaming market AMD would be pricing below MSRP and they wouldn’t need to be considering moving what little binned (geldings) like this and the R7 265 to feed the demand. None of these SKU’s should be here right now, but the market is burning through everything, and AIB are clamoring to get anything.

Just because you have a pulpit don't go abusing it.
Wha? Chip designs take YEARS.

For example, Maxwell went through trial runs in March 2013. Early silicon arrived at NVIDIA in JUNE. The design was originally approved in late 2012.

The first Hawaii silicon was already in AMD's hands as TITAN rolled out. Hawaii's initial design was cobbled together from the pieces of Sea Islands along with bits of Southern Islands.

Thinking that Hawaii was "commissioned" when TITAN came out is preposterous.

And no, this isn't conjecture. This is straight from contacts at TSMC.
__________________
Reply With Quote
  #18 (permalink)  
Old March 4, 2014, 04:33 PM
geokilla's Avatar
Hall Of Fame
F@H
 
Join Date: Aug 2008
Location: Toronto
Posts: 3,625

My System Specs

Default

Interesting how power consumption is higher. But other than that, hopefully the chips are "more mature" so we yield better chips and thus, better overclocking capabilities and lower temperatures. Please AMD, get your vBIOS right and not force us miners to flash a vBIOS modified by The Stilt.
__________________
Reply With Quote
  #19 (permalink)  
Old March 4, 2014, 05:32 PM
sf101's Avatar
MVP
 
Join Date: Jun 2012
Posts: 305
Default

I love AMD but who here also thinks they will ramp production on the next chip's only to see asic's destroy the gpu based mining leaving AMD holding a huge "Bag" of chips to sell.

The Hawaii release could have been a heck of alot better if . A. no black screens . B . more chips available.

I think right now performance wise i feel things are pretty similar.

Maxwell is next gen i Guess we will get to see how much performance nvidia squeezed out of things because if its just a efficiency release its not going to be as appealing as they may think in the gaming markets.

It would have been endlessly useful to AMD to have had more stock of the 2xx "290/x" series cards to invest back into R&D.
__________________
My 5 Current Rigs-Guest-Main-Work/Spare-HTPC-Laptop
I7930@4.0GHz-Guest
3770K@ 4.8GHz -Main
2500k @ 4.7GHz -work/Spare
3570K @ 4.5 -HTPC
MSI GT70-2OC/770m-998/4800MHz -Laptop
SF101 HW BOT
GPU's ("xfire"2x XFX290X@1150/1500"EKWB"|GIGAOC7970WF3@1250/1850|Asus480GTX@900/2000|
2xEvga260Core216's SLI in HTPC-Retired gpu's 9600ati|9800proATI|8800GTS"320mb&640mb"
|8800GTX|9800GT|9800GX2
Reply With Quote
  #20 (permalink)  
Old March 4, 2014, 08:03 PM
Shadowarez's Avatar
Allstar
 
Join Date: Oct 2013
Location: Arctic Canada
Posts: 635

My System Specs

Default

So Nvidia can hold off on Highend Maxwell till they see a need. From what iv read Nvidia could just release these efficient cards till they really see a need to from Amd. If that the case Htpc/Steam. Machines will boom for a few months.
Reply With Quote
Reply


Thread Tools
Display Modes

Similar Threads
Thread Thread Starter Forum Replies Last Post
Powercolor 7870 LE EZ Tahiti (aka 7930) igot6strings Price Check 2 December 9, 2013 05:04 AM
The VelociRaptor Lives On Soultribunal Press Releases & Tech News 35 April 9, 2010 07:13 AM