AMD CPU speculation... and expert conjecture

[Cazalan]

Why are they still using 32nm to make the next APU seem better by having people wait on 28nm, even wonder if that's the reason why the A10 trinty was clocked at 3.8Ghz instead of 4.1Ghz to begin with.

They had to redesign it for the TSMC process. That's the rumor anyway.

 

[JAYDEEJOHN]

Who are these others?

That kind of sentiment means nothing, and unless the second supplier can deliver reasonably good products, a one horse town could easily come about, even though not desirable for most people.

No sentiment involved, just sound business practice.
Not talking about people at all.
There currently is a level playing field in the console designs, M$, Sony both decided their best product was to be made using AMD products, then you also have the OEMs as well.
Not sure where youre getting this sentiment thing

 

[viridiancrystal]

No sentiment involved, just sound business practice.
Not talking about people at all.
There currently is a level playing field in the console designs, M$, Sony both decided their best product was to be made using AMD products, then you also have the OEMs as well.
Not sure where youre getting this sentiment thing

I just realized how good a marketing push AMD could do with all the consoles. They have done it with Nintendo's since the Gamecube, just have a sticker on the side of every one that says: "Powered by AMD." If someone likes their consoles quite a bit, they may just end up with a little bias towards AMD hardware in their computers 😉

 

[mayankleoboy1]

What does clock bump have to do with overall performance, Richland will be on a more mature process by that point so lower power,

Not sure how much maturing is left in 32nm tech for improvement.

 

[esrever]

Not sure how much maturing is left in 32nm tech for improvement.

its really mature now. And it will be even more mature 3 years from now when glofo gets 28nm off the ground.

 

[sarinaide]

AMD are limited to what GloFo and/or TSMC are capable of producing if it means perfecting their arch at 32 and 28 for the next year or then that is all that can be done. They don't have a FAB so it is almost pointless trying to argue that they should go to 22nm.

 

[mayankleoboy1]

its really mature now. And it will be even more mature 3 years from now when glofo gets 28nm off the ground.

AFAIU, this late in into lifecycle of a fabrication node, getting more mature means that the yields have improved much more than before. And not that the power/clocks have become better.
Power/clocks improvements are for maybe the first or second year only.

 

[de5_Roy]

They had to redesign it for the TSMC process. That's the rumor anyway.

iirc tsmc scrapped their 32nm node because it was too costly. remember the radeon 6k and fermi gpus were 40nm?
trinity and richland are both made by glo(!@#$ing)fo on their 32nm process.

I just realized how good a marketing push AMD could do with all the consoles. They have done it with Nintendo's since the Gamecube, just have a sticker on the side of every one that says: "Powered by AMD." If someone likes their consoles quite a bit, they may just end up with a little bias towards AMD hardware in their computers 😉

good call! 😍
put something like 'gaming evolved' or 'powered by hsa' or 'innovated by amd' on the box. only bad publicity is no publicity (or bulldozer publicity).
just don't put 'beats core i7 3970x' or something similar.

Not sure how much maturing is left in 32nm tech for improvement.

it's glofo after all. there's Always room for (massive) improvement. 😗

its really mature now. And it will be even more mature 3 years from now when glofo gets 28nm off the ground.

i saw their recent roadmap, shows that 28nm is ready since 2012...
scratch that, they've been ready since 2011(!!!).
http://www.xbitlabs.com/news/other/display/20120920113524_Globalfoundries_Accelerates_Roadmap_Expects_First_14nm_Tape_Outs_in_2013.html

btw, i've seen people speculating that ps4 might have a bd cpu. imo problem with that idea is that bd is made by glofo but the rumored customized jaguar(28nm) cpu as well as the sea islands gpu will both probably be made by tsmc. doesn't make sense for amd to use two different manufacturers. i mean, won't that increase b.o.m.? bd isn't even a low cost, high volume cpu (or any volume...).

 

[gamerk316]

btw, i've seen people speculating that ps4 might have a bd cpu. imo problem with that idea is that bd is made by glofo but the rumored customized jaguar(28nm) cpu as well as the sea islands gpu will both probably be made by tsmc. doesn't make sense for amd to use two different manufacturers. i mean, won't that increase b.o.m.? bd isn't even a low cost, high volume cpu (or any volume...).

Unless they expect yield issues, and have a contract where they are held liable if they can't meet product demand.

 

[JAYDEEJOHN]

Sometimes GF does remind me of Scotty, where he finally admits, never tell the captain how long itll really take

 

[Cazalan]

btw, i've seen people speculating that ps4 might have a bd cpu. imo problem with that idea is that bd is made by glofo but the rumored customized jaguar(28nm) cpu as well as the sea islands gpu will both probably be made by tsmc. doesn't make sense for amd to use two different manufacturers. i mean, won't that increase b.o.m.? bd isn't even a low cost, high volume cpu (or any volume...).

They've been using two manufacturers. TSMC has been making their GPUs and GF their APU/CPUs. AMD paid a boatload (~1Bil) to break their contract so their new APUs could be made at TSMC. This is why Krishna and Wichita were cancelled.

Things may change again at 20/14nm but that's a long ways off.

 

[noob2222]

iirc tsmc scrapped their 32nm node because it was too costly. remember the radeon 6k and fermi gpus were 40nm?
trinity and richland are both made by glo(!@#$ing)fo on their 32nm process.

TSMC is normally a half node manufacturer anyway and hasn't done a full node, at least not since 65nm. Aside from that, kaveri is the 28nm apu, it will be interesting to see how it does on bulk non soi process.

 

[Chad Boga]

Does anyone think they know what the GPU of Richland will be capable of?

To what discrete card could it be compared to as its equal?

 

[JAYDEEJOHN]

http://techreport.com/news/24277/leaked-richland-specs-reveal-higher-clock-speeds

Look for just a small bump compared to trinity, 10-15% my guess, higher clocks, better power utilization and a stronger cpu to drive it all.
This is very conservative as well

 

[jdwii]

Does anyone think they know what the GPU of Richland will be capable of?

To what discrete card could it be compared to as its equal?

After seeing the article JAYDEEJOHN posted thinking not much more maybe 10% at most from the clock speed increase(since its using the same design as radeon 6000 series looks like Amd doesn't even care this time around) take into fact that most laptop manufacturer don't buy high speed ram its going to be a troll release yes chad i used that word again.

5% boost in CPU performance on average and 10% boost in GPU performance. Wow just wow.

 

[noob2222]

If thats the case, the media will have another fun day laughing at AMD.

http://www.techspot.com/news/51433-amds-upcoming-richland-desktop-apu-lineup-detailed.html

AMD is set to replace the current generation A-series "Trinity" APUs with updated parts that are said to offer a performance boost of 20% to 40% over their predecessors

AMD is screwed if they overstated that much again and will lose all credibility. Time will tell if AMD's new marketing team is going to repeat their BD mistake and get fired again.

 

[JAYDEEJOHN]

But, you the customer should get the high end perf of trinity in the 65 watt models, which the OEMs will like
Im saying, in most at worst case scenarios, it will be 10-15%, it can and will be more in many other scenarios

 

[JAYDEEJOHN]

So, instead of once again looking at pie in the sky, and expect 20-40% thruout its overall perf range compared to trinity, the outliers will be there, but so too will those higher marks

 

[viridiancrystal]

After seeing the article JAYDEEJOHN posted thinking not much more maybe 10% at most from the clock speed increase(since its using the same design as radeon 6000 series looks like Amd doesn't even care this time around) take into fact that most laptop manufacturer don't buy high speed ram its going to be a troll release yes chad i used that word again.

5% boost in CPU performance on average and 10% boost in GPU performance. Wow just wow.

On a Half-release that could almost be classified as a stepping? Same process, same arch, and better performance/power numbers. What is the problem?

 

[jdwii]

On a Half-release that could almost be classified as a stepping? Same process, same arch, and better performance/power numbers. What is the problem?

The problem is Amd is not in their own little world with a release from Intel coming out this is one horrible upgrade and yes i'd like to know if Amd did say 20-40% boost(marketing guys should NEVER use numbers they need to go back to marketing i know this as i've read lots of marketing books) and no one cares about 1 test i care about all of them on average. Again their is know reason why they didn't put a better GPU besides a little speed bump, not to mention with the rumors about IGPU from Intel, Amd should be really trying to make a better GPU this time around or at least use the one from the radeon 7000HD series. This should be called Trinity 1.1 because that's all were getting. With Trinity I could recommend Amd again and have people get a A10 vs a I3 as a thought it was better overall now with haswell i don't know have to wait to see benchmarks. If Intel is right when saying their next GPU is on par with a gtx 650M then Amd has issues since even a 5670 is better then trinity.

The 650 is on par with a 6850

Hope Intel is pulling a Amd and exaggerating their numbers as well.

 

[jdwii]

Also what's wrong with TSMC why is Steamroller delayed? Steamroller plus GCN Architecture would of been a good fight with Intel. Heck even Piledriver+GCN Architecture would of been better.

 

[Cazalan]

The problem is Amd is not in their own little world with a release from Intel coming out this is one horrible upgrade and yes i'd like to know if Amd did say 20-40% boost(marketing guys should NEVER use numbers they need to go back to marketing i know this as i've read lots of marketing books) and no one cares about 1 test i care about all of them on average.

It is just 1 test showing 20-40%, 3DMark 11. PCMark 7 was 10-12%.
That's all AMD released for numbers so far which is clear in the footnotes of the announcement page.

Overall CPU clocks are up 5-8% (Turbo/reg) and GPU clocks up 5-10% (44/84Mhz) depending on model.

Official support for PC2133 is good if smart system builders take advantage of it. These APUs are bandwidth starved and you can see a 15% gain going from PC1600 to PC2133. AMD only tested with PC1600 so they could have put out higher numbers if they really wanted to.

http://hexus.net/tech/reviews/cpu/46073-amd-a10-5800k-trinity-needs-faster-ram/?page=3

 

[JAYDEEJOHN]

Trinity isnt selling huge numbers.
This will help, as a stop gap before real change, and its coming soon.
Why do you think AMDs management has been so shook up, with condensed releases by delay after delay, being tied into one scenario etc etc.
Its cost them dearly in some ways, in others, time will tell, it will finally work, and work well.

 

[Cazalan]

Also what's wrong with TSMC why is Steamroller delayed? Steamroller plus GCN Architecture would of been a good fight with Intel. Heck even Piledriver+GCN Architecture would of been better.

Steamroller was designed for GF process and AMD decided to switch to TSMC. GF and TSMC use a different process and it takes a good 18 months to switch.

So it's not TSMC causing a delay it's AMDs decision to switch fabs.

 

[Cazalan]

Its cost them dearly in some ways, in others, time will tell, it will finally work, and work well.

We can hope. TSMC has been great for GPUs. The biggest issue is Apple, and how much is TSMC going to bend over backwards to secure the massive volumes they need. AMD/NVidia are going to get stuck with the leftovers when it comes to the most advanced nodes.