AMD CPU speculation... and expert conjecture

[jdwii]

Spent 40$ for the Houdini Pro engine for chess my CPU beats the I5 haswell every time now happy. I'll recommend this engine to anyone who enjoys chess(i'll play you, i bet i can beat you i'll play you if you want)

 

[blackkstar]

You realize Hawaii GPUs have been sold for over a year, right? Before you get too caught up in the whole AMD is doomed and can't compete thing. We're looking at an over one year old GPU competing with a brand new one. We usually get new GPUs in the fall or in the spring (at least from Nvidia). AMD's fall high end GPUs are all MIA at the moment. We are just seeing staggered releases, and Nvidia like to do this with Tegra as well. But this time it's not Nvidia releasing early, it's AMD not releasing.

GTX 480: Spring 2010
GTX 580: Fall 2011
GTX 680: Spring 2012
GTX 780: Spring 2013
GTX 880: n/a
GTX 980: Fall 2014

5870: Fall 2009
6970: Winter 2010
7970: Winter 2012
R9 290x: Fall 2013

As you can see, AMD is behind and missing the latest high end update. And they are more likely to release something in Winter around January or February as opposed to in the spring. I'm guessing the delay from 6970 to 7970 was due to dropping VLIW and moving to GCN. AMD dropping price on cards is more than likely a combination of response to 900 series as well as clearing inventory for new products.

 

[szatkus]

Why would Nvidia need additional capacity for Maxwell? Where once they were producing a truckload of Kepler GPUs at 28nm, now they have switched production to Maxwell, no?

You can't just turn a switch and start producing 20nm. They are produced in other fabs than 28nm.

BTW after reading about 20nm I don't think that new Radeons will be in newer process or that we will see 20nm Maxwell.

If I'm right I'll link to this post, juan-style

 

[colinp]

Maxwell is 28nm.

 

[szatkus]

Maxwell is 28nm.

I meant 20nm chips in general, not Maxwells.

I shoul read the post before sending :/

 

[jdwii]

Spent 40$ for the Houdini Pro engine for chess my CPU beats the I5 haswell every time now happy. I'll recommend this engine to anyone who enjoys chess(i'll play you, i bet i can beat you i'll play you if you want)

I like a challenge, your on. pick a online chess and we will play

Just PM me whenever you want to play

 

[jdwii]

http://www.thg.ru/technews/20141108_130108.html

Production of 16-nm GPU Nvidia postponed to a later date

Guessing we won't be seeing anything like this for awhile perhaps Samsung and Amd can give us something else instead.

 

[Reepca]

Houdini Pro chess engine? The last time I heard "houdini" in a vaguely computer-related context was when my brother was researching how Frozen was made (referring to the modelling/animation/etc studio). I assume this is a different Houdini, though, and I'm feeling too lazy (as is my potato laptop) to google it just now... I suppose I would assume a "chess engine" is one whereby a computer is able to think ahead in chess (if I do this move, it will cause this next move, this the move after that, this 3 moves down... etc). Or maybe it's just an engine that drives a simple normal chess game. I dunno. Enlighten me!

 

[jdwii]

Houdini Pro chess engine? The last time I heard "houdini" in a vaguely computer-related context was when my brother was researching how Frozen was made (referring to the modelling/animation/etc studio). I assume this is a different Houdini, though, and I'm feeling too lazy (as is my potato laptop) to google it just now... I suppose I would assume a "chess engine" is one whereby a computer is able to think ahead in chess (if I do this move, it will cause this next move, this the move after that, this 3 moves down... etc). Or maybe it's just an engine that drives a simple normal chess game. I dunno. Enlighten me!

Sure thing man, yeah you are correct all chess engines use the worst strategy possible brute forcing. I remember getting my 8350fx and after i got done overclocking i fired up Arena 3 and lost horribly to my friends I5(locked 3.2Ghz haswell) i built him i was mad when i found out something as easy as predicting different moves ahead and putting each assumption on different cores was so hard to do then just recently i found Houdini Pro 4 and it claims it can use 32 different cores i know it uses all of mine and my friends little I5 cried. Although to be fair he can at least upgrade to something more powerful.

 

[szatkus]

http://www.thg.ru/technews/20141108_130108.html

Production of 16-nm GPU Nvidia postponed to a later date

Guessing we won't be seeing anything like this for awhile perhaps Samsung and Amd can give us something else instead.

http://www.digitimes.com/news/a20141105PM201.html

Anyone excepted new GeForces before Q3?

 

[blackkstar]

Isn't AMD next GPUs leaving TSMC and moving to GloFo? So Nvidia will be on TSMC and dealing with delays and maybe GloFo will execute a bit better and have a huge lead? If Maxwell is good in efficiency against Hawaii and both are 28nm, but if AMD's next chips are 20nm or 14nm Maxwell is really going to struggle.

 

[jdwii]

Isn't AMD next GPUs leaving TSMC and moving to GloFo? So Nvidia will be on TSMC and dealing with delays and maybe GloFo will execute a bit better and have a huge lead? If Maxwell is good in efficiency against Hawaii and both are 28nm, but if AMD's next chips are 20nm or 14nm Maxwell is really going to struggle.

We were hearing that Amd was leaving both of them and going samsung, samsung and nvidia will probably never work together for awhile over the lawsuit. I guess IBM did just sell their stuff to GF that might make them more competitive with TSMC in a few years.

 

[szatkus]

Isn't AMD next GPUs leaving TSMC and moving to GloFo?

http://ir.amd.com/phoenix.zhtml?c=74093&p=irol-newsArticle&ID=1914734
Doesn't sound like leaving.

Apple does similar thing with their A8. Most of chips are produced by TSMC and some by Samsung.

 

[szatkus]

We were hearing that Amd was leaving both of them and going samsung

But GF will share 14nm with Samsung.

samsung and nvidia will probably never work together for awhile over the lawsuit.

Like Samsung and Apple? Oh, wait.

 

[gamerk316]

Sure thing man, yeah you are correct all chess engines use the worst strategy possible brute forcing.

What most do now is they simply have giant databases of the most common openings, and the "correct" response to any "on book" variant. Likewise, at a certain point in the endgame, engines can perform perfect-play and force a draw/win as the board dictates. Where chess engines still STINK is when they are forced to go off book.

That's why 1: h4 is becoming a popular first opening. It's not a good move, but one White eventually wants to make, and essentially lets White play Blacks defenses with Tempo (which isn't always a good thing mind you). By going off book the first move, the chess engine falls back to controlling the center and making equal trades, which is where they still aren't very good. You can brute force only so many possibilities, but without the vast amounts of on-book knowledge to rely on, chess engines simply are no match for even moderately skilled humans.

That's also why I laugh at any "we're close to getting real AI" arguments. We're not. We're barely at the point where computers can beat humans when given the total knowledge of about every previous game ever played AND allowing for perfect play based on that acquired knowledge. There's no actual skill on the point of the computer, just replay of what has already been done.

Now, back to your regularly scheduled programming.

 

[8350rocks]

@colinp: Tonga is GCN 1.2 where Hawaii is GCN 1.1, what you are seeing is purely refinement of the first gen GCN architecture. It is competing the NVidia. However, the word I have from AMD is that when GCN 2.0 drops, "even a next gen titan will be behind...perhaps 2 years before you see anything competitive from NVidia."

 

[colinp]

If you believe everything a salesperson tells you, then I have some amazing snake oil for you. Cures any ailment. Honest, guv.

But seriously, I'm sure AMD do have some improvements to come. I would be less worried about whether they can beat a Titan and more about whether they can compete all down the range where real people buy devices. AMD are almost dead in the mobile dGPU space right now, for example. Similarly, they can't just keep throwing moar Watts at the problem, big desktop rigs just feel anachronistic these days.

That's why Maxwell is so compelling to me right now. I want a decent gaming laptop - budget isn't a constraint for me - and Intel/Maxwell is my only choice. I'll eventually want to upgrade my AMD 7870 in my micro-ATX case (and I want a rear-venting one for airflow, and not a sky-high TDP), and Maxwell is my only choice right now.

 

[gamerk316]

@colinp: Tonga is GCN 1.2 where Hawaii is GCN 1.1, what you are seeing is purely refinement of the first gen GCN architecture. It is competing the NVidia. However, the word I have from AMD is that when GCN 2.0 drops, "even a next gen titan will be behind...perhaps 2 years before you see anything competitive from NVidia."

We'll see. Can't help but suspect theres a "but only if you use it a certain way" buried in there.

 

[szatkus]

@colinp: Tonga is GCN 1.2 where Hawaii is GCN 1.1, what you are seeing is purely refinement of the first gen GCN architecture. It is competing the NVidia. However, the word I have from AMD is that when GCN 2.0 drops, "even a next gen titan will be behind...perhaps 2 years before you see anything competitive from NVidia."

What is GCN 2.0? As far as I know they AMD doesn't call their cores GCN 1.2 etc. but "3rd generation GCN" or something like that.

 

[colinp]

And of course there's another company other than AMD looking to beat Maxwell. That'd be Nvidia.

 

[gamerk316]

http://www.extremetech.com/computing/193868-powervr-series7-gpu-unveiled-bringing-desktop-class-graphics-to-the-mobile-masses

Take the performance claims for what it's worth. Still, a generation or two like that, and PowerVR might want to release a few mid-tier desktop parts...

 

[-Fran-]

I don't think PowerVR wants to enter the Desktop market again. It's saturated already and they won't offer anything special from what I can read. At best, a good low cost Video card and that's about it. I don't think OEMs will be very excited to get products for them if that's the case.

Still, for low power devices, it looks really cool.

Cheers!

 

[8350rocks]

If you believe everything a salesperson tells you, then I have some amazing snake oil for you. Cures any ailment. Honest, guv.

But seriously, I'm sure AMD do have some improvements to come. I would be less worried about whether they can beat a Titan and more about whether they can compete all down the range where real people buy devices. AMD are almost dead in the mobile dGPU space right now, for example. Similarly, they can't just keep throwing moar Watts at the problem, big desktop rigs just feel anachronistic these days.

That's why Maxwell is so compelling to me right now. I want a decent gaming laptop - budget isn't a constraint for me - and Intel/Maxwell is my only choice. I'll eventually want to upgrade my AMD 7870 in my micro-ATX case (and I want a rear-venting one for airflow, and not a sky-high TDP), and Maxwell is my only choice right now.

I do not talk to salespeople...I talk to engineers...

 

[logainofhades]

I do not talk to salespeople...I talk to engineers...

I talk to engineers every day, and sometimes they are the dumbest smart people ever. :lol:

 

[8350rocks]

@colinp: Tonga is GCN 1.2 where Hawaii is GCN 1.1, what you are seeing is purely refinement of the first gen GCN architecture. It is competing the NVidia. However, the word I have from AMD is that when GCN 2.0 drops, "even a next gen titan will be behind...perhaps 2 years before you see anything competitive from NVidia."

What is GCN 2.0? As far as I know they AMD doesn't call their cores GCN 1.2 etc. but "3rd generation GCN" or something like that.

Well, all GCN architecture to this point has been refinement of the initial GCN cores...there have been some advancements made, but only increasing numbers of cores with better power efficiency. They have not really done anything drastic to the main architecture (even buses have remained mostly the same). What you are about to see is the first changes to GCN beyond simple refinements in the next iteration. I cannot really talk beyond vague generalizations at this point, however, suffice to say that what is coming is going to be landscape changing.

As far as terminology, that is how it was related to me in terms of refinements...what we have to this point is a refinement. So all of this to date is basically GCN 1.x, the next bunch coming would be the equivalent of a 1.x to 2.0 jump in terms of significance of architecture.