AMD Piledriver rumours ... and expert conjecture

[Reynod]

We have had several requests for a sticky on AMD's yet to be released Piledriver architecture ... so here it is.

I want to make a few things clear though.

Post a question relevant to the topic, or information about the topic, or it will be deleted.

Post any negative personal comments about another user ... and they will be deleted.

Post flame baiting comments about the blue, red and green team and they will be deleted.

Enjoy ...

 

[tonync_01]

As for AM3+ boards getting bios update, pretty much all 900 series should be getting support for PD as it seems the 1000 series boards were put on hold (probably till steamroller)

I was wondering about 1000 series. The only information I found was from November of last year. Do you happen to a more recent source suggesting that they won't be released till steamroller and what features it might include?

 

[JAYDEEJOHN]

No question about it: Trinity's integrated graphics are fast. They're substantially quicker than even Llano's, and the contest with Intel's solutions is really no contest at all. From a seat-of-the-pants perspective, only the A10-4600M and Radeon HD 7760G are really playable at these settings. Llano is borderline, and the Intel offerings are just too choppy.

http://techreport.com/articles.x/22932/10

 

[de5_Roy]

Here's an extract from TechReport regarding VCE and QS2.0 and QS1.0:

http://techreport.com/articles.x/22932/8

Cheers!

EDIT: Take a look at the game benchies... The FPS's the HD4k gives don't tell the whole story, indeed.

nice. i hope trinity will only get better from now on.

 

[triny]

I understand why but they could have still included it and put a disclaimer instead of just leaving it out. I want to see the numbers stack up since IB will be Trinitys main competition, not SB.



It is still a 45w CPU vs a 35w CPU. I wouldn't mind seeing the 35w IB or even the 17w IB vs SB or each other.



Post some links. I would like to see QS 2.0 image quality comparisons as from what I have read, QS 2.0 is faster and better quality.

I believe the 4655 is 25watts so it's comparing trinity 25watts to IB 45watts
http://www.notebookcheck.net/AMD-A-Series-A10-4655M-Notebook-Processor.74883.0.html

 

[triny]

We will see who is closer in estimating Piledrivers performance Me or triny i say 15-20% boost OVERALL! Not 1 benchmark but 20+ benchmarks. And guskline to even think Piledriver will go toe to toe with a I5 ivy in terms of gaming means you have a LOT of hope! 😉

what is your prediction ? trinity a105800 over lano 3850

 

[-Fran-]

which means all the sites would have to re-review IB's IGP in games?

seriously, they should consider including image quality as well.
I remember back when I was using Intel's GMA, some pixels won't even render, as in Big chunks, no flames....etc

There are some sample screenshots in TechReport's article if you want to go look at them. The driver implementation has come a really long way (in a good way) for the Intel HD4k, but they still have to optimize it for some games. It seems they focused in UE3 in the first tandem, and now they should go for the rest of the engines I guess, but at least, they are now competitive.

Cheers!

 

[JAYDEEJOHN]

BW will still play a major part here.
Its currently set low, but true fusion will correct much of this, later

 

[truegenius]

Cheers!

why cheers always?

 

[noob2222]

I was wondering about 1000 series. The only information I found was from November of last year. Do you happen to a more recent source suggesting that they won't be released till steamroller and what features it might include?

other than the november articles, not anything solid. The thing is Read cleaned house and seems to be removing some of the stupidity, especially in marketing. looking at the 1090FX specs, there would be no reason for anyone to desire the board, it didn't bring enough of a benefit over the 990fx. Even if they do bring a 1090FX as the november article stated, it will be looked at as a "why bother?"

his new agenda seems to be more in-line with what needs to be done. heavy push on OpenCL (will make steamroller look good), and more development where its needed (not stop-gaps)

I was told that delaying the socket migration beyond the AM3+, C32 and G34 to new socket is a good move,since AMD can design more aggressive, rather than stop gap

Read more: http://vr-zone.com/articles/amd-to-survive-and-thrive-still-/15564.html#ixzz1v3gJ6R3g

this statement makes me wonder if steamroller won't first appear on some AM3+ boards after the initial APU. with a fully mature ddr4 board and cpu in 2014. This article could also be referring to the 1090fx.
Edit:

Cheers!

 

[-Fran-]

Yep, it's like saying "hi" to a person when you greet him/her. You can get by if you don't say it, but the atmosphere seems better when you do 😛

Cheers!

 

[truegenius]

He's being friendly ?
Cheers!

:lol:
Gotcha
seems a tough job as he never not forgets to cheers

http://xbitlabs.com/news/cpu/display/20120116163742_AMD_Describes_Piledriver_Architecture_Peculiarities_to_Software_Developers.html
and do anybody remembered this
4channel memory controller on pd
trinity is still duel?
Amd if pd is going to be, then maybe this 4 channel may provide another 5-10% boost in handpicked 😉 benches

and 5m/10c pd 😱
means multithreading multithreading multithreading all the way, 10c pd is coming around just crossing the bridge

CHEERS! 😗

 

[lilcinw]

I thought the quad-channel and 5 module parts got axed until steamroller so that everyone doesn't have to upgrade to a new socket again.

 

[-Fran-]

I'd say that's way too old as well, truegenius. I'd love that PD would embrace 4ch for DDR3, but I'd say the smarter move is to go forth into embracing DDR4 ASAP.

AMD needs a new platform: Chipset, CPU and RAM. Server wise, DDR4 will see rapid adoption, I'm sure. Most server apps are very memory intensive (specially AIX and VMs in farms). That could mean a rapid transition to Desktop and be next to Intel, at least, in adoption, haha.

Cheers!

 

[noob2222]

I think that quad channel memory was more geared towards the server boards. currently they are running 2x dual channel memory controller in order to keep the designs the same for server and dt, pd may be native quad for servers.

may be a bit naive, but I wonder if read is going to separate dt from server a bit instead of using server chips for dt (less l3 cache and the quad channel memory). at this point, who knows, its pretty silent outside of AMD.

 

[amdfangirl]

why cheers always?

Cheers is an acronym.

It stands for Chemical Hazard and/or Electrical Engineering Risk, Sorry?

People who are about to kill their workmates with an accident* use cheers .

Cheers.

*Thanks to the new workplace laws, as long as you warn a co-worker of impending doom, you are not legally responsible. Yay democracy!

 

[dogman_1234]

I like Cheers...good show!

 

[dogman_1234]

*Thanks to the new workplace laws, as long as you warn a co-worker of impending doom, you are not legally responsible. Yay democracy!

Koolio...

 

[palladin9479]

To be fair, thats because the features not implemented in Rasterization are implemented naturally as part of any Ray Tracing engine [refraction/reflection of light, for example]. The rendering equation, when fully implemented, is the only method for true photo-realism, but Ray Tracing is MUCH easier to implement overall then the full rendering formula would be.

I also note if you want to go into the whole "quality" argument, one could point out to minute quality differences between NVIDIA and AMD, simply due to differences in the implementation of DX. Implementation matters.

The difference between TR and Rasteriation is like the difference between Newton and Einstein. Newtons' laws of physics work incredibly well for day to day uses, yet their not exactly right. At the incredibly big and incredibly small levels they fall apart and that's when relativistic physics comes in. You can apply those to any level and they would be correct, but it would be a giant waste of time and newton's laws predict the same outcome until you get a dozen or more decimal places down the line.

So while rasterization works, it will never be perfect nor as good as Ray Tracing for quality. Ray tracing actually calculates out what happens when light photons (rays) bounce off objects and get diffused / refracted. It's ridiculously expensive computationally and thus not an option for any form of real time rendering. Even with modern hardware you'd need to limit the number of rays and cut down on the detail to get it to render at anything near real time. Studios employ multi-million dollar render farms to process the sheer number of calculations required to do Ray Tracing.

Also are you advocating purchasing a product for a feature that you can not use? No current rendering engines use QS (or CUDA / GPGPU) worth a damn, the only good engine has gone with OpenCL as their accelerator of choice. I do angry when either company (NVidia / ATI) use's shortcuts to cheat on quality.

 

[jdwii]

what is your prediction ? trinity a105800 over lano 3850

Piledriver over Bulldozer. Its not even much a prediction as we all ready seen Piledriver cores without L3.

15-20% more performance(On average not upto statements based on the majority of app's not 1 or 3) at launch compared to Bulldozer plus 10-20% more efficient as well.

 

[guskline]

Piledriver over Bulldozer. Its not even much a prediction as we all ready seen Piledriver cores without L3.

15-20% more performance(On average not upto statements based on the majority of app's not 1 or 3) at launch compared to Bulldozer plus 10-20% more efficient as well.

Will Piledriver's real test be its overall improvement over the 1100 Thuban? A lot of us jumped on the AM3+ bandwagon and bought 1090T or 1100T CPUs initially while hoping the Bulldozer would be a significant jump from them. When it turned out poorly, I think there is great hesitation about the Piledriver.

jdwii: I see your configuration on your AMD rig is very close to mine. How much difference from the 1100T to the Piledriver do you need to make the jump? I take it you passed on the Bulldozer?

 

[viridiancrystal]

Will Piledriver's real test be its overall improvement over the 1100 Thuban? A lot of us jumped on the AM3+ bandwagon and bought 1090T or 1100T CPUs initially while hoping the Bulldozer would be a significant jump from them. When it turned out poorly, I think there is great hesitation about the Piledriver.

jdwii: I see your configuration on your AMD rig is very close to mine. How much difference from the 1100T to the Piledriver do you need to make the jump? I take it you passed on the Bulldozer?

The PD cores in trinity are much faster than BD cores clocked similarly. (See Noob's posts from a short while ago). They also use much less power. So, PD looks to be a big improvement over anything AMD has to offer currently.

 

[guskline]

The PD cores in trinity are much faster than BD cores clocked similarly. (See Noob's posts from a short while ago). They also use much less power. So, PD looks to be a big improvement over anything AMD has to offer currently.

I'm cautiously optimistic. We shall see when the desktop Piledriver arrives.

 

[amdfangirl]

By my calculations, Piledriver will handly beat Thuban.

Then again I said the same about Bulldozer.

 

[gamerk316]

The difference between TR and Rasteriation is like the difference between Newton and Einstein. Newtons' laws of physics work incredibly well for day to day uses, yet their not exactly right. At the incredibly big and incredibly small levels they fall apart and that's when relativistic physics comes in. You can apply those to any level and they would be correct, but it would be a giant waste of time and newton's laws predict the same outcome until you get a dozen or more decimal places down the line.

So while rasterization works, it will never be perfect nor as good as Ray Tracing for quality. Ray tracing actually calculates out what happens when light photons (rays) bounce off objects and get diffused / refracted. It's ridiculously expensive computationally and thus not an option for any form of real time rendering. Even with modern hardware you'd need to limit the number of rays and cut down on the detail to get it to render at anything near real time. Studios employ multi-million dollar render farms to process the sheer number of calculations required to do Ray Tracing.

Also are you advocating purchasing a product for a feature that you can not use? No current rendering engines use QS (or CUDA / GPGPU) worth a damn, the only good engine has gone with OpenCL as their accelerator of choice. I do angry when either company (NVidia / ATI) use's shortcuts to cheat on quality.

I'm just pointing out that Ray Tracing has some areas where it does not have the same capability as a fully implemented Rasterization formula. Ray Tracing works GREAT for simulating the effects of light [which Rasterization coincidentally does very bad at]. But remember, for it to be effective, EVERY SURFACE has to properly reflect/absorb/bend light in a realistic manner, which is very hard to do. Phosperesence doesn't come naturally, and simulating the effects on light from outside sources [gravity] becomes insanely expensive to perform. All I'm pointing out is that Ray Tracing isn't perfect either, despite the immediate quality increase you get.

And if anything, Ray Casting [going from the eye out, rather then from each light source] would be implemented to make games playable.

 

[sarinaide]

counting on your fingers and toes..? 😛

How does that work, ... one, two, do I need a pedicure, or maybe some nail gloss, what was I doing again.


What is more interesting is the calculations part, is there a method to this madness, or madness to the method? 😛