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 ...

 

[jimmysmitty]

I was reading, a few days ago, about solid state drives and how the newer smaller process used for their NAND memory is actually a step down regarding quaility and longevity. Do you think this might apply to CPU processes also?

It is. Not because the process is bad but because the lanes for electrons is smaller and therefore more leakage is possible. Intel was talking about this as well with CPUs and had an idea for a CPU that had some inactive cores that will take over when an active core starts to fail.

They are also working on NTV (Near Threshold Voltage) which will allow the CPUs to operate at a minimum voltage, about 200mV according to Intel, while idle or under low use that way it will increase the CPUs life.

Of course there will be other things that help in the future such as new materials to use with the process of new transistor styles such as 3D Tri Gates, like In Intels 22nm coming up.

As for GF 32nm, the only benefit it does have is the density due to the way they are making it. But current yields and performance do not look nearly as good as Intels current 32nm and will probably not look any better compared to Intels 22nm.

 

[aford10]

Mine was a slight jab. But you need to have a sit down with Triny and his hypocracy. He goes around saying everyone is a this and that then doesn't like the tables being turned.

Please don't call out members in public. How would you feel if you were called out in a public thread? If you have an issue, you can PM one of us. Thanx!

 

[Chad Boga]

Please don't call out members in public. How would you feel if you were called out in a public thread?

He would probably feel as I would, and that is, being completely unperturbed.

 

[nforce4max]

got any light to shed on Optimus.?

Na just reading the writing on the wall, 50-75 years before modern civilization finally gives up the ghost if ways keep going the way they are. When it comes to the economic systems around the world they all are pretty modeled on the boom bust scenario but unlike times before the whole world is integrated into almost a single global economy. Take out a few small nations like Greece or Spain and most of the world will hardly notice except for the small nearly unexplained rise in everyday prices but take out the EU or the US and the whole system will implode. When that happens it wouldn't be long for basic services to break down if solutions to keep everything running within only a matter of a few days. The food distribution system is the most fragile depending on which country or region is in question but if fuel is limited or nonexistent then there is no trucking and the store shelves will be empty. If the trains that deliver the coal that powers many of the power plants around the world it only takes hours to a day or two before they run out of coal for the boilers so many will end up in the dark. Communications are more likely to have broken down early on due to governments switching off key routers and hubs that make up the net. So it doesn't look good when one thinks about it but the sad fact is that very little happens by happenstance but is the result of nearly limitless greed of the mega corporations and greedy parasitic speculators that ride their coat tail. Much of what is seen in the news of nations and their pity political and military posturing only helps to ferment wars.

I could go on :s

Back on topic does anyone know why AMD has yet to introduce a single FireGL model of Llano and Bobcat? If they would just introduce a few they would sell very well for professionals on a tight budget that need to be able to work on the go.

 

[aford10]

I can see the double standard because 'jimmysmitty' is a community reporter but he speaks truth in this matter.
I am at fault too and admit that.
he to was called out in public, and yes two wrongs do dot make it right but go back through this thread and see the 'hypocrisy' for yourself.
please.

It's not about truth, it's about professionalism. I've been watching over the thread, and this is something that isn't going to be tolerated. Let's all play nicely and there won't be any problems.

 

[king smp]

"AMD Will Push 'Piledriver' Beyond 4GHz Using Resonant Clock Mesh Technology

Saturday, February 25, 2012 - by Paul Lilly


Advanced Micro Devices (AMD) plans to use resonant clock mesh technology developed by Cyclos Semiconductor to push its Piledriver processor line to 4GHz and beyond, the company announced at the International Solid State Circuits Conferences (ISSCC) in San Francisco. Cyclos is the only supplier of resonant clock mesh IP, which AMD has licensed and implemented into its x86 Piledriver core for Opteron server processors and Accelerated Processing Units (APUs).

"We were able to seamlessly integrate the Cyclos IP into our existing clock mesh design process so there was no risk to our development schedule," said Samuel Naffziger, Corporate Fellow at AMD. "Silicon results met our power reduction expectations, we incurred no increase in silicon area, and we were able to use our standard manufacturing process, so the investment and risk in adopting resonant clock mesh technology was well worth it as all of our customers are clamoring for more energy efficient processor designs."




Resonant clock mesh technology will not only lead to higher clocked processors, but also significant power savings. According to Cyclos, the new technology is capable of reducing power consumption by 10 percent or bumping up clockspeeds by 10 percent without altering the TDP. But what exactly is resonant clock mesh technology?

"Cyclos resonant clock mesh technology employs on-chip inductors to create an electric pendulum, or 'tank circuit', formed by the large capacitance of the clock mesh in parallel with the Cyclos inductors," Cyclos explains. "The Cyclos inductors and clock control circuits 'recycle' the clock power instead of dissipating it on every clock cycle like in a clock tree implementation, which results in a reduction in total IC power consumption of up to 10 percent.



Source: design-reuse.com

Cyclos and AMD didn't go into too much detail about Piledriver, though they did say it will consist of a 4GHz+ x86-64 core built on a 32nm CMOS process. For all kinds of geeky details on resonant clock mesh technology, "

source - http://hothardware.com/News/AMD-Will-Push-Piledriver-Beyond-4GHz-Using-Resonant-Clock-Mesh-Technology/

cant we all just get along?

 

[nforce4max]

I was reading, a few days ago, about solid state drives and how the newer smaller process used for their NAND memory is actually a step down regarding quaility and longevity. Do you think this might apply to CPU processes also?

No, there are major differences and legally AMD, Intel, and other cpu manufactures have to maintain a minimum 10 year operating life span as the US and other militaries are major buyers of their products as well the business world. Electron magnation will likely be a problem towards the end of the shrink and till model comes to an end after all they can only get each transistor so small with reliable results.

 

[king smp]

http://www.semiwiki.com/forum/content/917-clock-design-socs-lower-power-better-specs.html

Dan Ganousis posted in our SemiWiki forums about a newer technique to lower the power consumed by GHz clocks on SOC designs and asked if I was interested to learn more, so we met today via WebEx. Dan is with a company called Cyclos Semiconductor, co-founded in 2006 by Marios Papaefthymiou, President and Alexander Ishii, VP of Engineering.


real good read on clock mesh being used in PD
I just skimmed it but will read it in depth soon

 

[jimmysmitty]

Please don't call out members in public. How would you feel if you were called out in a public thread? If you have an issue, you can PM one of us. Thanx!

Actually if you look back, I have been "called out" so to speak. I have been called a "fanboi" and short of defending myself I let it go. I admit I do enjoy companies when they make good products but to be called a fanboi is beyond myself.

He would probably feel as I would, and that is, being completely unperturbed.

Pretty much. Its as with all of life. I let it go. No need to stress over it.

It's not about truth, it's about professionalism. I've been watching over the thread, and this is something that isn't going to be tolerated. Let's all play nicely and there won't be any problems.

I am all for that. I guess I got a bit out of hand, and I will try not to. I try to keep at bay as I see no reason not to but honestly it gets hard not to point out someone who is saying one thing andacting the opposite.

Still, I apologize and will try to stay away fro it.

 

[jdwii]

Can anyone shed some wisdom and explain to me what happen to Enhanced Bob cat?

 

[esrever]

they are coming with new e series APUs after trinity not sure if they are enhanced bobcat cores or not.

 

[aford10]

Actually if you look back, I have been "called out" so to speak. I have been called a "fanboi" and short of defending myself I let it go. I admit I do enjoy companies when they make good products but to be called a fanboi is beyond myself.

As a community reporter, you have certain mod rights, and the responsibility to uphold a certain impartiality. You have the ability to keep others in check if they step out of line. I encourage you to use it, instead of public arguments.

 

[jimmysmitty]

Can anyone shed some wisdom and explain to me what happen to Enhanced Bob cat?

Possibly it went the same way as enhanced Bulldozer? I remember that AMDs plan was BD -> Enhanced BD (higher clocks) and then PD. But since GFs 32nm seems to not be working out as well as they had hopped, well lets get this correct the "Consortium" (As it was AMD, IBM, Samsung and about 8 other companies working on 32nm HKMG) they might have canned the "enhanced" nodes and just moved to full releases of "next gen" to hopefully improve more than slight clock speed bumps could.

So Bobcat might become a Trinity part instead as it should, if the 32nm is better now, have lower thermals. Just cut the core up a bit and there ya go.

of course its all speculation on my part but would make sense as we haven't heard anything about any "Enhanced" cores and just trinity/PD instead.

As a community reporter, you have certain mod rights, and the responsibility to uphold a certain impartiality. You have the ability to keep others in check if they step out of line. I encourage you to use it, instead of public arguments.

I understand. As I said, I normally keep it out of my head and just keep going. I will however respond to false claims. Not to call the poster out per say but to make sure people don't get the wrong info.

its like at my work. A sales guy had a customer who was going on and on about AMDs FX and everything he said was untrue, be it performance or power usage. Instead of making sure the customer knew the information he had was wrong he let the customer think it was all true. I can't stand for people being given misinformation.

 

[triny]

http://www.semiwiki.com/forum/content/917-clock-design-socs-lower-power-better-specs.html

Dan Ganousis posted in our SemiWiki forums about a newer technique to lower the power consumed by GHz clocks on SOC designs and asked if I was interested to learn more, so we met today via WebEx. Dan is with a company called Cyclos Semiconductor, co-founded in 2006 by Marios Papaefthymiou, President and Alexander Ishii, VP of Engineering.


real good read on clock mesh being used in PD
I just skimmed it but will read it in depth soon

thanks for the link
Did he confirm AMD will use it?

 

[triny]

http://hothardware.com/Reviews/AMD-A83870K-Unlocked-Llano-QuadCore-APU-/?page=1

great comparison between lano and SB

 

[jaguarskx]

http://hothardware.com/Reviews/AMD-A83870K-Unlocked-Llano-QuadCore-APU-/?page=1

great comparison between lano and SB

Except that the SB CPUs used have the slower Intel HD 2000 graphics core, not the faster Intel HD 3000 graphics core. Would it have killed Hothardware to use a Core i3-2125 or Core i5-2500k which do have the Intel HD 3000 graphics core?

Granted that the A8-3870k's graphics core would have beaten the Intel HD 3000 anyway, but the performance difference would not have been as much compared to the slower Intel HD 2000. If price was a consideration, then okay an Intel Core i3-2125 costs $5 more than a Llano A8-3870k. But then why toss in the Core i5-2500 which costs $70 more than A8-3870? All prices are from Newegg.

 

[triny]

Except that the SB CPUs used have the slower Intel HD 2000 graphics core, not the faster Intel HD 3000 graphics core. Would it have killed Hothardware to use a Core i3-2125 or Core i5-2500k which do have the Intel HD 3000 graphics core?

Granted that the A8-3870k's graphics core would have beaten the Intel HD 3000 anyway, but the performance difference would not have been as much compared to the slower Intel HD 2000. If price was a consideration, then okay an Intel Core i3-2125 costs $5 more than a Llano A8-3870k. But then why toss in the Core i5-2500 which costs $70 more than A8-3870? All prices are from Newegg.

would have changed the outcome but not enough to mean much
cpu wise its all Intel

 

[triny]

AMD game heaven mayhem
http://www.youtube.com/watch?v=ZP0EtOXGTac

 

[triny]

The A8 is four slightly beefed up (1MB L2 vs 512KB) K10 Star's cores but without any L3 cache. They perform right about the same as any Phenom II x4 would at 2.6Ghz. Can't really compare to the Athlon II because they have double the L2 and L2 makes a bigger difference then L3.

I'll see if I can run some generic bench's when I get some time (doing 12hr shifts right now). My impressions have been that the 6620G is plenty for running generic games at 1366x720, you won't be running Crysis 2 at ultra maxed settings, but stuff is definitely playable. The device itself works well as a mobile notebook for work / traveling with gaming as an option. Good for the enthusiasts who's job demands we travel and who's back refuse's to allow them something bigger.

http://www.youtube.com/watch?v=k-aBcaHLqSU

A6 3420 + 6520
playing crisis 2 extreme setting @1280x720 no OCing

 

[palladin9479]

this a good comparison.?
http://www.anandtech.com/bench/Product/399?vs=81
not mobile but looking for something to compare.

reynod - yes, we do..

While I don't typically trust anything off anandtech, that's a decent comparison on what the A8 is like. 32nm Phenom II x4 with an IGP. Most people freak when they see no L3 but they miss that it's got double the L2 of the desktop Phenom II x4. End up making the overall comparison more even. And as I've stated before, APU's don't make much sense in the desktop world unless we're talking HTPC / Mini-ITX stuff. They fit perfectly in the sub $1000 mobile segment and that's where I expect them to be popular. AMD needs to be more aggressive with their pricing, if they could lower the prices even further on the APU's then they'd sell even faster.

@Triny,

I specifically said the 6620G was decent for gaming @1366x768 with medium~ish settings. I know people freak when they see that resolution, but remember we're talking the 15 inch or under notebook market. The 6620G can do movies and other media at 1080p no problem, just don't expect smooth games on high settings at that resolution. That is why I recommend anyone who gets the 1920 screen should also get the dGPU add on. While you ~can~ lower the resolution to something the 6620G could handle, playing on a LCD at non-native resolution sucks hard core.

@Optimus
Optimus is nothing like Hybrid-SLI or ACF. Optimus is Nvidia's way of controlling the output frame buffer. It's both a software driver and implemented in hardware (need both for this to work). Windows Vista / 7 WDM (actually all of them since 2000) works by having a single "Primary" output surface. When a program makes a call to get a render target the WDM will place it on this surface. Optimus allows for seamless transition from the IGP as primary render surface to the dGPU as primary render surface. You can do this manually with a program or through a BIOS setting, but the goal is to get it automated.

AMD use's Dual View which is essentially the same thing, the only difference is AMD/ATI allows it to work with ACF to provide a bit more graphics power. There is currently an issue with micro-stuttering if the dGPU and APU are too far apart in performance, for this reason I would never recommend anything stronger then a 6750M / 7690M and to use DDR3-1600. The problem seems to be a driver issue where the ACF driver isn't managing the world load like it should.

 

[king smp]

thanks for the link
Did he confirm AMD will use it?

check out the other link I posted just before that one
AMD confirmed on the 25th that clock mesh will be integrated into PD seamlessly
that even this late in design that the Cyclos tech can be used in PD
PDs are expected to come out at 4+ghz stock
I think they mean standard clock and not turbo
so if they do that means 5+ghz OCs should be possible even 6ghz should be doable
that last statement is just my opinion
I could be wrong....

edit- here is the link- http://hothardware.com/News/AMD-Will-Push-Piledriver-Beyond-4GHz-Using-Resonant-Clock-Mesh-Technology/

dated Feb 25th 2012

 

[fazers_on_stun]

http://www.newegg.com/Product/Product.aspx?Item=N82E16834158224

There you go guys, a nice HP DV6.

OK, it has the Llano A8 and 6620G graphics. Looking at http://www.notebookcheck.net/Mobile-Graphics-Cards-Benchmark-List.844.0.html note that the 6620G ranks #158 on the GPU charts; the 560M used in the laptop I mentioned before ranks #37 with nearly 3X the 3DMark01 score of the 6620G. I won't even bother comparing the 2 CPUs..

Pretty obvious which laptop gets ' blown into the weeds' here 😛..

 

[-Fran-]

OK, it has the Llano A8 and 6620G graphics. Looking at http://www.notebookcheck.net/Mobile-Graphics-Cards-Benchmark-List.844.0.html note that the 6620G ranks #158 on the GPU charts; the 560M used in the laptop I mentioned before ranks #37 with nearly 3X the 3DMark01 score of the 6620G. I won't even bother comparing the 2 CPUs..

Pretty obvious which laptop gets ' blown into the weeds' here 😛..

The cheapest 560M I found was around USD$1200 last time I checked.... It has HW DTS output and was the one I wanted, but damn it is expensive to get.

Anyway, the 6620G is not that card's competitor; the 6750M would be the one, I think. Maybe the 6990M, lol.

Cheers!

 

[guskline]

check out the other link I posted just before that one
AMD confirmed on the 25th that clock mesh will be integrated into PD seamlessly
that even this late in design that the Cyclos tech can be used in PD
PDs are expected to come out at 4+ghz stock
I think they mean standard clock and not turbo
so if they do that means 5+ghz OCs should be possible even 6ghz should be doable
that last statement is just my opinion
I could be wrong....

edit- here is the link- http://hothardware.com/News/AMD-Will-Push-Piledriver-Beyond-4GHz-Using-Resonant-Clock-Mesh-Technology/

dated Feb 25th 2012

I took the time to read the white paper written by Cyclos about the resonant clock meshing technology. I don't own a Bulldozer chip but bought an Asus Sabertooth 990FX mb in anticipation for the Bulldozer release. I opted for the 1100 Thuban when the Bulldozer reviews came in. What makes me interested in this Cyclos information is the tacit admission by AMD that Bulldozer could only really perform overclocked significantly and this caused substantial power consumption and heat. I expect the Piledriver will have a number of improvements beside the resonant clock meshing. That technology alone might convince me to "upgrade" from the 1100 Thuban.

If you have the time go to the Cyclos web page and read the white paper on the resonant clock meshing. Some was way over my head, but I understood enough to convince me that this might be the answer AMD needs to right the Bulldozer ship.

 

[fazers_on_stun]

Please don't call out members in public. How would you feel if you were called out in a public thread? If you have an issue, you can PM one of us. Thanx!

OK, I'm pretty guilty of doing the same 'calling out'. However, reviewing the history of this thread since last week, the start of the flaming can be pinned down to Palladin's post where he took great offense to some comment + link I made. Since then he has calmed down for the most part. However Triny is merely here to troll, not learn anything or provide much useful information. Take a look at his fact-free post about a $1K Intel laptop getting blown into the weeds by an AMD laptop half the price. Entirely ridiculous assertion with zero proof. I note that he hasn't even bothered to respond, but then somebody deleted my post challenging his statement..

The mods simply cannot ignore such behaviour or else this is what happens. The TOS clearly state that flamebaiting and trolling are not permitted. I'm guessing that the rules mostly apply to Intel fans nowadays?

I vote this thread be closed since it is fairly worthless now..