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 ...

 

[dogman_1234]

Or, jump to 16 nanometres like I said before.

 

[noob2222]

If AMD jumps to TSMC for Krishna/Wichita, and seeing as a complete re-tooling design would take ~18 months according to the article, AMD should probably be designing for 20nm wich TSMC is supposed to have volume production in q4 2012, just in time for AMD's re-design. Gloflo won't be on 22/20 till 2013. I wonder where the silent fabbing companies are at (samsung, Freescale(motorola spinoff)) as far as capacity and production.

 

[AbdullahG]

Did someone already post this:

http://www.fudzilla.com/processors/item/24932-amd-trinity-performance-detailed

AMD's upcoming Virgo APU platform has been detailed a while back, but the performance figures have only been rumoured in regard to the leaked specs. Now, we have the first leak of actual performance compared to the previous generation Llano APU and the performance figures look promising.

As you already know, AMD's Trinity APUs will be based on the FM2 package and will feature AMD's Piledriver CPU architecture combined with VLIW4 architecture GPU. According to the info from Donanimhaber.com and those internal benchmark slides, you can expect a significant improvement over the previous Llano APU parts. The most noticable improvement is in GPGPU performance, mostly due to the superior GPU inside the Trinity APU.

AMD's Trinity A8-series part scores around 1000 points more in 3DMark Vantage performance benchmark when compared to its Llano counterpart. The Trinity A6-series scores a pretty much same scores while the least noticable improvement is with the Trinity A4-series. In "general performance" benchmarks the situation is pretty much the same, while in "compute benchmark", the Trinity A8-series is significantly faster.

According to the same slides, AMD claims a significant improvement in the "dual graphics" department as the new A8-, A6- and A4-series Trinity parts should benefit greatly when paired up with Turks Pro GPU, aka Radeon HD 6570. Of course, bear in mind that these are AMD's own internal benchmarks and we would rather wait for some real world testing to say if the Trinity is actually as good as AMD says it is.

You can find the rest of the slides in a video over at Donanimhaber.com.

 

[jimmysmitty]

Will it work? We’re dubious. Trinity should be strong enough to anchor AMD’s mainstream notebook parts through 2012, but this cancellation leaves AMD without a new ultra low-power product family at a time when its competitors are heavily focused on providing such solutions. The company’s ability to launch tablet parts or hold the ground it gained in netbooks is very much in question. At the upper end of its product stack, AMD needs a Bulldozer core that delivers the power efficiency and performance the company originally promised, but there’s no such chip on the horizon.

THIS.

A mistake of colossal proportions, given where the market is going. Read simply cannot be this stupid, so I have to assume there are other plans for the low-power market. And it's not too hard to guess what they are.

As I said before, hes the former CEO of Lenovo, a PC company. I did not see anything in him that could take the previous AMD CEOs position and do anything as good as or better than he did. Dirk Meyer took AMDs sicnking ship, patched it and got it on its way again after Ruiz pretty much slammed it into a giant avoidable ice berg.

He didn't focus on non-essentials, he focused on getting AMDs core product back into shape and then he would move to other areas. Now there is a CEO, who probably has no idea how a CPU truly works, at the helm who I think is probably going to do more damage than good.

Love the AMD BoD. A bunch of rocket scientists they must be.

Or, jump to 16 nanometres like I said before.

They can't AMD no longer FABs for themselves so they will now rely on others to research process tech for them. That was the whole point to "FAB lite". No R&D money for process tech or any money to FAB/upgrade FABs. Just R&D on CPU archs and buying the CPUs from GF.

I guess it sort of bites AMD in the arse since they have to rely on another company and have no control over QC or how fast said company ramps process tech or moves to the next node.

Kinda like I said when this whole thing was announced.

Did someone already post this:

http://www.fudzilla.com/processors/item/24932-amd-trinity-performance-detailed

AMD's upcoming Virgo APU platform has been detailed a while back, but the performance figures have only been rumoured in regard to the leaked specs. Now, we have the first leak of actual performance compared to the previous generation Llano APU and the performance figures look promising.

As you already know, AMD's Trinity APUs will be based on the FM2 package and will feature AMD's Piledriver CPU architecture combined with VLIW4 architecture GPU. According to the info from Donanimhaber.com and those internal benchmark slides, you can expect a significant improvement over the previous Llano APU parts. The most noticable improvement is in GPGPU performance, mostly due to the superior GPU inside the Trinity APU.

AMD's Trinity A8-series part scores around 1000 points more in 3DMark Vantage performance benchmark when compared to its Llano counterpart. The Trinity A6-series scores a pretty much same scores while the least noticable improvement is with the Trinity A4-series. In "general performance" benchmarks the situation is pretty much the same, while in "compute benchmark", the Trinity A8-series is significantly faster.

According to the same slides, AMD claims a significant improvement in the "dual graphics" department as the new A8-, A6- and A4-series Trinity parts should benefit greatly when paired up with Turks Pro GPU, aka Radeon HD 6570. Of course, bear in mind that these are AMD's own internal benchmarks and we would rather wait for some real world testing to say if the Trinity is actually as good as AMD says it is.

You can find the rest of the slides in a video over at Donanimhaber.com.

http://www.fudzilla.com/images/stories/2011/November/trinity_leak_1.jpg

Most important part. As said many a time. Even better is that BDs AMD "performance" slides were showing a different CPU than we have seen.

Now if you shall excuse me, I am off to play Batman AC. Been waiting a long time. Plus I hear there are issues with DX11 performance. Imma try it out and see if I get them, much like the issues with RAGE I never had.

I love knowing what I am doing.

 

[gamerk316]

Now if you shall excuse me, I am off to play Batman AC. Been waiting a long time. Plus I hear there are issues with DX11 performance. Imma try it out and see if I get them, much like the issues with RAGE I never had.

I love knowing what I am doing.

Let me know how you do; I'm getting close to playing the game [after i finish Skyrim, of course]

 

[Reynod]

People have already started being hospitalised from fatigue due to excessive Skyrim exposure.

Iv'e had to make my kids cut back a bit ...

 

[dogman_1234]

What the deal with Skyrim?

/Seinfeld joke.

 

[AbdullahG]

There's been kids out for a week at my school and some in NJ because they were playing Skrim.
It's like a plague.
A really awesome plague...

 

[jimmysmitty]

Now if you shall excuse me, I am off to play Batman AC. Been waiting a long time. Plus I hear there are issues with DX11 performance. Imma try it out and see if I get them, much like the issues with RAGE I never had.

I love knowing what I am doing.

Let me know how you do; I'm getting close to playing the game [after i finish Skyrim, of course]

I do decent. I get some stuttering but not enough to kill the game, and thats maxed with VSYNC off.

As for "finishing" Skyrim, not gonna happen. Its supposed to have 8-10 hour story but a couple hundred hours in side quests and stuff.

What the deal with Skyrim?

/Seinfeld joke.

Its flippin amazing. You get for kill Dragons. Or, like me, you fight a Dragon and two Giants at the same time because Mr. Dragon ran and aggroed them like a wuss bag....

Yea. Hes a happy guy.

 

[warmon6]

Well for those in the USA, Happy Thanks Giving!

 

[-Fran-]

http://www.phoronix.com/scan.php?page=article&item=open64_50_bulldozer&num=1

Well, looks like BD might catch a breath in the couple of months to come.

We'll need a lot of recompiling though, lol.

Cheers!

EDIT: Oh, yes. Happy Thanks Giving, guys.

 

[fazers_on_stun]

http://www.phoronix.com/scan.php?page=article&item=open64_50_bulldozer&num=1

Well, looks like BD might catch a breath in the couple of months to come.

We'll need a lot of recompiling though, lol.

Cheers!

EDIT: Oh, yes. Happy Thanks Giving, guys.

Looks like a few percent increase with the BD optimizations, in most of the tests. I wonder if this is all we can expect with Win8? Anyway, as has been mentioned before, it's unlikely a lot of software devs are going to recompile for a tiny percent of the market, so BD's best hope is to wait a year or two until a newer version comes out. Of course, the devs might not bother with BD if it remains at a tiny percentage.

 

[-Fran-]

Tiny as in non-existent in shelves? 😛

Well, devs are not even caring to do fine tunning these days, so it's pretty much a lost battle in the Desktop. Unless it has a server counter part, I hardly doubt there will be any BD optimization at all. Not that Barcie had any either AFAIK, lol.

Cheers!

 

[jimmysmitty]

I blame Apple. Only thing that makes sense to make is AMDs BoD seeing Apple rake in the dough with mobile devices, and they want the same. They boot out the CxOs who tell them to focus on their core business and not get into the mobile game. (Considering ARM, not a bad idea at all.) If AMD isn't careful they will end up a mild player in the mobile space and irrelevant in desktops/servers. And to me this is best case.

It's also been suggested that when it came time to release something new, BD was the most advanced thing they had in the lab that was ready to release. I certainly hope THAT is not the case either.

This is from the other now closed thread but he has some interesting things here.

I honestly don't think BD was the best they had in the lab. They have been working on BD for a long time. It was supposed to come out after K10 (Barcelona) as their next step and was set to be 45nm and feature CMT, kinda what BD has now. But it got delayed, Deneb went out and here we are.

I do agree that the BoD did see that. But you cannot take a leaking ship and steer it from its main soruce of income until that leak has been patched. Dirk had the right idea. Get the graphics back on track (which so far has been doing well) and get DT back on track. I honestly think that if Dirk was still there, the BD launch would have been better or he would have delayed it till it was a ready product.

As it stands BD was not ready and PD is set to only improve on it a bit, which may just be due to higher clock speed and not IPC increases. If the rumors of Intels top TDP being 77W for the 3700K IB CPU, then it wont matter as Intel will have tons of headroom for clock speed increases which would make PD moot.

Of course, as with BD, we will have little to no info until its pretty much set to release. Not sure if its a good idea to withhold again as with BD it didn't help at all.

And as we know, Chris who did the THG review checked his numbers with AMD before he published them because even he thought there was something off but AMD did not agree. They said it was as expected.

So here we are again. On the verge of a new Intel product and a "new" AMD CPU. We can only hope for the best.

 

[Reynod]

The BD results were so confounding Chris said he ran them serveral times.

There is obviously a significant issue behind this and hopefully one of the ex-AMD engineers recently downsized will spill the beans.

Secrets in this industry always come out ... in the end.

This stuffup is the primary reason why 10% of the workforce is being shed.

/waiting

 

[fazers_on_stun]

This is from the other now closed thread but he has some interesting things here.

Yeah, when I reported the flame posts to the mods I asked that thread to be closed. A shame we cannot have a civil discussion about the pros and cons of CPUs.

Sometimes I think THG needs to separate the AMD fans from the Intel fans, sorta like what XtremeSystems does with their AMD and Intel discussion areas. Intel fans would not be permitted to say negative things about AMD products in the AMD area, but they could in the Intel area. And vice-versa. Posters agreeing with each other doesn't make for any flame wars 😛..

This way, the idea behind the forums - a place where people can ask questions, speculate, etc without flamewars erupting and scaring them off - could be hopefully maintained..

Maybe I'll post this idea over in the forum feedback area. I realize the PD and SB-E threads are supposed to be that way, but the idea of each camp's fans not permitted to post negative stuff in the other camp threads hasn't occurred, assuming it was supposed to.

 

[BaronMatrix]

Yeah, when I reported the flame posts to the mods I asked that thread to be closed. A shame we cannot have a civil discussion about the pros and cons of CPUs.

Sometimes I think THG needs to separate the AMD fans from the Intel fans, sorta like what XtremeSystems does with their AMD and Intel discussion areas. Intel fans would not be permitted to say negative things about AMD products in the AMD area, but they could in the Intel area. And vice-versa. Posters agreeing with each other doesn't make for any flame wars 😛..

This way, the idea behind the forums - a place where people can ask questions, speculate, etc without flamewars erupting and scaring them off - could be hopefully maintained..

Maybe I'll post this idea over in the forum feedback area. I realize the PD and SB-E threads are supposed to be that way, but the idea of each camp's fans not permitted to post negative stuff in the other camp threads hasn't occurred, assuming it was supposed to.

And for all of the talk about me I have never posted in the SB-E thread. Not because of negativity but a lack of concern as I won't be buying it. If you don't like or plan on buying AMD what is your reason posting in an AMD thread?

 

[Chad Boga]

Poor IntEl is clueless on how to market to the enthusiast crowd.

:lol: Intel is selling chips between $300 to $1,000 to the enthusiast crowd, and yet Intel is the clueless company?

 

[piesquared]

:lol: Intel is selling chips between $300 to $1,000 to the enthusiast crowd, and yet Intel is the clueless company?

Yeah, enthusiasts should be falling over with joy that they get to pay $1000 for a cpu a fraction faster than a $300 cpu.

Here's an interesting article....


http://www.bbc.co.uk/news/technology-15869683

 

[Reynod]

If I had noticed the other thread earlier I would have closed it.

Have those discussions here but keep the personal attacks out of it.

How hard is that?

 

[jimmysmitty]

Yeah, enthusiasts should be falling over with joy that they get to pay $1000 for a cpu a fraction faster than a $300 cpu.

Here's an interesting article....


http://www.bbc.co.uk/news/technology-15869683

You are right. Intel holds a gun to the head of those who buy the $1K CPU. Just like AMD did, and would, when they had the Athlon 64 FX for $1K+.

A company is a company. They only want one thing, your money. You do not have to buy a $1K product if you do not want to. Hell for the most part I tend to suggest a maximum of $300 for a CPU. Right now, I push the 2500K if they want the best bang for their buck. Thats all backed by facts as well, not pointless bias.

When they want AMD, I suggest a Phenom II X4 or X6 where applicable, but not BD unless they are insistant, or until AMD kills off the Phenom line.

Also, thanks for the pointless link.

If I had noticed the other thread earlier I would have closed it.

Have those discussions here but keep the personal attacks out of it.

How hard is that?

Pretty hard it seems. Then again most people allow emotions to rule them rather than logic.

But hey, its a free country (well where I am anyways) so people can be who they want to be. Here at THG we do prefer people who stay away from that kind of stuff though.

 

[AbdullahG]

If I had noticed the other thread earlier I would have closed it.

Have those discussions here but keep the personal attacks out of it.

How hard is that?

NOOOOO!!! YOU CANT TELL ME WHAT TO DO EVEN THOUGH I ONLY POST IN THIS THREAD ONCE A WEEK!!! IM TOO BUST BEING PISSED OFF AND THROWING STUFF AT MY SCREEN RATHER THAN THINKING LIKE A NORMAL HUMAN BEING!!! RWAAAHH!!!

/endrage

 

[Reynod]

/puts cotton in ears

 

[viridiancrystal]

Okay, im going to jump into this tread here and figure some stuff out, im good with HW when it comes to knowing whats fast and whats not, so can someone help explain WHY bulldozer has a bad IPC and what AMD is trying to/needs to do to make it better?

 

[warmon6]

Okay, im going to jump into this tread here and figure some stuff out, im good with HW when it comes to knowing whats fast and whats not, so can someone help explain WHY bulldozer has a bad IPC and what AMD is trying to/needs to do to make it better?

Well im not sure if anyone exactly heard from AMD what is truly causing the low IPC but from what i hear/remember from all the discussions around the web, it's being caused by one of 2 things.

1. bad/slow cache (think it was L1 but could of been at L2).

2. To many things being shared in the front end of the cpu.

what AMD is trying to/needs to do to make it better?

Well, most people believe that Amd is going to fix the issues in there next cpu "Piledriver". PD should have a 5-10% increase in IPC.

But since we not 100% sure where the slow down is happening at and AMD (to my knowledge) hasn't mentioned what is the issue exactly, so we cant say how amd going to make it better.