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 ...

 

[-Fran-]

So after all the marketing BS about Fusion, which in theory should be well suited to smartphone markets, they should just become another ARM manufactuer?

Heh, RISC is more flexible than you might think; take a look at the co processing part in them, it's quite interesting. AMD right now has a lot of beautiful things to add into ARM designs on their own, IMO. They could always get licenses for their A line and get one hell of a strong Tegra SoC competitor.

Cheers!

 

[viridiancrystal]

so we all know that bulldozer is not nearly as good as sandy bridge in gaming and most threaded tasks, save the few that can use all of its threads. How does it compare in all out multitasking? I mean all out like playing a cpu hevy game, with music playing in the backround and skype conference going + other things one may have going during gameplay? I'm trying to get as real world as i can here. Would the extra 2-4 threads on a FX help over a 2500k?

EDIT: I'm really trying to back AMD here, i don't want to see intel selling cpus that cost an arm and leg to get. Looking for situations where bulldozer/pilerdrizer will be of better use than sandy or ivy bridge.

 

[gamerk316]

Why would this happen Amd is having their best year in a long time, Hell they may even make a profit this year as a whole. Every Quarter that's gone by Amd made a profit. Like i keep saying their graphics/APU's will keep them profitable if they don't go to crazy on spending. Also if them they team up with Arm they could make a little more money and even get into the Lower-end market more.

When a multi-billion dollar company is making less then $100 Million in profit, and then releases a product thats a swing and miss, AND runs into supply issues, its very hard to maintain a long-term profit. Lets see if AMD can maintain theres, I suspect they can't.

 

[jdwii]

When a multi-billion dollar company is making less then $100 Million in profit, and then releases a product thats a swing and miss, AND runs into supply issues, its very hard to maintain a long-term profit. Lets see if AMD can maintain theres, I suspect they can't.

Its sad that this is still their best year.


http://www.wired.com/cloudline/2011/11/amd-trinity-performance-leaked/

but it’s not a huge surprise that the boost is modest (around 13 percent). This is a bit less than the 20 percent forecast in some leaked AMD slides from October, but within the ballpark. Graphics performance is another story, though.

Trinity’s graphics performance is significantly higher than Llano’s—around 30 percent, which is exactly what the aforementioned leaked slides claimed. Much of this boost is due to the fact that Llano’s GPU was derived from AMD’s older Cypress (HD 5000 series) design, while Trinity will use the newest 7000 series (a.k.a. Southern Islands)

 

[noob2222]

looks like not all is lost for windows 7 getting BD patches.

6. It has been stated that Bulldozer will see improvements in performance with the Windows 8 scheduler. Would you elaborate?
...
We are also working with Microsoft on a scheduler update for Windows 7 that will be available soon.

Interesting Q/A, but most of the answers are somewhat sidestep tactical answering.

If the scheduler update can put the code properly into the shared l2 cache, I can see the advantage of loading a module over spreading out modules first then cores. Thats a big IF.

Edit: forgot link. http://www.hardocp.com/article/2011/11/29/hardocp_readers_ask_amd_bulldozer_questions/1

 

[jdwii]

looks like not all is lost for windows 7 getting BD patches.

6. It has been stated that Bulldozer will see improvements in performance with the Windows 8 scheduler. Would you elaborate?
...
We are also working with Microsoft on a scheduler update for Windows 7 that will be available soon.

Interesting Q/A, but most of the answers are somewhat sidestep tactical answering.

If the scheduler update can put the code properly into the shared l2 cache, I can see the advantage of loading a module over spreading out modules first then cores. Thats a big IF.

Good news i think this should put BD ahead of Phenom II x6 in almost all test's. Heck if they could just come out with a new stepping and lower the price dramatically it might be a considerable option. I say make the 8150 a 95 watt version and get rid of the 8120. Drop the price of the 8150 to 199$ and make a new 8170 for 229$. Also lower the 4110x to 99$ and make the 4170 119$. Also drop the watts on both the 4110 and the 6110 with the new stepping.

 

[gamerk316]

^^I doubt it. And lets not forget that loading one core in each module makes it MUCH harder for turbo to kick in...[wasn't that supposed to be a key advantage of BD?]

 

[dogman_1234]

Sounds like there will be no more AMD CPU's...or direct competition against Intel's x86 CPU's.

 

[fazers_on_stun]

ATi may return if it were to be split off and kicked out into the cold.

Hmm, I wonder if Intel would then be interested in acquiring ATI.. However if those leaked slides on Ivy Bridge's GPU performance are true, as posted in the IB thread, then Intel wouldn't need anybody else's tech. Which is why I don't believe them 😛..

 

[Mousemonkey]

Hmm, I wonder if Intel would then be interested in acquiring ATI.. However if those leaked slides on Ivy Bridge's GPU performance are true, as posted in the IB thread, then Intel wouldn't need anybody else's tech. Which is why I don't believe them 😛..

I don't recall Intel showing any interest a few years ago before AMD bought them.

 

[fazers_on_stun]

Good news i think this should put BD ahead of Phenom II x6 in almost all test's. Heck if they could just come out with a new stepping and lower the price dramatically it might be a considerable option. I say make the 8150 a 95 watt version and get rid of the 8120. Drop the price of the 8150 to 199$ and make a new 8170 for 229$. Also lower the 4110x to 99$ and make the 4170 119$. Also drop the watts on both the 4110 and the 6110 with the new stepping.

IIRC GF's partially-depleted SOI process has to have 'leaky' transistors for high clocks, which means either low-power and low-clocks or high-power and high-clocks. Doesn't look like they have the 32nm gate-first HKMG node figured out too well since BD failed to achieve the intended clocks (by 600MHz or more I think), while staying within a 130W TDP. Plus BD has 2 billion of those leaky transistors..

I think a 95W TDP for the 8150 would have to wait until 22nm, where GF is supposed to go to a fully-depleted gate-last HKMG SOI process.

The price drops, I agree with, although I don't expect to see them happen anytime soon. BD took over 5 years of expensive R&D and somehow AMD has to recoup those expenses..

 

[fazers_on_stun]

I don't recall Intel showing any interest a few years ago before AMD bought them.

Not for $5.4 billion anyway 😀.

I do recall some mention of AMD in a hurry to buy ATI because they thought Intel was interested..

 

[fazers_on_stun]

Sounds like there will be no more AMD CPU's...or direct competition against Intel's x86 CPU's.

It also sounds like that perhaps one of our CRs will have to find something new to rant about in the near future as well 😛..

 

[-Fran-]

how's that Crosshair of yours holding up.?

Working flawlessly, tyvm.

---

They will still make x86 CPUs, but they won't battle in the high end for some time; and it seems for QUITE some time. They're going to attack other markets and that means that the little capital they have, has to be invested in multiple things, so it's a very risky/bold move. If they actually pull it off, they could come back into the Desktop segment to battle the high end.

Basically, when they bought ATI, their path was somewhat clear and aligned. They were going to make APUs. They're still learning from that IMO, so this diversification might be something that was anticipated long ago and not a sudden change of heart. I do remember talking about this at that time. Also, the RAM move, that's quite interesting IMO. Hope they do get deeper into RAM technologies and improve it.

Anyway, what they do from now on, will hurt the enthusiasts (most of us) a little (hopefully xD), but it has always been a premium segment to live in, so I'm not quite shocked TBH.

Cheers!

 

[Mousemonkey]

Not for $5.4 billion anyway 😀.

I do recall some mention of AMD in a hurry to buy ATI because they thought Intel was interested..

I don't recall that but that don't mean it weren't so.

 

[noob2222]

One thing I will add on AMD's not competing with Intel, its more of an announcement on how stupid their marketing team was on trying to make people believe that BD would outperform SB-E.

One must hope that AMD doesn't drop out of the desktop cpu race alltogether, if they do, well, bye bye cheap cpus, seems to me Intel has no desire to make thier cpus faster as IB is the same clock speed as SB.

 

[de5_Roy]

i suspect that the amd pr guys prepared all their press notes, ads and stuff about how fx will outperform x58 platform in terms of price-performance, features, overclocking etc. but amd was unable to release fx at that time. after, when amd was about to release fx, the pr guys just erased x58, replaced it randomly with sandy bridge and x79 and pretty much read whatever was written on the papers. amd's (former)pr department could use some awareness.

 

[gamerk316]

Hmm, I wonder if Intel would then be interested in acquiring ATI.. However if those leaked slides on Ivy Bridge's GPU performance are true, as posted in the IB thread, then Intel wouldn't need anybody else's tech. Which is why I don't believe them 😛..

Wrong company. IBM.

Think about: IBM's triumphant return to the consumer desktop, with its brand new X86 based Cell processor. Now THAT would be something revolutionary. Plus, they get to expand into the GPU market, which will allow them to have a complete hardware platform for the first time since they lost control of the PC market.

 

[-Fran-]

Wrong company. IBM.

Think about: IBM's triumphant return to the consumer desktop, with its brand new X86 based Cell processor. Now THAT would be something revolutionary. Plus, they get to expand into the GPU market, which will allow them to have a complete hardware platform for the first time since they lost control of the PC market.

If I recall correctly, IBM didn't "lose" in the Desktop front, they just got rid of it because they didn't want to invest alá-Dell into fine tuning their Desktop selling and assembling process. That's why Lenovo was chosen and got that part of IBM's business. In fact, I've never ever heard, at the time, "bad" and "IBM Desktop" in the same phrase 😛

And IBM has a portfolio of their own with the Cell CPUs. I don't think they would care about getting ATI (or nVidia for that matter) to get back into the Desktop Business. Even more, nVidia would seem far more attractive in terms of money-making, thanks to CUDA and they server solutions. The Cell CPU is a world on it's own, so x86 in it doesn't seem quite right to me; the SPE's are RISC based and the PPE is a weird bug to me, lol. Besides, wasn't IBM supposed to kill the Cell CPU?

Cheers!

 

[gamerk316]

^^ IBM was run out of the market after they lost control of the underlying hardware [specifically, EISA bus winning against MCA bus]. From that point on, it was other companies that drove PC development [Both Extended and Expanded RAM were developed with no input from IBM, etc]

 

[-Fran-]

^^ IBM was run out of the market after they lost control of the underlying hardware [specifically, EISA bus winning against MCA bus]. From that point on, it was other companies that drove PC development [Both Extended and Expanded RAM were developed with no input from IBM, etc]

Are you talking about the developments for the PC in the Pentium MMX era?

Cheers!

 

[Cazalan]

One must hope that AMD doesn't drop out of the desktop cpu race alltogether, if they do, well, bye bye cheap cpus, seems to me Intel has no desire to make thier cpus faster as IB is the same clock speed as SB.

Do they really need faster right now? The key for IB is power consumption and energy efficiency. Getting the same speed for 1/2 the power is significant as it lets the processor move into smaller form factors. Cheaper power supplies. Less cooling needs. Longer life.

This is the first mass produced tri-gate chip. They will certainly be able to crank the speed if you raise the TDP.

The next step will be stacking 4GB of DRAM on the CPU. The power/cost savings will be massive.

 

[noob2222]

Do they really need faster right now? The key for IB is power consumption and energy efficiency. Getting the same speed for 1/2 the power is significant as it lets the processor move into smaller form factors. Cheaper power supplies. Less cooling needs. Longer life.

This is the first mass produced tri-gate chip. They will certainly be able to crank the speed if you raise the TDP.

The next step will be stacking 4GB of DRAM on the CPU. The power/cost savings will be massive.

which was pertty much the point I was trying to make. Without competition, Intel has no desire or reason to be faster, instead of putting money into R&D, just sit back and collect money, raise prices, and rake in the profits.

Intel still appears to be removing features like hardware virtualization on its K-models for whatever reason, as well as support for it trusted execution technology. It's funny how we end up paying more for less when it comes to Intel's high-end, overclockable CPUs

Read more: http://vr-zone.com/articles/even-more-ivy-bridge-details-leak/14082.html#ixzz1fEFOlv8l

And 1/2 the power ... lol, 95W TDP to 77W TDP, not seeing that one other than Intel's overhype of 50% less power (maybe they meant at idle.)

 

[fazers_on_stun]

I don't recall that but that don't mean it weren't so.

IIRC it's quite a common theory over on AMDZone 😀. But that doesn't mean it were so 😛..

Anyway, as mentioned previously, if Intel is really making as huge GPU performance jumps as alleged in the IB thread, then obviously they didn't need to buy a whole company to achieve it - just a few key engineers 😛..

Haswell is due out in about 15 months from now - if it really does have a 7X jump over SB's GPU, as alleged by Intel, that would put it in the current top-end discrete card range. There would be no reason for anybody to buy a discrete video card, except maybe for extreme gamers going for tri- or quad-SLI or XFire.

 

[fazers_on_stun]

One thing I will add on AMD's not competing with Intel, its more of an announcement on how stupid their marketing team was on trying to make people believe that BD would outperform SB-E.

One must hope that AMD doesn't drop out of the desktop cpu race alltogether, if they do, well, bye bye cheap cpus, seems to me Intel has no desire to make thier cpus faster as IB is the same clock speed as SB.

IMO AMD won't suddenly "drop out" of DT and/or server - they will just shift more R&D dollars to the new target areas at the expense of the former. So I would still expect PD and Trinity to appear, only not as fast as previously. Of course, AMD is not bound by previous statements - recall their promising that Barcelona would be backwards-compatible with QuadFX, then changing their minds and leaving those unfortunate (and few) mobo purchasers with no upgrade path. IIRC our AMD CR came dangerously close to buying a QuadFather mobo 😛.

Long-term, I think AMD wiill be evaluating their market position frequently and make adjustments as necessary to ensure profits. Which is PR-speak for watcing what happens and making new plans if needed 😛.

As for Intel, basically they are just competing with themselves already. They have to offer something significantly new every so often, in order to generate market interest and sales. Otherwise customers like myself would continue with their C2Q CPUs. I haven't bought a new DT CPU since my Q6700 purchase almost 5 years ago.