Could AMD die shrink thuban and add two cores?

[mosox]

Yep but many enthusiasts would not buy AMD even if it were 10% faster, 10% cheaper and 10% more power efficient. Intel has a stronghold on this market through the distributors, and it wouldn't be the first time AMD is better and the sales don't go up.

IMO AMD had too many things going at once, I don't remember any CPU manufacturer releasing in one year three platforms (Bobcat, Llano, Bulldozer). AND some new Deneb and Thuban CPUs, while Llano is mobile and also desktop which makes a total of 4 platforms in one year. And I forgot the server CPUs.

 

[BlueBlazer]

As a server guy myself, this is what I've been waiting to see. Unfortunately, I have not seen such reviews yet.

Agree, the SPEC benchmarks are nowhere to be found. :hmm:

They dont even need that much cache - if I was them I would probably investigate maybe giving each core 256 or 512kb of L2 cache but keeping the 1mb per core of L3 cache. That could save them precious die space.

End result would be something much faster than Zambezi at every task, while also being cheaper to produce and consuming less power.

Here's a good reason >> Clock for clock: Cinebench R11.5 64 bit.......

When the clock speeds are matched the performance differential between the Core i5 2500k and FX 8150 is very close, 7.29 points compared to 7.32 points, in favour of the FX 8150. It is worth reiterating that the FX 8150 has eight physical cores while the Core i5 2500k only has four.

Surprisingly, even at 4.3ghz, the last generation Phenom II X6 1100T outperforms both the Core i5 2500k and FX 8150 – scoring 7.51 points. This is particularly interesting on an architectural level as the FX 8150 has two extra physical cores when compared to the X6 1100T.

 

[jvroig]

because enthusiasts spend 41% of the total 16.6 Billion in hardware

Enthusiasts spent 41% of the total 16.6 Billion PC gaming hardware. This is not the total PC hardware figures, at least, not according to the slide. This does not take into account how big revenues are for the non-pc gaming portion of hardware revenue.

That's where we don't get each other.

You are saying they care because PC enthusiasts spend 41% of the total PC gaming hardware, and AMD seems to have pointed it out as important

I am saying that it is all marketing, and without the proper context (the entirety of the PC hardware market, not just gaming), we don't know if the enthusiasts are really worth bothering about or they are just a small blip compared to the chunk of low-end and mainstream that AMD serves well with Zacate, Ontario, and Llano.

From that 16.6B AMD would very much like to grab the most they can buy selling CPUs, Motherboard Chipset, GPUs and more. So yes they care about enthusiasts and that is why they show gaming benchmarks in High resolutions in their marketing Slides

You think it does, but it doesn't. At least, not for bulldozer. GPUs? Yes. But in the context of our discussion (Bulldozer not really made for enthusiasts), then no, it doesn't mean what you think it means.

First, as I've repeated a few times already, where is the breakdown that shows how much of the revenue is from enthusiast market? I doubt it can exceed or rival the low-end and mainstream, but who knows, so I'm looking forward to the data to show us.

Second, that slide does not distinguish between processors and cards. It's all meshed together. It may in fact be important to AMD, but not to their CPU division, for all we know. Again, without the breakdown, we can't say.

So the conclusion is the same: That slide is just same-old marketing. It tries to convince you of something, but it offers no proof at all beyond serious marketing effort for bulldozer. At no point does it tell you that "We at AMD CPU division hold enthusiasts in high regard because they contribute a significant XX.XX% of our revenue!". You think it does, but look at it closer. It offers you nothing but an estimate of the entire PC Gaming hardware, which CPUs only consist a portion of.

That is not proof that Bulldozer was designed to cater to the CPU enthusiast market because it is important to AMD.

 

[Red Storm]

16.6 Billion in hardware sounds nice, until you remember the CPU is only one hardware component out of many, and - going purely from my own personal experience - it does not get changed out as frequently as other parts. I'd like to see how much of that 16.6B is actually CPUs.

 

[nenforcer]

You said they (AMD) dont care about enthusiast space, i said they do care (lets agree at least for the Gaming market) because enthusiasts spend 41% of the total 16.6 Billion in hardware.

From that 16.6B AMD would very much like to grab the most they can buy selling CPUs, Motherboard Chipset, GPUs and more. So yes they care about enthusiasts and that is why they show gaming benchmarks in High resolutions in their marketing Slides

The first blog counter from the horses mouth "Our Take on FX"

http://blogs.amd.com/play/2011/10/13/our-take-on-amd-fx/

AMD is quoting benchmarks from the BF3 beta in anticipation of the launch in a couple of weeks and the numerous ugprades (GPU + CPU) it will likely inspire.

I know myself am waiting to upgrade shortly after.

 

[soccerballtux]

it turns out people won't have any trouble handling this much power.

 

[Ajay]

That said, we can't discount project momentum, group psychology, denial, and the universal trait that humans share called hope.

Who knows how long they languished, paralyzed like a deer caught in the headlights, before someone pulled together a Plan B. They may have simply not.

^^This sums it up for me. That, and the probability that the design team expected more out of GloFo's 32nm process, means they just soldiered on.

Their plan B appears to be just improve what they can in Piledriver. Though a year to get another 10-15%, with part of that coming from process improvements, simply isn't enough.

Again, this means that the Arch/implementation is just fatally flawed and they don't have the R&D $$s/manpower to do anything radical about it till sometime after 2014.

AMD is kind of stuck, like Intel was on Netburst (but without Intel's awesome FAB capability)

 

[pmccall2]

They could have, and I would have gladly accepted it if I knew BD was going to be this much of a turd.


I just hope AMD realizes thier mistake and doesnt put this peice of shit in Trinity.

 

[Ajay]

They could have, and I would have gladly accepted it if I knew BD was going to be this much of a turd.

Since it didn't happen, we'll never know how that would have turned out, though on the face of it, it would have at least come out much earlier.

I just hope AMD realizes thier mistake and doesnt put this peice of shit in Trinity.

It's too late to do anything about that without pushing out their schedule another 18+ months - unless they already did a K10 design and shelved it doing a BD design instead.

I've read somewhere in that Trinity will use Piledriver cores - that should help, plus the GloFo process will be more mature. This chip is more important to AMD since it will go into all the <$600 desktops that people are buying these days - so it could give AMD allot of volume.

It's pretty clear at this point that BD was aimed at the server market first and enthusiast market second. Hopefully BDII will bring a bit more to the enthusiast market than AMD's graphs show (at least that's what I've taken away from some of the speculation on the RWT forums).

 

[Maximilian]

I am very interested in how trinity turns out if it uses piledriver.

 

[IntelUser2000]

It's pretty clear at this point that BD was aimed at the server market first and enthusiast market second.

I don't think this was their aim, it just happened to be that way.

Speaking of enthusiasts and importance: Enthusiast markets are how companies differentiate themselves. Otherwise, Intel or AMD wouldn't matter. Everything will be same, with a different badge and will all be bland. Remember how Acer was number one in its strategy for going all out in the race to the bottom? Now they are losing marketshare, and I don't think its a coincidence.

Enthusiasts may be only 1&#37; in terms of population. But they buy most of the high margin, >$150 CPUs and >$200 graphics cards. Revenue share might be 5% rather than 1%. See how companies strive for that extra 1% in earnings reports to beat analyst expectations?

Chips for average joes: Its to meet the expectations for the company
Chips for enthusiasts: That extra difference to make them stand out

 

[hasu]

With massive multi-tasking capability, AMD should tune Linux to improve its performance specifically for their chip. AMD and Linux together can change the way people think about Servers. Microsoft will be forced to make Windows Servers better to maintain their server market share.

I see a great deal of improvement when I run VirtualBox 4.x on Linux Mint compared to Windows 7 (both 64bit on AMD CPU). Not sure if the change is only because of EXT4 vs NTFS on the host file system. I did not do any benchmarking though.

 

[RobertPters77]

They should've done that in the first place.

Or the very least not tease the consumer with BD until they new it was good and ready and could stand toe to toe with Intel's offerings.

If PD(Pile Driver) does turn out to be good, then why the hell did they not release PD in the first place?

Classic Amd for you. Self Destructive with no Sense of Direction.

 

[3DVagabond]

They should've done that in the first place.

Or the very least not tease the consumer with BD until they new it was good and ready and could stand toe to toe with Intel's offerings.

If PD(Pile Driver) does turn out to be good, then why the hell did they not release PD in the first place?

Classic Amd for you. Self Destructive with no Sense of Direction.

Shrinking Thuban might not have been an option at the time they realized BD wasn't going to achieve it's design targets. Piledriver isn't ready yet and they made the decision to release what they had now rather than delay any further.

I wouldn't label it as classic AMD. Other companies have done this before as well.

 

[GaiaHunter]

Lets face it - the percentage of people that can benefit from 8 cores over 6 cores in the desktop is very small.

So AMD wouldn't have to add 2 cores to Thuban - a die shrink and higher clocks (if allowed by GF process - Llano results are ambiguous, especially due to the TDP limiter but the Llano with the GPU fused seems more promising) in conjunction with lower prices achieved by smiler die size could be a compelling product.

I wouldn't find surprising if AMD studied a Thuban shrink while they researched/executed the Dened shrink - so maybe that would make it possible (in terms of a quicker production/validation/release).

But I doubt it is going to happen.

 

[3DVagabond]

Lets face it - the percentage of people that can benefit from 8 cores over 6 cores in the desktop is very small.

So AMD wouldn't have to add 2 cores to Thuban - a die shrink and higher clocks (if allowed by GF process - Llano results are ambiguous, especially due to the TDP limiter but the Llano with the GPU fused seems more promising) in conjunction with lower prices achieved by smiler die size could be a compelling product.

I wouldn't find surprising if AMD studied a Thuban shrink while they researched/executed the Dened shrink - so maybe that would make it possible (in terms of a quicker production/validation/release).

But I doubt it is going to happen.

I think most people who benefit from 6 cores could use 8 cores because if your apps scale to 6, they'll likely scale to 8. I am of the opinion that 4 cores/threads is all most people need.

 

[mrjoltcola]

It isn't like nobody is buying the Bulldozer. A few trips through Microcenter or Bestbuy in the pre-built areas and you overhear all sorts of uninformed customers asking what a core is, followed by salesmen who really can't explain it well. I think those people are prime sucker... er I mean candidates...

Don't underestimate sheer marketing to move those chips and recoup something.

What are they supposed to do? They had to ship something. Every month they didn't meant another month with less AMD options in front of a customer. AMD loyalists will buy it, or finally decide on the 1100T. Remember all of the people saying "well I'm waiting on Bulldozer". Well, they aren't waiting now. The uninformed will buy it, most will actually be happy with the performance (as long as they don't look at a CPU benchmark comparison) and in the end, it'll sustain AMD long enough to fix the problems. That is my theory anyway.

 

[Vesku]

It does tick a lot of brand name PC marketing boxes.

 

[RobertPters77]

@3dvaga

I would've definitely bought a 150$ Octo thuban.

I don't know of many other companies that release products that are inferior to their predecessors. Unless they do it for cost.

 

[3DVagabond]

@3dvaga

I would've definitely bought a 150$ Octo thuban.

I don't know of many other companies that release products that are inferior to their predecessors. Unless they do it for cost.

I would've paid more than that for it.

 

[pmccall2]

Shrinking Thuban might not have been an option at the time they realized BD wasn't going to achieve it's design targets. Piledriver isn't ready yet and they made the decision to release what they had now rather than delay any further.

I wouldn't label it as classic AMD. Other companies have done this before as well.

With as many delays and problems BD even had before launch im surprised AMD didnt think of doing it just in case BD got delayed again, especially since they were working on stars for Llano anyhow.

They gambled on BD improving before launch, and lost horribly. Delaying BD again would have been unpopular, but not nearly as bad as throwing BD out there, knowing that it was a turd, and relying on marketing to polish the turd as much as possible.

 

[BladeVenom]

I think most people who benefit from 6 cores could use 8 cores because if your apps scale to 6, they'll likely scale to 8. I am of the opinion that 4 cores/threads is all most people need.

And most people will never need more than 640kb of memory.

 

[3DVagabond]

And most people will never need more than 640kb of memory.

Fair enough, but I never said they'll never need more than 4 cores.

 

[RobertPters77]

I would've paid more than that for it.

It's not worth more than that.

Hey if Amd released a 100$ Hex core they'ed corner the damn market.

 

[Ajay]

I don't think this was their aim, it just happened to be that way.

I meant that it was more important for BD to hit it's server targets than the high end CPU targets. I'm sure they had hoped to hit both, but it didn't and they weren't going to sacrifice server performance for desktop performance (in terms of what was likely done in the last few steppings).