When will AMD take the crown again?

[Byte]

There really isn't much of a "value" crown. I enjoy a value much more than most people, however it just makes you look cheap. When AMD had the crown in the P4 days, Athlon meant you were an enthusiast. You knew what you were doing, and a MAVERICK. Now it just means you don't have the money to buy an Intel chip. I use to run nothing but AMD in the P4 era, but now, running AMD for anything than servers is just a bit sad. It gets worst in the notebook sector. There is pretty much no reason to go for AMD other than its cheaper. You get better battery life and performance esp with Penryns. The excitement of the i7 doesn't bold well for AMD. I'm really looking forward to it, but interested to see how many bugs the 1st gen will have. Not sure if i'll be able to wait for 2nd gen. Hopefully AMD will have some nice surprises with the Fuzion. Some CS4 suite acceleration is what i'm hopping for.

 

[Idontcare]

Originally posted by: harpoon84
Yes, that's what I meant, as in the new 'tock' or uarch.

Yeah, not that the difference invalidates the point you were making, I was just curious if you really were envisioning Bulldozer being AMD's flagship product to compete with Sandy Bridge.

Although when you think about, they are screwed. Here we are talking about whether Bulldozer will be out to compete with Nehalem or Westmere and AMD hasn't mentioned anything about what's beyond bulldozer to compete with Sandy Bridge.

 

[KingstonU]

What was the main cause to Bulldozer's delay by a year? R&D $$$ cuts? Will Bulldozer be on 45nm or 32nm?

 

[harpoon84]

Originally posted by: Idontcare

Originally posted by: harpoon84
Yes, that's what I meant, as in the new 'tock' or uarch.

Yeah, not that the difference invalidates the point you were making, I was just curious if you really were envisioning Bulldozer being AMD's flagship product to compete with Sandy Bridge.

Although when you think about, they are screwed. Here we are talking about whether Bulldozer will be out to compete with Nehalem or Westmere and AMD hasn't mentioned anything about what's beyond bulldozer to compete with Sandy Bridge.

Which is why I'm surprised every time someone steps out and states authoritatively that AMD will come back, using past successes in K7 and K8 as examples. I mean, seriously? The world has changed, the industry has changed, past successes mean nothing at this stage.

I think in the grand scheme of things, K7 and K8 will turn out to be exceptions to the rule, AMD got lucky and caught Intel napping, but now that the 'giant has awaken' so to speak its hard to fathom how they can give up this lead - unless of course they get a rubbish CEO and let marketing dictate development as you pointed out earlier.

 

[eternalone]

To me and to everyone it shouldn't matter who takes the crown as long as we have competition, we dont want a dictatorship in the Cpu market.

 

[Wuzup101]

Originally posted by: eternalone
To me and to everyone it shouldn't matter who takes the crown as long as we have competition, we dont want a dictatorship in the Cpu market.

Yup, this is exactly what I was going to say. I don't give a rats arse who's CPU I have in my machine and the machines I build (provided it works). However, we need competition to foster a healthy industry. Hell, I'd love to see another company start making processors (though that is HIGHLY unlikely at this time).

I just got my first Intel system since the P3 days up and running. It replaced my 5 year old AMD XP 2500+ (barton) based system that I built around the summer of 2003 (and to some extent my ancient G4 mac powerbook... which will still see some use). I have to say, this q6600 is very impressive, though since I made such a huge jump, pretty much everything was guaranteed to impress me.

In any case, the value market is important to the companies, even if it isn't important to the "enthusiast." Intel and AMD don't make the majority of their money selling their high end processors (even with the crazy mark-up Intel has on the X series stuff). They make the money in the value segment, selling processors for $300-700 desktops and notebooks. If you want healthy competition, you should be concerned about the value segment of the market (even if you will never buy a "value" processor).

 

[harpoon84]

The problem for AMD is that they simply cannot survive on 'value' CPUs alone - unless they restructure and become a '2nd tier' CPU manufacturer like VIA, which would mean drastically cutting down R&D costs and would basically make Intel a monopoly in anything but the absolute low end.

 

[Idontcare]

Originally posted by: Wuzup101
In any case, the value market is important to the companies, even if it isn't important to the "enthusiast."

Intel and AMD don't make the majority of their money selling their high end processors (even with the crazy mark-up Intel has on the X series stuff).

Truthfully it is your second sentence that invalidates the first, from what I have observed in a healthy majority of the types of discussions that exist in these forums.

That is to say most enthusiasts recognize Intel's performance dominance when it comes to what CPU is going to be inside their enthusiast DIY computer sitting under their desk today...and at the same time they recognize the fact that next year's or two years from now the processor they are likely to step into will be determined by the health of the balance sheet being generated by today's value-market.

Hence the persistent debate over the health and viability of AMD's current value-market relative to Intel. It's not because we enthusiasts are trying to argue AMD's current value chips are competitors to Intel's performance chips, but because we are trying to gauge the trajectory of both AMD and Intel based on the health (or lack thereof) of the cashflows (and thus today's R&D for next year's products) being generated from those value-segments.

So it is important to the enthusiasts for it determines how quickly the leader is going to respond (price-wise and flagship performance wise) to the competition. We wouldn't have sub-$200 Q6600's if it weren't for AMD's situation with Phenom, nor $300 Bloomfields if it weren't for the projected performance envelope of Deneb. This matters to us, it directly shapes our world as the dog (volume value segment) wags the tail (enthusiast, top-dollar low volume segment).

 

[v8envy]

Originally posted by: harpoon84
The problem for AMD is that they simply cannot survive on 'value' CPUs alone - unless they restructure and become a '2nd tier' CPU manufacturer like VIA, which would mean drastically cutting down R&D costs and would basically make Intel a monopoly in anything but the absolute low end.

Oh, they can survive -- they've done it before. Just need to swallow their pride, jettison a whole lot of executive and middle management. And cut down on marketing costs. Letting Hector go is a step down that path.

Unlike the K6 days everyone knows who AMD is now. AMD-powered boxes have been a staple product on retail store shelves since at least 2005. Dekstop consuers now view the CPU as a commodity.

It's not a question of whether or not the AMD cpu will do the work of an Intel CPU (unlike VIA, whose platforms have questionable stability and reliability). It's just a question of which brand offers the best performance at a certain price level.

ATI has shown that it's possible to succeed in the low end and mainstream segments without having a top tier product. I think that's a lesson AMD will need to apply to CPUs as well.

 

[taltamir]

Originally posted by: harpoon84
The problem for AMD is that they simply cannot survive on 'value' CPUs alone - unless they restructure and become a '2nd tier' CPU manufacturer like VIA, which would mean drastically cutting down R&D costs and would basically make Intel a monopoly in anything but the absolute low end.

well... the chinese are about to unveil their own x86 cpu....

 

[Extelleron]

I don't think we can accurately make a statement right now as to whether AMD will ever have the fastest CPU on the market - there are too many variables involved, and too much that we don't know/can't accurately predict. For one we don't know anything about what AMD is doing on the desktop after the Deneb refresh and the release of tri/dual-core 45nm CPUs; Bulldozer is nowhere to be found in the roadmap and it is highly unlikely that we will see a 12-core desktop CPU, although a 6-core Istanbul for the desktop seems possible.

With Intel's lineup, we know a lot about what they are planning, but the question is execution - so far they have done well, but IMO at least there are some signs that Intel is slowing down. The time period between the release of enthusiast Nehalem/ 2P server parts and the release of mainstream parts is going to be huge - around a year at the least. I have even heard rumors that mainstream Nehalem parts could be pushed back to 2010.

Given that, it also seems to me at least that Intel is going to be slow with the 32nm transition. If Intel is not releasing mainstream 45nm Nehalems until late 2009 or even early 2010 it seems unlikely that they will be bringing out 32nm parts in 2H 2009. Also Larrabee is coming out in the late '09 - 2010 timeframe, and it is based on 45nm technology. Same thing with the initial Atom SoC CPUs, which will be released in the 2009-2010 timeframe.... and Atom is something that demands the newest process possible because of its low power / small size focus.

So Intel isn't infallible, we don't know for sure yet but it sure seems to me like Intel is slowing down, at least for the mainstream market. Even now with quad-core CPUs, I don't find Intel's current offerings to be that impressive TBH. If the Q6600 didn't exist (and eventually it is going to reach EOL) I'd probably be recommending AMD Phenoms right now, because the Q8200 is absolutely pathetic. Same thing with the Q8300 when it comes out later this year/ early next year at the same price point as the Q6600.

AMD has a chance with Deneb to provide superior performance/$, they will not get anywhere near the performance of a Core i7 965, but if the Q8xxx series is the best Intel can come up with for the low end, then they have a serious chance to provide the better choice in the <$250 quad-core market. If Bulldozer ever comes out.... then who knows, it may be fast enough to provide Intel with real competition.

So my answer would be a cautious yes, I do think that AMD will eventually beat Intel in performance sometime down the road. Eventually Intel will make a mistake and with a completely new architecture in Bulldozer, AMD has the chance to really turn things around.

 

[taltamir]

AMD lost their best employees. And a lot of run of the mill employees.
A company isn't magical name with power. Research takes talent, and with a fraction of their talent from before, they aint gonna make it.

Also intel is now seriously playing, intel needs to proove to court the AMD can not compete... while before intel wanted to spend the least amount of money to milk a market they already controlled.

 

[sonoran]

Originally posted by: v8envy
It's not a question of whether or not the AMD cpu will do the work of an Intel CPU (unlike VIA, whose platforms have questionable stability and reliability). It's just a question of which brand offers the best performance at a certain price level.

Very true today. But what happens when Larrabee comes out? I can't help but believe it will have all kinds of snazzy graphics optimization instructions. Will the competition be able to compete with that? Will x-86 processor compatibility go out the window? Shoud be interesting times ahead...

 

[nyker96]

they are mysterious on Fusion products, so maybe they got a nice one and didn't want Intel spies to know and hope Intel will just sit back and relax for a while. :] yeah, no clue, I only hope they got enough to keep surviving at this point much less overtake Intel.

 

[fire400]

Originally posted by: Ocguy31
I thought it was funny that you brought up GPUs after you quoted me in that other thread, Myo.

I took that bait hook-line-sinker.

link to the GPU was hilarious.

Intel will probably take both crowns with CPU/GPU so far up, nVidia and AMD/ATI will merge, and later go bankrupt. Then IBM makes a come back from the 70's and 80's and buys out Intel.

Microsoft finalizes it's monopoly and takes IBM/Intel and makes another 3DFX company just to show off at how much money it has.

After they've really had it, they buy SuperValu and 3M, they become Fortune-1 USA.

It's over... it's been over since Bill Gates dropped out of college, guys. Face it, you can't beat Microsoft, "Microoked-soft."

Okay. Intel is so far ahead, Nehalem will destroy.

 

[nonameo]

Plan: Make specialized laptop CPU, eventually have a variant replace phenoms on the desktop. At the same time, phenom becomes even more server oriented.

 

[Idontcare]

Originally posted by: nonameo
Plan: Make specialized laptop CPU, eventually have a variant replace phenoms on the desktop. At the same time, phenom becomes even more server oriented.

Sounds a lot like Via's plan with nano, or Intel with atom.

Or do you mean fusion? That is definitely unique at this point, unless Intel beats them to punch.

 

[taltamir]

Originally posted by: fire400
Intel will probably take both crowns with CPU/GPU so far up, nVidia and AMD/ATI will merge, and later go bankrupt. Then IBM makes a come back from the 70's and 80's and buys out Intel.

mmm... i partially disagree.
Intel will take both the CPU and GPU crowns... and the SSD crown (eventually being the ONLY maker of HDDs), and also... and the chipset crown, and the system on a chip crown (for phones), and what else? I know there was another big crown they are taking.

Anyways. AMD and nVidia would be gobbled up by foreign interests.

China is about to unveil its x86 CPU, russia, the EU, and several oil rich arab countries would be ideal targets for gobbling up nvidia and AMD and making their own chips. (heck, they already own large shares in AMD)

 

[secretanchitman]

i dont think intel has the gpu market (at least mid and high end - thats where ati/nvidia dominate), but it DEFINITELY has the cpu and chipset market.

amd just needs to pull a core 2 duo/conroe and i think they'll have a winner. but then again, its looking like intel is going to pull yet ANOTHER core 2 duo/conroe with core i7...

 

[harpoon84]

Originally posted by: Idontcare

Originally posted by: nonameo
Plan: Make specialized laptop CPU, eventually have a variant replace phenoms on the desktop. At the same time, phenom becomes even more server oriented.

Sounds a lot like Via's plan with nano, or Intel with atom.

Or do you mean fusion? That is definitely unique at this point, unless Intel beats them to punch.

Isn't there a version of Nehalem coming out next year with an integrated GPU? That'd be roughly the same timeframe as Fusion right?

 

[taltamir]

Originally posted by: secretanchitman
i dont think intel has the gpu market (at least mid and high end - thats where ati/nvidia dominate), but it DEFINITELY has the cpu and chipset market.

amd just needs to pull a core 2 duo/conroe and i think they'll have a winner. but then again, its looking like intel is going to pull yet ANOTHER core 2 duo/conroe with core i7...

where are they gonna pull it out? their best engineers quit, meanwhile intel MULTIPLIED its R&D spending.

 

[quadomatic]

those Phenom's that are supposed to OC really well are supposed to come out soon right? Maybe those will get them back on top?

 

[harpoon84]

Deneb? It doesn't like like it will even beat Yorkfield... they won't be back 'on top' anytime soon, unless they pull a miracle.

 

[Cookie Monster]

I dont like the idea of integrating intel GMA IGPs into nehalem variants. What a waste of transistors..

 

[myocardia]

Originally posted by: Idontcare

Originally posted by: nonameo
Plan: Make specialized laptop CPU, eventually have a variant replace phenoms on the desktop. At the same time, phenom becomes even more server oriented.

Sounds a lot like Via's plan with nano, or Intel with atom.

I was thinking it sounded exactly like the Core/CoreDuo/Core2Duo/Core2Quad/Nehalem plan myself.