Advanced Micro Devices May Be Acquired – Analyst

[clue69less]

I have a very unique sieve called a bs sieve

Around here, it probably gets plugged up real often.

 

[zeezee]

We love AMD ,they just got a bit distracted,and failed to keep up.

You probably mean Intel finally caught up after four years

 

[zeezee]

We love AMD ,they just got a bit distracted,and failed to keep up.

You probably mean Intel finally caught up after four years

the only difference between cakes and doughnuts is the formfactor :lol:
It also depends on how it's served. Cold or HOT

 

[InteliotInside]

We love AMD ,they just got a bit distracted,and failed to keep up.

You probably mean Intel finally caught up after four years

Ummmm you are little off.... K8 debuted in Sept 03, so it was about 2.5 years of AMD holding a significant lead. The actual lead really widened up with the X2 series, that was a nice bit of design work going to the 3800+ and up.

However, if you think about it.... AMD never had a mobile competitor and if you clocked up a Banias or Dothan chip to K8 speeds then Intel actually had the better CPU --- why they did not scrap Prescott and socket up a Dothan to desktop is odd to say the least.

jack

Indeed. The Pentium 4 HT did perform on par and even better than single core A64s. The X2 line really thrashed NetBurst.

I think part of the reason why Intel never bumped Dothan up to desktop level is because of Intel's management. Back then Intel's management was primarily made up of people from the marketing sector, and they were the ones who wanted to keep clock speeds up to attract consumers to the Pentium 4.

Now however, after Intel reshuffled its board, it seems to be on a mad rampage for performance/watt and performance/dollar.

 

[InteliotInside]

We love AMD ,they just got a bit distracted,and failed to keep up.

You probably mean Intel finally caught up after four years

Ummmm you are little off.... K8 debuted in Sept 03, so it was about 2.5 years of AMD holding a significant lead. The actual lead really widened up with the X2 series, that was a nice bit of design work going to the 3800+ and up.

However, if you think about it.... AMD never had a mobile competitor and if you clocked up a Banias or Dothan chip to K8 speeds then Intel actually had the better CPU --- why they did not scrap Prescott and socket up a Dothan to desktop is odd to say the least.

jack

Indeed. The Pentium 4 HT did perform on par and even better than single core A64s. The X2 line really thrashed NetBurst.

I think part of the reason why Intel never bumped Dothan up to desktop level is because of Intel's management. Back then Intel's management was primarily made up of people from the marketing sector, and they were the ones who wanted to keep clock speeds up to attract consumers to the Pentium 4.

Now however, after Intel reshuffled its board, it seems to be on a mad rampage for performance/watt and performance/dollar.

I made a similar postulate a while back --- the product release and line up through the past few years until C2D seemed to be just dumb decisions --- my thought was this sounded like a board room full of marketers looking at a list of choices for next designs, and they sat their like a pack of cackling hienas 'give me speed, give me clock', and a loan engineer was probably in the back going 'this is a dumb idea', but was never heard... one thing P4 would do was scale to high clock speeds --- too bad it could only do 0.3-0.9 IPC.... at least the scenario I envision.

The decline in 2006 (first 1/2), and the ensuing struggle to sell P4s, is well deserved, it was a dumb move.... and a horrid product. On top of that, it was clear that AMD caught Intel off guard with the speed at which they got to dual core --- K8 is really a good design.

Yup yup.

I think DailyTech or TG daily reported that engineers from Israel were the ones who made the NGMA/Core arch, even before the Pentium 4 started to run aground in the server market. However, management encouraged all engineers to continue to ramp up clock speeds.

Intel got hit really badly in the server market and less so in the desktop market. I think Opteron took 50% market share away from Intel in just one generation, which was pretty much the nail in the coffin for NetBurst and the marketing execs who encouraged clock speeds.

Now we have Conroe and Woodcrest (forgot the laptop version, I need my coffee XD), and the Penryn derivatives coming up. This the best time to be an Intel fan since the Pentium III days. 😀

 

[InteliotInside]

To be fair, you'd look at AMD as the evil no gooders if they were in Intel's position. The only reason why AMD seem like the good guys is because they're the little guys who have to settle for less and more moderate goals.

2008 seems to be in Intel's home court, with the release of the new Nehalem architecture. 45nm high-k process, new Intel bus link (CSI tech), DDR3, native quad core design and either an IMC or a dual die octo core processor (they would opt for the latter, IMO). Also, don't forget that this is a new arch in general, not just a revision, so we may see HUGE performance gains.

AMD will just finish Barcelona this year, and I doubt they have the capacity to design a whole new architecture to combat with Intel. Not to mention that they don't enough FABs to pump out high yield wafers in huge amounts (which results in cheaper prices).

 

[InteliotInside]

Morally speaking all companies are the same. Shareholders want their companies to rake in the biggest net profits (NetBurst of money, anyone?), to receive higher dividends and investment rates.

Of course we need competition in every market, and I hope AMD and Intel stay in the game with each other. The same goes for NVIDIA and ATI, Nintendo and Sony and etc...

 

[BaronMatrix]

Baron, it is not simply 'who is qualified' to determine if something is infantile or demented --- it is really simply just plain obvious.

But you keep us entertained.

Baron does so much more than that. He is single handedly responsible for more first time users creating logins, more hostile posts, higher hit numbers for the site, longer pointless threads etc etc etc. While his words demonstrate lack of intellectual capacity and lack of social ability, the results of his words could easily indictate is is a schill, for any of AMD, Intel or THG.

After almost 14 months of being caught in his own mistakes and relentlessly proven wrong, abused by the vast majority of users and generally disdained by everyone, you would figure he would have went away. That he hasnt leaves one wondering as to whether he's being paid to be here or he has some extreme personality or intelectual disorder.


I told you if any aliens or any higher intellects come here, I don't want to be associated with you. That would be embarrassing. I do what I do to maintain at least a little bit of unbiased commentary. You only get mad cause I keep turning out to be right.

Baron, could you please provide a link from a time you were right?? I have very rarely seen it .... it is this form of narcissism that just cracks me up.

Thanks,
Jack

Searching for links is your job. If you search for platforms and me you will see I said AMD would release several platforms after QFX (can you say Live! Home XXXX? or the new SFF form factor?) I also said that with the improvements Barcelona would be around 40-60% faster than K8 (AMD reported 40% - 1.8x dual core)

Even farther back I said Charlie would be wearing a bunny suit and that Intel would see severe drops in profits - along with AMD.


Stalk someone else. Or remind everyone how no matter what materials you use you should always ground yourself and have the best drivers and HW.

 

[InteliotInside]

Nice finds, JJ, and thans to VD.

Hey, do you guys know if Intel is interested in undertaking anyone in the finance sector? My life goal is to work at Intel in the financing/management departments, but I don't know if they'll be looking for anyone like me in a few years. Stuff like economics, accounting, business management and stuff.

My friend wanted to work for NVIDIA (as an engineer), but after I told him that Intel are looking for engineers in the GPU department he decided to focus his time on that. XD

 

[wolverinero79]

I do what I do to maintain at least a little bit of unbiased commentary. You only get mad cause I keep turning out to be right.

Thanks for providing the comedy. It's the same reason I like to look at the opinion articles on things like the Michael Moore website :-D. People enjoy posting "information" that reveals their lack of knowledge or logical thought. Why is that constantly funny? 😀

 

[zeezee]

However, if you think about it.... AMD never had a mobile competitor and if you clocked up a Banias or Dothan chip to K8 speeds then Intel actually had the better CPU --- why they did not scrap Prescott and socket up a Dothan to desktop is odd to say the least

Isn't the above what Conroe pretty much is?

I happened to have used both Banias and Dothan laptops - actually, I still use the Dothan Laptop at work. Not that they are bad CPU's. They worked adequately in my case, particularly with their low power usage. They only lacked a little bit of stamina if you know what I'm saying.

I guess it took a while for Intel to finish the final tune-up of their design for a good desktop processor.

 

[qcmadness]

However, if you think about it.... AMD never had a mobile competitor and if you clocked up a Banias or Dothan chip to K8 speeds then Intel actually had the better CPU --- why they did not scrap Prescott and socket up a Dothan to desktop is odd to say the least

Isn't the above what Conroe pretty much is?

I happened to have used both Banias and Dothan laptops - actually, I still use the Dothan Laptop at work. Not that they are bad CPU's. They worked adequately in my case, particularly with their low power usage. They only lacked a little bit of stamina if you know what I'm saying.

I guess it took a while for Intel to finish the final tune-up of their design for a good desktop processor.

I don't think so.
Intel needs a better architecture than K8 so that they can win back the performance crown. P6+ (Banias / Dothan) just win marginally at similar frequency to K8.

 

[zeezee]

and it wasnt four years,its was just over 3. not even that if you count dec 05 core products and the narrowing of the gap with the last of the d series @65nm.32 bit core2 chips were keeping pace with amd making the spread just under 2 years;which even by moores considerations is pretty right on.

Well... I streched it to Willamette days but okay, let's say three years or two years. Absolutely fine with me.

BTW, as everybody here very well knows, they are still selling Netburst cores today No complaints. I bought a Smithfield the other day - dirt cheap. It doesn't create miracles but it does what it's supposed to do.

 

[zeezee]

Intel needs a better architecture than K8 so that they can win back the performance crown. P6+ (Banias / Dothan) just win marginally at similar frequency to K8.

That's where we disagree. Conroe as far as I'm concerned, is not a brand new design. Effectively, it's a Banias with muscles and accessories (and arguments like Conroe's prefetch algorithm is 10% more efficient than Banias/Dothan or it has one more FPU, ect. don't really change my point).

 

[gOJDO]

Searching for links is your job. If you search for blah...blah....blah...blah....blah...blah....blah...blah..... the best drivers and HW.

 

[qcmadness]

Intel needs a better architecture than K8 so that they can win back the performance crown. P6+ (Banias / Dothan) just win marginally at similar frequency to K8.

That's where we disagree. Conroe as far as I'm concerned, is not a brand new design. Effectively, it's a Banias with muscles and accessories (and arguments like Conroe's prefetch algorithm is 10% more efficient than Banias/Dothan or it has one more FPU, ect. don't really change my point).

The improvements:
1. Wider execution units
2. Shared cache systems
3. Better prefetch allgorithm
4. 64-bit support
5. Single-pass SSE units
......

The improvements are not brand new, but as massive as K7 => K8.

 

[zeezee]

you bought a what? a smithfield floor heater? :lol: :lol: hows that working out :lol: :lol:

It's out of my sight, in the attic running an IMAP server. I put about four fans in the case and inside the case the temperature is barable but outside... ...you don't wanna know I had to turn off the heating in that room and THIS IS NOT A JOKE

 

[zeezee]

The improvements:
1. Wider execution units
2. Shared cache systems
3. Better prefetch allgorithm
4. 64-bit support
5. Single-pass SSE units
......

The improvements are not brand new, but as massive as K7 => K8.

All of the above are better fine tuned features. Like you said, "WIDER" execution units, "BETTER" prefetch algorithm but architecturally, they are same/similar. As for 64-Bit support, even the late Prescotts had it.

In my head, the difference between the Core and Netburst is like a Gas and Diessel engine. Conroe versus Banias is more like a four-cylinder versus V8.

Following the same analogy, K7 to K8 is more like a new transmission on the same engine technology (OMC and HT/X-Bar though the latter is useless with single cores and/or desktops) - maybe a little increased engine volume.

64-Bit is like a sports spoiler. Looks good but useless in practice, at least for now.

Anyway, it's just a matter of definition. Conroes are far superior to Banias or Dothans whether or not the reasons are brand new features or improvements.

 

[Whizzard9992]

I'm sure there are a lot of undocumented features as well.

One thing I liked in particular is how C2D actually shuts down power to a portion of a bus (and associated resources) if it's not in use; even if it's just as short as an instruction cycle. For example, if an instruction uses 64 bits of a 128-bit bus, 1/2 of the bus will be "powered down" for the duration of the instruction.

That's definately not a tweak.

 

[BaronMatrix]

you bought a what? a smithfield floor heater? :lol: :lol: hows that working out :lol: :lol:

Oh and don't forget AMD has a robust SuperComputer business, XO is about to ship in volume, AMD is estimating 30% of the server market with Barcelona, and more which eans they will not need to bought out.

At least some of the Motley Fool guys realize what a sweet stock AMD is right now.

 

[qcmadness]

I'm sure there are a lot of undocumented features as well.

One thing I liked in particular is how C2D actually shuts down power to a portion of a bus (and associated resources) if it's not in use; even if it's just as short as an instruction cycle. For example, if an instruction uses 64 bits of a 128-bit bus, 1/2 of the bus will be "powered down" for the duration of the instruction.

That's definately not a tweak.

A lot of small "tweaks" are not easy to make.
Also some of them are new.

 

[halbhh]

[
Ummmm you are little off.... K8 debuted in Sept 03, so it was about 2.5 years of AMD holding a significant lead. The actual lead really widened up with the X2 series, that was a nice bit of design work going to the 3800+ and up.

However, if you think about it.... AMD never had a mobile competitor and if you clocked up a Banias or Dothan chip to K8 speeds then Intel actually had the better CPU --- why they did not scrap Prescott and socket up a Dothan to desktop is odd to say the least.

jack

Another poster mentioned what I've always thought: Intel has really only tweaked and improved on that architecture. It's great to talk about a *new* arch ever two years. It's been a lot longer than 2 years. In all fairness, it's very likely that most of the really quick payoffs in arch are now in designs. Not that nothing else could be done!! But at the moment, I'm guessing it would just be more of the same. Further expansions of what was expanded, etc., and modest tweaking of the tricks like out-of-order. Of course, on the physics side, dramatic stuff is possible, as both Intel and IBM have shown just lately! But about a "new" arch every two years.....

I'd believe that *after* I saw it.

 

[zeezee]

Let's take it one step at a time before throwing a bunch of links and techno-babble thus confusing everyone and sound right or at least knowledgable. After all this is a public forum and what we write here better be understood by others.

I've been reading your posts for a while. You have a funny tendency to accept anything as "correct" or "accurate" as long as you provide a link to it. I might as well back my opinion on a seperate web page and link to it to convince you.

Shortly, a link is just somebodyelse's opinion which might be right or wrong. I won't bother chasing your links or provide them to you to convince you. What I say is my opinion and if you have something to say, just say it.

You argument is particularly weak and certainly in the miniority as most people view the C2D as a new architecture.

And you would be the president of the argument evaluation comitee? Backed with "most people"? No need to scare me with the "Public Opinion". I have no problem with going against it.

Comparing C2D to netburst is like comparing a orange to a jalepeno and saying both are sweet.

I am comparing Banias/Dothan to Conroe. Netburst was an example to show what a structural or architectural change meant. Bottomline, I agree with your above remark, just don't twist it into something that I didn't say.

...a bunch of techno-babble to impress the audiance about architectural revolutions in the Conroe design.

Search Anand. I remember an Intel spokesman even admitting that Conroe was made up of the best parts of the Banias and Netburst designs to which Annand added, the only piece taken from Netburst was the prefetch.

Link....well no links. I have no time to find them just to look right. You might as well rate the above as "Poorly Supported" or "Incorrect" in your evaluation.

You do a great disservice to AMD's product by exclaiming that it had the utter snot kicked out of it by an ancient architecture with just a hand full of tweaks.

I don't care. I am not at AMD's service. Nor Intel's. Do you happen to know anyone who might be at either one's service? Who at least sounds like he is?

 

[mabaty]

The AMD vs. Intel posts I guess are never going to die....Anyway it is funny because you can tell who the people are who understand both the tech industry and wall street and you can tell who the nerds/fanboys are who dont understand how a company is run, etc..

 

[nilepez]

I would like to note again that Apple looked at IBM,AMD, and Intel for supplying it processors and went with Intel. Everything that has happened so far since then was pretty much known to be going to happen by Steve Jobs because he research all the companies and they all told them what they had comming in the next couple years. Just a thought, maybe this includes AMDs ability to pull off 65nm and the K8L performance.......

I suspect what happened was Jobs saw what Core2 could do and knew that for a couple of years (or more) Intel would have the advantage once it came out. Will that last beyond 08? Doesn't really matter. Apple certainly wasn't going to move to AMD and have the slower CPU for all but 8 of 24-40 months (that being jan-aug of last year).

I hope AMD can pull an rabbit out of the hat in next 2 years.