AMD's roadmap rumors

[dmens]

LOL, it's the folding monkey. Was wondering when you'd show up. Hey no offense, but there's a big difference between comparing benchmarks/overclocking and talking about the guts of the machine. No doubt duvie is very good at the former, but not the latter.

Oh, didn't know mark spoke for zebo. Too new to know that, I guess.

 

[robertk2012]

Cant we all just get along. Maybe we could buy some Conroes in bulk ;-)

 

[Duvie]

Originally posted by: dmens
You're right, you've switched more than me, the last intel I used was a P3-550 back in college. It's really too bad if you don't see the technical aspects of my posts, probably because you don't understand the details. The most fun I have on AT is dissecting bullshit people like you post. Fun times.

I dont usually see anything technical in your bibble....

What part of technical data have you introduced in this thread???

instead of your flamebait comments about AMD fanboys above while stating the obvious to most (maybe not you)....NO one said k8L will beat a Conroe!!!

You are just plain trouble and like I said add very little to the cpu forum....

I have just been taking a break recently from the forums (self-imposed) not to call you out for the pain you are...

I took the break cause I was BORED with all the X2 this and opteron that threads....I want something new and fresh just not a whole bunch of speculation...It is like we are all working for the inquirer around here of late...

 

[swtethan]

i have intel discount :x dont tell anyone

 

[dmens]

If you've been taking a break and glossing over my crap, as you admit, then you wouldn't be able to see anything I post, technical or not, hahaha. Go ahead and do a search. I'll leave you alone now since you sound pained at being challenged or corrected. Trouble indeed, LOL.

 

[Duvie]

Originally posted by: dmens
If you've been taking a break and glossing over my crap, as you admit, then you wouldn't be able to see anything I post, technical or not, hahaha. Go ahead and do a search. I'll leave you alone now since you sound pained at being challenged or corrected. Trouble indeed, LOL.

NO....I read your crap....I just chose to not comment in many of the threads....NO I didn't miss your flamebait and attacking the AMD fanboys becuase you were obvious de-masculated by AMD dominance....

I realized what you were about long ago....


Edit: And what have I been corrected on??? Go talk to PM and he will tell you what we discussed...this is not a new topic...I didn't quote the 75% comment...i dont know what factor it is off just that they are figured and reported differently...

So I am interested in what you have corrected me on and how you have enlightened me...

 

[openwheelformula1]

The summary of all dmens' posts:

Everything pro-Intel and total disregard of AMD's positives.
I knows more than anyone here arrogance.
Anything against dmens' argument is AMD fanboism.
Irrelevant architectural statements without ANY statistical data. (oh oh this is gonna set his face on fire)

 

[Zebo]

dmens is a fanboy - little doubt which "side" of fence he sits on when glosses over anything crappy about intel, which has been a lot, and thier fanboys but does'nt hesitate to point out AMD's problems or problems with AMD evangelist posts.. Why don't you tell us why conroe uses so much more power than yonah? Explain in technicals if you can 40 vs 65 TDP when using same 65nm process.

 

[dmens]

How absurd. If you wanted my honest technical criticism of some intel CPU or platform, just ask. I'd be more than happy to engage in that particular discussion to the best of my ability. I've dived into netburst flaws plenty.

Just to answer duvie and zebo, hell yeah I flamebait. A big reason I post here is to troll the crap out of the fanboys, because it's fun. Reminds me of the slashdot days. Once in a while there's a decent tech discussion... that's also good.

As for yonah vs merom power, here's a quick list:

1. More aggressive frequency target on a similar database = bigger gates
2. Much deeper internal buffers (ROB, RS, SB, LB, etc.) = more transistors
3. Wider datapath at most pipestage points (in general, 4 wide consistent front to back) = more transistors
4. New features and implementations (em64t, faster SSE stacks, extra dispatch port, extended fusion support, etc.) = more transistors

Given the new features, the jump to 65W is very good. Without the extensive power optimizations done on merom, it would have been worse.

There

 

[Duvie]

Originally posted by: dmens
How absurd. If you wanted my honest technical criticism of some intel CPU or platform, just ask. I'd be more than happy to engage in that particular discussion to the best of my ability. I've dived into netburst flaws plenty.

Just to answer duvie and zebo, hell yeah I flamebait. A big reason I post here is to troll the crap out of the fanboys, because it's fun. Reminds me of the slashdot days. Once in a while there's a decent tech discussion... that's also good.

As for yonah vs merom power, here's a quick list:

1. More aggressive frequency target on a similar database = bigger gates
2. Much deeper internal buffers (ROB, RS, SB, LB, etc.) = more transistors
3. Wider datapath at most pipestage points (in general, 4 wide consistent front to back) = more transistors
4. New features and implementations (em64t, faster SSE stacks, extra dispatch port, extended fusion support, etc.) = more transistors

Given the new features, the jump to 65W is very good. Without the extensive power optimizations done on merom, it would have been worse.

There

wow that wasn't so hard!!!!

 

[openwheelformula1]

:Q some insights.

edit: and please don't "try to flamebait the "AMD fanboys"" because that'll create too many useless threads. Especially since AMD currently has a far superior architecture. You can't move a mountain with a toothpick.

 

[Duvie]

more transistors in an ever-decreasing die size and you are going to have to deal ti the heat....yes voltage is going down but smaller and smaller parts are more sensitive...


dmens is Intel using a new material for its gates now??? And what was the magical cur eof 65nm in terms of leakage as compared to the horrible 90nm transition??

 

[dmens]

openwheel, I'll bait as I wish. In fact, the more fanboys the better. Baiting a large group of mutually-reinforced fanboys set in their worldview is far better than just one person. I've been having a great deal of fun picking apart your bs. The fact AMD currenly has a better product is totally irrelevant. What matters is people using bullshit and myths to argue a point. I'll jump on that in a second. It's a good destresser since it makes me laugh.

And duvie, if you read some of my posts before, you'd see my pov arguing leakage is not the main problem with prescott. It is the uarch and implementation.

 

[Duvie]

Originally posted by: dmens
openwheel, I'll bait as I wish. In fact, the more fanboys the better. Baiting a large group of mutually-reinforced fanboys set in their worldview is far better than just one person. I've been having a great deal of fun picking apart your bs. The fact AMD currenly has a better product is totally irrelevant. What matters is people using bullshit and myths to argue a point. I'll jump on that in a second. It's a good destresser since it makes me laugh.

And duvie, if you read some of my posts before, you'd see my pov arguing leakage is not the main problem with prescott. It is the uarch and implementation.

Well that goes against quite a few widely publicated tech articles that said the opposite...Had to be one of the shortest lineups following a die shrink I ever saw!!!

The added lengthof the pipeline was initially an issue at lower speeds but as it ramped above 3.4ghz it drew more in line with the northwoods

 

[Zebo]

Originally posted by: dmens
How absurd. If you wanted my honest technical criticism of some intel CPU or platform, just ask. I'd be more than happy to engage in that particular discussion to the best of my ability. I've dived into netburst flaws plenty.

Just to answer duvie and zebo, hell yeah I flamebait. A big reason I post here is to troll the crap out of the fanboys, because it's fun. Reminds me of the slashdot days. Once in a while there's a decent tech discussion... that's also good.

As for yonah vs merom power, here's a quick list:

1. More aggressive frequency target on a similar database = bigger gates
2. Much deeper internal buffers (ROB, RS, SB, LB, etc.) = more transistors
3. Wider datapath at most pipestage points (in general, 4 wide consistent front to back) = more transistors
4. New features and implementations (em64t, faster SSE stacks, extra dispatch port, extended fusion support, etc.) = more transistors

Given the new features, the jump to 65W is very good. Without the extensive power optimizations done on merom, it would have been worse.

There

Wow must have been hard for you..straight non evasive/attacking answers.. "There!" is what my 6 yr old says when I make him do somthing he's not fond of

All kiding aside I don't think you're a moron like Mark said I do.

 

[DAPUNISHER]

I see everyone has their hackles up this evening. I like demons, he is a PITA, but has the credentials to back it up. BTW, no one asked me to join the cabal, WTF?!?

 

[dmens]

Originally posted by: Zebo
Wow must have been hard for you..straight non evasive/attacking answers.. "There!" is what my 6 yr old says when I make him do somthing he's not fond of

All kiding aside I don't think you're a moron like Mark said I do.

Glad you liked it. The "there" was actually a typo... I was gonna type something witty, but forgot. Crap. It was more like, eyeroll, here, have some technical answers. Next time let me have fun baiting and mocking, then I'll get the the discussion, thanks.

 

[TraumaRN]

At the risk of sounding like an AMD fanboi I'll post this.

Considering the cost of a decently overclocking stable Intel mobo that supports current chipsets...I'm not so sure if I'm going to consider Intel as best bang for my buck considering my limited college income. Now if these Conroe cores come out and they OC good, AND the mobos are cheap....then I know what I'll upgrade too, otherwise it'll be AM2 and OC the hell out of it even if the perf is 5-10% less.

 

[Keysplayr]

Originally posted by: DeathBUA
At the risk of sounding like an AMD fanboi I'll post this.

Considering the cost of a decently overclocking stable Intel mobo that supports current chipsets...I'm not so sure if I'm going to consider Intel as best bang for my buck considering my limited college income. Now if these Conroe cores come out and they OC good, AND the mobos are cheap....then I know what I'll upgrade too, otherwise it'll be AM2 and OC the hell out of it even if the perf is 5-10% less.

Well, and FX60 CPU is about a grand these days right? (I'm not keeping an eye on the AMD prices, but I heard they were up there for the top end) The Conroe 2.66 like the one that was benched is "supposed" to debut at 530.00 ish. That would leave you with 470.00 to go and find the o/c Intel board of your dreams.

But seriously, fun to wait, yet again.

 

[Markfw]

Originally posted by: Zebo

Originally posted by: dmens
How absurd. If you wanted my honest technical criticism of some intel CPU or platform, just ask. I'd be more than happy to engage in that particular discussion to the best of my ability. I've dived into netburst flaws plenty.

Just to answer duvie and zebo, hell yeah I flamebait. A big reason I post here is to troll the crap out of the fanboys, because it's fun. Reminds me of the slashdot days. Once in a while there's a decent tech discussion... that's also good.

As for yonah vs merom power, here's a quick list:

1. More aggressive frequency target on a similar database = bigger gates
2. Much deeper internal buffers (ROB, RS, SB, LB, etc.) = more transistors
3. Wider datapath at most pipestage points (in general, 4 wide consistent front to back) = more transistors
4. New features and implementations (em64t, faster SSE stacks, extra dispatch port, extended fusion support, etc.) = more transistors

Given the new features, the jump to 65W is very good. Without the extensive power optimizations done on merom, it would have been worse.

There

Wow must have been hard for you..straight non evasive/attacking answers.. "There!" is what my 6 yr old says when I make him do somthing he's not fond of

All kiding aside I don't think you're a moron like Mark said I do.

Right. That was a typo, dmens is an asshole. And the "folding freak" is quitting folding for Anandtech. I spend all this money on electricity for a group of jackasses like dmens ? Not anymore. I have had it with all these fanboys and flamebaiters. If thats what this forum is about, and no moderators, then its not for me. I thought this was a forum that was for more serious people wanting to learn about computers, but I guess not.

 

[TraumaRN]

Originally posted by: keysplayr2003

Originally posted by: DeathBUA
At the risk of sounding like an AMD fanboi I'll post this.

Considering the cost of a decently overclocking stable Intel mobo that supports current chipsets...I'm not so sure if I'm going to consider Intel as best bang for my buck considering my limited college income. Now if these Conroe cores come out and they OC good, AND the mobos are cheap....then I know what I'll upgrade too, otherwise it'll be AM2 and OC the hell out of it even if the perf is 5-10% less.

Well, and FX60 CPU is about a grand these days right? (I'm not keeping an eye on the AMD prices, but I heard they were up there for the top end) The Conroe 2.66 like the one that was benched is "supposed" to debut at 530.00 ish. That would leave you with 470.00 to go and find the o/c Intel board of your dreams.

But seriously, fun to wait, yet again.

Oh I agree but if you notice the proc I have and the mobo I have, I spent a grand total of about 250 bucks for BOTH...Call me frugal but I dont see the sense in paying $150(or often times more for a decent OCing Intel board these days) for a freakin motherboard. When you only make 16/hr in college and have a car, apartment payments, food, gas etc I've gotta make wise choices

And as a note about spending a grand on a processor.....well my entire system as seen in my sig was ~1350 bucks

EDIT: Damn typos

 

[IntelUser2000]

A joint development with IBM. Rumor has it, AMD can tweak 40% more from their current cores. Well, even as a AMD fanboi I can't believe AMD can pull 40% out of it.

Umm, 40% is GOOD, but not GOOD if you know what they are talking about. They are comparing SOI, DSL against BULK silicon. Now that's good. Except they are using both SOI and DSL. Not so good anymore. Sure they can get few % but not 40%. A simple 40% created so much rumors and assumptions. If you scale it, its around as bad as the Conroe news creating all the assumptions/rumors.


Fanboy?? Or not?? Let's analyze, see the following
1. "Intel's 90nm sucks because it leaks since it runs so hot as shown by Prescott, while AMD's will be better because its better on the current CPUs"

Sort of Fanboy, maybe just ignorant or lazy. I assume this person also thinks that Dothan is using a AMD's process because it consumes little power?? Simple look at the architecture of Prescott shows why it runs so hot.

125 million transistors(!!). Let's calculate this. Northwood=55 million, 512KB L2, Willamette=42 million, 256KB L2. A theoretical Northwood 1M would be=81 million.

125 million(Prescott)-81 million(Theoretical Northwood 1M)=44 million

Remember also cache consumes very little power. Where did the 44 million go?? All power hungry part of the CPU. No wonder it runs so hot.

Die size of Prescott=112mm2, Die size of Northwood=131mm2
Prescott TDP=103W, Northwood TDP=82W

Thermal density increased by 46.9%!!! Of course we can't see that as linear increase in heat, but it will contribute significantly nonetheless.

2. Assuming that Core would run hotter than TDP because it does in Prescott

Ignorance and Uninformed

I have heard in forums where it was said that Dothan consumes lower power than TDP. However opposite is true for Prescott. It consumes more than TDP.

I assume Core Duo would be same. It would also be true for Core arch. Because Prescott's thermal management SUCKS. EIST, TM2, they are more advanced on PM/Core Duo than Prescott cores.

 

[BrownTown]

clock frequency has a big part to do with it too. Dynamic power increases linearly with frequency, so running at higher frequencies means using more power. Also, i'm not sure what the voltage of eac his, but power is proportional to the voltage squared, so voltage has a HUGE effect on power.

 

[hans007]

Originally posted by: Regs
Ok I'm sick of this loss of hope for AMD crap. You would have to believe a 4 GHz prescott beats a 2.2GHz A64 to believe this hype from Intel.

Yes, Anandtech has "inspected" the test systems for a whole hour to make sure there was no trickery involved however Anandtech has no access of making a independent test using their own systems for more than 6 months from now. When 5 FPS can mean 10-15% in one test, this will become more and more important.

Rumor one - AMD has been developing new architecture with licensed IBM and RAMBUS technology. They also have been working on Dual Stress Liner. A joint development with IBM. Rumor has it, AMD can tweak 40% more from their current cores. Well, even as a AMD fanboi I can't believe AMD can pull 40% out of it.

Rumor two - AMD is making 35w X2's for AM2 and Turion 64 for mobile parts. And this is using 90nm, not 65nm.

Rumor three - AMD is moving to 65nm soon. More room for more cache. More importantly however is how fast will the cache be and will it have lower latencies.

Rumor four - Intel has taken the bait and bit hard by advertising hardware that isn't due out until Q4 of 2006 or Q1 of 2007.

Hint - AMD's Henri Richards is commenting on NetBurst's "failure" and feeding the public that conroe is the proof of how badly everybody bought into NetBurst. They are trying to gain more ground in the retail market. So no matter what Henri said, even though how loose or completely bumpkiss it might be--AMD is prepared. Obviously a poor attempt to try to out play another poor attempt by Intel. To stall the market for a half a year or more.

rumor 1 is unlikely.

rumor 2 , they do have a joint partnership with ibm obviously. this is probably also the reason they arent getting 65nm fabs setup as quickly as ibm is having problems with it. ibm also makes the xbox360 and ps3 cpus... the xbox 360 cpus are at 90nm too... if they were 65nm we'd probably actually have xbox360s on the shelf.

rumor 3 is not a rumor, soon being relative, most reports say q1 2007 if they proceed on schedule. being on schedule for process changes / cpu introductions hasnt exactly been AMDs forte outside of the 90nm shrink (and they were still a good year behind intel on that one), the athlon 64 was even delayed, i think no one noticed because the p4 sucked so hard.

rumor 4 , henri richards is the sales and marketing VP. netburst was not that great, but well.. um that really doesnt matter now that it is in the past. sales and marketing guys um dont know anything in general. the only thing he said that mattered was that amd would have an answer eventually and eluded to the k8L but thats still probably over a year from now.

 

[DAPUNISHER]

Originally posted by: Markfw900

Not anymore. I have had it with all these fanboys and flamebaiters. If thats what this forum is about, and no moderators, then its not for me. I thought this was a forum that was for more serious people wanting to learn about computers, but I guess not.

You should have taken that fleet to BE months ago! The TeAm has a bunch of great people, and I was always impressed with how everyone would pitch in to help a member when they were down on their luck. I left to help start a DC forum elsewhere, where my pro bono tech support efforts for other members are far more appreciated by the management than they are here. The reasons I still hang around here, are to help out some of the inexperienced with their TSing, and because there are a bunch of people here I have gotten to know well enough to consider positive acquaintances, a few even friends, one I have even gotten to know and have a friendship with IRL.

I think it is impossible for a site to reach the size this one has, and not get the freak show effect. I choose not to acknowledge the "fanboys and flamebaiters", and disdain to engage them unless feeling particularly argumentive. There are some great people here, and though Dmens is aggressive and incendiary, I for one, think he brings more to the discussions despite that, than most here, including myself, can hope to. I swear the egos around here are as easily bruised as bananas....

I have nothing to bring to the table where technical discussions of microprocessors are concerned, and unlike some here, will not pretend to. This post is just a off-topic, observational, POV. So, If you are saying your done with these forums Mark, I'll see you at BE, where I still lurk but rarely post