AMD's Socket M, AM2 to launch before June 30th

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
In my humble fanboyish opinion AMD wouldn't exist now if Intel did lower prices 8 years ago. So AMD should be thankfull.
Click to expand...

Thankfull??? 8O
WTF are you talking about here?
If I'm not wrong, AMD had 20% of market share during the K7 days.
It was sake of Intel's "good and honorable" practices that put AMD down in the processor battle.

No,what i'm trying to explain here is that if Intel lowered their prices,AMD would have beem doomed. No one would have cared about AMD since Intel had a reputation and the best offer leading to AMD =broke.
Yonah in most tests beat the X2 3800(excluding games). Considering the VERY LOW energy consumption i consider that an improvement.
I dunno about performance but Intel CPU will certainly have a performance per watt advantage. And they'll probably OC nicely too.

Yeah, sure. It's like saying that more GHz makes your processor faster!
(Intel words a few years ago).
You'll be an @sshole to compare two different companies with different processes.
Anyhow, current 90nm Athlons are giving Intel's processors (which are based on 65nm) a real beating. Just look at the benchmarks.
Once AMD implements PD-SOI, the gap will widen even more.
Click to expand...
That is understandable. Intels Net burst is closing on a dead end.
EE 950 performs quite well IMO. Not to mention that 9xx are very good OCers.
But AMD is taking a much more humiliating beating at the mobile part. Both pentium M and Core Duo beat the crap out of Turion. That leaves high hopes for Conroe and Merom.
I put my money (k,not my money(IE flat broke),my Intel-ish pride) on the Conroe. I think Intel will reign supreme this year(it'd better :) )
Can't wait for reviews like:
Intel Conroe sweeping floors with AMD
AMD beaten in ALL areas
AMD taking a beating from Intel's new CPUs
AM2 proves to be unworthy competition for Intel's new CPUs
K,i exagerated though it would be nice 😀
 
Can't wait for reviews like:
Intel Conroe sweeping floors with AMD
AMD beaten in ALL areas
AMD taking a beating from Intel's new CPUs
AM2 proves to be unworthy competition for Intel's new CPUs
K,i exagerated though it would be nice
Click to expand...

Could be the other way too :)

Intel Conroe = CRAP!
Intel Conroe VS AMD AM2 = AM2 is superior
Intel Conroe failure results in IBM takeover!
Intel commits Sempuku with Conroe release!
MAC and Linux take over the universe and AMD gets stuck in a time rift!

We can all have our speculations and jokes about the future :wink:

On a serious note, I think everyone should stop debating about weather or not AMD can infact pit out the AM2 now or in a few months and just go back to their lives and wait for the damned thing to come out 8)
 
Intel commits Sempuku with Conroe release!

On a serious note, I think everyone should stop debating about weather or not AMD can infact pit out the AM2 now or in a few months and just go back to their lives and wait for the damned thing to come out 8)
Click to expand...
:lol: :lol: :lol:
You killed me with the Sempuku LOL.
Yeah,you're right,too much speculation. Once the two will become available then we'll have some FACTS/RESULTS to talk about.
 
AM-2 is crap http://www.anandtech.com/weblog/default.aspx

That's the only reason it's not out NOW! Performing less than 939 parts.

Conroe is going great, with a $530 2.67ghz chip sure to out-pace a FX-60 and use about half the power.

I hope AMD can fix this. I'm just rooting for Conroe. If all this is correct, AMD will have to lower it's top bins to the non-profit zone.

You talk trash, but what about a link?
 
With what did AMD have a marketshare of 20%? With a XP 3200+ that was pwned by a 2.4 GHz P4?
Some guys around here (especially BullSh!tter), but worst of all even AMD, think that S939 and AM2 will keep up with Conroe (*lol*, they even think they will own badly *lol*). I feel sorry for you all, cause you are so wrong, its not even funny...

Lets get your facts right.
As you know the only differences worth mentioning between AM2 and 939 is the DDR2 support (which also is highly overestimated) and a massive boost in power consumption. So, we might say that AM2 is far away from being a step forward and increasing competitiveness.
S939 CPUs hardly "own" HT-enabled offerings (thats what AMD users think) and get beatings by non-HT offerings when overclocked. Alas, they even get beaten by badly memory handicapped mobile offerings. Some benchmarks on clean testsystems MIGHT tell you otherwise (even dual-cores dont have that much of a performance boost on testsystems), but use your common sense, dont get fooled. Personal experience of millions of people shows parity between the processors AT BEST besides the improved system responsiveness.
Power consumption... well... your parents pay for it anywayz *roflz* Na, seriously. Noone can change the Prescott power consumption. The high clock speeds require the power, thats the thing. But this is where we come to AMD. With a much higher Watt/MHz AMD is just running into the Prescott trap. Since clockspeed is the most rewarding speedboost (cause it is related to the main AMD-advantage) it will get AMD into a really bad situation one day unless they drop their current architecture. The FX-62 has a max. power consumption 125W, what will come next?

Now imagine a Prescott with all details improved massively and the already very aggressive pricing... who is AMD?

In fact, theres nothing for AMD fanboys to get excited about... any processor will do your work, so why bother arguing? As for me, Ive had processors from both companies. And Ive always felt that I made the right decision after having bought the processor and experience proves that.
 
I'm an AMD fanboy compared to the above posters.
Yes i feel the same about the Conroe. I think it will kick some serious ass. AMDs only response will be 65nm and socket F .
Now AMD dominate but from Q3 things are about to change.(hopefully).
Anyway, i hope BOTH will bring something worthwhile. Remember it's best for us if both come up with a decent product so they can lower the prices to eliminate competition :lol: .
 
The original SOI implementation was targeted specifically to increase clocking room rather than temperature and power, resulting in the Venice core being slightly hotter and drawing a bit more power than their Winchester counterparts
Click to expand...
You have to be kidden about that. I'll suggest you re-read the reviews again.

would hope that PD-SOI 90nm performs adequately since recent news by X-bit Labs indicates that the 65nm process may not be available until 2007 which is later than the previous late 2006 introduction date. 2007 seems like a reasonable figure considering that AMD has long quoted that 65nm production won't begin until late 2006.
Click to expand...

I've posted an article that'll clear your doubts about that. (if it's the same article, then, don't bother reading it).
Check it out here
Does AMD have any roadmap for 65 nanometer CPUs?

What I said last November at our annual analyst meeting still holds. We have had 65nm preliminary silicon running in our Dresden Fab 36 since last June; we plan to begin 65nm volume production in the second half of 2006, and we plan to be substantially converted to 65nm in Fab36 by mid-2007.
Click to expand...
 
You have to be kidden about that. I'll suggest you re-read the reviews again.
Click to expand...
The reviews on the subject are quite clear.

http://www.xbitlabs.com/articles/cpu/display/athlon64-venice_5.html

The new Venice core is a little bit warmer than Winchester, which is actually reflected in its official specification: the maximum CPU case temperature claimed for Venice is higher.

This way, it would be incorrect to claim that Dual Stress Liner technology used for Venice processors reduced their heat dissipation level. The major gain is definitely obtained from the higher frequency potential, but in no way from lower heat dissipation.
Click to expand...
I read that article since you posted that in another thread. However, what I was commenting on was when AMD will actually be shipping 65nm parts. In the article, Mr. Hester was just reiterating the same stand that AMD has had for a few months, which is that 65nm production won't begin until H2 2006 and even by mid-2007 they wouldn't be fully converted. If they are just starting to produce 65nm parts in H2 2006, it's unlikely that they'll be able to have enough volumes to hard launch them in the same time period.

The Register has also stated that 65nm parts will be shipping in H1 2007.

http://www.reghardware.co.uk/2006/02/09/amd_roadmap_65nm_desktop/

It should be noted that this information is based on leaked slides with processor information that completely contradicts what we've heard from to date. The Inquirer has long reported that the FX-62 is a 2.67GHz part, while the Register is reporting that the FX-62 is a 2.8GHz part. While at first this may seem like an improvement, but in reality they appear to be completely different processors since the Inquirer's one has a 333MHz base HT while the Register's 2.8GHz one has the older 200MHz base HT.

It gets even weirder when you look at the AM2 5000+ from the Register which is a 2.6GHz 2x512k part while the one from the Inquirer is listed as a 2.4GHz 2x512k part. The Inquirer of course reported that the AM2 X2's will recieve a 400MHz PR rating boost making the base model the X2 4200+ while the Register (based on HKEPC's AMD slides) reports the X2 3800+ remains. Personally, I'm more inclined to believe the Inquirer's information since 7 AM2 X2 models + a dual core FX is far too many models.

I'm assuming you've read Conroe's (the forum member, not the processor) post. He had an interesting link from Anand following up on their recent article on the AM2.

http://www.anandtech.com/weblog/default.aspx

Why are there no performance numbers in that review? Unfortunately it seems that the CPU is quite early and the motherboards aren't ready for prime time yet, so performance of AM2 is significantly lower than Socket-939 today. I do expect that to change, hopefully by the end of this month, but that's why you won't be able to see any realistic previews of AM2 performance this early. It is a bit disappointing, since I expected the move to be fairly seamless and I honestly was hoping for production-level performance by now.
Click to expand...
I hate to say it, but AMD can hardly "deliver AM2 when they want" if the platform isn't even working properly, performs "significantly lower than Socket 939", and won't be in production until April/May. I would hope AMD fixes their performance problems soon, because it'll take a while to get production set-up and yields high enough to make a June launch, especially if they need to wait until May to start production.
 
With what did AMD have a marketshare of 20%? With a XP 3200+ that was pwned by a 2.4 GHz P4?
Some guys around here (especially BullSh!tter), but worst of all even AMD, think that S939 and AM2 will keep up with Conroe (*lol*, they even think they will own badly *lol*). I feel sorry for you all, cause you are so wrong, its not even funny...

Lets get your facts right.
As you know the only differences worth mentioning between AM2 and 939 is the DDR2 support (which also is highly overestimated) and a massive boost in power consumption. So, we might say that AM2 is far away from being a step forward and increasing competitiveness.
S939 CPUs hardly "own" HT-enabled offerings (thats what AMD users think) and get beatings by non-HT offerings when overclocked. Alas, they even get beaten by badly memory handicapped mobile offerings. Some benchmarks on clean testsystems MIGHT tell you otherwise (even dual-cores dont have that much of a performance boost on testsystems), but use your common sense, dont get fooled. Personal experience of millions of people shows parity between the processors AT BEST besides the improved system responsiveness.
Power consumption... well... your parents pay for it anywayz *roflz* Na, seriously. Noone can change the Prescott power consumption. The high clock speeds require the power, thats the thing. But this is where we come to AMD. With a much higher Watt/MHz AMD is just running into the Prescott trap. Since clockspeed is the most rewarding speedboost (cause it is related to the main AMD-advantage) it will get AMD into a really bad situation one day unless they drop their current architecture. The FX-62 has a max. power consumption 125W, what will come next?

Now imagine a Prescott with all details improved massively and the already very aggressive pricing... who is AMD?

In fact, theres nothing for AMD fanboys to get excited about... any processor will do your work, so why bother arguing? As for me, Ive had processors from both companies. And Ive always felt that I made the right decision after having bought the processor and experience proves that.
Click to expand...

It's wasteless talking with an Intel twat.
 
I read it once and it's clear.
Intel is dominating the mobile world with AMD 1 year behind.
You can't just compare a laptop CPU with a desktop CPU.
That same CPU using more power should be able to give more performance.
"
Honestly, as it stands today, if Intel can get clock speeds up, the only area that they will need to improve on is gaming performance to be competitive with AMD. We wouldn't be too surprised if the comparisons that we have shown today end up being very similar to what we encounter at Conroe's launch: with AMD and Intel performing very similarly at the same clock speeds, but with AMD's on-die memory controller giving it the advantage in gaming. "

Well here it says AMD will be equal to Intel but considering that Anand Tech has always been a little bias towards AMD(their benchmarks in rare cases reflect that so their presumptions are a bit kind to AMD) and that AM2 won't improve till next year(in fact, at the begining it'll be slower than the current 939 skt) you can say that Intel has a slight advantage.

Even if AMD=Intel at launch Intel still has got room for upgrades while AMDs 90 arhitecture will have limits. I don't expect something clocked beyond 3 ghz from AM2. The next socket (F ) might change that.

Poor AMD. They don't know what's waiting for them!
P.S. While Intel (as always) won't shine at gaming it will beat AMD at anything else IMO
 
Again with the lack of understanding.
While the venice core was released within the same TDP guides as the winchester, it's actual thermal output was markedly lower. The twit in the article you posted does mnot seem to understand that max case temp is about cooling requirements and OCability, and for thermal output, you need to look to TDP specks.
Amd has regrouped the 3000/3200 chips into a TDP of 57watts. The winchester chips are still in the 67 watt group. So are the lowest three X2 chips.
Interestingly enough, the dothans, when run on a desktop mobo, and without the laptop drivers, fall in about the same range.
Oh, and the 130 nano chips utilized SOI as well.
For those of you who think conroe will be an A64 killer, remember, conroe has more pipeline stages. More pipeline stages always means higher latency, and so lower IPC.
 
Again with the lack of understanding.
While the venice core was released within the same TDP guides as the winchester, it's actual thermal output was markedly lower. The twit in the article you posted does mnot seem to understand that max case temp is about cooling requirements and OCability, and for thermal output, you need to look to TDP specks.
Amd has regrouped the 3000/3200 chips into a TDP of 57watts. The winchester chips are still in the 67 watt group. So are the lowest three X2 chips.
Interestingly enough, the dothans, when run on a desktop mobo, and without the laptop drivers, fall in about the same range.
Oh, and the 130 nano chips utilized SOI as well.
For those of you who think conroe will be an A64 killer, remember, conroe has more pipeline stages. More pipeline stages always means higher latency, and so lower IPC.
Click to expand...

You don't understand Conroe, it has a 14 stage pipeline and a four issue wide core. All are complex issues that can decode all instructions, unlike AMD's three issue cores. Add in macro-opp fusion too, and IPC will be better than K8 could ever be, heck Yonah is at K8 level now and Meron/Conre will be 20% better.

Anyway, if AM2 was all going well, IT WOULD NOT HAVE BEEN DELAYED THREE TIMES!
 
According to this review, Venice is cooler than its predecessors:
http://www.lostcircuits.com/cpu/amd_venice/ .

Here’s another review about the Venice core power consumption:
[url=http://www.hothardware.com/viewarticle.aspx]http://www.hothardware.com/viewarticle.aspx [/url]
AMD's new Venice core is a welcome revision to the Athlon 64 line of processors, and hints to the progress AMD has made on their .09 micron manufacturing process. Not only does this refinement to the A64 core require less power, run cooler and overclock well, but AMD fixes a few issues that were present in the previous revision as well. If you remember from page one, the Venice core now incorporates SSE3 multimedia instruction sets, and its integrated memory controller has been updated as well.
Click to expand...
.

Another review from Tech Report.

The list of reviews keep growing. Do you still want more proof that Venice is cooler, or you still want to believe that crappy review from Xbit labs??

I'm assuming you've read Conroe's (the forum member, not the processor) post. He had an interesting link from Anand following up on their recent article on the AM2.

http://www.anandtech.com/weblog/default.aspx
Click to expand...

A coment posted on a blog from an Intel fanboy. No, I don't read blogs; I read reliable sources that gives me real facts and not someone's thought on something. :wink:
 
Poor AMD. They don't know what's waiting for them!
P.S. While Intel (as always) won't shine at gaming it will beat AMD at anything else IMO
Click to expand...

I've read the review and still have a diferent point of view than yours.

I admit that core duo might be cooler than the X2 3800 (or its 2.0GHz equivalent), but you've missed the whole picture here: Core Duo didn't beat the X2 the way it should. :wink:

Once dual core Turions are launched this year, then we can make an apples to apples comparisson.

If games could be well optimized for AMD's architecture, how come developers don't take that advantagte for common applications?
I'm really clueless about this.
 
Of the three articles you posted only one is relevent. The ones from HotHardware and Tech Report only compare 90nm SOI to 130nm SOI, so obviously Venice would be cooler and that's not in dispute. The article from Lost Circuits does contradict XBit Labs, showing Venice cooler than Winchester. There really doesn't seem to be many Winchester vs Venice comparisons with most of the focus going to Venice vs Clawhammer whose result is obvious.

A coment posted on a blog from an Intel fanboy. No, I don't read blogs; I read reliable sources that gives me real facts and not someone's thought on something.
Click to expand...
You're position on this issue seems contradictory. It was only only Tuesday, that you were lambasting WGruener on TGDaily's lack of AMD news and openness. One of the articles you threw in his face was the Anandtech article on AM2 that you used to show how fair reporting should be done. Now when Anandtech follows up on that article and clarifies why performance data wasn't included, they've suddenly become Intel fanboys and an unreliable source. Either the Anandtech AM2 article and its associated follow up is unbiased or its not, it can hardly be both.
 
holy shit wat a flame war. so many fanboys running amuck. i smell major intel noobs ahhahahah. AM2's performance wil be skeptical cuz its all 90nm process. same old stuff...so amd must be waiting for intel to make the first move.
 
You're position on this issue seems contradictory. It was only only Tuesday, that you were lambasting WGruener on TGDaily's lack of AMD news and openness. One of the articles you threw in his face was the Anandtech article on AM2 that you used to show how fair reporting should be done. Now when Anandtech follows up on that article and clarifies why performance data wasn't included, they've suddenly become Intel fanboys and an unreliable source. Either the Anandtech AM2 article and its associated follow up is unbiased or its not, it can hardly be both.
Click to expand...

It_Commander:

Sorry to say this but, are you real on this??

So, because it says Anand Lal Shimpi's Weblog it was him who posted that?

It won't make sense to me (as a reviewer) to post in my frontpage an incredible preview of socket AM2 and then later on post shit about it in a weblog.

I'm pretty sure it wans't him who've posted that since that's not Anand's style.
 
I'm pretty sure it wans't him who've posted that since that's not Anand's style.
Click to expand...
I can concede that it may not have been directly written by him, but it would have to have at least been reviewed and approved by him. Otherwise, what you are suggesting is that Anand let's anyone just make stuff up and post it in his name thereby allowing the destruction of his reputation. That would seem even less Anand's style, much less anyone else's. Besides, it's been out for 5 days, and if it was so false and contradictory then you would think someone would have taken if off by now.

I know I may be biased, but I wouldn't characterize a 3 page article, the last page of which is just a conclusion, incredible. Especially if they only showed socket and processor shots, most of which were focused on the new cooling cage. Of the six pictures in the article, 1 was of the S940, 2 were of the current S939 cage, and the AM2 cage picture was duplicated, meaning that the entire article only had 2 new pictures (1 of the AM2 pin out and 1 of the cage). They could have at least shown the rest of the motherboard.
 
Just so you know, the 4 issue input model only becomes usefull at about 20 stages. Optimal is about 6 stages per issue.
That is typical of Intel, to put in units that only become usefull later.
If you think Jonah is on par with K8, I suggest you look again. Jonah also has fewer stages than conroe will, so expect lower IPC.
 
I thought there were many factors that effect IPC such as the number of ALUs, not just the number of stages. Intel has stated that the number of ports for Conroe will be greatly increased. Of course we don't know the specific number but seeing that Yonah has 2 universal ports, Conroe will have at least 3. This would mean we are looking at 3 integer units and 2 full FPUs, which is up from Yonah's 2 integer units, and 1 FPU and 1 vector unit.

I've actually never heard of issue width needing to be proportional to pipeline length to achieve benefits so I'll have to trust you on that although I would think that additional decoder width would be beneficial to keeping the additional execution units filled. In any case, the main benefit is not actually as much the number of decoders as the type, since they will now be 4 symmetic complex decoders versus the current complex-simple-simple.

Features like expanded micro-ops fusion and new macro-ops fusion are also supposed to raise IPC.

I thought that the main reason longer pipelines are more inefficient was because it takes so long to fill them up again after a stall especially if it needs to access RAM. Since Conroe's 2 stage increase over Yonah isn't as crazy an increase as Prescott was to Northwood, it should be a lot easier to reduce the effect of stalls. Enhanced prediction, prefetching, and memory disambiguation should reduce the likelihood of stalls happening. Even if they do happen Conroe has direct L1-L1 cache transfer for quick recovery if the data is available from the other core, doubled the L2 cache, and even if it needs to go RAM the FSB is from 60% faster than Yonah to double its speed depending on whether it runs at 1066MHz or 1333MHz.

Granted the 2 extra stages would negatively effect IPC, but there are also other features that increase IPC including features in place to prevent stalls or reduce their effects.
 
The architecture on scotty was designed to prevent stalls. With the # of stages on Jonah, I wouldn't even give it a thought. Intel has, so I really dont see it as a concern.. At the same time, I dont see much use for the level of prefetch, and branch predition they chose to incorporate.
Think of the pipeline as a row of switches. The more switches in the row, the more time it takes to change all thier states.
Now, remember that those switches are three position, on, off, and neutral. If you have a line of runners set up to through those switches, they will still have to wait until the switch returns to neutral before they can switch it again. Add to that, that each operation may not use the same number of runners, or that some runners may have to wait, while one runner does a loop back.
Now tell me, does it matter if you have 3 or 4 starting gates? Maybe, if the pipe is fast enough. But how fast does it have to be?
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom