Help me catch up, been out the loop for the last year

[clarkey01]

Hi all, been out the loop for a while on the CPU side of things (been working on an old M5 plus university stuff)..

From the vibes I get it seems AMD are in trouble and conroe *spelling* is a real gem? AM2 is a unflattering from what I'm reading? whats on the horizon for both Intel and AMD?

 

[TejTrescent]

Well, the article up on this site pretty much explained it all.

For a memory switch with no other real changes other than lower power consumption (What? 35W X2 3800? Niice), the 2-10% performance jump of the AM2 switch is pretty impressive. I mean, if you have a 939 system, no real reason to switch yet--that'd be K8L's influence to do that, but until then, you're better off sticking with 939. AMD's not particularly in trouble--they've had better performance for a while now and finally starting to get some marketshare. They're just going to need to pricedrop, or K8L's going to need to be pretty awesome. Considering there's going to be a few month gap between Conroe and it's release, well, I'm personally just going to wait and see. 65nm should let them take up the clockspeeds pretty nicely, and that's all most of their processors need to compete on par with the Conroe showings that've been seen.

But like it says--with Conroe offering 20% gains over the FX60 in most cases.. It's not as impressive. A lot of people are attributing it to Conroe's cache--but there's just ES samples and few benchmarks out there so it's really hard to tell, considering they've had a lot of problems running it. Victor over at XS seems to be one of the few. Of course, if you wanna take Yonah/Core Duo as an example that can hold its own with everything but an FX60 and X2 4800 really, I can fully believe most of the benchmarks we have seen.

Personally, I just don't believe the price. But hey, Intel could really be going that cutthroat. Of course, I'd really hope Intel could atone for Netburst so it's good to see high offerings from Conroe and Yonah and the like. Either way, I'm waiting for both to be out there and sit for a while until A) prices drop or B) I can get a decent view of what's working out better.

 

[Keysplayr]

Originally posted by: TejTrescent
Well, the article up on this site pretty much explained it all.

For a memory switch with no other real changes other than lower power consumption (What? 35W X2 3800? Niice), the 2-10% performance jump of the AM2 switch is pretty impressive. I mean, if you have a 939 system, no real reason to switch yet--that'd be K8L's influence to do that, but until then, you're better off sticking with 939. AMD's not particularly in trouble--they've had better performance for a while now and finally starting to get some marketshare. They're just going to need to pricedrop, or K8L's going to need to be pretty awesome. Considering there's going to be a few month gap between Conroe and it's release, well, I'm personally just going to wait and see. 65nm should let them take up the clockspeeds pretty nicely, and that's all most of their processors need to compete on par with the Conroe showings that've been seen.

But like it says--with Conroe offering 20% gains over the FX60 in most cases.. It's not as impressive. A lot of people are attributing it to Conroe's cache--but there's just ES samples and few benchmarks out there so it's really hard to tell, considering they've had a lot of problems running it. Victor over at XS seems to be one of the few. Of course, if you wanna take Yonah/Core Duo as an example that can hold its own with everything but an FX60 and X2 4800 really, I can fully believe most of the benchmarks we have seen.

Personally, I just don't believe the price. But hey, Intel could really be going that cutthroat. Of course, I'd really hope Intel could atone for Netburst so it's good to see high offerings from Conroe and Yonah and the like. Either way, I'm waiting for both to be out there and sit for a while until A) prices drop or B) I can get a decent view of what's working out better.

So, through your eyes, the 2-10% (more like 5% actually) improvement of AM2 over 939 is impressive, but the 20% gains of a conroe over an FX-60 was not as impressive..
Am I getting this straight? What am I missing? Maybe it's just the way you worded it. If I am misunderstanding you, apologies. I highlighted specifically what I am talking about.
What does the "it" describe?

 

[SexyK]

Originally posted by: keysplayr2003

Originally posted by: TejTrescent
Well, the article up on this site pretty much explained it all.

For a memory switch with no other real changes other than lower power consumption (What? 35W X2 3800? Niice), the 2-10% performance jump of the AM2 switch is pretty impressive. I mean, if you have a 939 system, no real reason to switch yet--that'd be K8L's influence to do that, but until then, you're better off sticking with 939. AMD's not particularly in trouble--they've had better performance for a while now and finally starting to get some marketshare. They're just going to need to pricedrop, or K8L's going to need to be pretty awesome. Considering there's going to be a few month gap between Conroe and it's release, well, I'm personally just going to wait and see. 65nm should let them take up the clockspeeds pretty nicely, and that's all most of their processors need to compete on par with the Conroe showings that've been seen.

But like it says--with Conroe offering 20% gains over the FX60 in most cases.. It's not as impressive. A lot of people are attributing it to Conroe's cache--but there's just ES samples and few benchmarks out there so it's really hard to tell, considering they've had a lot of problems running it. Victor over at XS seems to be one of the few. Of course, if you wanna take Yonah/Core Duo as an example that can hold its own with everything but an FX60 and X2 4800 really, I can fully believe most of the benchmarks we have seen.

Personally, I just don't believe the price. But hey, Intel could really be going that cutthroat. Of course, I'd really hope Intel could atone for Netburst so it's good to see high offerings from Conroe and Yonah and the like. Either way, I'm waiting for both to be out there and sit for a while until A) prices drop or B) I can get a decent view of what's working out better.

So, through your eyes, the 2-10% (more like 5% actually) improvement of AM2 over 939 is impressive, but the 20% gains of a conroe over an FX-60 was not as impressive..
Am I getting this straight? What am I missing? Maybe it's just the way you worded it. If I am misunderstanding you, apologies. I highlighted specifically what I am talking about.
What does the "it" describe?

I think he meant to say that, because Conroe brings a 20% performance boost, the 5% gains AM2 brings are looking pretty unimpressive ... maybe?

 

[Keysplayr]

That's what I thought he could've meant. At least I hope so.

 

[aidanjm]

What did people make of this:

Conroe performance claim busted!

 

[HopJokey]

Originally posted by: aidanjm
What did people make of this:

Conroe performance claim busted!

Let's just say that the author, Sharikou, is not the most objective person...

 

[Keysplayr]

Originally posted by: HopJokey

Originally posted by: aidanjm
What did people make of this:

Conroe performance claim busted!

Let's just say that the author, Sharikou, is not the most objective person...

Yeah. His blog seems to have a lot of malice in it. And besides, look at how many other people have merom and conroe ES chips who do not have the "priviledge" of using the so called "black box" from Intel and have their own setups. Just about every one of them reporting substantial power from these CPU's. So, Maybe somebody should send sharikou a proper Conroe rig with fully supported mobo and memory the conroe was intended to run with. I mean, take Super Pi for example. There have been numerous threads in here that title "Post your best Super Pi scores here!". Everybody was into Super Pi. So I guess it was a worthy bench. Now that Merom/Conroe seems to be mindnumbingly fast at it, suddenly Super Pi is not indicitive of real world CPU performance. I give up.

 

[dmens]

sharikou is a total fraud with a phd in marketing from a diploma mill. The few technical posts he attempts stack up like a house of cards, LOL. I wish he'd post here so I can tear him a new one.

 

[EffeX]

What he meant is Conroe is next gen and its 20% better than the fx60 however am2 is pulling an extra 5-10% out of current generation. Its really not that hard to understand.

 

[Keysplayr]

Originally posted by: EffeX
What he meant is Conroe is next gen and its 20% better than the fx60 however am2 is pulling an extra 5-10% out of current generation. Its really not that hard to understand.

He said 2-10% actually, and I would like to hear from the actual poster. Funny how numbers morph from just a few posts difference. Now that's something thats hard to understand.

 

[n19htmare]

I don't know about you but a $500 chip (I would prefer the $325 conroe) or whatever it was had a 20% better performace than a to be released $1200+ fx62 then it's good enough for me to jump on the intel bandwagon.

As far as AM2 goes, it's just simple socket change. No real performace gains. UNTIL AM2 gets the next gen AMD chips, it's no comparison to the conroe.

 

[Furen]

Originally posted by: keysplayr2003
So, through your eyes, the 2-10% (more like 5% actually) improvement of AM2 over 939 is impressive, but the 20% gains of a conroe over an FX-60 was not as impressive..
Am I getting this straight? What am I missing? Maybe it's just the way you worded it. If I am misunderstanding you, apologies. I highlighted specifically what I am talking about.
What does the "it" describe?

Nice way to take him out of context.

For a memory switch with no other real changes other than lower power consumption (What? 35W X2 3800? Niice), the 2-10% performance jump of the AM2 switch is pretty impressive.

This means that for a simple memory-controller switch (if it can be called simple), which most people were expecting to actually decrease performance, a 2-10% performance increase is impressive, period.

The next part is split into two paragraphs, for some reason...

[blah, K8L, blah, blah]. 65nm should let them take up the clockspeeds pretty nicely, and that's all most of their processors need to compete on par with the Conroe showings that've been seen.
But like it says--with Conroe offering 20% gains over the FX60 in most cases.. It's not as impressive.

So this is what he says: The Top-of-the-line Conroes are about 20% faster than an FX60 (plausible enough) which is a current generation chip (3 years in the market, no less) on a 90nm process. Considering that the 65nm AMD shrink is coming and that K8L will give SOME performance increase, the 20% performance advantage is not all that great.

Obviously, the highlighted "it" means Conroe.

I dont actually agree with all his points though. AMD cannot afford to start a price war, at least not until its 65nm process is significantly ramped up, since it could find itself with much more demand than its ability to supply it. Also, destroying your margins for damage control is not very smart (just ask Intel, and they went through roundabout ways to do it, via rebates and the like) and it would be better served by doing mild price drops (to maintain market share... and only if needed, since I'm sure Intel's Conroe ramp up will be nice and slow during the first few months) and then phasing in K8L (when possible) in a top-to-bottom fashion (like it's been doing for the past years).

Also, a 20% performance advantage is huge. While it's true that 65nm will allow higher clocks, I wouldnt be surprised to see a 2.93GHz Conroe part before the year is out (or at least before K8L), which could negate any advantage from AMD's 65nm process. Then it'd be up to K8L (which should only bring double the FP units and whatever frontend improvements are needed to use them properly) to make up a 20% across-the-board performance deficit. I think that, realistically, what we'll end up with is Conroe being an Int monster whith K8L being its FP counterpart, so we'd be picking our chips according to our needs rather than a compeltely superior chip.

 

[Duvie]

Originally posted by: dmens
sharikou is a total fraud with a phd in marketing from a diploma mill. The few technical posts he attempts stack up like a house of cards, LOL. I wish he'd post here so I can tear him a new one.

I am sure he is shaking!!!! LOL!!!!

 

[dexvx]

Originally posted by: aidanjm
What did people make of this:

Conroe performance claim busted!

That guy has been debunked by:

1) The creators of ScienceMark @ Aceshardware Forums.
2) Arstechnica Forums
3) HardOCP Forums
4) Xtremesystems Forums

 

[Keysplayr]

Originally posted by: Furen

Originally posted by: keysplayr2003
So, through your eyes, the 2-10% (more like 5% actually) improvement of AM2 over 939 is impressive, but the 20% gains of a conroe over an FX-60 was not as impressive..
Am I getting this straight? What am I missing? Maybe it's just the way you worded it. If I am misunderstanding you, apologies. I highlighted specifically what I am talking about.
What does the "it" describe?

Nice way to take him out of context.

For a memory switch with no other real changes other than lower power consumption (What? 35W X2 3800? Niice), the 2-10% performance jump of the AM2 switch is pretty impressive.

This means that for a simple memory-controller switch (if it can be called simple), which most people were expecting to actually decrease performance, a 2-10% performance increase is impressive, period.

The next part is split into two paragraphs, for some reason...

[blah, K8L, blah, blah]. 65nm should let them take up the clockspeeds pretty nicely, and that's all most of their processors need to compete on par with the Conroe showings that've been seen.
But like it says--with Conroe offering 20% gains over the FX60 in most cases.. It's not as impressive.

So this is what he says: The Top-of-the-line Conroes are about 20% faster than an FX60 (plausible enough) which is a current generation chip (3 years in the market, no less) on a 90nm process. Considering that the 65nm AMD shrink is coming and that K8L will give SOME performance increase, the 20% performance advantage is not all that great.

Obviously, the highlighted "it" means Conroe.

I dont actually agree with all his points though. AMD cannot afford to start a price war, at least not until its 65nm process is significantly ramped up, since it could find itself with much more demand than its ability to supply it. Also, destroying your margins for damage control is not very smart (just ask Intel, and they went through roundabout ways to do it, via rebates and the like) and it would be better served by doing mild price drops (to maintain market share... and only if needed, since I'm sure Intel's Conroe ramp up will be nice and slow during the first few months) and then phasing in K8L (when possible) in a top-to-bottom fashion (like it's been doing for the past years).

Also, a 20% performance advantage is huge. While it's true that 65nm will allow higher clocks, I wouldnt be surprised to see a 2.93GHz Conroe part before the year is out (or at least before K8L), which could negate any advantage from AMD's 65nm process. Then it'd be up to K8L (which should only bring double the FP units and whatever frontend improvements are needed to use them properly) to make up a 20% across-the-board performance deficit. I think that, realistically, what we'll end up with is Conroe being an Int monster whith K8L being its FP counterpart, so we'd be picking our chips according to our needs rather than a compeltely superior chip.

I did not take him out of context, and I don't see how it would be nice if I did. I wanted to make sure I understood his context. I'll just wait to hear from the poster himself, but thanks for the translation anyways.

 

[Dadofamunky]

Originally posted by: n19htmare
I don't know about you but a $500 chip (I would prefer the $325 conroe) or whatever it was had a 20% better performace than a to be released $1200+ fx62 then it's good enough for me to jump on the intel bandwagon.

As far as AM2 goes, it's just simple socket change. No real performace gains. UNTIL AM2 gets the next gen AMD chips, it's no comparison to the conroe.

Yeah, based on the value proposition it's no contest at all....

 

[jiffylube1024]

Originally posted by: clarkey01
Hi all, been out the loop for a while on the CPU side of things (been working on an old M5 plus university stuff)..

From the vibes I get it seems AMD are in trouble and conroe *spelling* is a real gem? AM2 is a unflattering from what I'm reading? whats on the horizon for both Intel and AMD?

Depends on when you're looking to buy.

In the here-and-now, AMD's Opterons (both single and dual core) are just dominating everything.

All performance previews indicate Intel's Conroe to be a winner when it eventually comes out, which is supposed to be ~3-5 months from now. We won't know for sure until it's out.

Reports also indicate AM2 is underwhelming, and AMD won't have a significant performance improvement over current socket 939 chips until 65nm rolls out late this year and through next year.

 

[imported_Crusader]

Originally posted by: aidanjm
What did people make of this:

Conroe performance claim busted!

Amazing.
Looks like I was right, this is AT BEST another Northwood. Parity + MAYBE a bit more.
Overall, another intel flop unfortunately.

I call the Nwood/AXP situation the best Intel is going to pull off here. Not worth saving your pennies for, prob best off saving up for a AM2 rig as AMD will be using that after their major architechural changes next year.
Rather than being stuck on Conroe (doesnt look like Intel has found stable footing, even yet)

 

[coldpower27]

Originally posted by: Crusader

Originally posted by: aidanjm
What did people make of this:

Conroe performance claim busted!

Amazing.
Looks like I was right, this is AT BEST another Northwood. Parity + MAYBE a bit more.
Overall, another intel flop unfortunately.

I call the Nwood/AXP situation the best Intel is going to pull off here. Not worth saving your pennies for, prob best off saving up for a AM2 rig as AMD will be using that after their major architechural changes next year.
Rather than being stuck on Conroe (doesnt look like Intel has found stable footing, even yet)

Anybody who is in the know knows that particular blog is extremely AMD biased and is to be taken with an extreme grain of salt regarding Intel negativety.

 

[TejTrescent]

Wow!

I'm amazed I started such a controversy. Sorry, I had been up for hours writing an essay the night, and my wording WAS horrid in the post (hopefully not the essay!)--I've been away at a programming competition all weekend. Just really surprised to see that I had.. so many dissections of my post.

Yeah, I meant that with Conroe posting 20% better than the FX60, it's really not that impressive that AM2 is a 2-10% jump. It's STILL impressive for a memory switch (did I really put in the word simple? I know it's not, but it IS not.. too much of a change, unlike the radical whole architecture thing), but just.. it pales in comparison. It IS a lot better than most of the predictions that AM2 would kill performance over 939, and it DOES make AM2 a good choice if planning to build an AMD-based system this summer thanks to the ability to upgrade later on without sacrificing any performance instead of going dead-end.

I'm aware AMD can't afford the pricedrops until 65nm is in full swing, sadly. I'd love to see them--if the 4800 X2 was in my price range I'd drop the money for it in seconds. The lack of fabs compared to Intel (It's 2 vs 8, isn't it? Not counting the production deal AMD has to get access to that one other) hurts their ability to produce in volume, though, so hopefully that money getting poured into Dresden can make a bit more of a difference against demand and just chip prices in general. It'd be nice.

I'd like to clarify that I never meant Conroe is not impressive, ESPECIALLY since we've never seen the top of the line one benchmarked, if I remember correctly. Victor has the 2.4, not the 2.66 (it's around there) I think. I could be speaking crazy talk from sleep deprivation thanks to two 6 hour drives, so don't quote me too badly there, have some mercy. But, hearing that an Opteron running at 3.4+ can match it, well, I've got some faith in AMD being able to stay competitive from a performance standpoint for the gap of time leading up to K8L, where they can (hopefully) do a bit more than a quick fix of clock speed ramping, so that they CAN stay in the running with a 2.9 or even a 3.2 Conroe. Not counting Intel OR AMD overclocks, of course, since we have no idea how those are going to work out yet with Conroe or K8L.

I'll be one of the first to say, if Conroe can overclock like the Dothan-based M's can.. Wooo. AMD's stock configs and overclock options better be nice and cheap or I'm sadly switching camps on my main gaming machine for the first time since the K6-III, which really is sad to say.

 

[Viditor]

Originally posted by: TejTrescent
Wow!

I'm amazed I started such a controversy. Sorry, I had been up for hours writing an essay the night, and my wording WAS horrid in the post (hopefully not the essay!)--I've been away at a programming competition all weekend. Just really surprised to see that I had.. so many dissections of my post.

Yeah, I meant that with Conroe posting 20% better than the FX60, it's really not that impressive that AM2 is a 2-10% jump. It's STILL impressive for a memory switch (did I really put in the word simple? I know it's not, but it IS not.. too much of a change, unlike the radical whole architecture thing), but just.. it pales in comparison. It IS a lot better than most of the predictions that AM2 would kill performance over 939, and it DOES make AM2 a good choice if planning to build an AMD-based system this summer thanks to the ability to upgrade later on without sacrificing any performance instead of going dead-end.

I'm aware AMD can't afford the pricedrops until 65nm is in full swing, sadly. I'd love to see them--if the 4800 X2 was in my price range I'd drop the money for it in seconds. The lack of fabs compared to Intel (It's 2 vs 8, isn't it? Not counting the production deal AMD has to get access to that one other) hurts their ability to produce in volume, though, so hopefully that money getting poured into Dresden can make a bit more of a difference against demand and just chip prices in general. It'd be nice.

I'd like to clarify that I never meant Conroe is not impressive, ESPECIALLY since we've never seen the top of the line one benchmarked, if I remember correctly. Victor has the 2.4, not the 2.66 (it's around there) I think. I could be speaking crazy talk from sleep deprivation thanks to two 6 hour drives, so don't quote me too badly there, have some mercy. But, hearing that an Opteron running at 3.4+ can match it, well, I've got some faith in AMD being able to stay competitive from a performance standpoint for the gap of time leading up to K8L, where they can (hopefully) do a bit more than a quick fix of clock speed ramping, so that they CAN stay in the running with a 2.9 or even a 3.2 Conroe. Not counting Intel OR AMD overclocks, of course, since we have no idea how those are going to work out yet with Conroe or K8L.

I'll be one of the first to say, if Conroe can overclock like the Dothan-based M's can.. Wooo. AMD's stock configs and overclock options better be nice and cheap or I'm sadly switching camps on my main gaming machine for the first time since the K6-III, which really is sad to say.

A few corrections here if I may...

1. Availability this year of Conroe - Please note in this foil that in Q3, Intel will only be producing enough Conroes to cover 10% of their performance desktop parts. Note that Celeron and the value segment isn't listed there...in Q4, they will be able to produce 20%.

Performance desktop parts is probably the only sector where AMD is selling more chips than Intel (remember that these parts are usually not part of the business desktop sector, Celerons are). In all the other sectors (value/business, server, mobile, embedded, etc...) Intel still has a much larger marketshare than AMD.

Conroe will only have 10% of this very small (for Intel) sector available, which means that the bulk of AMD's competition will still be Netburst through the end of this year (and you guys thinking of upgrading to Conroe will probably have to speak to Dell about it unless you want to wait several months).

Because of the low availability of Conroe, AMD should still rule the desktop (financially) for the rest of the year...(which BTW should put them into 65nm parts when Intel's availability becomes an issue).

2. AMD no longer has any production capacity problems at all...even though Intel has more Fabs, not all of them are super-Fabs. In addition, AMD will have tripled or more their production capacity by year's end (they are already at double from a year ago). Remember that Intel's Fabs produce many other things besides CPUs...

 

[imported_Questar]

1. Availability this year of Conroe - Please note in this foil that in Q3, Intel will only be producing enough Conroes to cover 10% of their performance desktop parts.

Interesting, I look at a foil that says nothing about Intel being the author, or any kind of date, and I see 15% and 25% where you see 10% and 20%.

 

[TejTrescent]

I don't know, it's a lot closer to like 12.5% and 22.5%. Hard to tell off a graph, it's in a small range of that, though. Somewhere between 10-15 for the first time it's available, and 20-25 for the second. Either way, okay, that's something I hadn't seen. That does put K8L and just AM2 offerings in general in a POSSIBLY better light if AMD's production isn't bad. Huh. Good to know.

I don't know the CPU production capabilities of most of Intel's fabs, though I probably should've taken for granted it'd be a bit more even or AMD would be in a tighter spot. I do remembering hearing somewhere they still make even the ol' Pentium for embedded and such, and they do handle a lot of chipset stuff so geez, don't know what I was thinking really.

Ah well. Learn something new every day I suppose.

 

[dmens]

Originally posted by: Duvie
I am sure he is shaking!!!! LOL!!!!

Haha, he definitely should be. I was glancing over some of his commentary and I came across this gem:

Sharikou, Ph. D said...
With a 4-issue core and 14 stage pipeline, Conroe will have 56 instructions in flight at any moment, not much less than a Northwood Pentium 4.

LOL, what a doofus. The only thing funnier is the sight of people agreeing with him...

Originally posted by: Duvie
I actually think his cache theory has some value.....

ROFLMAO! You're in good company!

The difference between jason at techarray (who rightfully got trashed) and sharikou is that people who think they know more than they really do (hiya!) are afraid to criticize sharikou because he tacks a "PhD" onto his name, even though he is an obvious fraud.

Well, then there's people who actually agree with the fraud, in which case, they know nothing at all! :laugh::laugh::laugh: