What does Conroe mean for AMD?

[Vinny]

This is something that's really been bugging me. I know that for the past few years that AMD has eaten a large share of Intel's market and now they want it all back.

Since Core Duo completely smashes (when you consider the price of the two CPUs) AMD's best, what can AMD do to fight back? I can't see them slashing the price of a $1000 CPU to match the Core 2's price, which I'm hearing will start around $180-$250.

I'm just wondering if AMD will reveal something they've been working on later this year that'll be even better than Core Duo or if they've been slacking off all this time. I just can't imagine that a small company like AMD would let something like this come around and have nothing to fight it with- you'd think they would have something since they were doing so good against Intel.

I haven't kept up with CPUs too much because I was sure I'd go with an AMD CPU but that's become a scrambled thought now. I really can't make a mistake in choosing a socket type, I need one that'll last me a while.

BTW, I'm a virgin CPU brand bias so excuse me if something does sound biased. I've only used Intel before but I never really cared about who what what before because I thought they all performed the same.

 

[BaronMatrix]

Basically Core 2 will improve their image after 3 years of a really crappy architecture. Because AMD has locked up the high end server market, more CIOs will be wiling to allow AMD desktops.

Pretty much by the end of the year AMD should have 25% of the market. There have been questions as to how many Core 2s will be available for the next few months without buying a system.


ALso if you do some research the last few years AMD has been handing Intel it's a s s, so Intel has to catch up to their own chips first. If there are 3 Core 2s and 7 PDs, it makes for a sticky situation.


AMD has a better product differentiation right now so it's easier to decide.

In other words, AMD will be fine if they have a "trick up their sleeve" or not.

 

[Pain]

AMD is better today, Intel tomorrow, who will be better next week? It really goes up and down and I honestly don't think it means anything to either one of the companies except to tell them they need to work on their technology. They will both be in business 5 years from now.

 

[chuckshissle]

True that. It's just like a wheel when one is on top and the other is on the bottom. So with Conroe, AMD is gonna be more competitive to keep up with the market share on desktop cpu's. Intel's Conroe will dominate probably for a while but we shall never underestimate AMD for eventhough it is a small company they have shown Intel that they can level with them as they have produce high performance cpu and had the throne for the last three years. So now it's Intel's turn to sit on the throne but how long that would be will depend on it's arch enemy.

 

[Heyyou27]

I hate these threads.

Core 2 means faster processors!

 

[Mex]

I hear you; I was in the exact same situation you are in and held off from building a new PC because of Conroe. Now, you're looking at Conroe like it's unexpected or serious trouble for AMD. In reality, this is just the natural progression of the market. Just look at how often ATi and nVidia leapfrog each other. This isn't bad; it's normal.

 

[rodney_ws]

Well, for the past 2 or 3 years, "normal" has been Intel getting served by AMD... so I believe it IS a valid question. In the past AMD seemed like the little guy with a better product trying to gain market share... now AMD is about to be the little guy with the inferior product. That's bound to change the dynamics of this mess.

 

[Mex]

Well, for the past 2 or 3 years, "normal" has been Intel getting served by AMD... so I believe it IS a valid question. In the past AMD seemed like the little guy with a better product trying to gain market share... now AMD is about to be the little guy with the inferior product. That's bound to change the dynamics of this mess.

When I go back and think about it, I have to object to yours and my usage of the world "normal" because normal is a subjective point of view. Some on these forums try to have us believe that AMD has always been on top while others would want us to think that this is just a blip on the radar of Intel's continuous ownage. A better word would probably be "trend," because the general trend of the industry shows AMD consistantly coming out on top of Intel. When I said "normal," I was more pointing to the entire industry in general, hence the ATi/nVidia example. Oh well, that's just my take.

Of course, now you probably just think I'm nitpicking.

 

[BaronMatrix]

Well, for the past 2 or 3 years, "normal" has been Intel getting served by AMD... so I believe it IS a valid question. In the past AMD seemed like the little guy with a better product trying to gain market share... now AMD is about to be the little guy with the inferior product. That's bound to change the dynamics of this mess.

The problem is that we reached the "threshold" of PC perf for the SW. X2/FX will still blast through games, encoding, etc. That's why they will keep their market share an continue to get more. The average consumer won't know the difference between PD and Core 2.

IF the PD is cheaper they'll buy it. That will get less perf than AMD - even though again most people won't notice a few seconds or 245 fps vs. 295 (~actual numbers for Core 2 and AM2).

Considering that the 5000+ is an EXCELLENT processor for the money things are looking good for AMD.

If the K8 was luck then yes AMD is in for problems, but if it was a design plan by the guys they got from Alpha, I think the K8L will reclaim the top spot. It is also planned for dual core as AMD may differentiate servers from desktops by amount of cores.

They have said they have samples already and with the additions such as an extra fp unit, a more aggressive prefetch, an extra complex decoder, perhaps enhanced DMT - assuming it exists L3 cache, enhanced SSE128, etc. it will improve K8 by at least 40%.

65nm will allow for 30% greater clockspeed and AMD can play the clockspeed game this time. That should take them to 4GHz.

It is still a two horse race. If the next architecture is an improvement on K8L, Intel will be under a lot of pressure for their next arch ( that's assuming K8L delivers on it's specs).

 

[turpit]

Well, I will give you another take on it..... then will strengthen up after a strong product refresh.

Nicely said Jack

Well, for the past 2 or 3 years, "normal" has been Intel getting served by AMD... so I believe it IS a valid question. In the past AMD seemed like the little guy with a better product trying to gain market share... now AMD is about to be the little guy with the inferior product. That's bound to change the dynamics of this mess.

I agree with what you’re saying Rodney, but I wouldn’t use the word inferior to relate AMDs products to Conroe. Its semantics, I know, but inferior implies "crappy", and AMD has hardly been that, especially relative to some of the arch Intel has pushed out over the past 3 to 4 years. If Conroe lives up to the hype, it will out perform AMD's current best by no small margin, and if Intel keeps the relative prices below AMD’s, then any individual in the market for a DTPC would be a fool to buy AMD. However, if AMD can take/keep a lead in the performance/dollar fight, even with lower performance, then the question of which processor to buy will be much more difficult to answer. That’s a huge if, and unlikely, I realize. Although, thinking about it, it does present itself as a potential marketing gimmick. The dollar per Mhz/Ghz ratio fell out of favor a long time ago, the dollar per watt is valid but not that exciting. Maybe its time to use the GPU standard of dollar per FPS for CPUs as well, or dollar per second of encoding time?

 

[ethernalite]

The problem is that we reached the "threshold" of PC perf for the SW. X2/FX will still blast through games, encoding, etc. That's why they will keep their market share an continue to get more. The average consumer won't know the difference between PD and Core 2.

Funny, I didn't see any AMD fanboys saying that back when the A64 uArch reigned supreme.

IF the PD is cheaper they'll buy it. That will get less perf than AMD - even though again most people won't notice a few seconds or 245 fps vs. 295 (~actual numbers for Core 2 and AM2).

That's average framerate. At low resolution. Min framerate can be much lower. Either way, GPU's are more important for the most part. For someone on a budget, it's still much more important to go for a better GPU and then buy a mid-level CPU. With a couple of notable exceptions, it wouldn't really matter if it's from AMD or Intel, unless you're going with a $900 GPU setup. But still, K8 gained popularity with enthusiasts for having that 3.2FPS advantage at 1600x1200 on $6000 systems, maybe Intel will see the same thing happen with the C2D.

Considering that the 5000+ is an EXCELLENT processor for the money things are looking good for AMD.

You would probably be better off saying the X2 3800+ is a good deal. The $100 cheaper E6400 beats the 5000+ in most everything. And that's not including the cost of the $300 memory. :roll:

If the K8 was luck then yes AMD is in for problems, but if it was a design plan by the guys they got from Alpha, I think the K8L will reclaim the top spot. It is also planned for dual core as AMD may differentiate servers from desktops by amount of cores.


They have said they have samples already and with the additions such as an extra fp unit, a more aggressive prefetch, an extra complex decoder, perhaps enhanced DMT - assuming it exists L3 cache, enhanced SSE128, etc. it will improve K8 by at least 40%.

40% is being a bit generous, don't you think? Maybe in a couple of select benches, but for the most part I think it's a bit much to say 40%. The only thing in that list that will produce massive speedups would be the SSE128 in certain benches. Don't get me wrong, I expect the K8L to put up a fight with the Conroe - especially with the possible speed bumps AMD might be able to accomplish with 65nm - but 40% on a clock for clock basis is an outrageous claim.

65nm will allow for 30% greater clockspeed and AMD can play the clockspeed game this time. That should take them to 4GHz.

Extremely unlikely. A modest 400-600Mhz from the transition (starting from 3Ghz) is likely, unless the K8L contains specific modifications in an attempt to increase clock speeds. In addition, it isn't like speed gains will come instantly - the process needs to mature. Even if they could, it would be stupid to. It would mean they would be without a new product to launch until the 45nm transition, or K10, whichever comes first. :?

 

[Vinny]

Well, looks like Core Duo is pretty much the 'safe' choice for the moment... too bad we don't know how long that moment will be.

 

[The_Abyss]

What does Conroe mean for AMD?

Basically, skid marks followed by brown shorts.

 

[turpit]

People -- the dielectrics are thinned to the point of diminishing return, the performance curve is flattening out -- do not expect huge things from AMD's 65 nm process in terms of speed -- power yeah, that will be good, but gate delay (i.e. clock speed)... not much.

So true. It wasnt that long ago that Intel was expecting 130 to be as small as they could go, then 90, then 65, then 45. Ive seen some stuff about someone toying with 25, but even if Intel or AMD could get down to 25, as you said, its "diminishing returns".

What is interesting is what they are going to do when they finally hit the wall. As I recall, the whole point to multi core, when they first started working on it was not for multi-tasking, but to continue increasing performance when die/transistor size could no longer be reduced.

But with the multi core, and the whole issue of RHT (beaten to death in another thread), if RHT is nothing more than a series-parallel data circuit (over simplified, I know) , then you simply keep adding cores in parallel to increase performance, without all the mucking about of have to optimize software for multithread. Again, as I recall, this was originally the whole point when Intel started toying with the multi core concept.

 

[clue69less]

So now it's Intel's turn to sit on the throne but how long that would be will depend on it's arch enemy.

Arch enemy... Was that an intentional pun?

 

[ethernalite]

So true. It wasnt that long ago that Intel was expecting 130 to be as small as they could go, then 90, then 65, then 45. Ive seen some stuff about someone toying with 25, but even if Intel or AMD could get down to 25, as you said, its "diminishing returns".

Even if transistor switching performance doesn't increase, the die size still decreases (almost) by half, allowing more logic and more cache. It also reduces power, and the increased leakage is going to be taken care of (hopefully) by High-K dialetrics in the 32nm process.

 

[turpit]

Totally off subject, forgive my curiousity, but are you planning on purchasing a Conroe when they come out Jack?

 

[BaronMatrix]

The problem is that we reached the "threshold" of PC perf for the SW. X2/FX will still blast through games, encoding, etc. That's why they will keep their market share an continue to get more. The average consumer won't know the difference between PD and Core 2.

Funny, I didn't see any AMD fanboys saying that back when the A64 uArch reigned supreme.

IF the PD is cheaper they'll buy it. That will get less perf than AMD - even though again most people won't notice a few seconds or 245 fps vs. 295 (~actual numbers for Core 2 and AM2).

That's average framerate. At low resolution. Min framerate can be much lower. Either way, GPU's are more important for the most part. For someone on a budget, it's still much more important to go for a better GPU and then buy a mid-level CPU. With a couple of notable exceptions, it wouldn't really matter if it's from AMD or Intel, unless you're going with a $900 GPU setup. But still, K8 gained popularity with enthusiasts for having that 3.2FPS advantage at 1600x1200 on $6000 systems, maybe Intel will see the same thing happen with the C2D.

Considering that the 5000+ is an EXCELLENT processor for the money things are looking good for AMD.

You would probably be better off saying the X2 3800+ is a good deal. The $100 cheaper E6400 beats the 5000+ in most everything. And that's not including the cost of the $300 memory. :roll:

If the K8 was luck then yes AMD is in for problems, but if it was a design plan by the guys they got from Alpha, I think the K8L will reclaim the top spot. It is also planned for dual core as AMD may differentiate servers from desktops by amount of cores.


They have said they have samples already and with the additions such as an extra fp unit, a more aggressive prefetch, an extra complex decoder, perhaps enhanced DMT - assuming it exists L3 cache, enhanced SSE128, etc. it will improve K8 by at least 40%.

40% is being a bit generous, don't you think? Maybe in a couple of select benches, but for the most part I think it's a bit much to say 40%. The only thing in that list that will produce massive speedups would be the SSE128 in certain benches. Don't get me wrong, I expect the K8L to put up a fight with the Conroe - especially with the possible speed bumps AMD might be able to accomplish with 65nm - but 40% on a clock for clock basis is an outrageous claim.

65nm will allow for 30% greater clockspeed and AMD can play the clockspeed game this time. That should take them to 4GHz.

Extremely unlikely. A modest 400-600Mhz from the transition (starting from 3Ghz) is likely, unless the K8L contains specific modifications in an attempt to increase clock speeds. In addition, it isn't like speed gains will come instantly - the process needs to mature. Even if they could, it would be stupid to. It would mean they would be without a new product to launch until the 45nm transition, or K10, whichever comes first. :?


Intel did somethign similar with Core 2 from Core and go MORE THAT 40% so I guess you're sayign K8 was a fluke and they could never make it happen.

Can we have any discussions here that recognize AMD as the No 2 CPU maker in the world. True I don't have inside info on their processes but from what was said at Analyst Day Brisbane will get damn close to 4GHz.

Back when K8 was eating Intel's lunch you would have given Intel the benefit of the doubt but now that they have eclipsed a 3 year old arch, AMD is now crap?

I bet that Dell doesn't think that. Besides the way the Woodcrest launch went retail may not see a Conroe until Oct. so the 5000+ will be the top of the hill for awhile for CPUs under $600.


WHen Core 2 is available it will be dethroned, but until then.....

http://www.newegg.com/Product/ProductList.asp?N=2000340343+1051720996&Submit=ENE&SubCategory=343


All the AM2s are available at Newegg, including the FX62.

 

[BaronMatrix]

Just a quick point in your rebuttal above: AMD's 130 nm process topped out at 2.6 GHz, 2.4 GHz being common. The 90 nm process has topped out at 3.0 GHz, with 2.8 GHz being common. This is a mediocre 400 MHz from one generation to the next.

The data AMD has made available to date is suggesting around 15 % improvement from todays median bin, so about another 300 or 400 MHz gain, though I expect they may push this an extra 200 MHz so somewhere between 3.4 to 3.6 GHz is where the 65 nm will top out at best.

People -- the dielectrics are thinned to the point of diminishing return, the performance curve is flattening out -- do not expect huge things from AMD's 65 nm process in terms of speed -- power yeah, that will be good, but gate delay (i.e. clock speed)... not much. Unless, like you say -- the K8L extends the pipeline and streamlines the logic.

Considering that the 5000+ is an EXCELLENT processor for the money things are looking good for AMD.

You would probably be better off saying the X2 3800+ is a good deal. The $100 cheaper E6400 beats the 5000+ in most everything. And that's not including the cost of the $300 memory.

This is true --- none of the AMD high end is looking good as the E6600 and E6700 pretty much exceed that of an FX-60 and FX-62 class processor respectively, adding to that significantly cheaper and 1/2 the power.


The E6xxx DOES NOT EXIST AND CAN'T BE PURCHASED EVEN IN A SYSTEM.

The AM2 is in ~full stock at Newegg.

That condition didn't stop millions of people from heating their offices with a PD while AMD was grooving 1GHz less and more perf.

Intel fans are pathetic.

 

[pat]

Let say you have a bakery and you are making the best cake in town for the last few years. Everybody want your cake, and you're making full of it. Now, your competitor don't sell that many cake right now. So they start working and announce an improved receipe that test show superior to yours, but not available yet. What do you do?

Sure enough, you have secret cake that should be better than your competitor new ones, but you still have lot of money to make from the good old receipe, because with the time, production cost by cake is less than it was when the first cake was released. You still have lot of current cake to sell and frankly, buyer still want them.. So, all in all, you are making good money anyway, no matter what the competitor say or do.

Or, you release your secret cake now, loosing money when cutting the established product out of sale, rushing the cake market with a product that is not needed nor ready to be accepted because it needs new plate and ustensil to properly eat it. You should have wait your competitor cake to see what new feature people like from their cake, and maybe adapt them in your new secret receipe to make it more attractive. You sure enough make some improvement to your current receipe, making sure that it will taste almost the same to not really surprize and hurt your already good market.


All this time, you had more cook working at the new receipe, and less needed to improve the old one, as it was a good one from start. Lesson learned.


This is AMD side now..

Now, have a look at the cake competitor view.

You once made good cake, but the little town bakeryjust surpassed you. And it is stealing customers from you. Worst, your cake sales are not as good. You put some rushed product base on the old receipe to make money with the now well established and rentable netcake architecture, as redesigning a new one will means research cost, and bakery improvement that will add to the production cost of a new cake. Happily, your cook were working on a new receipe all this time, and now, it is time to announce it. It wont hurt anything now, as your sale are already declining and your cake, while being good, are not that good.. just good enough. You had to cut price to make them more attractive to customer too, but they are still prefering your competitor' cake.

The new announced cake make great impression and future look good now. All you have to do is to make sure that you have enough cake when ready to send to your big customer. Sure enough, the walk-in customer may not have it at first, but loosing some walkin customers is not as bad as loosing a customers like a grocery... so you need to priorize those, leaving the walkin customer market a bit behind.. Make sure that the new plate are ok too. You don't want you new cake to melt the new plate like when you release your new dual flavour cake to be used with current plate. Happily, plate maker respond fast enough to limit the disaster..

Sure enough you know that your competitor was probably working on a new kind of cake and that's why you don't want to miss that cake release. You don't want to rushed it to the market, so you take the time to see if every one is ready before releasing it. You don't want to rush it like the other cake.. Lesson learned.

When you have your mind set on one brand, it is hard to see the whole thing. The "game" AMD a Intel are playing is not a new and exclusive one. It is played everyday, at larger and smaller scale. On all product.

If you fanboy has enough brain left from your brain washing (sorry ycon, it is too late for you.. your brain has been replaced with an answering machine one with the message Intel is good, praise Intel) just read about marketing on the internet. That will hurt, because you'll see that you are only puppets..

 

[ethernalite]

Intel did somethign similar with Core 2 from Core and go MORE THAT 40% so I guess you're sayign K8 was a fluke and they could never make it happen.

Going from Core Duo to Core 2 is only 10-20% performance gain per clock, as per reported by Intel (Yonah -> Merom). And even then, they did a lot more than just add an extra FP unit and 128bit SSE instructions.

I'm talking about overall, just not a couple of select benches.

 

[gOJDO]

Conroe for AMD means:
1. Lost profit and lost market share
2. Lost reputation and crown of performance
3. Pain in the fat @$$, and moving it to bring something better to compete

 

[BaronMatrix]

Let say you have a bakery and you are making the best cake in town for the last few years. Everybody want your cake, and you're making full of it. Now, your competitor don't sell that many cake right now. So they start working and announce an improved receipe that test show superior to yours, but not available yet. What do you do?

Sure enough, you have secret cake that should be better than your competitor new ones, but you still have lot of money to make from the good old receipe, because with the time, production cost by cake is less than it was when the first cake was released. You still have lot of current cake to sell and frankly, buyer still want them.. So, all in all, you are making good money anyway, no matter what the competitor say or do.

Or, you release your secret cake now, loosing money when cutting the established product out of sale, rushing the cake market with a product that is not needed nor ready to be accepted because it needs new plate and ustensil to properly eat it. You should have wait your competitor cake to see what new feature people like from their cake, and maybe adapt them in your new secret receipe to make it more attractive. You sure enough make some improvement to your current receipe, making sure that it will taste almost the same to not really surprize and hurt your already good market.


All this time, you had more cook working at the new receipe, and less needed to improve the old one, as it wasn a good one from start. Lesson learned.


This is AMD side now..

Now, have a look at the cake competitor view.

You once made good cake, but the little town bakeryjust surpassed you. And it is stealing customers from you. Worst, your cake sales are not as good. You put some rushed product base on the old receipe to make money with the now well established and rentable netcake architecture, as redesigning a new one will means research cost, and bakery improvement that will add to the production cost of a new cake. Happily, your cook were working on a new receipe all this time, and now, it is time to announce it. It wont hurt anything now, as your sale are already declining and your cake, while being good, are not that good.. just good enough. You had to cut price to make them more attractive to customer too, but they are still prefering your competitor' cake.

The new announced cake make great impression and future look good now. All you have to do is to make sure that you have enough cake when ready to send to your big customer. Sure enough, the walk-in customer may not have it at first, but loosing some walkin customers is not as bad as loosing a customers like a grocery... so you need to priorize those, leaving the walkin customer market a bit behind.. Make sure that the new plate are ok too. You don't want you new cake to melt the new plate like when you release your new dual flavour cake to be used with current cake. Happily, plate maker respond fast enough to limit the disaster..

Sure enough you know that your competitor was probably working on a new kind of cake and that's why you don't want to miss that cake release. You don't want to rushed it to the market, so you take the time to see if every one is ready before releasing it. You don't want to rush it like the other cake.. Lesson learned.

When you have your mind set on one brand, it is hard to see the whole thing. The "game" AMD a Intel are playing is not a new and exclusive one. It is played everyday, at larger and smaller scale. On all product.

If you fanboy has enough brain left from your brain washing (sorry ycon, it is too late for you.. your brain has been replaced with an answering machine one with the message Intel is good, praise Intel) just read about marketing on the internet. That will hurt, because you'll see that you are only puppets..

Finally a CPU fanboy.

HOORAY!!!

 

[BaronMatrix]

Intel did somethign similar with Core 2 from Core and go MORE THAT 40% so I guess you're sayign K8 was a fluke and they could never make it happen.

Going from Core Duo to Core 2 is only 10-20% performance gain per clock, as per reported by Intel (Yonah -> Merom). And even then, they did a lot more than just add an extra FP unit and 128bit SSE instructions.

I'm talking about overall, just not a couple of select benches.


Ok so I don't have inside info as to the EXACT increase of Core to Core 2 the point was that AMD is an ENTERPRISE supplier of CPUs and shouldn't be thought of as dead because Intel has 1-2 Core 2 for every 10 chips until at least Oct.

It's great that they have something competitive. They did leap ahead with Core 2 and I'm sure that AMD will introduce somethign to push the envelope even more before toolong. They have 2 years before the next arch and Core 2 is showing 3.6GHz on air so improvements will be needed to get higher before Nehalem.


The ball is still in AMDs court because as DDR2 gets cheaper and faster - I'm sure that production levels have shifted in every mem fab from DDR to DDR2 - AM2 will look even better. Especially if AMD stockpiled some of the F2 revs and are onto F3 for the next major shipments.

And Rahul Sood of VoodooPC ALMOST said that "yes AMD has this but who knows how much it will do" in terms of Dynamic Threading.

Core 2 means that when it comes out it will be the fastest thing out by at least 20% overall. Hopefully GMA X3000 will provide a better experience than their previous crap - I mean IGPs.

 

[ethernalite]

Ok so I don't have inside info as to the EXACT increase of Core to Core 2 the point was that AMD is an ENTERPRISE supplier of CPUs and shouldn't be thought of as dead because Intel has 1-2 Core 2 for every 10 chips until at least Oct.

Don't put words in my mouth. I said nothing about AMD being dead. I'm not retarded. And I'd hardly call the comparison between Merom and Core Duo as "inside information." Besides, I thought you were an analyst? Or something. Shouldn't you know stuff like that? :roll:

Core 2 means that when it comes out it will be the fastest thing out by at least 20% overall. Hopefully GMA X3000 will provide a better experience than their previous crap - I mean IGPs.

Yes, because we all know that those 2D toolbars really slow an IGP to a halt.