AMD executing with more proficiency while Intel Atom cliffdiving and I7 flatlining

[jww20]

AMD may be much smaller than Intel but they are recently executing pretty well:

"AMD saw its market share in processor shipments reach 22.3 percent during the first quarter of 2009, gaining 4.6 percent of the market compared with the fourth quarter of 2008. Intel lost 4.7 percent of the total market to reach a 77.3 percent share, IDC said in a survey.

AMD saw its market share increase because of a pricing advantage over Intel and a strong increase in desktop shipments, said Shane Rau, research director at IDC. Sequentially, AMD chip shipments increased 13 percent while Intel's shipments declined 16 percent.

Intel's sequential decline was partly due to suppliers holding back on purchases as they tried to clear up excess inventory of mobile processors, especially Atom processors for netbooks. Shipments of Atom processors recorded a sequential decline of 33 percent in the first quarter of 2009."

It seems that AMD execs were right when they decided not to utilize limited resources in the high end desktop CPU market. Fudzilla reported that Intel indicated that Core I7 sales are at 1% (near flatline)! If you add in the hundreds of millions Intel has spent on "incentivizing" DDR3 memory manufacturers then net sales must in the negative territory.
You can hear the shouting in the Intel headquarters about how they are going to position the Core I7/i5/i3/i1 CPUs, because at current levels they have no future with Intel or motherboard manufacturers.

http://www.fudzilla.com/index.php?option=com_content&task=view&id=13686&Itemid=1

 

[oldpablo]

People would rather have 5 awesome computers then 2 amazing ones for the same price, news at 11.

 

[Shane_c82]

Now if they can just turn a profit.

 

[LAiN_OfTheWired]

This just in: the economy is shit and nobody can afford something they don't need!

 

[funkydmunky]

In a way Intel kinda hurt themselves. The Atom took the ultra low end by storm, forcing Intel to position their budget chips like Celerons and low end C2D's in a tough position compared to a aggressively priced AMD line up that features dual and triple and some quad cores in budget positions, some with low power options. Of course AMD has no answer to the Atom and soon to be Atomx2, which I thought AMD would have hedged its bets towards as they lost their grip on the high end (there are only so many rabbits you can pull out of you ass before the massive $$ backed machine Intel is, comes a knocking) but they all but abandoned the ultra mobile market. AMD has done very well considering the chips they are dealt (pun intended) when fighting a Goliath such as Intel. As it stands they are also the only player that is competitive in CPU, and GPU at enthusiast levels. That's pretty impressive in itself.

 

[DarkLegacy]

Intel will be fine, The hole AMD dug them selves into is much deeper than they've ever been in. They'll need something much larger than what the AMD 64 was to ever become a real threat to Intel again. A simple price drop across the board of Intels I7 chips will put AMD right back to where they were in 2008. Intels shareholders may not approve of that kind of decision but Intel is in the position to absorb the losses if it would end up hurting AMD's market share.

AMD is hoping to garner profit in the form of mass sales while Intel garners profit per sale. The only reason AMD is even still around is the generosity, or stupidity, of investors dumping wads of cash to keep them operating. AMD CPU's & GPU's are nearly just as expensive to produce but sell for quite a bit lower than the competition. Taking this route has lead to the losses they've been having the last few years. You need to have a massive amount of sales to turn a profit when a single product that sales does not turn a profit on its own. There little jump in the market share is cute at best but they"ll need a whole lot more to become someone Intel should worry about again.

 

[Reus]

Intel will be fine, The hole AMD dug them selves into is much deeper than they've ever been in. They'll need something much larger than what the AMD 64 was to ever become a real threat to Intel again. A simple price drop across the board of Intels I7 chips will put AMD right back to where they were in 2008. Intels shareholders may not approve of that kind of decision but Intel is in the position to absorb the losses if it would end up hurting AMD's market share.

AMD is hoping to garner profit in the form of mass sales while Intel garners profit per sale. The only reason AMD is even still around is the generosity, or stupidity, of investors dumping wads of cash to keep them operating. AMD CPU's & GPU's are nearly just as expensive to produce but sell for quite a bit lower than the competition. Taking this route has lead to the losses they've been having the last few years. You need to have a massive amount of sales to turn a profit when a single product that sales does not turn a profit on its own. There little jump in the market share is cute at best but they"ll need a whole lot more to become someone Intel should worry about again.

I was waiting for the obligatory holier than thou Intel zealot post. You can never post anything even remotely positive about AMD without 'em!

Oh but who am I? Some less fortunate bum, I should go panhandle for an i7......

 

[DarkLegacy]

Ah don't get me wrong, I've most likely built & owned more AMD based rigs than a huge majority of the members of this forum have. I still have the motherboards & CPU's to back up that claim since I held on to all of them. I'm simply not blind to the crap they've been dishing out the past few years. I see all of the praise they get for having such low priced parts that offer great performance & all that good stuff in between. It just comes off as " whatever " to me, still doesn't out perform Intel parts so I'm not caring. The [H]ard|OCP community has simply changed since I was first around. I get that Walmart forums feeling now with everyone looking for the cheapest stuff to get. Yeah, it's a personal thing with me =P.

 

[facebook]

A simple price drop across the board of Intels I7 chips will put AMD right back to where they were in 2008. Intels shareholders may not approve of that kind of decision but Intel is in the position to absorb the losses if it would end up hurting AMD's market share.

They will not do it, i7 is more expensive and manufacture compared to Phenom. AMD can lower if Intel wants a price war and the Phenom platform is much more flexible compared to i7.

 

[LAiN_OfTheWired]

DarkLegacy is totally right on this. Sorta. As in, I get exactly what he means, but I don't entirely agree with how he's saying it.

Basically, AMD hasn't really been competative. I hear ya. BUT also look at this: Intel created the i7, a needlessly powerful processor. AMD created Phenom II, a reasonable amount of power (especially for gamers) and they're not charging an arm and a leg. This is helping their market share get back to where it should be in "troubled" times.

What will be interesting is what each company does next, but we can be assured Intel's retarded budget will be used in another attempt to kill AMD's marketshare. AMD really doesn't have much to fire back with, and they haven't even been competing in some parts of the market just so they can try to compete in the main section.

I'm still waiting for a new PPC from IBM to go personal, because when you get down to the Intel/AMD fanboy arguing performance per clock and overall clock speed come into play but um... IBM had Core i7 beat what, like 3 years ago? Maybe 4? You know, talking single-chip. If a new PPC CPU went personal and was priced to compete, I'd abandon both of these jerks. To shed some light: Cell is a general purpose core (if you want to argue this point, you've never run a Linux distro on your PS3) and it performs at 1 TeraFLOPs when clocked at its original 4GhZ (the PS3 is clocked at 3.2 I think). Core i7 965 at 4GhZ is roughly 87 GigaFLOPs. So, even if you cut the Cell in half, IBM's Cell still outperforms i7 by 5 1/2 times.

Having said that, I'm very interested in seeing what all this hubbub between IBM and AMD is about. AMD is definitely behind NOW, but Intel has never been faster to develop new shit than IBM, and IBM is helping AMD to the 28nm line. IBM is shooting to one-up Intel on this front for sure, and they say they can do it by a whole year. So what AMD does now may not be too exciting, but what AMD is fueling with this extra bit of marketshare and all their new funding could be well worth it here in the next few years.

Sorry, had to play devil's advocate, I'm very satisfied with my Phenon II

 

[pxc]

I'm still waiting for a new PPC from IBM to go personal, because when you get down to the Intel/AMD fanboy arguing performance per clock and overall clock speed come into play but um... IBM had Core i7 beat what, like 3 years ago? Maybe 4? You know, talking single-chip. If a new PPC CPU went personal and was priced to compete, I'd abandon both of these jerks. To shed some light: Cell is a general purpose core (if you want to argue this point, you've never run a Linux distro on your PS3) and it performs at 1 TeraFLOPs when clocked at its original 4GhZ (the PS3 is clocked at 3.2 I think). Core i7 965 at 4GhZ is roughly 87 GigaFLOPs. So, even if you cut the Cell in half, IBM's Cell still outperforms i7 by 5 1/2 times.

Oh dear.

The PS3 Cell has one PPE (the part that runs the OS) and 7 SPEs (the number crunchers). The 3.2GHz PPE is an in-order execution design with roughly the performance of a Pentium 3 1.2GHz CPU. The SPEs are not general purpose enough to run full PPC threads, meaning that your typical non-SIMD threads have to run on the PPE. Even the triple PPE core Xenon CPU in the XBox 360 (which IMO is much better suited for PC-type apps than the Cell's single PPE) lags *far* behind either AMD's or Intel's lowest speed quad core CPUs (and overall likely well below even low end Core 2-based dual core CPUs too).

IBM has been pushing Cell derived chips as co-processors and I guess that's fine if you have a specific use for one (something likely to change when universal OpenCL compatibility builds). But it's not a good general purpose CPU.

 

[Mac Truck]

Having said that, I'm very interested in seeing what all this hubbub between IBM and AMD is about. AMD is definitely behind NOW, but Intel has never been faster to develop new shit than IBM, and IBM is helping AMD to the 28nm line. IBM is shooting to one-up Intel on this front for sure, and they say they can do it by a whole year. So what AMD does now may not be too exciting, but what AMD is fueling with this extra bit of marketshare and all their new funding could be well worth it here in the next few years.

That would be a SICK merger. IBM's Cell processor as part of DAAMIT's Fusion platform?

 

[Atech]

They will not do it, i7 is more expensive and manufacture compared to Phenom. AMD can lower if Intel wants a price war and the Phenom platform is much more flexible compared to i7.

Really?
http://techreport.com/articles.x/16147

"This new fab process has allowed AMD to fit many more transistors into a Phenom II die—an estimated 758 million, versus 463 million for the Phenom—while reducing the die size from Phenom's 283 mm² to just 258 mm². Interestingly enough, the Phenom II's basic specs sound remarkably similar to those of the Core i7, which weighs in at roughly 731 million transistors and 263 mm². "

Now I would like to see your math...

 

[pxc]

They will not do it, i7 is more expensive and manufacture compared to Phenom.

Actually, the Core i7 is cheaper to manufacture, and not only because of scale. AMD uses SOI wafers and has more metal layers, both of which increase manufacturing costs vs bulk Si and fewer metal layers. So not only does each good Phenom 2 chip cost more to manufacture, AMD sells nearly the whole line (X3 and X4) for less than the lowest end i7.

 

[jimmyb]

I would be shocked if any serious conclusion could be made about the relative manufacturing costs based on the number of metal layers and the use of SOI. There are so many other important factors that affect the cost of an IC (oxidation, pattering, diffusion & implantation, deposition, etc.). I think the only reliable way to compare costs is to find out what they are.

That said, I suspect Intel does it cheaper.

 

[LAiN_OfTheWired]

Oh dear.

The PS3 Cell has one PPE (the part that runs the OS) and 7 SPEs (the number crunchers). The 3.2GHz PPE is an in-order execution design with roughly the performance of a Pentium 3 1.2GHz CPU. The SPEs are not general purpose enough to run full PPC threads, meaning that your typical non-SIMD threads have to run on the PPE. Even the triple PPE core Xenon CPU in the XBox 360 (which IMO is much better suited for PC-type apps than the Cell's single PPE) lags *far* behind either AMD's or Intel's lowest speed quad core CPUs (and overall likely well below even low end Core 2-based dual core CPUs too).

IBM has been pushing Cell derived chips as co-processors and I guess that's fine if you have a specific use for one (something likely to change when universal OpenCL compatibility builds). But it's not a good general purpose CPU.

Come back to me with that argument when you run Linux on your PS3. There isn't a single day to day operation that can be performed faster on a Core i7 965-based system, and relatively few operations overall that can be done faster on the i7. I mean, we'd really have to scrounge. Oh, and the cell does it with regular DDR, no need for triple-channel DDR3.

You can try to discount me all you want, but IBM had overshadowed Intel in the calculations-per-clock arena for at least the past decade (Macs have not needed clock speeds as high as Intel-based PCs since Mac OS 8 from 1997 to achieve equal or higher performance). IBM coupling with AMD in the broom closet says to me there's a good chance of something good coming from the AMD end in the next couple of years.

 

[rflcptr]

Oh, and the cell does it with regular DDR, no need for triple-channel DDR3.

No, the PS3's Cell uses 256MB XDR clocked at 3.2GHz as cache.

There isn't a single day to day operation that can be performed faster on a Core i7 965-based system, and relatively few operations overall that can be done faster on the i7.

Could you define both of those underlined phrases for clarity?

 

[pxc]

I would be shocked if any serious conclusion could be made about the relative manufacturing costs based on the number of metal layers and the use of SOI.

Those are both related to direct costs (significantly higher initial cost of the wafers and more deposit/polish manufacturing steps), but of course there are other factors. It doesn't help that AMD has other higher costs there either, especially in the number of strain techniques used.

The costs, like yields, are kept secret by both companies. But best estimates made by engineers I've read show a distinct cost disadvantage for SOI and more metal layers vs bulk and fewer layers, on the order of up to thousands of dollars more per processed wafer. And with immersion lithography in AMD's 45nm process, costs have likely creeped up even more.

 

[stealthy123]

IBM and AMD have been "helping" each other for years, there is nothing new to the two companies working together.

 

[pxc]

Come back to me with that argument when you run Linux on your PS3. There isn't a single day to day operation that can be performed faster on a Core i7 965-based system, and relatively few operations overall that can be done faster on the i7. I mean, we'd really have to scrounge. Oh, and the cell does it with regular DDR, no need for triple-channel DDR3.

You can try to discount me all you want, but IBM had overshadowed Intel in the calculations-per-clock arena for at least the past decade (Macs have not needed clock speeds as high as Intel-based PCs since Mac OS 8 from 1997 to achieve equal or higher performance). IBM coupling with AMD in the broom closet says to me there's a good chance of something good coming from the AMD end in the next couple of years.

I can tell you really love the Cell. Too bad you don't understand how it works. The core "running" Linux on your PS3 is the single PPU, a general purpose in-order CPU capable of running 2 threads at once. The Core i7 will blow it away in any measurable way on "day to day operation" as far as I understand what that implies as someone who uses a computer day to day.

However, bring in the 7 SPEs with appropriate workloads (heavy SIMD math) and it will whup the i7 6 ways from Sunday. But those aren't typical day to day operations for most people. You're into video encoding and or folding or other workstation workloads, great. The other 98% of people won't be using the SPEs much.

Funny you should bring up the PowerPC performance myth. There was a brief time around 1997 where both clock speeds were pretty comparable between PPC and x86, and things were slightly in PPC's favor. Then within a cople of years (this part is funny) Hector Ruiz of Motorola's semiconductor division decided to slow down PPC development and set Apple as a low priority customer. Far from having the lead for the past 10 years, things fell apart for the PPC alliance before 2000. Excellent coverage of what happened: http://www.realworldtech.com/page.cfm?ArticleID=RWT051400000000&p=3 IBM picked up the slack, but the performance deficit was permanent, unless you were drinking the Apple benchmark kool-aid (which were *always* contradicted by 3rd party testing).

At the end of Apple's PPC era, it was using 2 x 180W dual core water cooled processors to compete with single dual core Intel processors. Altivec was good, but Apple was lagging by sticking with PPC. It recognized that and switched to x86.

 

[cyclone3d]

Ah don't get me wrong, I've most likely built & owned more AMD based rigs than a huge majority of the members of this forum have. I still have the motherboards & CPU's to back up that claim since I held on to all of them. I'm simply not blind to the crap they've been dishing out the past few years. I see all of the praise they get for having such low priced parts that offer great performance & all that good stuff in between. It just comes off as " whatever " to me, still doesn't out perform Intel parts so I'm not caring. The [H]ard|OCP community has simply changed since I was first around. I get that Walmart forums feeling now with everyone looking for the cheapest stuff to get. Yeah, it's a personal thing with me =P.

I've also owned a lot more AMD systems than Intel systems... In fact, I have never actually purchased a new Intel processor.... the only Intel processor I ever bought was a
486DX2-66, and it was used when I bought it.

The main reason I have my Q6600 is because my little bro had an extra one that he gave to me.

Before that I had an Opty 170, 3700+, Athlon XP 3200+, 2700+, 2500+, 1800+, 1Ghz Socket A T-bird, Several different Slot-A Athlons.. from 500Mhz, up to 750Mhz.. All which were overclocked from 650 to over 1Ghz, Numerous K6-2s... fastest was a K62-550 that I ran at 660 on the best Socket 7 board ever.. the ASUS P5A.(I probably fried 2 or 3 K6-2 chips in my quest for overclocking.. but they were super cheap so I really didn't care.) Before that I had a couple regular K-6 chips, and before that I had an AMD 5x86-133 that I ran at 160.

And.. I still have an old laptop that I stuck a K6-2+ 450 that is at 550 on below stock volts and I also have my retro gaming rig which has a Slot-A Athlon running at 1080Mhz cause my board has a sucky fsb limit.

 

[LAiN_OfTheWired]

I can tell you really love the Cell. Too bad you don't understand how it works. The core "running" Linux on your PS3 is the single PPU, a general purpose in-order CPU capable of running 2 threads at once. The Core i7 will blow it away in any measurable way on "day to day operation" as far as I understand what that implies as someone who uses a computer day to day.

However, bring in the 7 SPEs with appropriate workloads (heavy SIMD math) and it will whup the i7 6 ways from Sunday. But those aren't typical day to day operations for most people. You're into video encoding and or folding or other workstation workloads, great. The other 98% of people won't be using the SPEs much.

Funny you should bring up the PowerPC performance myth.

At the end of Apple's PPC era, it was using 2 x 180W dual core water cooled processors to compete with single dual core Intel processors.

I wrote an awful lot, but I think its easiest to sum it up like this: I think ur effin jealous. haha, just kidding, but seriously, here's a list of some things Cell does faster than Core i7:

Word Processing
Video Encoding
Web Browsing
Java.

So... why do you need a Core i7? Your processor costs as much as my console, but my console does more, and faster. In fact, the only thing I think the Core i7 (and Phenom II for that matter) have over Cell is that they can run Windows and DirectX games (which is the whole reason I also have a PC is PC gaming > Console gaming IMO).

Somebody brought up the XDR thing... yeah, the PS3 uses "XDR." Cell was built to initially work with DDR, though. Sony eventually chose to use XDR from Rambus because of its superior bandwidth-per-pin, but Cell actually works with DDR and DDR2 as well.

We're not even getting into the performance myth thing. Intel can't compete in the calcs/clock arena against IBM, or Intel-based super computers would fare better on the supercomputer top500 list (they have 1 spot in the top 10, IBM has 3 and AMD w/ IBM assistance has 2 as of Nov. 2008 list). Plus, a dual PPC G5 "feels" smoother than its equivalent Core2-based Mac. it's just like The Phenom II X3 vs the Core2 E8600, the Core2 slightly beats it in benchmarks, but the X3 "feels" smoother all-around.

And yes, I know IBM and AMD have been working together for a long time, but this will be the first time IBM actually pushes technology at AMD (the 28nm manufacturing process). They've often lended a helping hand, but this time IBM is sorta just letting AMD take the easy way out on it.

yes, its still a lot, but you shoulda seen how much I wrote before lol

 

[ordovician]

On a side note, I think IBM could really, really make AMD go places, if it decided to bite...

 

[pxc]

On a side note, I think IBM could really, really make AMD go places, if it decided to bite...

Why would IBM want to get back into x86 again, especially without a PC division? Even in servers, only 3 out of 17 System x series models have AMD CPUs. Other than to prop up AMD with more cash (a benefit to AMD), what is the expected payout to IBM? IBM already gets $100 million a year from AMD to share process tech. If IBM is interested in anything AMD has, it will probably wait for the pieces to go up on the block.

 

[ordovician]

Well, I've just heard that IBM would be the most logical choice to buy AMD, if it came to that. If IBM wanted to enter the x86 desktop market, buying AMD would be a pretty wise move.

 

[rflcptr]

Well, I've just heard that IBM would be the most logical choice to buy AMD, if it came to that. If IBM wanted to enter the x86 desktop market, buying AMD would be a pretty wise move.

I'm not sure they'd be able to keep the x86 license, though.

 

[Malakai]

I think AMD is doing great and will continue to do so. They have some problems, even some serious ones, but they have a powerful well priced lineup. And according to sites like newegg Phenom 2's are selling like crazy.

Same goes for ati. They have both the fastest (or tied for the fastest) top end cards and by far the best budget cards (4830/4770).

With the economy the way it is, I'm hoping people continue to see that we are lucky to have a great company like AMD taking a "sell lots to everyone at a great value and small margin" approach over intels 'charge a huge premium for anything near the top end' one.

 

[pxc]

I'm not sure they'd be able to keep the x86 license, though.

IBM had and still might have an x86 license. Back in the 1990s it was cloning and collaborating with Intel on licensed 486sx/dx derived chips. IBM got out of that businesss a long time ago, after losing much money.

---
Malakai: AMD used to sell $1000 and $1100 desktop processors. I'm sure they'd love to do so again, if they had anything that could pull those margins. They haven't for years.

 

[Malakai]

IBM had and still might have an x86 license. Back in the 1990s it was cloning and collaborating with Intel on licensed 486sx/dx derived chips. IBM got out of that businesss a long time ago, after losing much money.

---
Malakai: AMD used to sell $1000 and $1100 desktop processors. I'm sure they'd love to do so again, if they had anything that could pull those margins. They haven't for years.

All I really care about is the here and now. I am in love with AMDs current pricing for the performance. In the past AMD has had a faster flagship than intel, does mean they are still faster now
I'm not faulting Intel, $280 is not a bad price for the i7 920. Overall a P2 platform build can be done for slightly more than 1/2 a base enthusiast i7 build and has 8gb of ddr2-1066 vs 6gb ddr3-1600, and the P2 can be dropped into many existing AM2/+ boards currently providing a home to slower AM2 chips.

I like ddr2 over ddr3 right now for price reasons as well. Great quality low timing ddr2-1066 is only $50-60/4gb compared to ~$100+/4gb moderate to high timing ddr3-1600.

 

[facebook]

Really?
"This new fab process has allowed AMD to fit many more transistors into a Phenom II die—an estimated 758 million, versus 463 million for the Phenom—while reducing the die size from Phenom's 283 mm² to just 258 mm². Interestingly enough, the Phenom II's basic specs sound remarkably similar to those of the Core i7, which weighs in at roughly 731 million transistors and 263 mm². "

Now I would like to see your math...

Harvetsting
You have probably seen the X3's and other variants of Phenom

 

[SmokeRngs]

A simple price drop across the board of Intels I7 chips will put AMD right back to where they were in 2008.

This is where your argument falls apart. It has been obvious almost since the release of the i7 that the high price for an i7 system is not the processor but the whole platform. At first all three main components were expensive. Intel dropped the price on the low end i7 and that alleviated that problem. DDR3 came out in larger quantities which dropped the price on the second component considerably. The only part left and the only part which hasn't fallen in price much is the motherboard.

The X58 chipset is marketed as the highest of the high end for chipsets. Because of that, Intel will not drop the price of the chipset. I don't have current figures but historically, Intel has always had high prices on their chipsets and does not like to lower those prices. This is a large part of the cost of the X58 motherboards and since Intel has a monopoly on i7 chipsets, there is no reason for Intel to drop the price on them.

Basically, Intel has put itself in the corner concerning the platform price of the i7. Even in the socket 775 arena, Intel rarely lowered prices on the x line of chipsets even after a budget chipset was released which had 95% of the features and performance of the top of the line chipset. x38 chipset based boards were still selling for a premium even after the P45 and x48 chipsets were released.

Taking a look at history, Intel will not get into a price war with the high end chipset. The only way the i7 can get into a true price war is by lowering the price of the CPUs as well as the chipsets to bring the total platform cost down.

That said, Intel doesn't actually need to put the i7 into a price war. The best AMD Phenom II only matches the current socket 775 line of CPUs in performance. The i7 stands unopposed at the moment. It's in the low end to midrange market where AMD is currently hurting Intel. That's the part of the market where marketshare battles take place and the largest volume is sold. Profit margins per piece are much lower than the high end but volume makes up for the difference.

 

[oldpablo]

Guys as ridiculous as it sounds, performance is no longer the end all be all these days. Because unlike before, computers are just plain "fast enough" for the average consumer. This is why AMD is playing a great game and i7 is just not taking off. We're hitting a brick wall in some ways in terms of clock speeds so other factors are starting to weigh in with importance. More cores, platforms, software, ease of use, total cost, etc. If all i7 has going for it is "it's really fast" then it's just not a priority anymore except for enthusiasts. And enthusiasts frankly are the minority. Ferarri's are also "really fast" but people just don't need them and are perfectly happy with their bang for buck/efficient cars. As I said earlier, I'd rather spread my money across multiple machines then dump it all into one and realize all I did was make it a couple seconds faster then my last one.

 

[TechBoy]

Guys as ridiculous as it sounds, performance is no longer the end all be all these days. Because unlike before, computers are just plain "fast enough" for the average consumer. This is why AMD is playing a great game and i7 is just not taking off. We're hitting a brick wall in some ways in terms of clock speeds so other factors are starting to weigh in with importance. More cores, platforms, software, ease of use, total cost, etc. If all i7 has going for it is "it's really fast" then it's just not a priority anymore except for enthusiasts. And enthusiasts frankly are the minority. Ferarri's are also "really fast" but people just don't need them and are perfectly happy with their bang for buck/efficient cars. As I said earlier, I'd rather spread my money across multiple machines then dump it all into one and realize all I did was make it a couple seconds faster then my last one.

very true but it really DEPENDS. The i7's would not just make it a few seconds faster but a good 150% faster or around an avg. of at least 3 mins per task such as video encoding over what i got. Not just that, but alot of todays proc's are indeed more than what is needed for a single thread or program but honestly with today's monitor prices and need for higher productivity who the hell just runs one program anymore? I run at least 7 different things at once lol, and keeps adding because i like to have the convenience for the ability to keep multiple things open and have them running all at ample speed. Not that the P2s cant do that but for about $50 more for the proc I can grab a processor that will prbly blow the P2s out of the water with a i7 doing 7 threads and programs than the P2.

As much as I love AMD (I have only built on Intel Comp compared to the 5 AMDs that I have in the past) the P2s arent really anything compared to the i7s I only wish the motherboard prices were cheaper so I can get one now.

Sum it up, I can get a Core i7 for 230 and a P2 X940 for like 170? 40-50% faster in the majority of stuff I do for 60? Ill take that even if the motherboards cost a little bit more.

 

[funkydmunky]

very true but it really DEPENDS. The i7's would not just make it a few seconds faster but a good 150% faster or around an avg. of at least 3 mins per task such as video encoding over what i got. Not just that, but alot of todays proc's are indeed more than what is needed for a single thread or program but honestly with today's monitor prices and need for higher productivity who the hell just runs one program anymore? I run at least 7 different things at once lol, and keeps adding because i like to have the convenience for the ability to keep multiple things open and have them running all at ample speed. Not that the P2s cant do that but for about $50 more for the proc I can grab a processor that will prbly blow the P2s out of the water with a i7 doing 7 threads and programs than the P2.

As much as I love AMD (I have only built on Intel Comp compared to the 5 AMDs that I have in the past) the P2s arent really anything compared to the i7s I only wish the motherboard prices were cheaper so I can get one now.

Sum it up, I can get a Core i7 for 230 and a P2 X940 for like 170? 40-50% faster in the majority of stuff I do for 60? Ill take that even if the motherboards cost a little bit more.

As many have pointed out it is a platform cost, not just the cost of the CPU. A $50 difference is absolutely a best case scenario. Newegg for example has a $289 to $190 price list. If I can rely on my simple math that is a $99 difference. Not sure where you shop, but apparently newegg is going out of business. As for you wishing MB prices were better, keep wishing. Is the i7 faster? Sure. But we most all have a budget, and for some a bigger HDD, faster GFX card, more RAM, or a larger monitor, are more desirable then %20 lead in heavy render tasks.

 

[Socrilles17]

Agree with many of the comments:

Basically AMD and ATI have taken a new strategy vs. Intel and Nvidia respectively.

Go for cheap (not just manufacturing but also development costs which need to be recouped) platforms, and get performance by scaling ie. multiple cpu cores, or multiple gpu cores. We have already seen that this worked very effectively against Nvidia (radeon 4870, 4850, 4830 etc vs gtx 280, 260, 250). Processor wise AMD is definitely competing quite well with the socket 775 processors - just look at the change in marketshare posted at the start of the thread, and although AMD may not have much of an advantage on the manufacturing cost, it definitely has a huge advantage on the R&D side of things (phenom core is about 3 yrs old now?) and this will continue as they scale to six core and 12 core processors which the HT technology makes this easy (I believe however I'm no expert). Finally AMD/ATI makes amazing chipsets for the average consumer - the graphics included are perfect for the everyday user (blu ray, and mild gaming). If AMD just released a platform in which they really linked a decently performing low power dual core processor (AMD neo needs more performance IMO) with a slightly better chipset then they have available (integrated graphics that can easily do blu ray/modest gaming - ie play most games on lower res settings) they would kill the atom, and also be making profits on both the chipset and processor sales. And mobo manfacturers have to be hating intel (new chipset, new memory, new socket type constantly) whereas although not perfect backwards compatible AMD definitely makes far fewer changes. Although Intel could just drop the price on i7 and even their chipsets at any moment to regain marketshare the platform cost is still much greater (who doesn't have 12 gb of ddr2 laying around that they would like to keep using since ddr3 doesn't seem to offer much of an advantage). I don't know about the other people on this board, but I am definitely heading in the multiple PC direction (one for each TV, a file server, and my main PC) and except for my main PC there is no reason at all to consider intel as AMD/ATIs integrated graphics/processing power is perfect for HTPCs

 

[Dan_D]

Well, I've just heard that IBM would be the most logical choice to buy AMD, if it came to that. If IBM wanted to enter the x86 desktop market, buying AMD would be a pretty wise move.

IBM will never buy AMD. This I can almost guarantee. IBM makes over 80% of its revenue from non-hardware based sales of software and IT outsourcing services. IBM's most recent strategy has them moving farther and farther away from hardware sales. IBM still does a ton of R&D, but my guess is they'll continue to develop technologies and sell off that technology. They've been doing this for more than two decades now.

 

[Pylon]

Personally I think AMD really needs to get 45nm mobile processors ASAP. Their current 65nm Turions get easily smoked by Core 2. The only advantage these days to an AMD mobile platform is that in general you would get better graphics(HD 3200 or GF8200) for a low end AMD compared to low end Intels(X3100, X4500). What they need is a better mobile processor to go along with their better mobile chipsets and graphics.

Their 45nm process is proving to be quite mature, considering that they have 45W quad-cores on the road map. Adapt them to mobile use and cut TDP just a tad and they could easily kill Intel's mobile Core 2 Quads. They should also make 45nm Turion X2s. 25W TDP shouldn't be too hard to achieve and most likely they will outperform the 25W Pxxx Core 2s, and a set of regular 35W Turion X2s to compete against the Txxxs. Nehalem won't show up in mobile processors for at least the next 6-9 months and AMD could potentially dominate this field.

 

[JimmiG]

Actually, the Core i7 is cheaper to manufacture, and not only because of scale. AMD uses SOI wafers and has more metal layers, both of which increase manufacturing costs vs bulk Si and fewer metal layers. So not only does each good Phenom 2 chip cost more to manufacture, AMD sells nearly the whole line (X3 and X4) for less than the lowest end i7.

This is a huge problem for AMD. Their $200-ish CPUs are more expensive to manufacture than Intels $500+ i7's, not to mention the smaller Intel Quads. Not only is Intel selling more CPUs, they make more money from each sale too. The worst is still ahead for AMD because they are engaged in a price war they can't afford to lose, or win. What happens when they're forced to drop the price of their highest clocked X4's to under $140-ish?