Print 94 comment(s) - last by Justin Case.. on May 28 at 11:08 PM

AMD details its desktop plans for the next year, though be it a quarter later than originally planned

AMD this week took the covers off its quad-core desktop CPU codenamed Agena FX in Sunnyvale, California. The processor was featured in single and dual-socket configurations -- the dual-socket configurations mark AMD's first demonstration of an 8-core desktop. 

Next week Monday, AMD will reveal the brand name for this Agena FX-based brand, Phenom FX.  The official launch date for Agena is still undetermined.

"Quad-core, codenamed Barcelona will launch later this summer, in the July, August kind of time frame -- followed by [Agena FX] on the desktop," stated Robert Rivet, AMD executive vice president and CFO.

AMD demonstrated the power of this Phenom FX system by encoding a 1080p movie trailer into H.264 in near-realtime.  The company would not reveal the clock frequencies to DailyTech, though previous company guidance indicated Agena FX will debut at 2.6 GHz clock frequencies. 

Agena FX is just the halo product for the desktop AMD K10 lineup.  Agena (non-FX) will constitute the bulk of AMD's Phenom offerings.  Agena is essentially identical to Agena FX, though the non-FX processor does not support symmetric multiprocessing -- Agena FX can support up to two sockets at this time.  AMD will also announce its dual-core Agena bins, codenamed Kuma, shortly after Agena.

Intel is slated to launch its 45nm Penryn processor family in late 2007.  "Penryn will be shipping in 2007," stated Intel representative Nick Knupffer.  "We are launching server versions of Penryn in the second half of 2007," he added.

Intel's desktop variants of Penryn, dubbed dual-core Wolfdale and quad-core Yorkfield, surfaced last month.  The server counterparts to these chips, dubbed dual-core Hapertown and quad-core Yorkfield will suceed the existing Clovertown and Woodcrest Xeon components.

AMD's Rivet makes one promise; we'll see quad-core desktop components in 2007.  "That'll launch a little after the server part, it'll be called Agena. You'll see that in the Christmas line-up," Rivet claimed.

Comments     Threshold

This article is over a month old, voting and posting comments is disabled

Nehalem Just Over a Year Away...
By EndPCNoise on 5/10/07, Rating: 0
RE: Nehalem Just Over a Year Away...
By kamel5547 on 5/10/2007 7:58:47 PM , Rating: 5
*Shrug* Seeing as I ahve a hard time finding apps that will stress 2 cores I think 8 is relatively pointless outside of high-end servers and workstations. Not that it isn't a nice achievement, but I'd rather see speedier dual core chips (or more software).

Personally I'm more worried about reports that beyond 8 cores we are likely to see a performance decrease if the architecture doesn't change. I sure hope Intel doesn't make the mistake it made in the past and push cores the way it pushed MHZ.

RE: Nehalem Just Over a Year Away...
By EndPCNoise on 5/10/2007 8:12:18 PM , Rating: 1
For some time now, software companies have been developing optimizations for multicore cpu's. Software will "automatically scale" (I apologize for lack of a better term) for n-cores.

Please refer to this article from AnandTech concerning multicore capable software:

RE: Nehalem Just Over a Year Away...
By Justin Case on 5/10/2007 10:49:28 PM , Rating: 4
No, it won't. SMP optimizations aren't a magic dust that you can simply sprinkle on top of software to make it run faster. A lot of problems simply cannot be broken down into multiple threads.

Even software with years of multi-core optimizations (like 3D rendering and compositing applications) have some single-threaded stages, and adding more CPUs or more cores won't speed those up, so the scaling is far from linear (even on AMD platforms, that aren't bottlenecked at the bus).

In any case, most people's CPU stays idle for 90% of the time, so all this is a bit irrelevant. Having two CPUs (or two cores) does make a noticeable difference in system responsiveness. Beyond that, you're only likely to benefit in very specific tasks (video encoding, 3D rendering, some compilers, busy database servers and not much else).

People need to understand that multithreading will not necessarily speed up "traditional" tasks, but it does give developers the ability to do more things at the same time. In other words, a dual-core system at 2 GHz probably won't run a game faster than a single-core system at 2.2 GHz, but it can have much improved AI, for example. You get the same FPS (or even slightly less), but a better gaming experience.

RE: Nehalem Just Over a Year Away...
By EndPCNoise on 5/10/2007 11:48:50 PM , Rating: 1
Did you even bother to read the AnandTech article about multicore optimized software with the link I provided?

Software in the near future will greatly benefit from multicore systems. Is it linear scaling, not exactly, but it's close enough. The performance increase is very significant.

Software engineers are working hard on breaking down typical single threads into smaller threads which can be spread over multiple cores.

If you took the time to read the short 10 page article you would have learned about these new developments.

Even AMD's own representatives have publicly stated (addressing their losses) that they regret not developing quad core cpu's much earlier on.

I have nothing against AMD. My last three systems were AMD based, and I am using an AMD system now.

Intel has also recently stated that quad core systems will become mainstream in the near future.

If you choose to believe that multicore systems are for workstations, servers, etc., than stick with your single or double core system. The mainstream will move on without you. Barcelona is coming, and Intel will be offering quad cores at $266 later this year.

RE: Nehalem Just Over a Year Away...
By Flunk on 5/11/2007 12:15:00 AM , Rating: 4
What about smaller developers who can't afford to write the significantly more complex code the sort of optimizations you are talking about require?

Multithreaded code is much harder to conceptualize and build, not to mention horribly difficult to troubleshoot.

Couple this with the fact that in order to be scalable in the way you want the program would need to be written to spawn new threads on a continual basis that returned information to the main thread (for output) which would decrease performance on chips with only one core.

Multithreaded applications are not a magic bullet. Programming multithreaded applications is difficult, time consuming and expensive.

This is NOTHING like linear scaling, but if done right it will improve performance.

RE: Nehalem Just Over a Year Away...
By EndPCNoise on 5/11/2007 12:43:36 AM , Rating: 1
Multithreaded code is much harder to conceptualize and build, not to mention horribly difficult to troubleshoot.

That's why multicore software development kits, and new multicore game engines are being created. These will then be licensed out to other companies/developers.

This is NOTHING like linear scaling, but if done right it will improve performance.

A quad core's performance will not be equal to that of two dual cores, but it will be over 90% of the performance of two dual cores. That is very significant. Not linear, but close enough.

Please read this short 10 page AnandTech article. It will address each of your questions/concerns.

RE: Nehalem Just Over a Year Away...
By gramboh on 5/11/2007 2:13:44 AM , Rating: 4
As Justincase says below, you are grossly overstating the implications of the article which discusses what Valve (a single game developer) is attempting to do with multi-threading. You are overestimating the 'impact' multicore CPUs will have on desktop users in terms of software in the short-term. Dual core has been around in the mainstream for a few years, dual socket has been around for 10+ years (with desktop mainboards, although only a tiny % of users had SMP in the old days). It's not an easy problem to just use a devkit to 'break down single threads into multiple threads'.

RE: Nehalem Just Over a Year Away...
By TomZ on 5/11/07, Rating: 0
By Amiga500 on 5/11/2007 10:39:06 AM , Rating: 2
Yeap, broadly agree with Justin Case etc... for conventional computing, it will soon reach a bottleneck in terms of computing power/returns.

Heck, in some apps we are there a long time... how many CPUs do you need for openoffice again? :-)

Not that I care if AMD/Intel pile 528 cores onto a die - I'll use them for my CFD work :-D

AMD/Intel get the majority of their profits from the workstation/server area do they not? Workstations simply cannot get enough power (as long as communication times between CPUs are kept down - quite important that), parallel processing is a great way of achieving that.

Expect the CPUs 2 generations down the road to have 4 or 8 cores on them... half of which will be GPUs ;-)

RE: Nehalem Just Over a Year Away...
By Proteusza on 5/11/2007 7:29:19 AM , Rating: 2
The article has nothing about thread scaling or breaking down threads, its about valves distributed computing service!

Breaking down a thread into smaller threads suffers similar problems to using superscalar out of order architectures (issueing multiple instructions of a thread on different engines in a different order to what they were intended to be executed), but only worse.

Its not that it isnt worth doing, its that its very very difficult, and it may be better worth making each thread really fly on each processor than having lots of small threads, which might worsen overhead due to things like context switching.

There are some things that, no matter what, have to be done sequentially.

By GoatMonkey on 5/11/2007 3:25:26 PM , Rating: 2
That's true, there are lots of sequential processes happening in a computer. And a multi-core CPU can handle many of those sequential processes at the same time. Why worry about whether or not your apps are multi-threaded? Your OS does multi-tasking in a way that takes advantage of all of your cores. I currently have Visual Studio, Eclipse, Firefox, Internet Explorer, Outlook Express, and Textpad open and 4 things running in the systray, not to mention all of my running services and background processes. I wish this PC I have at work had more than 1 core.

RE: Nehalem Just Over a Year Away...
By goku on 5/14/2007 3:08:42 PM , Rating: 2
exactly, people just don't get it.. Some tasks simply can only be run on one processor at a time, balancing equations between two processors and having them come out in an orderly and timely manner isn't easy. Sure the processing capability is near linear but only in circumstances such as simple number crunching which is a best case scenario, otherwise, exponentially speaking you'll be gaining only by 1. So for example if you add 4 cores, it's really only +2 performance, *2 and then it's +1 *2 and then it's +1 etc. Thats the way I see it anyways, wouldn't be surprised if it was wrong. Anyways it just needs to be made clear that performance increases aren't going to be easy as adding another core.

RE: Nehalem Just Over a Year Away...
By Justin Case on 5/11/2007 12:23:51 AM , Rating: 4
With all due respect for Anandtech (and interpret that sentence as you will), I don't need to read an article on a website to understand a field that I've been involved in (both as a user and a developer) for almost 10 years. You see, I am a software engineer.

In any case, I did read the article, when it was originally published. Mainly because I like Valve's games and wanted to know what they're up to. But it's not exactly an advanced technical document about SMP development.

RE: Nehalem Just Over a Year Away...
By darkavatar on 5/11/07, Rating: -1
By clayclws on 5/11/2007 2:52:13 AM , Rating: 4
What Justin Case said is very true. When YOU graduate, you'll start to realize his points. I don't need to go on because clearly you don't understand his points.

RE: Nehalem Just Over a Year Away...
By Proteusza on 5/11/2007 7:30:56 AM , Rating: 2
He didnt say he wasnt writing multi threaded software already, just that the article presented is in much less depth than he works. he's an expert in the industry, who are you to say he doesnt use multithreaded code?

RE: Nehalem Just Over a Year Away...
By TomZ on 5/11/2007 3:25:00 PM , Rating: 1
he's an expert in the industry

LOL, where did you get that from.

By the way, most "experts" in the field of engineering have 20, 30, or more years of experience.

Also, just because someone is a software engineer, that doesn't tell you anything about the area they work in. That field is very diverse, anything from programming 8-bit single-chip micros in assembly on up, and across nearly every possible area of human activity.

RE: Nehalem Just Over a Year Away...
By KernD on 5/12/2007 1:16:07 PM , Rating: 2
Scaling in my field, which is the same field as Valve, is not linear, at work(an Activision studio) we have a great game engine that can keep a xbox360 fully busy at all time(all 6 threads) and the PC version has caused all but one Pentium EE (2 core with HT) to over heat to the point of shutting down.

Scaling is not linear simply because not all your code can be threaded.

On 2 core you would have work load distributed in such a way.
On a quad it would look like this
The lenght of the work line for quad is not half the lenght of the duo because some work can't be done on an other thread.

One very important work load in game is rendering and sending commands to the device doesn't work any faster with 2 threads because you have to use thread safe device which will take a critical section all the time. And one thread can already, when well coded, saturate the rendering hardware.

RE: Nehalem Just Over a Year Away...
By Justin Case on 5/13/2007 8:31:23 PM , Rating: 2
If the system is "overheating to the point of shutting down" (with any code), maybe it's time to upgrade the cooling, no...? ;-)

P.S. - Okay, I know, "adequate cooling" for a NetBurst "EE" Pentium probably means a continuous supply of liquid helium. :-P

RE: Nehalem Just Over a Year Away...
By KernD on 5/13/2007 10:10:26 PM , Rating: 2
You should never be able to overheat a computer with work loads... only with overclock or unplugging a fan...

Stock Intel fan are crap, the thing is that the whole case got too hot, so we replaced the processor fan with better ones and added a large fan at the front of the case just under the DVD drive... so now I have cool blue led at the front of my computer at work!

RE: Nehalem Just Over a Year Away...
By GlassHouse69 on 5/13/2007 10:57:06 AM , Rating: 1
lota posts!

I play several games that work faster on dual core systems, slightly optimized for the 2 cores.

when another program is running in the background now, now that the cores are optimized, I get slow downs! I need a 3rd core. Not a fourth core necessarily, but, a third core at least. I wish I could find something that can tell 2 cores to work for the game, 1 core for i/o, hd use, (worst slowdown ever) and the 4th core for other backround aps. if that works one day well, 4 cores will be great.

8 cores are useful if you have more than 1 program that utilizes more than 1 core at once. not likely but :) never know

RE: Nehalem Just Over a Year Away...
By Proteusza on 5/13/2007 3:39:24 PM , Rating: 2
there is a program that tomshardware used to permanently set processor affinity but I cant remember what it is called.

Yes hard drives and cd roms cause slowdown, but an extra cpu wont help there. hard drives and cd rom drives are a few thousand times slower than main memory, hence your cpu has to wait for them.

windows is a multiprogramming OS, so if you have a game that can use a dual core cpu effectively and another program running in the background, windows will manage your cpu time so that all of the programs get some time. your game will just have two or more threads, so both of them can run at the same time. my pc has 620 threads running at the moment, and it isnt even doing anything (those threads are mostly windows threads and my sql server etc).

so, what is happening on your system is that the game's threads are continually being swapped out for all the other threads on your pc. having more cores will help this, but it isnt something you should worry about - it does this so fast and so often you shouldnt notice.

By Justin Case on 5/14/2007 8:05:42 PM , Rating: 2
I keep reading about that THG program but never saw it. Anyway, the resource disk that comes with server versions of Windows includes a tool called "imagecfg.exe" that can be used to set a program's affinity. If you search for that on the net you'll find both the tool and some instructions.

P.S. - I seriously doubt you have 620 threads actually running. If you're simply adding up the threads in Task Manager, remember that completed threads are kept in the counters (in other words, a program that says it has 50 threads has started 50 threads, most of which have probably been closed since). Databases will generally run only 3 or 4 "base" threads plus one thread per active client.

RE: Nehalem Just Over a Year Away...
By honeg on 5/10/2007 10:09:04 PM , Rating: 2
*shrug* Seeing as I write threaded code for a living, having more cores available on my desktop makes me much happier. I'd rather have 4 or 8 marginally (10-20%) slower cores than a single or dual core.

RE: Nehalem Just Over a Year Away...
By elpresidente2075 on 5/10/07, Rating: -1
RE: Nehalem Just Over a Year Away...
By Justin Case on 5/11/07, Rating: -1
RE: Nehalem Just Over a Year Away...
By James Holden on 5/11/2007 1:16:07 AM , Rating: 5
Bill Gates' statement was quite accurate at the time.

Except, it wasn't his quote anyway:

By Justin Case on 5/28/2007 11:08:53 PM , Rating: 2
I never assumed it was a direct quote. Gates is too smart to say anything like that. But he effectively sanctioned that limit in the original MS-DOS / IBM PC design, so in a way he did "state" that.

And while 640 kB were enough for anyone in those days, limiting the system to that was dumb.

AMD's K8, for example, uses "physically" 40-bit memory controllers, but the AMD64 architecture was designed to suport 64-bit addressing. In other words, although 40 bits is "enough for anyone" at this time, the system can theoretically be expanded to support 17 million times (2^24) more RAM than that, without any changes to the instruction set.

RE: Nehalem Just Over a Year Away...
By EndPCNoise on 5/11/07, Rating: 0
RE: Nehalem Just Over a Year Away...
By johnsonx on 5/11/07, Rating: -1
By johnsonx on 5/11/2007 5:02:15 PM , Rating: 1
Oh, come on, a -1 for a harmless little pot-shot at Justin Case? He's a bit of an arrogant know-it-all, yet rarely seems to actually know anything. If his comments in this thread aren't enough for you, search DT for Rydermark and read his asinine comments about the photo-chop scandal. He's well earned it.

By Proteusza on 5/11/2007 7:39:26 AM , Rating: 2
Wow, somebody disagrees with your unfounded and uneducated opinion and you start with the sarcasm?

Firstly, dont put all of your faith in these companies. Intel made the Netburst architecture - anyone remember how bad that was? Yet they did what they thought was right. So dont think Intel, IBM etc never ever make mistakes, they do.

Secondly, what if the forum writers work for companies that routinely develop multi threaded apps? Are they not allowed to comment? Is it impossible that one of them knows as much about multi threaded apps as an employee from IBM? Are all IBM employees magically better than everyone else?

Thirdly, I suggest you read a few books on distributed and parallel computation, and you will realize why it is a difficult thing to do. Very few algorithms fit into a distributed model perfectly, they will always have a sequential component, which can only be executed by one core. That sequential component could be anywhere from 1% of the app to over 90%, depending on how lucky or unlucky you are. Amdahls law then kicks in and says that although your parallel parts are really flying, overall performance isnt that great (I'm not going to explain amdahls law here, wiki it).

Also, remember that Cell wasnt only designed for the PS3 - it will see lots of users in servers (highly parallel), science computing (largely highly parallel), possible audio and visual (process multiple audio streams, multiple pixels you get the idea).

Also, the multi core approach started partially because having multi cores lets you run multi threads, but also because they couldnt raise clock speeds much higher using current technology. Prescott didnt do well at 3.8 GHz, they just couldnt keep going, so they turned to parallel architectures.

RE: Nehalem Just Over a Year Away...
By defter on 5/11/2007 3:00:56 AM , Rating: 2
I think 8 is relatively pointless outside of high-end servers and workstations.

Duh, the 8-core/16-thread Nehalem is naturally aimed at servers and workstations... They probably make also 8-core EE version for desktop for those who have too much money :)

RE: Nehalem Just Over a Year Away...
By Aikouka on 5/11/2007 9:01:51 AM , Rating: 2
Kamel, I understand your point quite well, but I think your viewpoint is simply too narrow. Open it up and think about the other aspects of computing. I'm not even referring to number crunching (SETI/FOLDING@HOME), but simply multi-tasking.

At this point, a dual-core is pretty darned efficient for multi-tasking, because our applications are fairly non-threaded or a separate thread doesn't tax the system much to be noticeable. Increasing the available core count can constitute the same difference we saw in single to dual core when developers start going crazy with threads. Now 8 cores may still be a bit too much, even for what I'm talking about, but the advantage I see is 8-core processors being available can possibly drive down the quad-core processors to an even better price. Based on previous news reports, quad cores should be down to about $500-600 relatively soon, so it would be nice to see them even further ;).

RE: Nehalem Just Over a Year Away...
By zigzagcom on 5/11/2007 12:45:48 PM , Rating: 2
I am reading this thread with much interest; I might be a bit off the mark, but what about virtualization on the desktop. I run several OS's on different HDD's, but with multicores, it will become even easier and more efficient. Not your average user, but many folks are dual booting 'twixt Win & Lin. Any thoughts...

RE: Nehalem Just Over a Year Away...
By cochy on 5/11/2007 3:00:57 PM , Rating: 2
Multi-core will definitely benefit those who want to run a few OSes in VMs. However total system memory will be more important for performance.

Bend Over and Kiss Your Butt GOODBYE!!
By TheBeagle on 5/10/2007 8:11:43 PM , Rating: 1
Does the saying "Too Little, Too Late" have any meaning here? It is rather incredible to read just how far down the slippery slope AMD has fallen in this battle with Intel. I was always an AMD fanboy, but these last six months or so have left me speechless.

I could have never imagined that the company that stole the food right off Intel's table just a few years ago has now descended into a painful spiral of self-destruction. Talk about being caught with your pants down. This is downright painful to watch. It is vitally important to have at least TWO processor manufacturer competitors in the marketplace to ensure decent growth and technological advancement. I fear that once the 1200 lb. gorilla (Intel) takes hold of the playing field, we (consumers) are likely to pay a helluva price for that monopoly. I sure hope I'm wrong on this.

By LatinMessiah on 5/10/2007 8:23:49 PM , Rating: 2
Nice summary of other people's comments.

RE: Bend Over and Kiss Your Butt GOODBYE!!
By spluurfg on 5/10/2007 8:58:21 PM , Rating: 2
Intel is a much larger company... AMD was doing well for a while, but part of their success was that Intel was not really innovating but cruising on their marketing machine -- they were pretty much ignoring AMD for a while since AMD couldn't get any market share anyway.

Now that AMD started a tech war, how do they intend to compete with a company that can massively outspend them on fabs and research?

Innovation on the level of microprocessors requires a lot of people, time, and investment.

By cochy on 5/10/2007 11:25:02 PM , Rating: 3
Now that AMD started a tech war, how do they intend to compete with a company that can massively outspend them on fabs and research?

Yup that's the question of the year. Personally I was a big fan of AMD when they were offering quality products at very good prices. AMD = value. Intel was priced higher because they had the brand recognition. Ever since AMD took over the technology lead they slowly became equally priced with Intel. That worked until Intel woke up from their Netburst nightmare. Now where's AMD? Should they move back into the bargain value position? Or try to compete head to head? I guess we'll see when K10 comes out. They will get my dollar again if they bring value back to AMD.

RE: Bend Over and Kiss Your Butt GOODBYE!!
By cooperaaaron on 5/11/07, Rating: 0
By zsdersw on 5/11/2007 7:01:50 AM , Rating: 2
The facts of your first sentence have nothing to do with the predictions in your second sentence.

By LatinMessiah on 5/11/2007 12:53:34 PM , Rating: 2

RE: Bend Over and Kiss Your Butt GOODBYE!!
By hoyanf on 5/10/2007 9:11:30 PM , Rating: 2
I couldn't care less bout falling behind as long as i get to upgrade my cpu on my same board, bet you Intel cant make it happen if you were using 915/945/955/975 chips... For sure you'd hav to fork up sumthing for upcoming Intel cpu's...

By Chadder007 on 5/10/2007 9:31:26 PM , Rating: 3
But its still up to the mobo manufacturer to release a BIOS that will support the updated CPUs. There is currently not a guarantee of that from any of them that I know of.

RE: Bend Over and Kiss Your Butt GOODBYE!!
By Goty on 5/10/2007 9:16:38 PM , Rating: 2
AMD isn't going anywhere soon. Even though they've been getting their butts kicked left and right for the past year and there's a small chance that Barcelona won't be enough to make them competitive again, they'll just fall back into the same stance they were in back in the Athlon/AthlonXP days.

RE: Bend Over and Kiss Your Butt GOODBYE!!
By nerdye on 5/11/2007 1:07:03 AM , Rating: 2
AMD is still close to Intel currently in server performance despite their current lack of desktop performance/laptop performance in comparison to Intel. Why is AMD releasing their native quad-cores on the server market first? Because it will be most successful there! Will native quad-core desktops put the hurt on Intel, that is yet to be seen, but don't count AMD out just because they reached the highroad "finally" in market share with A64 architecture and are not in the lead now against core 2 duo. AMD will live on, whether people will be bragging about sporting them anytime soon, that is negligible.

RE: Bend Over and Kiss Your Butt GOODBYE!!
By cochy on 5/11/2007 1:56:12 AM , Rating: 2
Does it tire the fingers at all to keep typing "native quad-core"? :P

I think it's about time people start using NQC or something.

RE: Bend Over and Kiss Your Butt GOODBYE!!
By James Holden on 5/11/2007 2:23:44 AM , Rating: 2
And more importantly, what's the actual difference between "native quad-core" and "non-native quad-core."

By Viditor on 5/11/2007 5:33:39 AM , Rating: 4
what's the actual difference between "native quad-core" and "non-native quad-core."

Interchip communication...
For an MCM (like Intel's quads), performing a cache trace between the 2 halves of the CPU must go through the FSB instead of "right next door" to the other cache. This adds a great deal of latency and increases traffic on the FSB.
The cache of the "native quad-core" is all directly connected and much faster (as is the shared cache on C2D dual cores).

This is mainly why even though the Intel Woodcrest cores are faster than the current K8s, performance on a K8 system at greater than 4 cores is faster than it is on a Woodcrest system.
This effect will be drastically greater with the new K10 chips as internal bandwidth is doubled as well (not to mention all of the other enhancements). The "40% faster than Clovertown" claim by AMD has been fairly evident even before they announced it...but remember that this is on servers.

By CyborgTMT on 5/11/2007 2:48:17 AM , Rating: 2
And what 'stance' was that? Let me refresh your memory with my own buying experiences. The first AMD proc I purchased was a Athlon T-Bird. With a nice copper heatsink it easily overclocked to 1.4 GHz while Pentium III was still having trouble getting out of triple digits. Even with an expensive HSF it still cost less than the Pent. Next up came the XP Thoroughbred which again beat the young Pent 4 at a lower cost. It wasn't until the Northwood based Pent 4 that AMD didn't have a comeback with the XP line. But by then the Athlon 64's were in line to be released. After that was the first time in many years that we've seen a dominant CPU hold the performance crown for an extended period of time with the 64 line.

So if you meant to say by 'stance' - competitive products that keep the technology advancing and the prices down, I also hope things go back to the old ways.

whats going on at AMD?
By someguy123 on 5/10/2007 7:36:51 PM , Rating: 2
it seems like they weren't prepared for any sort competition from intel for 2005~08. their new acquisition of ati doesn't seem to be fairing well either, as the top of the line r600 (xtx) seems to be underperforming when it comes to it's specs compared to the g80 line.

i desperately hope amd atleast equalizes the performance difference so that there will be more of these endless price cuts ive been loving after the release of the c2d.

RE: whats going on at AMD?
By Teletran1 on 5/10/2007 9:03:18 PM , Rating: 2
Does it really matter? Are quad cores even mainstream? AMD needs a competitive mainstream desktop processor and they need it 6 months ago. Who cares about a core pissing contest?

RE: whats going on at AMD?
By someguy123 on 5/10/2007 10:01:04 PM , Rating: 1
what are you talking about. i was referring to the fact that amd has nothing out to compete with intel, and neither does ati with nvidia. it also seems like they were not prepared for any advancements from intel, as amd's future lines look more like small updates and core increases.

i never said anything about amd being behind in multicores; I honestly could care less about that. I have no use for quad and neither will anyone else not using their computer as workstations until developers develope software that scales per core. what i cared about is that it looks like amd can't compete for a while, meaning the market could become intel's and we'll soon start seeing less price wars, which have been VERY nice to us consumers.

RE: whats going on at AMD?
By Justin Case on 5/10/2007 10:58:21 PM , Rating: 4
Last time I checked, they do have competitive mainstream processors. After the latest cuts, they'll actually give you slightly better "bang for the buck" than Intel CPUs of the same price. Lower power consumption on idle, too (slightly higher under full load, but most PCs stay idle for 90% of the time, so...).

Intel has an advantage mainly in mobile processors and high-end single-socket systems (at 2 sockets they're more or less matched, at 4 sockets AMD wins thanks to hypertransport + NUMA).

For high-end gaming systems, right now Intel is the clear winner. But "mainstream"? It's very much a two-horse race.

If I was going to buy a home / office (desktop) system today, it would probably be a low- or mid-range Athlon X2.

Some people's perception of reality is very easily swayed by blog headlines...

RE: whats going on at AMD?
By rbuszka on 5/10/2007 11:12:22 PM , Rating: 5
Numa, numa, yay; numa numa numa yay.

RE: whats going on at AMD?
By someguy123 on 5/11/07, Rating: 0
RE: whats going on at AMD?
By Crazyeyeskillah on 5/11/07, Rating: 0
RE: whats going on at AMD?
By coldpower27 on 5/11/2007 12:26:51 PM , Rating: 4
Justin Case's post is pretty accurate overall, when talking about mainstream overclockers aren't factored into the equation, if your talking about the best processor for the enthusiast then that still remains Intel for both overclock ability and overall performance.

What mainstream refers to is stock processor sold to Dell and Grandma.

RE: whats going on at AMD?
By crystal clear on 5/12/2007 7:48:31 AM , Rating: 1
But "mainstream"? It's very much a two-horse race.

Just-in-case do you have some independent (figures) to back up this claim ?

RE: whats going on at AMD?
By Justin Case on 5/13/2007 8:23:40 PM , Rating: 2
Sure, what numbers do you want? Game benchmarks? SPEC? MP3 encoding times? System power consumption? Just divide the results by the price, and you'll see. AMD leads in 4S servers, Intel leads in 1S gaming systems and workstations, everywhere else there's a choice.

Why don't you simply look for the numbers on the net? Can't get much more "independent" than that. Unless you think there is some huge communist conspiracy to sap and impurify your precious bodily fluids...

RE: whats going on at AMD?
By defter on 5/11/2007 3:05:46 AM , Rating: 2
I think 8 is relatively pointless outside of high-end servers and workstations.

Well, Intel and AMD of course.

The catch here is that AMD cannot make quad cores in volume until 45nm process and H2 2008 (Barcelona has a huge 283mm^2 die on 65nm). That's why Intel wants to push market towards quad cores and push dual cores (the ones that AMD is capably of manufacturing in volume) to the low/mid-range segment in order to minimize AMD's profit margins.

RE: whats going on at AMD?
By coldpower27 on 5/11/2007 12:28:47 PM , Rating: 2
With their MCM approach and soon their 45nm process Intel will have a major advantage manufacturing Quad's. It will be interesting to see.

RE: whats going on at AMD?
By Proteusza on 5/11/2007 1:35:24 PM , Rating: 2

I was deciding whether to buy an Intel Core 2 Duo or AMD dual core barcelona (I know it uses a different core name but I cant remember what it is, its kuma or something).

But now it looks like AMD has made up my mind for me. They wont have new dual core parts out by then, so my only option is core 2 duo, because although the AMD's are competitively priced at the moment, they also are much older.

AMD should have released this 6 months ago, and next time their competitor releases a good product they shouldnt sit on their laurels for so long, they should have been quick to get this new core out, do whatever it takes.

Mind you I could buy a cheap X2 and next year...

RE: whats going on at AMD?
By just4U on 5/13/2007 3:02:43 AM , Rating: 2
Oh come on. 95% of those who use computers DO NOT OVERCLOCK THEIR SYSTEMS. It's a non issue. It's not until the 6600 that intel really starts to pull away at stock speeds from Amd's line up and .. Even then the 5600-6000 X2's are fairly competitive performance wise.

Even in the Budget Overclocker's have something to smile about in the amd Camp. Look at the 3600+ x2 at what .. $60 American. Damn thing clocks up to 2.7Ghz on air. You want value/performance that's quite the kicker.. and a definite option opposed to the 4300.

So yeah, Overall? Amd has a comparable lineup until you get into the High end desktop systems at "STOCK" speeds which most of the computing world operates on.

Does it hurt?
By Runiteshark on 5/11/2007 2:00:39 AM , Rating: 1
You know, I really wonder, with AMD's fantastic performance the first quarter, you would think they would do anything they could to rush out their r600 graphics cards and new barcelona agena, and kuma chips.

But have you noticed what they've been doing? They've been pushing them back, farther and farther back. When those benchmarks came out for the R600, the XT did well against the GTS (And hopefully its cheaper siblings will do as well against the rest of the 8xxx series) but they still haven't been released.

No real factual proof of the performance of the Barcelona parts, or any other parts for that matter have been given out to anyone. When Intel had its Core 2's about ready for launch, I remember that they were bragging about performance, and how much they beat AMD's current lineup (Which at the time, was priced quite a bit higher}.

During that time, FUD was about everywhere, AMD fanboys panicked, and forums were ablaze with supposed benchmarks of the C2D.

Where is AMD with its benchmarks about its upcoming chips? All we get told is that its "40% faster", and other random statements.

This leads me to believe that the Barcelona will be just like the R600, It will be able to compete equally with the C2Q 6600, but Intel will still hold the performance crown. Now I do understand that the top end chips don't get sold the most, but for marginal improvement over their previous k8 architecture (and k8l for the low power models), you must consider the amount of capital put into developing these chips.

I dunno, But I don't want AMD to slip back into the old days of Netburst. Sure it was good when it started, but it was beat to death, and I hope that Intel doesn't do the same thing over again, but this time instead of ramping up clockspeed, cores instead.

RE: Does it hurt?
By balokbok on 5/11/2007 3:31:04 AM , Rating: 2
RE: Does it hurt?
By James Holden on 5/11/2007 3:32:38 AM , Rating: 2
Holy linkfarm batman. Is that a real site?

RE: Does it hurt?
By James Holden on 5/11/2007 3:33:48 AM , Rating: 2
Oh wait,
according to

RE: Does it hurt?
By CyborgTMT on 5/11/2007 3:54:54 AM , Rating: 2
Just wait, pretty soon we'll be getting Rydermark scores from an 8 core FX system with quad-crossfire'd HD2900's from Fudo.

Fuad and the truth are like oil and water.

General Performance Comparison
By MikeyJ79 on 5/10/2007 11:03:36 PM , Rating: 2
AMD demonstrated the power of this Phenom FX system by encoding a 1080p movie trailer into H.264 in near-realtime.

How significant is this achievement? How close are we already to accomplishing this with today's processors? The fact that it's encoding a hi-res trailer into H.264 in near-realtime sounds fairly impressive in itself, but how does this compare to what we already have?

RE: General Performance Comparison
By HotdogIT on 5/11/2007 11:50:43 AM , Rating: 2
You know, I've wondered that too. It keeps getting tossed around as if it's a big accomplishment (and it may well be), but I have nothing to compare it to. How fast, for example, does a QX6800 do it now? Or an Intel V8 system? Or an X2 3600+? It's obviously a hint at performance, but come on, throw me a bone here :D!

RE: General Performance Comparison
By Amiga500 on 5/11/2007 12:27:35 PM , Rating: 2
Does DT (or anyone here) not have access to any Intel quad systems at the moment?

Just fire any movie through to give an idea of what a (very rough) comparable time to encode is.

By TechLuster on 5/10/2007 11:17:13 PM , Rating: 2
dubbed quad-core Wolfdale and dual-core Yorkfield

Wolfdale's the dual-core part, and Yorkfield's the quad. This is the second time I've seen this mistake on DailyTech, so is it perhaps not a mistake at all, in which case everyone else is wrong?

And by the way, I sincerely hope that when Rivet says, "You'll see that in the Christmas line-up," he's simply giving a conservative upper bound on the launch date.

RE: Typo
By KristopherKubicki on 5/10/2007 11:40:38 PM , Rating: 2
You are correct, I've fixed the typo.

RE: Typo
By rtrski on 5/12/2007 10:40:19 PM , Rating: 2
And by the way, I sincerely hope that when Rivet says, "You'll see that in the Christmas line-up," he's simply giving a conservative upper bound on the launch date.

I'm hoping he means we'll see Quads in the Xmas lineup of OEM machines from HP, Dell, Compaq, and the like...

...which by necessity means the processors were available at least a quarter before in some sort of volume for engineering, and might be available to the retail channel in the same timeframe as well.

Barcelona Debut Speed
By Dactyl on 5/10/2007 11:46:56 PM , Rating: 2
previous company guidance indicated Agena FX will debut at 2.6 GHz clock frequencies.

Yes, but they had a re-spin and were able to get it up to 2.9 GHz

RE: Barcelona Debut Speed
By James Holden on 5/10/2007 11:49:36 PM , Rating: 2
As I've said about the inquirer before, it's pretty easy to claim just about anything when you've predicted every possible outcome.

I'll personally trust the DT numbers until I see something that says otherwise from a reputable source (DT, Tech Report, Anandtech)

RE: Barcelona Debut Speed
By Shintai on 5/11/2007 1:20:06 AM , Rating: 2
Theinq also said the R600 would lunch many times between novemember and now. Along with being 1Ghz etc core part. We all know how that went...

And theinq is still the only one claiming it. And if it was a IMC issue. Im sure AMD would have made a clock difference to the rest of the core long ago.

Video encoding 1080p or 720p
By liquidaim on 5/11/2007 5:08:26 PM , Rating: 2

A question for the writers of the article:

Was the video encoding at 1080p or 720p? I ask because Tomshardware's news site reported it as 720p.

Do you have dual socket, quad core woodcrests sitting around in your labs? Why not attempt to replicate what AMD was showing as closely as possible to get a comparison. (If only for Sh*ts and Giggles).

Thank you

RE: Video encoding 1080p or 720p
By Zurtex on 5/11/2007 9:28:43 PM , Rating: 2
I've been trying to find out that one since the start, one website said 1080p, tomshardware said 720p.

After a bit of asking around I found out it didn't really matter. Without knowing what program was doing it, whether or not SSE4a was being used in the encoding and if the 2900XT was helping out or not, it's all up in the air and you can't run any comparative tests at all. Especially which program was going it, apparently some are very fast but encode crappy.

So while 720p / 1080p make a big difference when real benchmarks come out. All they did today was show that it could run fine and it was able to run an intensive application alright.

AMD just sucks these days...
By Blood1 on 5/10/2007 11:05:58 PM , Rating: 2
I'm an AMD fan. I have my current rig with my AMD 2500+ running with an 1950 Pro. I'm happy with it but I've been waiting to upgrade to see what AMD is going to bring to the table.
Seriously it's been forever since AMD released anything! WTF! F-in release the dam ATI 2xxxx series already and get a new f-in CPU out there that competes with the C2D already...
Sorry just had to rant b/c I've been, we've all been wait since Nov for the 2xxx series video cards and something to put Intel in it's place... (:-/

Core Clocks Freqs Irrelevant?
By teckytech9 on 5/11/2007 1:21:01 AM , Rating: 2
Frankly, I'm ashamed at AMD for postponing this product launch till x-mas, or later?
Give me a break, is this a marketing ploy?

Is this do or die for AMD to base its product revenue in its next generation chips till the end-of-year? Need a socket change too?

I'll be blunt, a true 4x4 is in the making, and I'm glad AMD is still standing-up
for future rounds with the big gorilla. Without AMD will have the big gorilla monopoly,
with no future price-cuts in the works.

Indeed the pc architecture is archaic if I may say. Hyper that, PCI-express this,
DDRX that, etc... Its the bandwidth in the drams where the bottleneck always is.
Give me DDR4 and 1066mhz X 4 then will talk about system responsiveness.
Quad core is app specific, not too many apps can take advantage of it today.
Big wow, I can see all webcams being converted to 1080P.

People always say
By Treckin on 5/11/07, Rating: 0
RE: People always say
By Accord99 on 5/11/2007 5:00:30 AM , Rating: 2
Overclocked that thing at 14x220fsb using a phasechange cooler. with the serial ATA and ddr running at 220, the thing SMOKEDED every p4 I ever came across. How much waas that Proc? $98 on newegg. The P4 EE was the only thing I ever saw that posted higher PC marks than I, and it was like $1200. That will be AMD return to glory. They need to offer great product at scary low priced.

The scary low prices are precisely why AMD suffered massive losses then and why they are suffering massive losses now.

Predictions vs. Performance
By Emryse on 5/11/2007 1:53:32 AM , Rating: 2
I want to see a healthy competition just as much as the next serious technology enthusiast...

What's must frustrating about the current situation of events is that I don't want to hear about what they're going to be rolling out - I want to see what they actually bring to the market.

Intel has had a much better track record of producing results; once AMD begins to bring their products out on time or at least not behind the competitive curve, then I'll be interested to hear what they're planning down the road.

Such as it is, I'm not impressed, and they've got a long ways to go... Oh, one more thing. I'm tired of all the "just wait until K10 because that's going to make all the stars in the galaxy align..." responses or the "AMD is better value for comparitive specs" mantra that I'm seeing every time AMD vs. Intel comes up.

This is not a question of value - that's for the consumer to decide (and they're deciding Intel, by majority). What is of concern is the advancement of technology; right now, Intel is leading that advancement. When I'm concerned about what processor to buy, then I consider the performance over cost factor. When I'm concerned about what company to invest in as a stockholder, then I'm concerned about that company's ability to deliver new and innovative performance-driven technology to the market ahead of the rest, and in a timely manner.

So before you go flaunting AMD or Intel as your "most favorite ever bestest biggest", go ahead and take the time to identify yourself as a technology enthusiast, investor, or die-hard preferential consumer... that way I can avoid wasting my time to read the rest of your post should you be the latter.

Future Xeon's.
By coldpower27 on 5/11/2007 12:33:48 PM , Rating: 2
"The server counterparts to these chips, dubbed dual-core Hapertown and quad-core Yorkfield will suceed the existing Clovertown and Woodcrest Xeon components."

This isn't correct, Harpertown is the successor to Clovertown, with Wolfdale-DP being the successor to Woodcrest.

Harpertown is a Quad Core meant for DP operation.

By alanvn on 5/12/2007 12:57:20 AM , Rating: 2
I usually upgrade my desktop system every other generation, its good to see AMD actually show working hardware rather than just slide shows. Another option to look out for and wait for before making the purchase / upgrade decision. Competition is good for the consumers.

By crystal clear on 5/12/2007 5:41:15 AM , Rating: 1
mark AMD's first demonstration of an 8-core desktop

The official launch date for Agena is still undetermined.

Yes ! 8 core desktop sounds good from the marketing point of view.
But is there a market for 8-core desktops ? (2007/08)
Do we really need it? Do we really have enough software to exploit these cores?
AMD should ask itself a simple question?

Does the market(user) really need it this year (07)?

What are their(user) current needs/requirement?

What are their(user/buyers) budgets for upgrading(approx)

Game addicts are a small percentage of the whole market.

Most can do very well with duo & quad far a starter !

AMD should follow the stratergy of -
ALL MARKET DEVELOPED(AMD) CPUs-Right now its DUO for 07 with
Quad in for (08+).

"Quad-core, codenamed Barcelona will launch later this summer, in the July, August kind of time frame -- followed by [Agena FX] on the desktop," stated Robert Rivet, AMD executive vice president and CFO.

Yes sounds fine-But
1)You handover the whole LAPTOP market to Intel as a "gift"
2)The growth potential for laptops is huge ! users are replacing the desktops with laptops.Ultimately the buyer decides the succes or failure of a product..
3)Does this time frame(July/Aug) give the OEMs/ODMs/ & most important of all "the motherboard manufacturers" enough time to get their products into the market by christmas.
4)Is the manufacturing - able/geared to satisfy market demand-ensuring "NO SHORTAGES"
AMD's Rivet makes one promise; we'll see quad-core desktop components in 2007. "That'll launch a little after the server part, it'll be called Agena. You'll see that in the Christmas line-up," Rivet claimed.

This time around NO MISTAKES-they got to get it right.
The whole package-prices/availability/quality & the rest.
Your last christmas was a disaster-I certainly hope this one will be a success.
We the buyers need both AMD/Intel around to benefit from quality & price.

Thats why I titled this comment-


By crystal clear on 5/12/2007 5:45:07 AM , Rating: 1
"Nowadays you can buy a CPU cheaper than the CPU fan." -- Unnamed AMD executive

Courtesy-Daily Tech.

Server Penryn Chips
By AskAboutComputers on 5/14/2007 1:20:24 AM , Rating: 2
The server counterparts to these chips [desktop Penryn], dubbed dual-core Hapertown and quad-core Yorkfield

The dual-core is Wolfdale DP--not to be confused with the desktop chip of the same codename. The quad-core server chip is Harpertown.


If you read it-now see it.
By crystal clear on 5/12/2007 11:56:51 PM , Rating: 1
If its good to read the article(DT) its equally good to see those boards.Here they are to see-

Slideshow: AMD shows off new chipsets and boards

"My sex life is pretty good" -- Steve Jobs' random musings during the 2010 D8 conference
Related Articles

Copyright 2016 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki