AMD has unveiled its first set of quad-core processors, three months after its original launch date. This ‘complicated’ design that resulted in the delay and puts the chip maker a full generation behind its archrival in terms of chip manufacturing processes.
…in the game, at least for now.
Reading about Barcelona in the tech news, you’d swear that it’s a dud or something. ZDNet Asia has an article discussing the Barcelona benchmark results compared to Intel’s quad core offering and It ain’t looking as good as AMD had tauted. Still, early day’s yet as AMD is sure to bump up the clock speed on these beauties.
It remains to be seen if AMD can use this architecture and run with it, hopefully building up an advantage and regaining some market share. Sure, Intel got there first but Intel’s current quad core architecture is a bit of a hack, as it’s only really two dual core processors ‘stuck together’. Sure, Intel is developing a solution for both this issue and the lack of an onboard memory controller, but AMD have already resolved these issue.
In the end, I’m glad AMD chose a elegant design. We have all seen elegant fail compared to inelegant hacks, as inelegance tends to be cheaper. On the other hand, it seems that by choosing elegance, AMD are pushing Intel onto the same path.
Here is to competition!
I like the “elegant” design as well however, I’ll take the “hack” if it turns out to be the better performer. We’ll see how things go.
I haven’t looked at any bench marks yet, but I wonder if AMD pulls ahead as you scale up with additional processors?
I’m kinda thinking the same thing. Alone the speed of minimizing the distance between the cores should be a factor, but you never know how these things pan out.
Intel has decided to go for the integrated design as well and if they can release a working integrated four core chip in the time frame that they have stated, AMD will be one full generation behind them, something we haven’t seen for a long time.
But if Intel starts having the same problems as AMD had with Barcelona, we will probably see a good eight core processor release by AMD first.
No matter how it turns out though, a sixteen core x86-64 is only about three to five years away. I wonder how software houses are going to deal with the resulting processor based licensing nightmare?
It’s all really exiting at the mo!
You can take the hack now, if it works with four cores.
But what happens when we move to 8? Does the hack scale better? Does the elegant design scale better?
Ars Technica has an interesting look at the 4-socket, 4-core offerings from Intel and AMD http://arstechnica.com/news.ars/post/20070905-intel-launches-tigert…
For setups that are mostly CPU-bound with FP ops, the Intel setup rocks.
For setups that are mostly memory-bound, the AMD setup rocks.
The graphics in the Ars article are the most telling. Until Intel comes out with CSI, AMD has it in the memory bandwidth department, and is gaining in the CPU processing department.
Single socket CPU setups, Intel rocks.
Dual-socket CPU setups, it’s mainly Intel.
Quad-socket CPU setups, AMD rocks.
a little OT, but from TFA:
If only this attitude were actually prevalent in the modern IT world and the market was allowed to play out as such
It is true in practice. AMD had more market share with their new technology not that long ago. But the Core stuff from Intel took the lead back. Enthusiasts don’t have much brand loyalty, they’ll buy whatever is best!
How true!
I used to be a big ChipZilla fan until it was time to upgrade. Before hand, I ran Athlon and AlthonXP processors and was very happy with them. I decided I wanted to go 64bit so I started to look into all options. At the time, Intel’s Core2Duo range where the best chips on the market as far as performance, power efficiency and price where concerned. So I bought an E6600 and haven’t looked back.
The fact remains that if AMD manages to beat Intel come time of my next major platform upgrade, I’m going to jump right back on the AMD wagon. No matter how much brand loyalty ChipZilla generated, or Intel generate, I’m still gonna spend money on the best solution come the time.
To do anything else would be just plain silly.
At the time, Intel’s Core2Duo range where the best chips on the market as far as performance, power efficiency and price where concerned.
Erm… the E6600 is more expensive than an X2 5000+, and uses more power?
At the time it was more expensive than the X2 5000+.
At the time, you would have compared E6600 to 5000+, E6600 using 33w/hr average, 5000+ using an average of 26w/hr for the same work. The 5000+ even has a lower “max power draw” than the E6600, with 5000+ sitting at 63w and E6600 sitting at 66w. Am I the only one here who knows how to do math?
average 33>26
max 66>63
min 15.78>7.73
It gets even worse if the computer just sits there all day, E6600 would use 379w just sitting there doing nothing, while the 5000+ would use only 186w.
Even that aside, the X2 5000+ is practically equal in performance to the E6600 for most tasks, and better at molecular research and other science related tasks. Given that the “lower performing” 5000+ beats or comes close to matching the E6600 in everything other than video games, I’m not sure claiming that “the E6600 wins in performance” is a proper statement to make?
Certainly the E6600 excels in games, but most CPUs being produced aren’t stuck into gaming rigs, they are put into computers used to do other things… things that, frankly, AMD is still better at, even with their old design.
Other than “I don’t know how to read power dissipation tables, and only bother to read game-related benchmarks” and “I want to whore out my bad purchase as an awesome idea” what was your point in suggesting the E6600 was the best at the time?
I’m sorry my friend, but you did not buy “the best CPU at the time” if those three things were your factors in deciding a purchase. You bought the CPU that the market hype told you to buy, and nothing more. Even as a gamer, the 5000+ won (at the time) in two of the three categories, hands down… not to mention that, as of right now, it is 1/2 the price.
Hell, I could build a dual CPU opteron 270 box for only a little more than a single E6600, and less than an E6700. I’d end up with 4 cores and still have better power usage since most of the cores would be close to idle all day.
“If only this attitude were actually prevalent in the modern IT world and the market was allowed to play out as such”
What do you mean? There are atleast 5 versions of Vista isn’t it? 7 maybe?
One of the great advantages of AMD K10/Barcelona is ease of upgrade, you just need to switch CPUs and all the rest stays the same.
… a little video from AMD here:
http://youtube.com/watch?v=SYvupXNmaKQ
This is my sticking point with the AMD vs Intel battle. With Intel, they are good about changing sockets everytime the release a new chip. AMD tends to be a lot easier on the pockets when it comes down to motherboards. You can often get quite a few level processors on a board before it has to be sent out to pastor. Not so bad now, as back in the P3-P4 days where you had to pull out the manual to to figure out which procs would work on your board.I wonder how many sockets/slot type both sides have racked up over the years?
That’s what I believed too when they release the socket 939… and now we have socket AM2 and soon socket AM3…
…still, it’s better them Intel’s way of new sockets all the time, but still…
This ‘complicated’ design that resulted in the delay and puts the chip maker a full generation behind its archrival in terms of chip manufacturing processes.
Drop the that and the while adding an its and the sentence becomes complete.
This ‘complicated’ design resulted in its delay and puts the chip maker a full generation behind its archrival in terms of chip manufacturing processes.
Quad-Core AMD OpteronTM processor launch celebration
Monday, September 10, 2007 6:30 p.m. PT
http://phx.corporate-ir.net/phoenix.zhtml?p=irol-eventDetails&c=740…
I’m excited to see amd is back. be afraid intel.
Intel won’t be afraid, they’ll just fall back on the old MS > Intel > Dell love triangle and ignore AMD.
That triangle is no longer a reality I’m afraid. That’s one reason why Intel rushed out they’re quad core design in the first place.
While Intel was touting the Itanic as the only 64bit solution you would ever need, AMD brought out the x86-64bit extensions. What happened? Everybody, including Dell and MS, jumped on board the AMD train.
Now Dell, IBM and Sun sell Opteron based servers, it seems like MS is scaling down support for the Itanium and Intel has been rushing out x86-64 chip designs to try and regain some of the mind and market share they lost to AMD.
Sure, AMD are not in a position to take first place from Intel. But they certainly have taken away much of what used to be a rock solid Intel market.
Sure,
Microsoft really didn’t jumped on AMD64, they seemed to pretty much ignore it until Intel went in and started using the instructions aswell. Sucks for AMD and pretty bad by Microsoft but it’s what happened.
I hate when people say “omg AMD are so dead now, Intel rules!”, wtf, Intel just have the lead for a short while, it’s not that big of a deal. Also AMD have been in way worse positions earlier considering the started from nothing, didn’t they? So sure it might not be as good of a position as it was 1-2 years ago but it’s still better than what they started with. This is not the end of the world as we know it.
That’s not really true. Although MS didn’t really put out a product until there was quite a bit of hardware out, Intel’s original plan after seeing AMD64 was to create their own alternative like they did with SSE vs 3DNow. MS basically told them that they were only going to support 1 64-bit extension to x86, and that since they had already started on AMD64 that Intel could either use that or go without support from MS. Intel caved.
Edited 2007-09-11 04:34
Intel won’t be afraid, they’ll just fall back on the old MS > Intel > Dell love triangle and ignore AMD.
Those days are over: dell now offers AMD based computers not only with Windows but also with Linux installed. Intel should be afraid becasuse they will not be able to become a total monopolist
where they can jack up the prices as high as they want,not with AMD around, they will make sure of that.
If you read Anandtechs take on the new Opteron it scales better than the Xeon as the core count goes up. Anand points out that NUMA is required to get most performance out of the Barcelona architecture.
“Be aware though that you need the Enterprise edition of Windows 2003 to see this kind of performance. The 32-bit Windows 2003 standard does not support NUMA and the bandwidth hungry AMD quad-core does not like that at all. Performance was up to 14% (!) lower, showing only 73 fps instead of 85 fps.”
The new Opteron platform is a winner as far as the server market is concerned. It uses less power than the equivalent Xeon platform while having more performance. But that is mostly due to Intel’s choice of using power hungry FB-DIMMS instead of cheaper and power saving DDR2.
Unfortunately for AMD though Intel’s Q6600 @ $270 is hard to beat on the desktop platform. The Q6600 easily overclocks on air (with the intel provided hsf) to 3.2GHz simply blowing the Barcelona architecture out of the water at current speeds.
I would like to see benchmarks from a neutral compiler (gcc?) compiled code just to be sure about who is the leader now on clock-by-clock base.
As a developer, I can say that any of them is pretty good for my needs and probably will buy my next machine with a huge electrical cost on mind as my main machine stays on most of the time.
If energy efficiency is your thing why not go with mobile components?
http://www.heise.de/newsticker/meldung/95680
It’s going to take AMD a lot longer than a year to catch up to Intel in terms of performance. AMD sat on, basically, the same old design for years adding nothing but what, essentially, amounted to compatibility “features”. They milked their relative performance throughput and power for too long without really do anything new. (Tying GPU functions to the CPU doesn’t interest me in the slightest, especially if AMD’s big deal is to go as far as tightly coupling their processors to their new ATI subsidiary GPUs…)
I really don’t see AMD’s current design crew turning out anything matching, let alone, surpassing, current Intel designs within a year, and probably even longer. Now all AMD has going for it is power use efficiency, but from power consumption benchmarks those are only relevant if you have a server with a great deal of dead time, which is something I certainly wouldn’t want to have unless it was some type of server for a small/medium sized office, as at full throughput the two companies’ chips are close enough in terms of power utilization, and Intel’s throughput efficiency added to that gives them a clear win in a business setting. As to my home machines, I game, and do development and other things at home, so I want the throughput giving another loss to AMD.
Bottom line, even though I haven’t really looked at it, I’d hazard that AMD’s only real hope for decent sales ATM would be to push their mobile processors.
Beyond that I’m still waiting for Intel’s new bus design to show up, which should give them a few more percentage points in overall performance, that is unless I missed it’s release somewhere in the last year and it turned out to really be a dud.
@tyrione:
Attention grammar police, you missed something yourself.
Edited 2007-09-11 16:29
I suppose that instead to focus merely on the marketing department, they were spending a lot in the R&D department, which not prooved any or little efficiency face to Intel’s offering. Up to now yet ! The HyperTransport stuff is fantastic… on the paper, but Intel prooved that some hacks can achieve good performance. Like demomakers do superb working with the software, Intel’s engineers have made a wonderful job with the Core2 Duo/Quad processor, you cannot miss it. Not only the overall performance is good, the power consumtion is pretty low (compared to what we’d used to know with the Pentium-D serie) and the price tag is very affordable.
However you’ve to see all of this on a larger scale. Intel choose to react fast on the market, but have to ‘hack’ each next processor to maintain its steps foreward. On the other side, AMD took its time improving the HyperTransport, tweaking things thightly, using the Opteron to check things on a real-life incarnation of their thoughts, selling less ‘advanced’ K8 chips up to now just to make enough revenue to ensure being able to sustain the R&D for their next generation chips.
And I bet they’re reaching the final step and are close to be ready to lauch something that’ll blast everybody’s mind, like they did before with the Athlon serie of processors ! But now they heard what everybody was expecting from them and their chips, so I guess they’re taking their time to make things worthy and reliable, to define a new level of confidence in their product line…
Plus add the whole chipmaker experience from ATI, about memory bus driving (HyperMemory, doesn’t it sounds familiar ? HyperTransport, HyperMemory) and parallel shaders ALU logic, so you can bet that the new chips will be better integrated, etched with a finer grain, consume less (Cool’n Quiet experience), so…