Intel "Harpertown" Xeon vs. AMD "Barcelona" Opteron
by Jason Clark & Ross Whitehead on September 18, 2007 5:00 PM EST- Posted in
- IT Computing
What's new with the Harpertown Xeon
Although Harpertown represents a "tick" (or minor update according to Intel's nomenclature), a lot has changed. Harpertown not only includes a variety of micro-architecture changes, but it also is also based on a 45nm manufacturing process. You'll notice that most of the tweaks that Intel has introduced focus on keeping the processor from going out to main memory. Below is a list of the main highlights of what's new:
45nm
The new Xeon is a 45nm part, which lowers power consumption, reduces die size (and increases transistor count), and helps Intel reach higher clock speeds. Harpertown will top off at 3.2GHz at launch, but higher clock speeds are rumored to follow. For an in-depth look at Intel's 45nm process, read the following article.
1600MHz FSB
With the new Stoakley platform, the Front Side Bus (FSB) now tops out at 1600MHz. This bus increase should help Intel fight off their bus speed bottleneck woes awhile longer until QuickPath (Intel's on-die memory controller) makes its debut.
12MB L2
Each set of two cores has a total of 6MB of L2, which brings the total L2 cache up to 12MB. Again, this will allow Intel to stay out of memory as long as possible which should increase performance.
New SSE4 Instructions
Harpertown includes Intel Streaming SIMD Extensions 4 (SSE4) instructions, the largest unique instruction set addition since the original SSE Instruction Set Architecture (ISA).
High-K Process Technology
In order to extend Moore's Law, Intel uses a new material in their transistors which is a combination of high-k gate dielectrics and metal gates. This new technology increases the switching speed of the transistors and helps reduce power consumption to allow Intel to continue to deliver faster processors that consume less power.
Although Harpertown represents a "tick" (or minor update according to Intel's nomenclature), a lot has changed. Harpertown not only includes a variety of micro-architecture changes, but it also is also based on a 45nm manufacturing process. You'll notice that most of the tweaks that Intel has introduced focus on keeping the processor from going out to main memory. Below is a list of the main highlights of what's new:
45nm
The new Xeon is a 45nm part, which lowers power consumption, reduces die size (and increases transistor count), and helps Intel reach higher clock speeds. Harpertown will top off at 3.2GHz at launch, but higher clock speeds are rumored to follow. For an in-depth look at Intel's 45nm process, read the following article.
1600MHz FSB
With the new Stoakley platform, the Front Side Bus (FSB) now tops out at 1600MHz. This bus increase should help Intel fight off their bus speed bottleneck woes awhile longer until QuickPath (Intel's on-die memory controller) makes its debut.
12MB L2
Each set of two cores has a total of 6MB of L2, which brings the total L2 cache up to 12MB. Again, this will allow Intel to stay out of memory as long as possible which should increase performance.
New SSE4 Instructions
Harpertown includes Intel Streaming SIMD Extensions 4 (SSE4) instructions, the largest unique instruction set addition since the original SSE Instruction Set Architecture (ISA).
High-K Process Technology
In order to extend Moore's Law, Intel uses a new material in their transistors which is a combination of high-k gate dielectrics and metal gates. This new technology increases the switching speed of the transistors and helps reduce power consumption to allow Intel to continue to deliver faster processors that consume less power.
Facebook
Twitter
RSS
77 Comments
View All Comments
Justin Case - Tuesday, September 18, 2007 - link
Anyone else feel that the first image...http://images.anandtech.com/reviews/it/2007/barcel...">http://images.anandtech.com/reviews/it/2007/barcel...
...looks somewhat... er... phallic?
TA152H - Tuesday, September 18, 2007 - link
Oh my, you're absolutely right.That's really foul. Even the area between the Tick and Tock looks like the urethra. It's so wrong. Is that really the only way they could have presented the information? I mean, if they wanted to get pornographic, couldn't they have used a woman's breasts? Right one for Tick, left one for Tock? It's much more attractive than this.
Regs - Tuesday, September 18, 2007 - link
Marketing geniuses. Intel at its best. A better product, with a bigger...In all do seriousness, It's no surprise AMD can't compete with an architecture that's been out for over a year. AMD needs more tweaks and needs more clock speed. I just hope they don't disappoint again like they did with the K8. 4-5 years of stagnation.
TA152H - Wednesday, September 19, 2007 - link
I think it comes down to Intel being wiser than AMD. They were always smarter, as evidenced by their much more advanced processors like the P7 and Itanium. But AMD was wiser, and chose an easier path that also performed better. Intel had all the great technology, super-advanced trail blazing stuff that just didn't work that well. AMD made the same mistake by going native quad-core before they were ready. Consequently, they have a poor performing part compared to what Intel has, today, and promises for tomorrow. Obviously, the extent of their failure isn't as deep-rooted as the Pentium 4 was and at least the Barcelona can be improved (mainly by clock speed) more quickly, but the big problem is that the Barcelona is getting raped by Intel processors using FB-DIMMS. You add clock speed to the Barcelona, and the power goes up (everything else being equal). You change FB-DIMMS out, and you get better performance and lower power. So, the future doesn't look that bright for AMD, despite the fact they should gain clock speed pretty quickly. It's unlikely to help their power/performance much. Intel using more appropriate memory will to a great extent. Also, if AMD does manage to get close to Intel in performance, Intel will just release a higher performing part. They can hit much higher than 3.2 with their G0 stepping, so it's really a matter of whether it makes marketing sense.But, it sure sounds good to have native quad-core, and they sure were smart to do it. Right? Just like Intel was to come out with trace-cache, double-pumped ALUs, and super-pipelining and unheard of clock speeds.
But all that aside, if they can get the clock speeds up to a reasonable amount, and increase the size of the pathetic caches (yes, I know they are limited by the IMC and it limits it, but still 512K????) and in a release or two get full memory disambiguation, they will have a really good product. It will at least be competitive.
Justin Case - Tuesday, September 18, 2007 - link
Any reason why the AMD system had 16 GB of RAM (8x2GB) while the Intel system had only 8GB (4x2GB)?Also, any reason for the big differences in cooling (AMD system had 7 fans, Intel system had 3)? If the Barcelona system actually uses <i>less power</i>, as your numbers show, surely it can't dissipate <i>more</i> heat.
When you're measuring the power consumption of the whole system (and extrapolating that to the power efficiency of each CPU), you should try to make the configurations match as closely as possible, no? Not to mention that the amount of RAM can have an influence on the actual system performance.
I could understand different configurations if you were testing systems at a specific price point (and couldn't "afford" more RAM for the Intel system due to the more expensive CPUs, for example), but that wasn't the case here.
Xspringe - Wednesday, September 19, 2007 - link
I would really like to see updated benchmark scores as well! It only seems fair to add more ram to the xeon, for it might improve the benchmark scores and would also increase energy usage (which would be beneficial to the barcelona).Final Hamlet - Wednesday, September 19, 2007 - link
Yuk!I really would like to see an explanation from an editor on this critique...
Justin Case - Wednesday, September 19, 2007 - link
Add the unusual choice of benchmark and fact that Harpertown isn't actually due to be launched until November, and I think this is one (more) article we can file under the "iNandtel" section.Speaking of that, anyone know what happened to GamePC's "Labs" section? Along with the Tech Report they were probably one of the last sites with a steady output of meaningful, objective reviews of PC hardware.
JarredWalton - Wednesday, September 19, 2007 - link
IMPORTANT UPDATE INFORMATION:There was a typo/error in the original config. We apologize for the confusion - I should have verified with Jason/Ross earlier. The Opteron setup was running 8x1GB, not 8x2GB. Sorry to pop all the conspiracy theories (again), but the systems are a lot more similar than you would apparently like to believe.
Note also the update at the end: 2.5GHz Barcelona is on its way and will be tested shortly. We'll see how that compares with the higher clocked Harpertown.
Proteusza - Thursday, September 20, 2007 - link
With the last Quad Core Comes to Play article, and now this, I've completely lost faith in Anandtech's benchmarks.These guys are too clever for them to make a mistake like that, and if they did I'm sure they would see the mistake and rebenchmark.
No, I think these benchmarks were just paid for by Intel, in anticipation of its November launch to steal AMD's thunder. I'm not accusing the entire site of constant bias towards Intel, but rather a bias towards advertising. AMD has probably done the same thing in the past, and I'm sure Anandtech has been happy to oblige.