• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Xeon64 or Opteron

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
LTC8K6
For the millions time Its under NDA. No company in their right mind allows benchmarks of a BETA Application on a BETA OS.

I said it about 2-3times faster than my p4 and 4 times faster than my athlon xp. What more do you want.
 
"There is an OS for Itanium there isnt a relased one for x86 64 from Microsoft"

has he nto been told about the 64-bit windows and longhorn supporting 64 bit ?


Hands up Nemesis and others if you'v got the 64-bit windows OS for Itanium..

Yeah thought so ....
 
It's okay Nemesis. I just didn't realize you were practicing a comedy routine. Now that I know, I have gotten out the popcorn to await your next round of hilarity. 😀 :beer:
 
If you read the source files on the x86 64bit windows OS disk its funny because it reference the Itanium all over the place.

Anyone with a copy of Windows 64 can easily verify this.
 
AMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMDAMD..............


Sorry got a little carried away there.

me ? fan boy ? get out of town

"yes hector , duel core is the way"
 
clarkey01
I just hate to see AMD crumble to Intel for not pushing the envelope. For years AMD had to adopt Intel technology with MMX SSE etc. 3Dnow never took off. With AMD being ahead of Intel causes Intel to drop prices because AMD cant compete with Cheap CPU's. AMD still makes all thier current money from memory not from CPU's. I want AMD to realize they need to compete with Intel but not lose sight of staying profitable. Intel has Money that AMD does not. If AMD can create a technology that Intel will have to liscense then AMD can survive with some cash flow from Intel. We still win. But if Intel comes through and wipes out AMD because AMD doesnt Innovate then chip prices stay the same, Technology stays the same, and eventually the prices increase. We Lose.

I like to see what others think. Some take it gracefully and some not so. I understand if you dont agree with me just tell me why. I am open minded. Much like there are things in Linux I really enjoy and wish windows had and there are things in Windows I wish linux had. Nothing is perfect and there is no right or wrong answer.

I am guessing by the penguin your a linux person. Whats your opinion from a linux standpoint in all of this? Do you favor any paticular chip? What would you like to see from Linux in all this chip war? I see you like AMD over Intel. Why not share why you like AMD despite the hell some Intel people will tear you apart from.

To me a forum is for sharing knowledge, experience, and general thought. Its not for trying to tear others to threads because they dont agree with you. Forums are not anger mangement.
 
Nemesis you keep claiming that intel is out innovating AMD. How? HT was already invented. To go faster they has to mask problems with the architecture so they added more cache and used your plan of making it more functional with SSE3. The truth is that Netburst was designed for speed above anything else. They can't really change that now without starting over. When netburst was conceived, everything was about speed. Now that isn't so. Amd has proven it time and again. Also you say that Amd will have problems when it hits the 4-5 GHz ceiling of intel. Why? They use completely different mfg. processes. Soon amd will be doing strained silicon on insulator. What happens then, We might see the Athlon 64 FX-65 without any further modicfications.
 
Originally posted by: Ardrid
I registered just to address Nemisis' post. You need to do some serious research before you start spouting off comments. First off, HT (HyperTransport) is used for more than just SMP setups. HT is effectively the I/O link of the entire system. It's what the processor uses to talk to the chipset, buses, etc. However, it truly shines when it's talking to another processor because each processor has its own bandwidth. That means total system bandwidth increases as you add more processors, unlike the Xeon and Itanium. I encourage you to read up.

Secondly, Intel did not, I repeat, did not invent HT (Hyper-Threading). SMT, as it's rightfully called, has been in existence and used long before Intel decided to "create" it. You might also be surprised to know that 3DNow was in existence before SSE. That being said, it was Intel who stole AMD's idea, much like they're doing with AMD64. And in case you didn't know, 3DNow was expanded to include SSE2 with SSE3 to follow. Including separate instruction sets would be pointless because there's little chance of them being adopted. And the business world could care less about the marketing names AMD/Intel give their technology. You think ppl are complaining about how ridiculous NetBurst is?

Expansion is innovation in this case. AMD took an old architecture and extended it, while maintaining complete compatibility. They didn't try to revolutionize the entire ISA and shove it down the industry's throat (*cough* IA-64 *cough*). Rather than being foolish, they did what was logical. Let's not forget the on-die memory controller as well, arguably one of the greatest features brought to the desktop arena. I've already touched on the multimedia instructions, so I won't do that again. Moreover, if you've been paying attention to the video card scene recently, you'd realize that both NVIDIA and ATI are developing methods to completely take the stress of encoding/decoding off the CPU and place it on the GPU. That being said, such instructions are unnecessary. As to the Opteron 939, it doesn't exist. Extra features will come when AMD debuts their 90nm processors.

And finally, AMD "missing the boat". Not happening because Intel already missed it. Look how long it took for them to swallow their pride. Now, add to that the fact that ppl in the industry are saying Nocona's 64-bit performance sucks. It's not too hard to see why AMD is the leader right now. Every Tier 1 OEM is behind AMD save Dell, and you can count on them joining sooner as opposed to later. And vid card manufacturers aren't doing anything like what you suggested. Almost every feature they create is an open standard (3Dc, MXM, Axiom, etc.). The only thing that's manufacturer specific is the architecture.

I'll end with this. AMD has been the undisputed leader for the past year and they look to continue that trend for the next 18 months or so. They've forced Intel to adopt AMD64, a model number system, and completely revamp their roadmap, effectively killing the P4 and NetBurst. Goliath may not be down but he's damn sure reeling.
Click to expand...


Welcome to the forums. Excellent post laying out AMD's superior think tank for at least 3 years running. Please post more often because it's nice to see people who talk facts rather than supposition, hyprboyle and are long winded with nothing to say like nemisis.
 
or better yet, Nemesis what does intel bring to the table with Nacona that will make it better than anything on the market.?
 
Ok, there are a number of misconceptions floating around in this thread, the first of which is that having deep pipelines makes the Xeon faster than the Opteron. That couldn't be farther from the truth as you'll see here and here. Having a deep pipeline is what hurts the Xeon, even more so now that it's using the Nocona core. I'm sure you've seen how bad Prescott performs in a number of situations. If there's one thing server apps don't like, it's deep pipelines and pipeline stalls. Secondly, what exactly do you do for a living that would allow you to beta test Nocona? I find that hard to believe unless you're associated with a publication or hardware site. Until I see it mentioned elsewhere, I'm sticking with what ppl are currently saying in the industry: Nocona's 64-bit performance "sucks".

I'm glad you think HT is an innovation. I'm sure you'll be pleased to know that AMD holds the patent for that technology. So, who's innovating again? 🙂 As to the dual-core issue, don't expect anything more than that. AMD may surprise you, but given the size of a dual-core Opteron, adding features that won't be used is wasteful. You would be correct in assuming that there will be things that the Xeon is good at and the Opteron is good at. It's always been like and will most likely continue to stay like, at least until the 64-bit transition takes place, as we've seen a number of huge boosts in the encoding realm from moving to a 64-bit environment. And AMD is thinking beyond x86. When they made AMD64 they threw out a lot of unnecessary stuff while extending the architecture. What a lot of ppl don't seem to realize is that x86 is here to stay. There have been numerous attempts to dethrone it and it's still going strong. The only way x86 would ever disappear is if an ISA was capable of doing near-perfect emulation. Chances are that's not going to happen.

Now, you speak of delays. How many times was the Itanium delayed? What about Intel's 90nm process and the "fastest ramp in product history"? How about Prescott, Nocona, and Dothan? Everyone has delays, it's a fact of life. To single out AMD as being the only one is asinine. You also need to take manufacturing into account. AMD has one fab, that's it. Intel, conversely, has around 10 or so devoted to just processors. In addition to this, both companies, one more so than the other, can't just ramp up clock speeds whenever they feel like it.

Btw, they can't die shrink it any further, or rather, they can, but they won't anytime soon. It's the die shrink that caused all the problems in the first place. I place the blame partially on SSi and partially on Intel being too stubborn to use SOI. By the time Intel's 65nm process comes online, the P4 will be dead and the PM will be the flagship.

Now, on to the Itanium. It's not going to run at 3GHz anytime soon. If you've noticed, Intel has pretty much been capped at 1.5GHz or so with that processor. Even the third generation version, Madison, is only getting a 200MHz bump. Intel continues to do nothing but throw large amounts of cache on the chip. The only real change that Madison is receiving is that it will be dual-core. Otherwise, it's just the same old formula and the Itanium is big enough as it is. Do you realize how much heat it puts out and you expect Intel to pull 2-3GHz with it? Not happening, at least not until they do a serious die shrink. Itanium may have been a threat had it actually gathered momentum but it's as good as dead in the water. Ppl are quickly abandoning it now that they see Opteron is more than a capable solution. And that's supposed to threaten the Opteron? It should also be mentioned that Itanium is not the only processor that can link 128 processors. Opteron has the same capability. I need only mention Red Storm and other clusters to indicate that. Itanium is a multibillion dollar failure despite what proponents and Intel will tell you. Intel has barely managed to sell any of them. Hell, the Opteron has already out sold in and is continuing to gain momentum. And you know what the final nail in the coffin is? That it's been said that MS will be dropping IA-64 support completely. Now, why would they do that? Because there's only room for one 64-bit architecture and Tier 1 OEMs have chosen it.

And, finally, AMD isn't going anywhere. They hit the bottom so many times I've lost count. And each time they've come back, with this most recent one being the strongest. AMD isn't the only one adopting the competition's technology, as I've mentioned elsewhere. Intel has adopted AMD64, what will arguably be the ISA for years to come. It should be said that despite this, AMD, nor Intel for that matter, will ever receive money from one another because of their licensing agreement. Despite what's been said year in and year out, AMD is here to stay.
 
I wish that were true bobbyk.

SOI has less current leakage than strained silicon however there is a electron tunnelling effect where the electrons can change state. I read about it a few week back. Trying to find the article might have been the Inq. Not sure if AMD will have this problem but I suspect they will when they start playing in the area Intel is having trouble at. But it looks like Intel may be close to a work around. All AMD has said is they arent having the problems with soi.

I expect the die shrink to get AMD up to P4 Speed 3.4-3.6ghz but then I expect some of those unusual problems speed problems to start appearing.

Its easy to say AMD has an 18month lead. But Did anyone see the Intel Celeron D coming? Anyone expect that? What makes anyone so sure Intel isnt coming out with anything else?

Everyone has thier thought on Xeon64 and minds seems to be made up on its performance. From what?

I see a lot of back talking me but I dont see anyone with technical reasons why except Ardrid.

Ardrid was the only person to outline any good information.
 
bobbyk I was unware there are SSE3 Instructions in the 939 opteron. Is this true?

Simply why cant AMD come up with thier own instructions instead of waiting to see what Intel adds?
 
I never said that amd is using SSE3. I was commenting about Intel. They use SSE3. Widespread adoption of the technology, however useful it may be, is not there at the present time. SSE2 seems to be able to handle everything that graphics cards can't.

I also strongly believe that increasing the pipeline makes things slower. That is why intel masked the performance hit with the cache and added functionality. If there was no performance hit, intel wouldn't have increased the cache because then more dies can be made from each wafer thus increasing company profits.

You point out the Celeron D came out of nowhere. That is almost true. THe only problem is currently they are comparing it to Amd offerings. Sempron might not be a huge hit but it will do better than the athlon xp. It will again look like celeron is crap. They compared next gen to current gen technology.

Until intel does integrated memory controller with xeon, opteron will probably kill it. If intel doesn't release a new chipset for servers then the new xeons are dead anyway. The support for ddr 400 will be non existent like it is now. They will be stuck with slow memory. Memory is one of the biggest factors affecting performance. Who cares about clock speed, not intel anymore even. Look at their new numbering scheme and the P-m that should answer that point.
 
Opteron WILL be using SSE3 (probably the 90 nano SOI parts), that's the right thing to do, software companies stated that they will use SSE3. The same thing happened with AMD64. Microsoft and a bunch of other companies support it, so Intel HAD to join in (we'll soon see if their solution is faster or not).
Hmm... inovations.
How about a integrated FPU (Floating Point Unit), integrated L2 cache, 3DNow! (SSE is 3DNow! on stereoids, 3DNow! is older and Intel just improved it, but becouse Intel had much more market share they could make it INCOMPATIBLE and forced software vendors to use SSE - it's not about technology and inovation, it's about marketing.

And Nemesis2038 -> if you were a real expert, then you SHOULD know that the longer the pipeline, the slower the processor and ALL improvements (better HyperTreathing, better branch prediciton, more L1, more L2) could not make up for the deeper pipeline (Prescott's integer pipeline - 31 stages , Opteron's pipeline - 12 stages, Northwood's pipeline - 20 stages !!!).
 
Maybe. I am not sure what route Amd will choose there. I do not know if it will make the next core revision. It would be good, but not neccessary.Gpu's today are extremely powerful, more so than cpu's. They can handle most of the multimedia things that SSE3 addresses. However, marketing will play and amd might be forced.
 
Are you sure that this is talking about the revision after CG. C0 Was the first one. Now the processors are made on Cg. I was wondering about the next one. If this is dated last september, that was when the processors came out. That might be talking about CG instead of C0.

obviously i didn't watch the whole movie at this time. It is rather long to do in one sitting looking for a specific part.

If it is about the next one, i am not sure what it is called, then Amd is way ahead as far as innovation goes. They are just bringing it around now, that would leave a 9 month time where they can work on the one after. That would be good, but it wouldn't allow them to follow the market trends and work on the problems or suggestions that people suggest.
 
According to Microsoft about a month ago, performance on AMD's 64-bit implementation was faster in 64-bit code than intel's. Does this mean that they had one of these new Xeons in their possession? I'll leave that for you to decide.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top