Intel double cheeseburger quad is a dog
Read Tomshardware benchmark of the Kentsfield double conroe. Look at the multithreaded games (Quake 4, Call of Duty 2, Oblivion). In Quake 4, a Kentsfield overclocked to 3.33GHZ is only 14% faster than Athlon 64 X2 FX62 (2.8GHZ). For some reason, Toms didn't produce a Call of Duty 2 result. If you look at non-overclocked result, the 2.66GHZ Kentsfield is only 9% faster than FX62.In iTunes test, the FX62 took 106 seconds, the quad core 2.66GHZ Kentsfield took 109 seconds. The Intel quad is 3% slower than AMD dual.
posted by Sharikou, Ph. D. @ 9/10/2006 02:39:00 PM
60 Comments:
Of all the benchmarks, you only mention the GPU bottlenecked gaming benchmarks and iTunes:)Way to skew it in your favor without showing the Kentsfield's strength:)
"Compared with the already not-too-shabby Intel Core 2 Duo/Extreme, the Core 2 Quadro can give performance a mighty tweak - but only for specific applications. In the best-case scenario, performance can even be doubled. However, this depends on the particular program. Software makers have yet to initiate the needed optimizations for multiple physical CPU units. The table below lists a hodgepodge of applications that benefit from four cores right away.
The future belongs to HD content. If we take our benchmarks into consideration you can no longer get by without a quad-core processor. Test results with the software packages Main Concept with H.264 encoding and the WMV-HD conversion make this very clear. We noticed performance jumps of up to 80% when compared to the Core 2 Duo at the same clock speed (2.66 GHz). A Core 2 Quadro at 2.66 GHz and higher is the answer for HD video (editing and rendering) at full HD resolution (1920x1080)."
Kentsfield is a bit overkill, but if you do HD, have the cash, there is nothing else that can compare.
I find it hard to keep up with Intel's names. What is the difference between Kentsfield and Cloverton?
Ehm, smartguy. Quake 4 doesn't take advantage of four cores. In fact, no game on the shelf these day do so. Quake 4, COD and so on only support DUAL-core. not Quad.
So it's the same wheter you have two dual-core opterons, 4x4, two dual-core Woodcrests or a Kentsfield. The games can only use 2 cores. Capice?
Looking at the other scores, they're quite impressive.
Take a look how many cores those tests use and everything becomes clear.
Forgot to clarify the errors.
"Two problems cropped up during the tests. The first was that the game Call of Duty 2 (version 1.03) would not launch. The program crashed with an MP_Error (multicore):
The second was that the Microsoft WMA Encoder is so slow that execution takes ages. Both of these problems should soon be solved with a core allocation. This is not only the case with Intel; it would likewise occur with any AMD system with more than two CPUs."
So the reason why some things didn't work was that the software was buggy.
Kentsfield have 8 mb cache and consist of 582 million transistors.
How much does it cost to produce this?
It cant be cheap
Thats for sure:P
Finally real benchmarks. Proof of exactly what I said. Only a 10% difference in performance at 200mhz differences. Too bad the quad core is useless because software limits its power. With 2 dual core opterons scaled. This would put quadro at only a 10% difference. Or worse when AMD quad cores come out on 2 sockets. 4 cores can not beat 8 unless your trying for a 100% OC, if software could use the 4 cores. Such as I said on MMM's bio.
And because the quadro can't, its power is useless. But a opteron can use 2 cores in a single cpu means even tho it has 2 cpus on one mobo. Its better then a quadro unless the quadro can use 2 cores for the OS and the other 2 for software that can use the 2 other cores. Then its power would be well managed. Such as opterons on dual sockets usaly are managed in this way to perform better. Because each cpu can do different if assined tasks such as the dual core cpu in socket 1 does the OS wile socket 2 does the games. The software doesn't become a limiting factor. But its impossible to utallize quad core for a single app at this time.
Good title btw. lol Its really true. Quadro is not native. AMD's is true quad core because its on one die. I can't understand why Intel glewed 2 conroes together. Just like the mistake of making pentium D's. Doesn't make any sense. So its no different then conroe. Not new arcs. For example a native AMD dual core was better then a pentium D because one main reason it was native. Pentium D's are only 2 pentium 4's on a single cpu board. AMD's K8's on dual cores are slightly different arcs then single cores.
This 2 dies on one cpu board can make some performance limitations. But toms hardware deff proves there is only about a 200mhz clock speed difference at best. Say anything else is only wishful thinking, fan talk, or simply ilogical BS.
Way to show that high IQ Sharikou...
"For some reason, Toms didn't produce a Call of Duty 2 result."
Did you read the article or just go straigt to the benchmarks like you said you do and draw your own conclusions?
Anonymous said...
"...AMD's is true quad core because its on one die. I can't understand why Intel glewed 2 conroes together..."
Well 4 Cores is 4 Cores, native or not.
As far as glued together, well yields would probably be a big factor considering 8MB of cache.
The yields of K8L will not likely be very good initially due to the complexity of the design.
Like I said not benchmarks done using multiple aplications running at full load! I wonder how can you market a processor saying it's great if yo want to do alot of thing at the same time but never benchmark how well can it acualy do it.
Isn't supposed to be a multi core's strong part, the ability to run multiple aplications at the same time with greater eficiency? Nontheless nobody does these kinds of tests!
And with the release of Kensfield this year it shows that only Intel of all is capable of shooting itself in the foot "twice"!
Even if better that Core 2 Duo kensfields will have a miserable price/perf ration and again sky high power consumption. So back to square one for Intel it seems. The "Netburt" lesson it seems was all to quickly forgoten!
"Of all the benchmarks, you only mention the GPU bottlenecked gaming benchmarks and iTunes:)Way to skew it in your favor without showing the Kentsfield's strength:)
"
These are the most important ones for an enthusiast coincidentaly!
"Of all the benchmarks, you only mention the GPU bottlenecked gaming benchmarks"
lol this was funny and retarded...
"Ehm, smartguy. Quake 4 doesn't take advantage of four cores. In fact, no game on the shelf these day do so. Quake 4, COD and so on only support DUAL-core. not Quad."
lol this was also funny and retarded...
"Even if better that Core 2 Duo kensfields will have a miserable price/perf ration and again sky high power consumption. So back to square one for Intel it seems. The "Netburt" lesson it seems was all to quickly forgoten!"
Performance is basically same as X6800, but vastly improved for threaded applications like video encoding. And it comes out at the same price as X6800. So what would I rather have...:)
Basing performance on games eh?:)Do you expect massive gains on 4x4 for gaming? You forgot to mention the huge gains in threaded apps, the purpose of more cores, like the H.234 encoding test or the HD rendering test.
If it makes the AMD fanboys feel better, yes the kentsfield psuedo-quad core has a lot of room for improvement. The bad news is that the need for improvement is in the power consumption only. Intel should be able to ride this product for quite awhile. This seems to be far from a final production chip, although the cpu-z gives a B1 stepping.
Does anyone have any idea what the consumption is going to be for 4X4?
Then of course the answer to everyone's question of who is buying the excess Intel legacy stock. The answer is people like me. I picked up a reman dell 2.53 celeron with an 80GB 7200 rpm drive cdrw/dvd drive for 208 shipped.
Price it out: Any chip, MB, ram, case, drive, Power supply, and oh yes the operating system. Anyone got a better deal than that.
That comp is not for me, its for my parents that surf the internet, email, IM and that's it.
"What is the difference between Kentsfield and Cloverton?"
It's Clovertown. Clovertown is the server(Xeon) version of the quadcore Kentsfield. So it'll support at least 2 sockets for at least 8 cores total.
http://www.xtremesystems.org/forums/showthread.php?t=115384
http://www.xtremesystems.org/forums/showthread.php?t=111251
Early look there.
That's a cute burger.t
Sharikou you forgot to mention one thing / disclaimer :
Please ignore all other benchmarks on the test (especially ones that use multi-threading like encoding). - these tests are biased as Tom's is a paid Intel pumper. Of course the game data is all real and Tom's is not a paid Intel pumper for gaming benchmarks only (and any other benchmarks which show an FX near to the Core2/Kentsfield performance)
I wonder how the 4x4 will do on a Quake 4 only benchmark - I think it was 80% better, no? (or maybe there was some disclaimer of "up to 80%" and the only on certain benchmarks - like heavy mutlitasking?) I can't remember with my selective, AMD rose colored glasses memory! I'm sure it must mean 80% on everything....
Pop Catalin Sever said...
"...if yo want to do alot of thing at the same time but never benchmark how well can it acualy do it."
Its called multi tasking and it would be extremely rare to use all 4 cores at 100%.
"And with the release of Kensfield this year it shows that only Intel of all is capable of shooting itself in the foot "twice"!"
Its an extreme chip, how are they shooting themselves in the foot, if anything they wanted something to go against 4x4 from AMD, and to not have a whole in there product list.
"Even if better that Core 2 Duo kensfields will have a miserable price/perf ration and again sky high power consumption. So back to square one for Intel it seems. The "Netburt" lesson it seems was all to quickly forgoten!"
WOW!!! Those are simply uninformed comments.
Price/performance will be bad...lmao, read the article again and look at the benchmarks.
$1000 for 4 cores and benchmarks that rival the dual opteron scores from link.
/quote Toms...
Running at full load, the Core 2 Quadro is at the same level with the old Pentium EE 965./quote
Four (4) cores vs two (2), I do not see Netburst all over again.
"These are the most important ones for an enthusiast coincidentaly!"
Not every enthusiast is a gamer, did you think of that?
And if all you do is game then even your beloved 4x4 is going to disapoint you, as will K8L.
Did you see that?!
Overclocked to 3.33 GHz Kentsfield is not much faster than 3.0GHz. It means that The Core is not scalable anymore! 10% faster clock doesn't mean 10% faster execution!
To produce faster CPU, Intel must extend the cache memory to another megabytes. And I think Dirk Mayer knew that...
Why do I read differently from many people here? I see Kentsfield about the same as Conroe at the same clockrate. In the end, Kentsfield is really just a dual Conroe packed in one MCM, with poor scalability. What's so great about it?
Not even the "multitasking" tests can save its face. Kentsfield is ~10% or less faster than a same-clocked Conroe in those tests which would supposedly show the quad-core advantage best.
What about even heavier and more demanding multitasking, you ask? Well, I bet the reason Toms didn't do those is that if you go heavier multitasking, memory bandwidth & latency kicks in, and large cache becomes less effective, and you'll make Core 2 duo/quad look bad.
The core of Core 2 is a decent performer, compared to K8, we all knew that. The Kentsfield OTOH is really not very impressive (except its power usage and perhaps price). Now spare us the fanboys talks and find us some good reasons that, based on Toms benchmarks, you Intel cheerleaders think otherwise?
For most desktop use currently both Kentsfield and AMD 4x4 will just be using using electricity to produce more heat.
In the server market obviously different, here AMD will be more competitive
Would AMD 2 chips on a socket 4x4 be equivalant to the Intel "2-chip glued together" quad-core?
-Longan-
Ehm, smartguy. Quake 4 doesn't take advantage of four cores. In fact, no game on the shelf these day do so. Quake 4, COD and so on only support DUAL-core. not Quad.
Not suppording/using four cores is no reason for the application stop working. Pentium EE 8xx and EE 9xx as been around for some time with 4 threads...
Lazy software developers or lazy microprocessor tech?
"Too bad the quad core is useless because software limits its power."
BS stupid ass. You are useless with your insight to this discussion. Operating systems scale way beyond 4 core and it's pretty damn important piece of software I would say. As to the Tomshardware's benchmarks - you clearly ignored 3DStudio Max, DivX/XVid/H.264 encoding results. Coincidence ? I think NOT!
"This would put quadro at only a 10% difference. Or worse when AMD quad cores come out on 2 sockets."
So Intel 4 core useless, while AMD 4x4 (dual socket expensive solution) will beat quadro ? What is this sh!# heh ?
"Good title btw. lol Its really true. Quadro is not native. AMD's is true quad core because its on one die."
AMD doesn't have 4 core cpu (to be precise Intel doesn't "have" it either, but it will in a month).
"I can't understand why Intel glewed 2 conroes together."
Becouse it's cheaper that way. The comromise is certainly reasonable, considering that you will get two "native" dual core in this cpu. So it's not entirely "gluless". I certainly wouldn't mind AMD selling ->now<- 4 core cpu's in one socket (2 dual core). The price would probably start at about 300$, and would be available today.
Finally - it doesn't matter if it's glueless, glued or whatever as long as it's faster (in respective application - your spreadsheet won't benefit it any time soon) and can compete in performance/price ratio or power consumption. You don't need it ? Who cares ? Million will gladly put their fingers on such gems drooling all over.
OH! - and I _am_ AMD fanboy, but not as clueless, blindless and reckless retard such as yourself.
The Benchamrk didnt's show heavy multithreading with gaming (I think Anand once did it in respect to dual core) - that would probably do wonders to show what 4core's are meant for.
Stupid paid pumper Tom has published another biased Intel benchmark.
Really, he chose all the stupidest benchmarks to show that Intel won, and ignored any important benchmark that AMD won.
I bet that by the time Kentsfield comes out, it will get its ass whipped by even Semprons.
As for 4x4, the competition between 4x4 and Kentsfield is really going to be funny. Even the 4x4 with two EE 3600+s is going to beat the hell out of the OCed Kentsfields.
All Intel suporters are retards. Just like the people who work at Intel.
Then of course the answer to everyone's question of who is buying the excess Intel legacy stock. The answer is people like me. I picked up a reman dell 2.53 celeron with an 80GB 7200 rpm drive cdrw/dvd drive for 208 shipped.
For 208 USD? Man you were robbed! Crappy specs and with a celery on top! what a waste of money.
Sharikou, I have a couple of questions. What is different in K8L from K8 ? If it was something radically different that it boosts performance like 80-90% why was not it named K9 ?
The basic issue with the tests -- and the state of the OS/apps/games world today -- is that for the most part, there is no scaling past two threads.
So dual core gets the job done. Even on the Mac, tests of Core 2 Duo iMac vs. the Mac Pro show the same thing. Having more cores doesn't help you with most of today's apps.
It's nothing new for the hardware to be ahead of the software. This is the way it has been for a long long time. Perhaps this in a way defines the Microsoft-defined PC. The software always lags.
What is interesting is that high-end consumer machines will very shortly all have 4 cores. This will be good for the industry whether those cores are Intel's "Quadro" or AMD's "4x4".
And high-end developer/professional machines will soon have 8 cores, as soon as Intel ships the Xeon version of the "Core 2 Quadro" with AMD to follow a bit later.
So we see that we are in the time period where capital expenditures should be limited until 4-core/8-core systems come out.
Even if the apps don't take advantage of your hardware today, you want to get a real speedup when they do. Which should be soon after Vista and Leopard ship.
You don't want to buy a machine today and then in 6 months see the same machine with double the number of processor cores. That will give you tech buyer trauma like nothing else.
Hence the smart buyer is waiting for the now imminent delivery of quad core computing to the consumer market and octa-core computing to the developer/pro market. Bring it on!
Stupid paid pumper Tom has published another biased Intel benchmark
Alright guys ..did some one investigate why Tom's hardware totally bans posts with say something negative about Intel..
This cannot continue ..it will be reduced to a website visited by a few morons!!
The HP xw9400 workstation is tested with next generation Opteron 2200se processors running at 2.8GHz as well as with one Conroe XE. The rest of the subsystem is top notch with SAS Drive and Quadro video cards. The benchmarks do not lie. AMD is quickly swapping places with Intel. The latter dominates most benchmarks which is outrageous as two Opterons can barely hold kick the ass of the Conroe CPU.
Imagine how much power those 2 opterons use. CPU's by themselves are going to be using 250 watts, twice what the kentsfield is using. Even with that they can barely beat 1 Conroe. Amd better be praying that K8L is some kind of miracle chip.
The basic issue with the tests -- and the state of the OS/apps/games world today -- is that for the most part, there is no scaling past two threads.
GUI hardly scale pass one-thread. The only apps/threads that needs more number of CPUs are those running in background. But you just don't buy processors that cater for today needs, right?
Hence the smart buyer is waiting for the now imminent delivery of quad core computing to the consumer market and octa-core computing to the developer/pro market. Bring it on!
Here is what I am going to do: I am building 2 computers, one with PD 805, and one with C2D. The PD805 is for my parents, and the C2D is for myself. When C2Q becomes more mainstream, I will move my C2D to my parents' computer, will put in a C2Q in my computer, and live happily ever after. Long live LGA775 and P965!!
The Kentsfield OTOH is really not very impressive (except its power usage and perhaps price).
The 4x4 will be WAY worse base on what you said :) later when there is review base on 4x4 (if it ever come out), let's look at how it perform base on the same app... and its power envelope as well as the system price. Do not give me the crap that it can go as low as 1000, if you wish, compare the said system to the kentsfield :)
And when AMD finally come out with its quadcore, just compare it again, using the same test sets :) wanna add heavy memory bandwidth application to the benchmark, go ahead. But remember to look at all sort of applications and take the average (there are really little desktop apps are memory bandwidth bound). Else, what's makes your view different from the so called 'bias' view from tomshardware.
" S said...
Sharikou, I have a couple of questions. What is different in K8L from K8 ? If it was something radically different that it boosts performance like 80-90% why was not it named K9 ?
7:22 AM, September 11, 2006 "
the reason is k9 = DOG ( ka-nine)
it could cause confusion, so they went to k8L
besides the "k9" core would be a k8 derivative with double FP and more cache, also quadcore native.
In my opinion, they should have gone for K9..
and use a core name such as "CERBERUS" or another famous dog,
because they're exhausting the city names. at least the important and recognized ones.
Sharikou, I have a couple of questions. What is different in K8L from K8 ? If it was something radically different that it boosts performance like 80-90% why was not it named K9 ?
K8L is the "code" name for K9, more or less. That name isn't even official yet. Plus K9 sounds like canine. Why would you name something dog. That opens the door for biased reviews from Tom's to say K9 performs like a dog, etc. It's a marketing nightmare name, it's just being renamed K8L or whatever, but it is a new architecture. After K8L (or whatever they call it) AMD will bring K10.
Hey Sharikou:
Intel on Intel:
http://www.viewfromsiliconvalley.com/id258.html
"Sharikou, I have a couple of questions. What is different in K8L from K8 ? If it was something radically different that it boosts performance like 80-90% why was not it named K9 ?"
Because the K9 is a real dog of a processor. Woof, woof!
How will the 4x4 address the lack of performance improvement in single and double threaded applications differently than Kentsfield? Will there be architectural features that allow it to skirt this problem, or will it be bound by the same limitations?
References please... no fairy tales.
Most people aren't even familiar with K7, K8. If K8L will be competitive with Core 2, I think it deserves more than just K8+L. Example, Windows went from 3.1, to 95, to 98, to 2000.
http://www.squarefree.com/2006/08/09/firefox-to-use-fibonacci-version-numbers/
Firefox is going to go from 1, 1.5[1], 2, 3, 5, 8..
Will there be architectural features that allow it to skirt this problem, or will it be bound by the same limitations?
4x4 will benefit from reverse hyperthreading. It is a very simple concept. Go back and read some of the older posts on this blog.
Intel is heading to bankruptcy sharikou -- 09/11/06
o.o
The omen is coming>.<
Pointer said: "The 4x4 will be WAY worse base on what you said :) later when there is review base on 4x4 (if it ever come out), let's look at how it perform base on the same app... and its power envelope as well as the system price."
This is so very a 10-th grade response, pointer. I wasn't even talking about 4x4, and yet your reply to Kentsfield's poor scalability is "others will be the same"?! Besides, 4x4 will scale better than Kentsfield (see below). So based on Toms benchmark numbers, there's simply no reason to cheer on Kentsfield's performance, because it doesn't perform/scale well. PERIOD. Toms cheering on Kentsfield purely reflects its pumper role.
For the same benchmarks, 4x4 will likely behave about the same. It'll scale slightly better due to direct connect between two CPUs (IIRC, Kentsfield's two dies have to go FSB to talk to each other), but just slightly. HOWEVER, for real-world heavy threading applications, where memory latency and bandwidth comes into play, 4x4 will scale more efficiently. We all know how Opteron compares favorably against Woodcrest on some real server benchmarks - quad core will only signify this favor more.
"How will the 4x4 address the lack of performance improvement in single and double threaded applications differently than Kentsfield?"
For 4x4, you can at least buy just one CPU with half the money, and think about upgrade later; for Kentsfield, you have no choice.
You might be right, that today's software do not take advantage of four cores very well. Well, you can always run many apps at the same time - very useful/important for servers and workstations, probably not so for desktops and laptops. Or you can buy a dual-core and upgrade later.
If you're the former, that you run a server/workstation, Kentsfield's big cache will be less significant, and 4x4's high bandwidth will likely give you better performance for the money. If you are the latter, that you do not run many apps on the desktop/laptop at the same time, then Kentsfield is pretty useless, based on Toms benchmark results.
Spin the results however you like, but one thing is true: Core 2 does not scale, and Kentsfield does not perform. The quad-core "improvement" (if any) over Conroe seems even less than AMD's socket AM2 over socket 939.
I read the article too... what I got from it was that Intel's first quad core will provide a significant performance boost over the already outstanding Core 2 Duo line.
I wonder why our perspectives of the same article are different? Maybe it's just tough to be unbiased while giving AMD head?
Hey everyone I am hoping you could help explain this...
I have been searching Spec.org looking for FX62 and X6800 scores and have found some interesting numbers and have also come up with a few questions for you...
IBM System X 3455 (AMD Opteron (TM) 2220 SE)
Fujitsu Siemens Computers
PRIMERGY TX300 S3, Intel Xeon 5140 processor, 2.33 GHz
How is it that this could be, I am not trying to argue, but understand that a lower clocked, half the ram setup is beating an AMD Opteron.
Is it the hard drives?
What am I missing?
Also...
Asus M2N32-SLI Deluxe, AMD Athlon (TM) 64 FX-62
Dell
Precision Workstation 390 (Intel Core 2 Duo processor E6700, 2.66 GHz)
This time Intel has twice the ram, but a slower hard drive, again what am I missing?
Please explain.
Thanks
4x4 consume more power then a quadro? Ok so how can 2 35w chips consume more power then 65watts when my single X2 EE 35w consumes only 31w on full load. When quadro consumes 130w and AMD's quad core only consumes 89w. I must be lost here. *Rolls eyes*
"This time Intel has twice the ram, but a slower hard drive, again what am I missing?
Please explain.
Thanks"
Enumae, check my blog -- there's answers there.
"I read the article too... what I got from it was that Intel's first quad core will provide a significant performance boost over the already outstanding Core 2 Duo line."
Then you probably need to improve your scientific/engineering reading skill.
Look at the benchmark results closely. Except HD content creation, Kentsfield perform all benchmarks 10% faster or less than Conroe at the same clockrate.
This is so very a 10-th grade response, pointer. I wasn't even talking about 4x4, and yet your reply to Kentsfield's poor scalability is "others will be the same"?! Besides, 4x4 will scale better than Kentsfield (see below). So based on Toms benchmark numbers, there's simply no reason to cheer on Kentsfield's performance, because it doesn't perform/scale well. PERIOD. Toms cheering on Kentsfield purely reflects its pumper role.
save your nineth grade comment and let's see when it comes out. I did not only talk about the 4x4 only, I talked about the AMD quad core too. Basically what i wanna point out is that Intel kentsfield (no matter how you wanna pitch on its scalability), WILL out perform AMD's 4x4 or quadcore, on average, taking into account for all the applications. Since Sharikou picked the limited test set and if you didi agree on his choice, go ahead and look at the same tests when 4x4 and AMD quadcore is out and I'll laugh :).
Edward said...
Look at the benchmark results closely. Except HD content creation, Kentsfield perform all benchmarks 10% faster or less than Conroe at the same clockrate.
as what i have said in my previous reply, if you want to limit yourself to those no yet heavily multithread apps test, please remember to look at the SAME test set for AMD's coming quadcore or 4x4. Do not point your finger to the benchmark site later and say unfair, intel pumper etc. :). Here is my prediction, the outcome is 10.00001% faster and overall performace, performace /$, perf / watt lose out to intel's :)
Jesus Pointer, you REALLY don't know what you're talking about, do you?
Jesus Pointer, you REALLY don't know what you're talking about, do you?
which one that you refer to? :)
i just read over my post again, and i admit i have the tendency of leaving word out when typing. so, i re-iterate and 'enhance' the potential confusion here:
Here is my prediction, the outcome is 10.00001% faster and overall performace, performace /$, perf / watt lose out to intel's :)
The AMD 4x4 or quadcore, under the same test sets chosen, will be faster than it's AMD dual core counter part by 10+%. It is the test nature, not the 4 core nature. I do believe it will have substatial gain in heavy threaded program, but so do the kentsfield. And then put everything into average, the kentsfield system (or any other better chip that intel has by the time of AMD release), will still out perform AMD's 4x4 or quadcore offering in term of performance, perf/$, perf / watt.
Pointer: "The AMD 4x4 or quadcore, under the same test sets chosen, will be faster than it's AMD dual core counter part by 10+%. It is the test nature, not the 4 core nature."
[Pops a Cig in his mouth, sits back, gets ready....]
What?....
Alright, let me analyse this one mo-gen. Running Single-Threaded tests, a Quad-Core AMD64 CPU will be 10% faster than a Dual-Core? Why? Because of Majik?
Maybe you're talking about Multi-Threaded...so you think adding 2 cores only gives them a 10% increase? Either way, your post makes about as much sense as the 9/11 commisions report.
Pointer: "And then put everything into average, the kentsfield system (or any other better chip that intel has by the time of AMD release), will still out perform AMD's 4x4 or quadcore offering in term of performance, perf/$, perf / watt."
[Puts another Cig in mouth, sits back again, gets ready...]
Uhm...what?
You think that Kentsfield can beat any Quad-Core by AMD? Why? Cuz they're special? Please, stop talking. There's nothing else to say, just stop talking.
I saw on some tests at XS, the Kentsfield had shown a scaling of up to 91.5% especially on multihreaded 3D rendering test such as CineBench. Processor speedup 3.66, divide by 4 and multiply 100% = 91.5%.
That's better than Opteron's 86% scaling.
Alright, let me analyse this one mo-gen. Running Single-Threaded tests, a Quad-Core AMD64 CPU will be 10% faster than a Dual-Core? Why? Because of Majik?
Maybe you're talking about Multi-Threaded...so you think adding 2 cores only gives them a 10% increase? Either way, your post makes about as much sense as the 9/11 commisions report.
I refer to the same tests as chosen by Sharikou as those tests are either not threaded well or GPU bound. You question this because i said about AMD quadcore, but why not you question that when Sharikou talk about Kentfield on the same test set in Tomshardware? :) Because of AMD Majik? :)
Uhm...what?
You think that Kentsfield can beat any Quad-Core by AMD? Why? Cuz they're special? Please, stop talking. There's nothing else to say, just stop talking.
yes, that's what i think. Kentsfield (or its successor at the time of the AMD release its quadcore) will definitely beat AMD quadcore in desktop arena. Not because it is special, just because it is better. The desktop space basically is 1P base. The better scalability doesn;t apply here. if you argue that Intel kentsfield will have to share the 1066/1333 MHz bus, so do the AMD has to share the single IMC. If you have nothing solid to talk about this and just simply thrashing by words wtihout facts, just shut up.
I have a question...Phd Pretender
Intel cheeseburger is out..
Where is the AMD processor? Is it on the grill yet, at the butcher, fattening on corn feed?
I heard a rumor it is still born calf and they have a new TO coming.
You are right that Kentsfield doesnt look that strong at apps optimized for DUAL core or even single threaded.
I do have a feeling that a processor with 2x cores just CANNOT be faster if the software doesnt provide enough threads, dont ya think?
You might consider having a look at the Moltitasking Benchmarks but I know you wont, it will make your world fall apart.
Just a little tidbit I heard the other day.
Seems a couple of techs at a retailer were installing a service pack from a popular OS manufacturer on an E6600 with not more than 2 GB of memory.
They usually allot half an hour for anything P4/K8 based. C2D did it in 6 minutes - they were absolutely astounded.
Also, the Turion X2 does indeed represent the "glued together" approach that y'all disdain so much.
But let's dissect this 'glue' thing. Each pair of cores of the Kent do share cache and can swap instructions posted there rather than going to main memory to do so a la PDP8/9XX, so you get a partial benefit. A dual socket has none of that benefit.
Looking at all the gaming benchmarks that C2D wallops anything AMD, I think if you play games, get C2D. If you encode media for a living, you get Kent and get a mere 3.5x the performance of anything AMD.
It would appear that everyone has missed the point here. 32 bit performance is limited to 1024x768.
A gaming resolution 10 years old!!!
I play F.E.A.R. in 1600x1200 Resolution as do most all AMD gamers do. 1024x768 only pats the Intel 32 bit fan boys boys on the back. You know, an ataboy on the back side. Typically Intel only has half the frame rate at this resolution. It's an insult to my intellect for Tom to Bull Shit me like this.
One low resolution gaming resolution to compare? You’ve got to be kidding!!!!!
you sir, have no clue.
benching in 1600x1200 is gpu benching.
benching in 1024x768 is cpu benching.
And sure, i want to bench a gpu in a cpu bench scenario....
But i agree that, noone plays at that resolution, but it shows what the cpu is capable of, without beeing limited by the gpu. For myself i would like to see test @ 640x480 to eliminate the gpu totally
I said it above:
Turion X2 is a glued-together solution AND NO ONE REFUTED ME AT THIS SITE. It must be true, as I know it to be true.
Selective comparisons are a hallmark of this site. When Intel glues together two processing cores it's bad, when AMD does it, it fails to get a mention...
So partially shared caches is a good thing in Kent, but it looks like the 'native' quad core will come out mid 2007. and AMD will still have nothing to compete.
"Selective comparisons are a hallmark of this site. When Intel glues together two processing cores it's bad, when AMD does it, it fails to get a mention..."
Except that AMD didn't glue anything to make Turion64 X2. You are just a stupid lunatic whose silly words don't even worth refuting by many's standards.
Oh so you think AMD uses MCM to package a dual-core Turion, too? Yeah it is true that moron like you will support Intel. ;-)
Post a Comment
<< Home