10.09.2006

Intel Kentsfield: General Observations...



Some weeks ago while in the middle of the HP-Voodoo acquisition we weren't able to dedicate resources to testing certain new products. Understandably many things were happening at once, so we couldn't focus too many resources on one area. When Larry Clark at Intel insisted that Kentsfield would "shock us all" we went ahead and built a couple of platforms for fun.

Let's just say Larry wasn't lying. This is a copy of the article I wrote for CPU Magazine for the most recent issue. For those of you who follow the hardware industry I would encourage you to grab a subscription of this magazine.

INTEL KENTSFIELD ROCKS THE HOUSE

Many weeks ago our FedEx driver arrived with a surprise: Intel had shipped us a few Kentsfield CPUs. The minute they arrived we quickly assembled some bench configurations to test the processors capabilities.

Overall the assembly succeeded without a hitch - it was effortless to put the configurations together and get them to post. We installed our standard 650 watt power supply, a Voodoo heatsink, some Corsair memory, and a pair of ATI Radeon X1900XTX in Crossfire. It was only until we started to load test it that we thought there was something wrong with the CPUs.

To give you an idea of what we witnessed you must understand that our techs test hardware using some of the most rigorous methods in the industry. We normally start by softwaring the machine from scratch followed by a custom benchmark suite called “Haiti”. Haiti was designed to bring any and all hardware down to its knees as it continuously executes multiple games and benchmarks with escalating demands. In the case of the Kentsfield platforms, they didn’t seem to have a problem running through the suite over a 24 hour period. This is considered normal behavior for us in our final testing prior to shipping a machine. If our systems do not pass a full run of Haiti we will not ship the machine to a customer.

I then asked the team to kick it up a notch and do whatever they could to sweat the hardware. I knew that if we killed a CPU it wasn’t a big deal as we were working with engineering samples – it’s always important to understand the limitations.
Our lead tech on the project decided to run four instances of Prime 95 in a “torture test” while running Farcry in a looping “max settings” demo. Still no trouble, so he ran four instances of Prime 95 while viewing a DVD with multiple applications sitting idle in the background to eat memory. The system still had 10% of its CPU resources left to spare.

To put this in perspective, usually two instances of Prime will cripple any dual core CPU at any level of performance. Not only did this machine run the through the most intense benchmark without breaking a sweat, it blew away our expectations with the DVD test.

To make things interesting I had the tech remove the fan from the CPU and leave the heatsink on. Under normal tasks everything seemed okay and the temperature hardly increased. This is where I thought for sure something was wrong with the hardware (because most systems would have crashed by now), but further tests proved that the CPU was running at full throttle the entire time. After about thirty minutes of tinkering the system finally tapped out. To ensure that the hardware wasn’t to blame for the odd “no fan” behavior we attempted the same thing on the other configurations - All of them yielded more or less the same results.

There is clearly something amazing about this processor - it easily takes place as my most wanted CPU for 2006. Not only does this machine run cooler than the AMD Athlon FX or Core 2 Duo Extreme, but it seems to run more applications at one time without breaking a sweat.

We all know AMD has some strength with multitasking being that they have an integrated memory controller, but Kentsfield left us wanting more. In fact, I would say based on the performance of the platforms we tested that it shouldn’t really matter about the architectural advantages or disadvantages – the proof is in the benchmarks.

Now imagine the possibilities with an advanced liquid cooling system, some high end memory, a top shelf motherboard, and a bit of overclocking. We were all surprised to see the flexibility of Kentfield – granted these were all engineering samples so the production modules may yield different results – at this point I will base our opinion only on what we have seen in our labs.

Needless to say Intel Kentsfield is a killer processor, and we are certain that at most of our high end OMEN machines will ship with it when it becomes available.

Intel has done good, again – It’s time to give the Israeli engineers a raise man!

Now if game developers would take note and start developing their games to take advantage of multiple cores we will probably witness a new evolution in gaming. With the idea of graphics and cpu computations being handled on one chip the possibilities are truly awesome. It's time for the developers to get on board.

60 comments:

Anonymous said...

After all that 4x4 hype..Yet this Kentsfield article makes it sound that good:D
http://voodoopc.blogspot.com/2006/09/amd-creating-new-category-of.html
You say AMD needs 'NUMA' and massive multithreading in the past[as compared to Intel?]..And just stated that Kentsfield should be in most of your OMENs. Kentsfield > 4x4 or 4x4 an even high class?

Darth Solarion said...

They say Alienware will have a few 4x4 machines for demo soon. Do you have anything on that? Thanks.

Anonymous said...

Intel execs lacks the basic IQ. Pat Gelsinger made himself a total retard by typing "I hate AMD" on screen. It's despicable behaviour.

As I pointed out, Conroe fatally wounded Intel and Kentsfield will finish the job. AMD can sell a few hundred K8L now, but it doesn't do it. Why? Because AMD is smart . The few hundred K8Ls will make all other processors worth little. That's what Intel is doing here. With Conroe, AMD basically takes the mainstream market, 95% of Intel's processors are total junk because of Conroe. Now, with Kentsfield, Conroe will be reduced to a $150 chip and Pentium D will be sold at $65, and Athlon X2 will take the sweet spot.

The high end market is tiny. AMD only makes 10000 FX CPUs per quarter. The money is in the middle, and DELL is selling millions of X2 CPUs. I noticed that Alienware is hyping the 4x4 right now. Expect DELL to go 100% AMD by the end of 2008.

AMD is having a major capacity problem right now, once FAB36 finishes ramp by the end of year. Intel will be half dead.

mariuz said...

I think intel learned an lesson or two from open+free source development

"Release early. Release often. And listen to your customers."

product doesn't have to be perfect but workable with it

in my opinion i would rather have an quad core now than one year after but on the other hand an 4x4 would suffice too ;)

I'm just testing some dual core serve and is amazing how smooth this machine runs even if is "just" an athlonX2

I would trade my desktop (and old sempron 754) with any of the new dual/quad cores (minus p4 based ones)

Anonymous said...

Thanks for allowing Sharikou's post to go through, you let the whole world know what an idiot he is XD

Anonymous said...

The crazy doctor overdosing on his own meds is back. Between your cracked up theories Sharikou you are truly living up to the repuation of being the joke of the net. Keep up the FUD buddy you aint fooling no one.

Anonymous said...

The crazy doctor

I guess Pat Gelsinger hates me too for predicting what would happen exactly. But, I am really trying to sae his job. He will be lucky if he has a job in 5 quarters as Intel struggle to survive. I told Intel folks to base their strategic calculations on the assumption of a duopoly. They refused to take hint. I told them they would have to lay off large numner of workers. Look where they are at now. 10K workers already lost their job, and more will be jobless in the coming quarters as FAB36 and Chartered FAB6 cranking out 65nm parts.

Anonymous said...

As I pointed out, Conroe fatally wounded Intel and Kentsfield will finish the job
yes, as everyone knows, excellent performance, price, and popularity will kill any company. you see what happened to AMD? their A64 and X2 lines kicked ass. great performance, great price, and popular with the crowd. what happened to them? they're now bankru... oh, wait... scratch that, need to find a better example...

Anonymous said...

No matter how you slice it Sharikou, Intel is beating out AMD left and right these days. X2 is no good relative to Conroes, nobody in their right mind should pay the same price for an X2 as they would for a Conroe when the Conroes soundly kick the crap out of X2's performance.

Heck, even the FX's are no match for midrange Conroes. At what, half the price? It's shocking that AMD is even selling 10000 FX's a quarter.

Kentsfield is on its way out soon - K8L will be negated long before it's even released.

Mobile CPU's are pretty much all Intel. Or again, any sane person would not get an AMD mobile CPU.

The clearout price of Pentium D's and even the low end Conroes should knock the crap out of Sempron sales as well. We can see budget 3800+'s being pushed and semprons and celerons drifting into oblivion. But hey, for a few bucks more you can pick up a Conroe, so why bother with anything else?

What does AMD have left? Opterons. Unless AMD can skip the whole 65nm process and go straight to 45nm, they haven't got a hope. They're so far behind in such a quick amount of time you have to wonder if AMD can really hang in there. They got their pants pulled down this year. I like the gumption of AMD but it's no longer making products that compete.

Anonymous said...

You didn't predict. I still remember your "prediction" that Intel would be bankrupt by the end of this year. Wow, totally true!
I'd bet a search of your previous posts would reveal your stupidity and and bias, but then again since you edit your posts I'm just gonna stick with what everyone else is saying: you're wrong, Intel is on a roll and no one can deny it, and besides we all know you're just a disgruntled ex-emloyee anyway.

satosphere said...

Good to see a good processor come out of Intel these days.

Any idea on the pricing and when they are coming out? Is there a server line similar to Kentsfield?

Anonymous said...

Methinks Sharikou got his Ph D from the same school as Derek Smart. :)

I for one, can't wait for Kentsfield. Should make for one heck of rendering box. It will be interesting to see how they stack up against a similarily clocked dual processor Woodcrest system.

Anonymous said...

No matter how you slice it Sharikou, Intel is beating out AMD left and right these days. X2 is no good relative to Conroes, nobody in their right mind should pay the same price for an X2 as they would for a Conroe when the Conroes soundly kick the crap out of X2's performance.


You guys are ignoring the reality, sad. Look at DELL, its dumps VIIV and goes AMD Live! Core 2 Duo is killing Intel. Before Core 2 Duo, most people thought Pentium is good, as good as Athlon, but now both Intel and AMD say Pentium is a piece of crap, hyperthreading is crap. So, people realized AMD was telling the truth, GHZ is meaningless. Since Intel was telling lies all the time, people lost trust on Intel, coupled with Conroe's tiny volume, people go AMD, as Michael Dell found out.

ashenman said...

Sharikou does make a point that Intel's executives are of the less intelligent variety. Not saying that sales execs are stupid, but Intel should take a look around and notice the obvious fact that extremely successful companies, both outside and inside the field of technology, are being run by engineers. Engineers are brought up to solve design problems that don't have concrete parameters, but that need concrete results, both of which Intel has historically had a hard time dealing with. Ya, there should be some people at the top who are from marketing, but not almost all of them.

sreeram said...

AMD-Dell relationship has its upsides and downsides. Upsides is AMD would increase its Market Share. Downside is its ASP would go down. Already so many distys are unhappy that AMD has given most of its chips to Dell. Also Dell would consume only low end semprons/athlon/x2 at discount prices.I cant see AMD making money when Dell is selling desktops for $319. Lets wait for AMD Q3 results. Numbers would be ugly for AMD.

Anonymous said...

Intel should take a look around and notice the obvious fact that extremely successful companies, both outside and inside the field of technology,

Intel was started by engineers. Moore was an egineer, he took the CMOS stuff from Fairchild semi and started Intel as a DRAM company. Andy Grove was a semiconductor physicist...

Today's Intel are run by people who aren't very sophisticated. Look at Pat Gelsinger, typing "hate amd". It's a shame.

Surya said...

So Sharikou, how exactly is Intel hurting because of Conroe? Can you explain that to me? You have a PhD and I don't so I am missing something obvious here.

Anonymous said...

Downside is its ASP would go down. Already so many distys are unhappy that AMD has given most of its chips to Dell. Also Dell would consume only low end semprons/athlon/x2 at discount prices.I cant see AMD making money when Dell is selling desktops for $319.

People like to talk about price war, they forgot the most important factor in a price war is cost.

AMD's cost per chip is $50 below Intel. When AMD sells a CPU at $100, it laughs to the bank with $59 gross profit. When Intel sells a chip at $100, it loses $40. Just look at the head count, AMD has 10K people and is taking 25% of the mkt. Intel has 100K people.

AMD will soon cross 33% mkt share and approach 40%, leaving Intel and 55%. At that point, Intel will sustain $2.5 billion losses per quarter and will have to raise prices trying to stay afloat, but AMD is ramping up like crazy. If I were Hector Ruiz, I will sell chips at cost just to BK Intel.

Anonymous said...

So Sharikou, how exactly is Intel hurting because of Conroe? Can you explain that to me? You have a PhD and I don't so I am missing something obvious here.



It's very simple, just try compute Intel's revenue due to the effect of Conroe. On one hand, you have about 10% Conroe chips selling at $200, on the other hand, you have 90% chips selling at an average of $80, most of them selling at $35 (celerons). Your ASP drop to about $90, a massive drop from the $150 Intel used to enjoy. The Pentium used to be a premium brand, now it's crap. Customers lost confidence in Intel, because they now know Intel was lying to them on Netbust and Hyperthreading.


DELL went AMD because of Conroe -- that should be good enough hurt.

You say Conroe will ramp up. However, by the time Conroe reaches 50% of Intel's production. AMD's K8L will frag Intel real good.

Anonymous said...

Intel has done good, again – It’s time to give the Israeli engineers a raise man!

Rahul, of course the Israeli engineers are very brilliant to come up with Core 2, no doubt about that. But by saying the above you are implicitly suggesting the rest of Intel is nothing. Sure the other team may have screwed up somewhat on P4 by not taking power into account (remember that P4 was supposed to scale to 10GHz and never got a chance due to power), but the rest of Intel isn't exactly doing nothing either. Now you are known for taking sly shots at Intel but IMO you should respect good talent anywhere (in Intel Israel, Intel US and AMD), especially when you aren't exactly a brilliant engineer yourself.

And before the rest of you guys start talking about the relative sizes of AMD and Intel remember that Intel has many many more fabs, chipset teams, cpu design teams, compiler groups and does much more for the software ecosystem than AMD ever has done. All this takes people. Sure they may be a little fat right now but its nowhere as much as people like to believe.

Ok, another topic, can you give some more details on your test suite "Haiti"?

Finally, do you have any thoughts on the 45nm Wolfdale and 65nm Antares face off next year? There has been some stuff in the news, do you have any more details or inside news?

Anonymous said...

One harmless comment about giving the Israeli engineers raises set some idiot off. Priceless.

Anonymous said...

I have only one comment.

Ban the Phd Pretender! He has no PhD and adds nothing!

Anonymous said...

Er, Sharikou, I think you're the one who is "ignoring reality" here, in your own words.

You said that Intel would be bankrupt by the end of this year.

Well, it's the "end of this year". Oh wait, it's still October. I suppose 2 more months? Should I give you 2 month's worth of grace period to shut it up?

Anonymous said...

Er, Sharikou, I think you're the one who is "ignoring reality" here, in your own words.

You said that Intel would be bankrupt by the end of this year.

Well, it's the "end of this year". Oh wait, it's still October. I suppose 2 more months? Should I give you 2 month's worth of grace period to shut it up?

Anonymous said...

Dear Mr. Sood:

From your writing, I think you have learned one thing - "Sell the Sizzle." Your combination of solid tech + real-world anecdotes made me absolutely salivate for a Kentsfield. Keep up the entertaining and precise writing.

It's good to know that Intel has unleashed their engineers to do great things again.

Anonymous said...

Actually the PhD pretender does provide laughter therapy to everyone :)

Mike said...

Well, Rahul, as for your last paragraph there, Remedy (I'm sure you know the guys behind the fabulous Max Payne series) is developing Alan Wake (which, again, I'm sure you know). According to an article on a gaming website, the game is designed to take advantage of at *least* four cores in a CPU. One for physics, one for AI, and the other two for things I don't quite remember. And I've heard rumours that Unreal Tournament 2007 will also take advantage of multi-core machines. So at least a couple of developers are on board, and it's good to see.

Anonymous said...

i sincerely think sharikou, Ph.D is an idoit and its obvious with his immature comments.

Anonymous said...

You said that Intel would be bankrupt by the end of this year.


No. I said Intel will BK by 2Q08. This is based on the projection that it will post losses from $1 billion to $3 billion in the next few quarters. Resulting in its inability to meet debt obligations. The end will be massive chpater 11 restructuring. You will see Intel looms around just like SGI.

Anonymous said...

Moderator Please remove the PhD pretender. He has is own little space to spew overhimself. Lets preserve some space for sane minds and mature people who's interest are technology and who's mind processes data and thinks logically.

We're tired of Sharikou's "P"iled "H"igh and "D"eep.

Romantic Planet Vacations Inc. said...

Hey Rahul!
Long time no talk!
I heard about the HP acquisition! Great job! Your incredible success has never surprised me! I hope everyone and all is well. My blog is http://romanticplanet.blogspot.com/
Take Care! Laurie Keith

Anonymous said...

PhD in this case stands for Permanent head Damage :D
Guys, just DO NOT write the whole nick because then Google will rank his pages higher and we don't want that no no no.
*Waiting patiently for Kentsfield*

Anonymous said...

At Friday, October 13, 2006 10:09:07 PM, Romantic Planet Vacations Inc. said...

Hey Rahul!
Long time no talk!
I heard about the HP acquisition! Great job! Your incredible success has never surprised me! I hope everyone and all is well. My blog is http://romanticplanet.blogspot.com/
Take Care! Laurie Keith


Blog spam! Ahh:(

Anonymous said...

Rahul, Romantic Planet is spam..I wish I could report to Google, but they're anti anonymous spam reporters..

dufuss said...

"Intel has done good, again – It’s time to give the Israeli engineers a raise man!"

Yes, Please give Isreaeli Engineers a raise...they rock :)

ashenman said...

I don't think the general populace that matters to either companies market share realizes that the pentium sucked when it was released. They do realize that it's old, because Intel and Dell are telling them that. Dell is also telling everyone that AMD is their "new" product, and that it's "64 bit", "vista ready", and in most cases "dual core." Buzz words=awesome to people who don't know better, and Intel still has a product line that's at least 80% of the pentiums they're telling everyone are obsolete, and that's a liberal estimate.

As odd as some of the conclusions sharikou has come to are, he still stirs up conversation. I only wish people would argue with logic instead of using Sharikou's poor reputation to skip past having to do that. Besides, we have yet to see what happens, when we have the ammunition to show Sharikou's error, then we'll throw that it him instead of childish name-calling.

ashenman said...

Also, what for of stress testing did you perform when you pulled off the fan? I seriously doubt a kentsfield that has been tested to consume about 100 more watts than an x6800 runs cool without one. I'm not doubting you, I just think the results could be better specified.

Anonymous said...

ashenman, THG has been the only one to have gotten 100W over X6800. That would esentially be 175W on the desktop for CPU alone o.o
Dailytech and VR-Zone have gotten 20-40W over.

ashenman said...

Ya, I've seen those articles, and they were retarded. I'm not saying there results don't make sense and that they're lying, I'm saying their results make perfect sense cause they're not even trying to really load the cpus. Daily tech doesn't even say how they test it, but says that they used a 1000 watt power supply, which is retarded, seeing as their load is only 1/5th of that. If you do much power supply efficiency testing, you'll find that the efficiency is very parabolic in nature, hitting its maximum at around 2/3 the maximum load of the power supply. So at the levels of consumption tested, their 1kw power supply would use a much larger amount of excess power with the 6800 system than with the 6700 system, giving very noticeably inaccurate results. Not only that, but they don't even tell you what they do to load the processor, so I'm betting they've got 2-3 cores running dormant when they do their "load" testing. The VR Zone testing method is even more retarded. They only ran one instance of CPUmark06, which is at best only double threaded, and even then, probably not very. I have very little confidence that THG's review of the product will be perfectly in line with the final product's results because of the Idle measurements, but I do think that such a massive reduction in power consumption is impossible.

You may ask, "what's the point in testing a full, extremely strenuous, load on all the cores if that's not likely to happen ever?" Well, you choose a heatsink based on the worst your system will ever have to endure, even if it's not likely to happen. You also make sure that if some poorly programmed application ever ends up looping itself infinitely repeatedly, it doesn't melt your computer because you didn't take the latter into consideration.

ashenman said...

And by the way, that would be 190, because Intel's TDP numbers are bull.

Anonymous said...

I think Voodoo's Kentsfield test is flawed. If you try anythng requires some I/O instead of mere Prime5 looping, you will find Kentsfield spent 50% of its time competing for the FSB. Four cores fighting for a FSB for a breath is not a pretty scene. Essentially,each Kentsfield has less bandwidth than a K7 with 266MHZ bus.

Anonymous said...

Pathetic Humdinger Dimwit says "I think Voodoo's Kentsfield test is flawed. If you try anythng requires some I/O instead of mere Prime5 looping, you will find Kentsfield spent 50% of its time competing for the FSB. Four cores fighting for a FSB for a breath is not a pretty scene. Essentially,each Kentsfield has less bandwidth than a K7 with 266MHZ bus."

sharikou, I sense much AMD (Advanced Mind Degeneration) in you. "Some I/O" also equals updating the video display (video playback, Prime windows, etc), getting data from the IDE/SATA drive (reading DVD for playback).

Pop Catalin Sever said...

Man you couldn't find a better way to skew up test results? than loading super pi threads that actually fit into processor cache and also decoding DVD content which has an input bit rate of 5.7 Mb/s or 9.8 Mb/s and which is also probably(read definitely) accelerated by the video card and also uses DMA memory access for transfer an doesn't tax the CPU, yea...great results!!!

You know what a real stressing test would be?? the test where you use double number of encoding applications or quadruple it the compare the tests with the ones with lower thread count to see real processor scaling. Why? well because in an OS there are typically hundreds or thousands of active threads and if you want to show a processors strength in multi threading then you have to stress it badly not loading each core with one thread!!! that's not a real threading test! it's only the first part of the test and it should be followed by other tests to draw the conclusions.

Why is this kind of test necessary? well to start with applications are beginning to become threaded and once they start their threads all of them will start an number of threads equal to the number of cores or higher so running multiple threaded applications should yield a higher much higher number of threads than the number of cores...get it?

Anonymous said...

Yeah right! Ye same olde cache argument again. Try squeezing the OS, pre-emptive task schedulers, memory resident DLLs, application programs and data into that 8MB cache. Heck! Whatever for we need 256MB to run Windows XP nicely? We don't run MS-DOS nowadays.

BTW, did you see AnandTech's Woodcrest Vs Opteron test on server type applications? Pwned!

Pop Catalin Sever said...

"BTW, did you see AnandTech's Woodcrest Vs Opteron test on server type applications? Pwned!"
You didn't see latest Sun offerings:

http://biz.yahoo.com/prnews/061017/sftu070.html?.v=72

Woodcrest pwnded!!!!!!!

"Additionally, under full load, the 2-socket Sun Fire servers
demonstrate better scalability and higher performance than similarly
sized Intel "Woodcrest"-based servers, with the ability to consolidate
up to 22 legacy servers using VMware."

Better scalability than Woodcrest under full load? doesn't that sound like Conroe has an Achilles heel?

Anonymous said...

SUN Micro:
"Additionally, under full load, the 2-socket Sun Fire servers
demonstrate better scalability and higher performance than similarly
sized Intel "Woodcrest"-based servers, with the ability to consolidate
up to 22 legacy servers using VMware."

Clearly, Woodcrest, while good at SuperPi, can't handle serious server load, its FSB bottleneck is well known, in a PC environment, it's not obvious, but on servers, the FSB bottleneck chokes Intel

Anonymous said...

Ummm, you guys realize that's a PR press release, right?

Anonymous said...

Yups! Its another PR FUD from Sun (and from the FUD doctor himself). Why not check out http://www.spec.org/ ? Where the heck is X4100 M2?

ashenman said...

As bad as referring to a press release for product superiority claims, going to anand is just about as bad. I'm pretty sure most people with an ounce of sense trust him about as much as they trust THG or the government. That being said, Woodcrest and Opteron are neck and neck. Woodcrest has good FP performance, opteron has better memory bandwidth performance. Both can be crucial pieces to a server environment and, depending on the application, could make either one a horrible or amazing choice for a businesses servers.
Ye olde cache argument is very valid in this case, because even though you can't fit the operating system's required components on the cache, they also hardly take up any bandwidth on the fsb, and are probably all going to one core if they ever become strenuous. Superpi can then run almost completely independently of the fsb because of how small the program is. For heating up the cores, I don't think it's a bad program, but you'd need to run more instances than rahul did to actually do that, since it is, on its own, a very non-strenuous task.

Anonymous said...

ashenman, grow up and stop bashing all the benchmarkers you and all your camp used to reference when AMD had the performance crown! the key word is "had"

Anonymous said...

Ash, I see 600W from Dailytech and 560W from VR-Zone..

ashenman said...

Sorry, I skimmed the dailytech article (I've been kinda busy lately, and wasn't aiming the power supply comment at the vr-zone article). Though my point is less valid now, it still effects the results. That's also not a 6700 that they're testing, it's a 6600, though I realize that that will still only slightly reduce power consumption. I actually didn't post on any blogs or actually really talk that much about technology back when AMD still held the performance crown. I'm also not saying that current AMD processors still hold the performance crown and that better testing would show they did. I'm starting to warm to the idea of using low res to give an idea of how strong a processor is, but stress testing does require a very specific set of parameters to prove valid. Noting THG's Intel bias (even if it is or isn't very strong, it's there) I find that strong disparity between their results, and dailytech's (i'm not including VRzone because they obviously didn't load all cores, and I think Rahul's system could use some refining) a sign of strong yield performance disparity, or that they've significantly changed the way they're manufacturing the processors, which would be pretty cool.

Anonymous said...

I don't believe in using low resolution benchmarking soley for big numbers, but since most people can't afford 1600x1200:)

Romantic Planet Vacations Inc. said...

In response to:

Anonymous said...
Rahul, Romantic Planet is spam..I wish I could report to Google, but they're anti anonymous spam reporters..

Get a life Anonymous. I go way back with Rahul as we went to school together. I was just saying hi.

Rahul Sood said...

I went to school with Laurie from Romantic Planet for years. She dated one of my friends for a number of years...

She's a great girl, and not a spammer :)

Sharikou, Ph. D said...

Rahu,

Why did you delete all my posts while keeping all those empty insults? Not that I care, but it's a matter of principle for you to consider why you deleted my posts. If it was because I questioned your rather primitive way of testing the Kentsfield, then you need to learn to take criticism.

Anonymous said...

The reason for paying the same price for a am2 x2 instead of a similar priced conroe is actually quite simple.
To get performance from a conroe you have to put it on a $200 to $350 motherboard with antique intel chip sets.
If you buy a am2 5000+ instead of a e6600 you can put it on a $50 mother board and still get ddr2 800 plus top ATI video performance.
You intel fanboys must not have tried running google earth, it does not run on intel chipsets without ati video add ins but I can run it just fine on a lowpriced amd-ati stable platform motherboard and you cant do that with conroes and intel.
Intel conroes plus the needed hardware to support them is just too expensive and not remotely competitive with AM2 AMD and the ATI onboard stable platform.
Intel has really missed the boat, there is not a single decent low priced mother board for conroes only expensive expensive and expensive intel chip sets that dont run google earth.

Anonymous said...

Idiotic sharikou anonymnous and his sock puppets making flames and "ash" again here. November's coming very soon. We'll see how Kentsfield stack up to 4x4. AMD has been very coy with their 4x4 FUD - no benchmarks, no Crysis, no hands on, etc at DigitalLife.

Anonymous said...

Great blog you have here Rahul, sorry to see how all fanboy destroys it. Keep it up, and concrats on HP. :)

Shakira, go back to lala-land. You hating Intel is no excuse for Kentsfield beeing fast.

And to the poster about Google Earth and 200-300$ mobos:

Reality-check, mate. Where do you get that from? Any 965 based mobo supports C2D, and is a fast chipsets. Price isn't 200-300$ either, atleast not for the low-end ones.

ashenman said...

I'm not saying AMD made perfect or reasonable demonstrations. In fact, I'm not saying 4x4 will be better. I'm saying that we lack sufficient results from either to speculate. I'd really like it if Rahul would rerun some multi-tasking tests that aren't just instances of p95 and one or 2 other applications. I realize that wouldn't be realistic usage, but we wont see performance that matters on quad core till late q1 07 (when programs have enough threads to validate the purchase of either platform).

Also, what's the deal with "and 'ash'"?