Not showing benchmarks is either a really futile attempt to make Intel THINK they got something really powerful in the works (thus delay intels counter) or it's just them acknowledging they've lost again and they need to have low prices to enter the market in the hopes of reclaming the market.

Either way IMO they shot themself in the foot.

"The fact that you basically just stated opinion gives nothing to prove and (no offense) shows you have no idea what you are talking about. Like what has been said already. No one knows what AMD's new processor is capable of so stating that it will wipe the floor with Penryn is just plain ignorant and you have no facts to back that up."

Barcelona benchmarks have been released at 1.9ghz already so there are already plenty of "FACTS" on the internet as to it's performance and scalability and one can easily extrapolate those results and come in with a rational estimate of it's performance at 3ghz! Definitely not guaranteed, but not ignorant either as you so claim. I change my statement and replace "will" with "will probably". Now it's a more PC statement.

ko ko said:

I thought those benchmarks were crushed by a single core even in Multithreading Application or something... anyone got sources?

ko ko,

Are you off your rocker man? The poor thing got annihilated. I highly doubt that raising the frequency to 3.0GHz (if they can even get it that high) will increase it's throughput and performance by a large margin.

And there's nothing to say Intel won't do the same. The moment a competitive AMD chip shows up, expect 3.2-3.6 GHz (at least) Intel models.

You want more proof, then go research and prove this fact wrong. FACT, Barcelona is the fastest per clock x86 processor in history! The only reason it sucks is because it is at an extremely low clock speed of 1.9GHZ you dense people!! So "IF" AMD does have monolithic quad cores coming in at 3GHZ or higher, then "they will" kick some major ass (my prediction based on scalability). If they don't release it, then they were caught lying by displaying a working 3ghz Phenom at the meeting.

Here are some links that match up with the AMD 3GHZ demo at the analysts meeting.
http://www.fudzilla.com/index.php?option=com_content&task=...p;Itemid=1

http://www.fudzilla.com/index.php?option=com_content&task=...p;Itemid=1

Lastly, the point of this entire thread was to explain the business strategy of NOT releasing benchmarks of your 3GHZ products to prevent cannibalizing your launch product speeds which are much lower and keep Intel guessing so they can't begin a counter attack. The articles author completely missed the strategic aspects of WHY AMD would hold off on benchmarks of a 3ghz Phenom and I have just told everyone why they would hold back and why holding back is the proper strategy to not kill Barcelona 1.9ghz sales. I suspect that only once the initial 1.9ghz and 2ghz Barcelona's are sold and launched will we see AMD release Phenom benchmarks and a faster clocked Barcelona with real benchmarks behind it.

Duhhhhh, common sense people. Sell your crappiest products first, then tell them about price cuts or the availability of new better models at faster clock speeds, otherwise noone will buy the junk at 1.9ghz!!!!!!! (Marketing 101). You reading this mister editor and founder? I think it's about time you write an update and state how while you were dissappointed about not seeing benchmarks of a Phenom at 3ghz, you can now understand why AMD may not have disclosed the benchmarks or included it in their roadmaps yet. And if you want, you can credit the Ko Ko for bringing AMD's possible rational to you.
Cheers

ko ko said:


Care to show a source confirming that statement?


If it was, and your theory on scalability is correct, it would still lose the cinebench test released a while ago, if even clocked to the same speed as the Intel counterpart.

It would need a 50% increase in clock speed to match the Intel CPU used, but its 58% slower in the benchmark, meaning (if the bench scales properly with clock speed) it would still lose if clocked at the same core frequency.

Seeing you capitalize the word 'fact' really reminds me of someone on these forums ... . Sure you're not buddies or something?

ko ko said:

Saying Barcelona is the fastest per clock x86 processor in history is quite the bold statement. Source please? Or is this something you figured out yourself?



Yes, but the only problem with this strategy is this: AMD has p**sed off their shareholders and they want to see something physical, material, anything that works with the five senses and not something written on paper and therefore, theoritical. Right now, their stock has been quavering at around $13ish while Intel still going strong at $22ish +. Besides, it has already been said that Intel has not enough time to change anything about their processors in time. That or they have to delay it which means AMD gets a chance to rebound which is good, but not likely.



Sell your "crappiest" products first? That's just professional suicide. There is a reason why the graphics companies always sell their flagship stuff first. People love seeing physical results. If they see that the company's flagship works extremely well, it builds a sort of "wait-and-see" for what they might release later that is more mainstream and budget. You do not sell your s**ttiest stuff first if you can help it, if you don't want to watch your stock go down the toilet.

Just because you give price cuts on a s**t product doesn't mean people will buy it. It sucked at the start, why buy it now just because it's cheaper? No. People, LOVE price cuts if the product is good. Case and point? Intel quad-cores such as the Q6600. Notice how there was a bit of a buzz when Intel announced price cuts?

Also, FX series and Extreme Edition processors usually hit the market at the same times as their Opteron/Xeon counter parts because they are the pinnacle of performance from the core. And, because corporations might buy a hundred thousand Xeon/Opteron chips in one swoop, they are usually the first out the door. It's easier to aim high and fill in the voids below than aim low and fill in the voids above.

I am not used in expressing my opinions here (as you can see from my post toll ) but... planning a corporate strategy is not as simple as many of us think. I must (otherwise will be foolish of me) believe that any desicion made by AMD was based on market analysis and research taking acccount of the company image strategy. On the other hand, being an electronic marketing manager myself the past few years I could not help but noticing how AMD has "drifted" down in sales. To show some mere facts to support their new processor (benchmarks are one of them) would boost the consumer morale and would keep, at least, the market waiting for the new chip to come out in stores. Could not help but remembering me as a pc sales advisor some years ago that whenever a new "faster than others" processor (supported by facts or a glimpse of "facts") came out we advised the customer to wait a couple of months to make his choice. (By customer I mean companies and not home users / single pc users).

So, if the point is whether some benchmarks must be published or not then i say it matters. These things "open our appetite" for upgrade or buys. For me personally, a benchmark would be nice, but some other form of presentation of the chip's capabilities would do almost the same work for the market.

Does the KO in ko ko name stand for knockout. There is absolutely no indication that the Barcelona will knockout the Core2. I'm sure if AMD had the winning performance numbers they would release them as a teaser, just so people would hold off buying any more Core2's. It would help them to release benchmarks like Intel did.

Not sure if anyone mentioned this yet, but, not only does releasing the benchmarks create a buzz about the upcoming AMD product, it will give Intel purchasers reason to pause and possibly not buy an intel right now if they can wait a couple more months to get the new AMD. It's not going to stop the fanboys or slow the mainstream pc market, since most prebuilt machines are sporting intels. AMD should've struck while the iron was hot. Sadly they didn't and even if their new quad cores live up to the hype, all intel will have to do is release their benchmarks of their next processor, give a reasonable release date, and stick to it. It will be like AMD quad core never exsisted.............kinda like AM2

for all those insomniacs out there.....
http://www.youtube.com/watch?v=SYvupXNmaKQ
lord give me strength. i get sleepy just hearing the guys voice......

the thing that people seem to forget is that ALL of the upcoming AMD chips have been designed, and optimized (by actually working closely with microsoft) for Longhorn, (which is exsclusively 64bit) and vista (which will be dropping 32bit support steadily)

As it stands comparing currently available AMD chips with currently available Intel chips, two things hold true.

1) Even with slower RAM speeds, AMD still manages BETTER memory performance
2) The majority of AMD Desktop chips continue to have lower power consumption, and not only do AMD's server chips alone have a significantly better performance per watt rating, HP did an unreleased study comparing the total power consumption of AMD servers and Intel servers, and found that the top of the line AMD server, consumed 40% LESS power then the top of the line intel server. With AMD still maintaining a significant performance advantage over the intel server.

Intel caught wind of that, and somehow "convinced" HP not to release its findings. But an independant Tech journalist heard of this and decided to conduct his own study as intel holds no sway over him.

Also, and i feel like this is the thousandth time i've said this...intel has a shoddily designed 2 dual core in one package "quad-core" that is primarily benchmarked in 32bit OS, using 32bit software. I don't care how well 64bit cpu's perform in 32bit apps, nor do i care how well DX10 cards perform running DX9 games. With it being 4 years and change since 64bit cpu's became widely available, i don't know why anyone else does either.

C2D's are optimized to run in a 32bit OS running 32bit apps. AMD 64's are optimized...amazingly enough...to run in a 64bit OS running 64bit apps. A side note, Nvidia 8000 series cards are based on DX9 with kind of token DX10 support. Ati HD2000 series cards, were designed from scratch to be DX10. Which is why when you take a AMD chip into 64bit vista, intel doesn't have such an impressive performance gain, despite double the cores, double the cache near double the memory speed and 65nm vs 90nm core. Much the same as how the HD2900, the ATI mid-range card, performs neck and neck, or better then the high end 8800GTX in 64bit vista running DX10 games. Not to mention that 64bit Vista Crossfire wipes the floor with 64bit vista SLI...when it works that is. But hey...what can we really expect from nvidia who had their cards out 7 months prior to ATI: or from intel, the undisputed performance leader....so long as you use it in a certain kind of OS...and certain kind of software. Functionality? In the OS and software for which they were supposedly designed? Why that's insanity!

Software optimized to run on an Intel CPU will...usually run better on an intel cpu. Games optimized for nvidia graphics cards...will usually run better on nvidia graphics cards. Who really gives a damn if the cpu or gpu that just came out performs the best on the previous generation of software?

Amusingly enough, in Q1 of 2008, when Microsoft switches server OS support to exsclusive 64bit, I.E longhorn, intel server chips can't actually run any 64bit server software. Even the simple things like, i don't know SQL...they can't run that. But hey, what server uses SQL these days?

Lets do some simple math here
Core 2 Extreme QX6850 Athlon x2 6000 % Difference
Core Kentsfield Windsor
CPU Freq 3000mhz 3000mhz
FSB/HTT 333mhz 200mhz
Ram Spd 800mhz 750mhz Intel 16% faster
# of cores 4 (2x2) 2 (2x1) Intel 50% more cores & Dies
L1 Cache 64x64KB 128x128KB Amd 50% more L1 Cache
L2 Cache 2x4MB 2x1MB Intel 75% more L2 cache
Mem Ctr Motherboard On Die
Fab Tech 65nm 90nm AMD 27.78% larger die size
Die Size 286mmSq 230mmSq
Transistors 582million 227.4million
TDP Rating 130 125 Intel Uses 3.85% More Watts
Power Consumption
Idle 124 96 Intel uses 22.59% More Power
In 3Dmark06 Cpu Test 2 225 210 Intel uses 6.67% More Power
Price $1499 $170 Intel Costs 88.66% More

Benchmarks
PCMark 05 V 1.2
Cpu 9679 6158 Intel performs 36.38% Better
Memory 4510 4153 Intel Performs 7.92% Better

Sandra XI

Arithmetic ALU 55945 21938 Intel Performs 60.79% better
Arithmetic MFLOPS 38712 18544 Intel performs 52.10% better
Memory Integer 6389 8600 AMD Performs 25.71% Better
Mem Floating Pt 6389 8597 AMD performs 25.69% Better

3D Mark 06 v1.1
All 3Dmark, Games &
@ 1280x1024 32bit
CPU 4346 2231 Intel performs 48.67% Better
Graphics 11765 10182 Intel performs 13.46% better

Prey 124FPS 107FPS Intel performs 13.71% better
Quake 4 132.8FPS 105FPS Intel performs 20.94% better
Serious Sam 2 167.5FPS 152.1FPS Intel performs 09.20% better
Warhammer 51.2FPS 32.5FPS Intel Performs 36.53% Better
Supreme Commander 59.0FPS 37.5FPS Intel performs 36.45% Better
Unreal Tourn '04 91.9FPS 79.9Fps Intel Performs 13.17% Better

All cpu benchmarks were from THG interactive cpu charts, power consumption i found in another tomshardware article.

Ugh ok, i'm bored of copying stats and figuring percentages. It's quite clear that the QX6850 is a strong cpu. Perhaps more in theory in some area's, but still strong.

Unless you want to take into account the memory performance. I chose the x2 6000 because it's the fastest available at the momeny, and while you can get a 5600+ that actually will run DDR2800 and overclock to 3Ghz as easily as going into the bios and smashing your head on the keyboard for 30 seconds, I didn't want to make assumptions in the scores. Besides, it makes intel memory performance look even more horrible this way. AMD memory was running 16% slower...and the best the intel chip could do was beat it by 8%...with an OCed 5600 chip, that would most likely mean that it would only be 8% slower then the AMD chip, but hey, that was just PCmark 05.

Sandra, still shows the memory running 16% performing 25% better then intel..In fact if you go to the CPU charts you'll see that the QX6850 ranks 29th on the Sandra memory benches...with 28 AMD chips ahead, the FX-62 running 2.8ghz beating it by 32% i belive, or close to it. To my knowledge all of those scores were acheived in 32bit Vista, that is conviniently not specified.

So to recap, the QX6850 has
twice as many cores
4 times the L2 Cache
16% memory speed advantage
65nm core compared to 90nm core
Consumes more power, without having the on die memory controller the AMD does
(though i didn't look for specifics i belive reading air cooled QX6850s will achieve above 60c under load, compared to the AMD which stays in the low to mid 40's)
The best of any real world benchmarks i found, recorded or not still did not push the intel chip above 50% performance gain.

All this, and it only costs a mere 9 times more then the AMD chip it was compared with.
The best gaming performance boost was 36%, and an average around 13%. Overclocking is limited at best, most of what i've seen has shown a 300mhz gain if you're lucky.

The Q6600 same chip essentially, but only 2.4ghz, still $300 and 20% performance loss over the QX6850. They can overclock, but that's always a gamble. But $300 isn't insane to pay for. $1500 is, and at that price, with twice the cpu processing power, 4 times the L2 cache, which that alone should guarentee the performance, i expect the cpu to perform a minimum of 60% in every application, in every area.

But it doesn't. Not even close. Memory performance has a HUGE impact on overall system performance and intel's blows. So even if AMD were to take the 6000x2, keep it at 90nm, and do the half-assed 2 dual cores in a single package and call it a quad core BS, it would obliterate intel.

But that isn't what they're doing. They spent more then 6 months designing the chip
They've had the benifit of working with microsoft to get an inside look at vista, where intel has not.

AMD fans usually know a bit better then the rest. They actually used to do things like look at the Clock speed on the cpu compared to intel, the speed and latency of the ram, the fab size of the die, and power consumption. More recently take into account the number of cores in regard to performance. Like the fact that the E6850, the highest performing dual core intel has out, is a 65nm chip with a 4mb L2 cache. Compared with AMDs 90nm chip with a 2mb L2 cache, and a synthetic benchmark performance difference of 18% on the cpu.

But it should be easy enough to see that synthetic CPU benches don't mean a hell of a lot when it comes to real application, because i certainly don't see intel besting AMD by 60% anywhere, occassionally 30-40...but usually 10-15. In 32bit....with crappy memory performance...which impacts the cpu performance...which means intel still cuts corners.

AMD would have to have replaced it's R&D department with a pack of lobotomized drunks to fall behind intel with the release they've been working on almost as long as they worked on the opteron series. During which time they did the exact same thing of releasing a steady stream of mildly better processors to keep some revenue coming in and a bit of attention on them.

When they've stated the fact that the new chips will be 4 cores and memory controller on a single die, 2mb L2 per core, with a shared L3 cache and 65nm fab. Not to mention that the AM3's will work in existing AM2 boards, though without DDR3 support, and will work as AM2 chips in AM3 boards with DDR2 memory so you don't have to waste money upgrading ram that performs worse and costs consideranbly more. I guess they look at it a bit the same as me when it comes to benchmarks.

Why give intel time to get the jump on designing something to soften the blow with releasing performance of chips that won't be widely availble across the entire line, when the benchmarks of current AMD chips against Intels really speak for themselves.

That is if you actually do the math. But that's always been intel's problem when it comes to product design. They've never been good with the hard numbers, they just rely on consumer ignorance and impatience. To bad it's worked again, especially when no one needed the damn things to begin with..

Wow. Someone's got quite a bit of free time on their hands.

*claps*

Im tired, I dont think Im giong to pick out every detail, but if I do then so be it. Here I go.

Dan Boak said:
And yet, C2D's and C2Q are 64-bit CPU's and run 64bit OS's just fine. Many of us here do that.

Dan Boak said:
I agree with the CPU part, but can you honestly say that there really is a DX10 game out there that we can bench against to give us a rough idea on how well a DX10 card does with a DX10 game? Best thing we can do is bench its DX9 capabilities with DX9 games.

Dan Boak said:
And yet, the HD2xxx cards still really cant compete with the G80 series cards. With the GTS models yes, but not the GTX or Ultra. ATi has already stated that they do not plan to compete in the high end market. Just budget.

Dan Boak said:
Do you have any next gen software to test the next gen hardware on? If so, Im sure we'd all like to see how well they stack up.

Dan Boak said:
And yet the desktop CPU's can. Hmmm, this doesnt seem right. I find that statement hard to believe.

Im not even going to bother with the math. Not much of a math wiz and even if I were, im to tired to even attempt it.

Dan Boak said:
Ive never even herad of a QX6850. Just an E6850.

Dan Boak said:
And yet, an Intel E6600 or so would probably wipe the floor with that AMD 6000+ for just a bit more money at a slower clock rate. I think my Q6600 would wipe the floor with that at 2.4Ghz (stock).

Dan Boak said:
And yet, you are comparing apples to oranges because the C2D line dominates anything AMD has to offer. AMD has no competing chip for the C2D line. Their competing chip is Barcelona (server) and Phenom (desktop) which is due in Q4 07 and Q1 08 respectively. So, seeing how you are comparing old tech to next gen, there really is no comparison.

Dan Boak said:
You are comparing two Intel quad cores why? Your telling me that you do not expect the QX6850 to be faster than the lowest budget quad core? Umm.....Id rather get the budget core and overclock it to the QX6850 specs.

Dan Boak said:
Pure opinion. It isnt fact because that has not and will not happen. You have no idea if that would be the case or not.

Dan Boak said:
Got a link proving that? The whole benefit of working with Microsoft part?

Dan Boak said:
18% on which side? AMD or Intel? If you say AMD, I wanna see the benchmark.

Dan Boak said:
And AMD doesnt? In the great amount of applications, I would say Intel would definately take the lead. Not in all, but in the great majority. Things that require a lot of CPU attention like video, streaming, and whatever else (to tired to think of s**t that takes advantage).

Dan Boak said:
Or maybe, just maybe, their competing chips arent doing as well as they are making them to be. No one really knows. Need benchmarks to prove or disprove it.

Dan Boak said:
Or AMD just knows that people are tired of them making new sockets which really arent necessary. Look at the LGA 775 socket. It has seen so many processors compared to AMD's sockets its quite interesting. With the release of AM3 chips that are backwards compatible, what makes their other CPU's so different? What really hurt AMD was cutting S939 off to soon.

Dan Boak said:
Because AMD is in a world of hurt and they know it. You said it yourself "AMD would have to have replaced it's R&D department with a pack of lobotomized drunks to fall behind intel with the release they've been working on almost as long as they worked on the opteron series." Like I said, AMD is in a world of hurt and they know it. They shrugged the C2D away and misjudged it. They had no idea it was going to be this good.

Dan Boak said:
The more cores the better. People that do video editing, ripping, encoding, things like that really take a shine to multi-core processors. It really helps speed things up. Compare encoding and ripping and what not with a dual core and a quad core. Then a quad core against a faster quad core. Times get better. Not worse.

Mark Allen said:

OR this website doesnt have much to offer to come back. lol

Im sorry but no part of this internet niche has moderators or users that are any better than another site like extremeoverclocking. Sander just sounds like an impatient little twerp/shareholder like the rest i've read in every other forum.

Fun to write opinions, but this site was not very professional one bit for me to come back after i write this....
insert bored icon.

ko ko said:

Okay, products are out, and what did I tell you, exactly, thanks very much.

Best regards,