Intel's Conroe Resurfaces, Benchmarks Strong

MojoKid writes "Intel has been occasionally leaking performance results of their upcoming Core 2 Duo processor for the desktop, code named Conroe. At this years IDF select members of the press were allowed to get hands-on access to test systems for benchmarking. Now, coincident with this week's Computex show in Taiwan, Intel has seen fit to show us just what their soon to be released CPU can do, yet again. Select press members got together with Intel in New York city for another round of testing with Conroe. HotHardware has a performance showcase posted with scores from a Core 2 Duo E6700 machine and a 2.93GHz Core 2 Duo Extreme Edition X6800. The results, compared against the backdrop of an overclocked 2.8GHz Athlon 64 FX-60 system, look very impressive indeed for Intel."

Comments

Intel's a bit wierd now.

[Anonymous Coward]

Now that Intel has it's first non-technical CEO, all they can talk about is vaporware of furture unreleased chips, while Shares of Intel have fallen 33 percent since Otellini succeeded Craig Barrett in May last year. Should the board/shareholders really allow someone with a background like Otellini's to run a company like Intel? You see how well medieval studies people worked out at HP. IMHO they need to get the tech people back in charge at Intel if they want it to compete in a tech market. At least in the past they succesfully defended their market share with their *existing* products even when they were inferior. This new strategy of basically saying "don't by our current stuff because our roadmap is even nicer" could only come from a MBA.

Re:Intel's a bit wierd now.

[Anonymous Coward]

Now that Intel has it's first non-technical CEO, all they can talk about is vaporware of furture unreleased chips, while Shares of Intel have fallen 33 percent since Otellini succeeded Craig Barrett in May last year. Should the board/shareholders really allow someone with a background like Otellini's to run a company like Intel? You see how well medieval studies people worked out at HP. IMHO they need to get the tech people back in charge at Intel if they want it to compete in a tech market. At least in the past they succesfully defended their market share with their *existing* products even when they were inferior. This new strategy of basically saying "don't by our current stuff because our roadmap is even nicer" could only come from a MBA.

Since the job of a CEO is to manage a firm, not to do technical work, an MBA is a more appropriate degree than, for example, an MSc in a technical area. A mediaeval studies degree is even less relevant than a technical degree (which probably at least includes some theory relating to the organisation of systems, which can be applied to a firm's organisation), and suggesting that both economics/business and mediaeval studies are equally relevant by virtue of being non-technical isn't a valid argument.

I've been an AMD user for some time, and would still buy AMD over Intel for desktops/servers, but Intel by a huge margin for laptops. If Intel's next generation desktop/server CPUs will be better than AMDs, versus its current solutions which are arguably inferior, convincing customers to delay purchases until Intel has the upper hand is actually a very wise move. If the Core Duo desktop/server CPUs are as good relative to AMD as the Core Duo notebook CPUs, it'll definitely push me away from AMD, and back to Intel.

To sum it up, when you're at a disadvantage versus your rivals in the market, in terms of product offerings, but expect to have an advantage over them in the near future, convincing customers to delay purchases is precisely the approach that should be taken. If you're at a disadvantage, and expect to remain so, hyping your inferior products can help sustain market share, but at a loss of confidence as customers realise your over-hyped products are actually inferior to less-hyped products from competitors.

On Intel built and Intel controlled boxes.

[Visaris]

These benchmarks were run on boxes that Intel built. Even the AMD box was built and configured by Intel. Trusting these benchmarks is abit like trusting a study funded by the oil industry claiming that global warming isn't real. There have been a good number of independant tests of the Conroe and these put the top of the line Conroe around 12% faster on average than a FX-62. The results from the Intel benchmarks show a much bigger performance delta, and to be quite honest, I don't trust them one bit. Somewhere around 15% is much more reasonable.

Re:On Intel built and Intel controlled boxes.

[TheRaven64]

One big difference likely came from the RAM. The AMD box had DDR400, while the Intel box had DDR2-800. DDR2 has a greater latency in terms of cycles than DDR, but when the DDR2 is twice the speed of the DDR then this disappears in absolute terms and you are left with the RAM in the Intel box having about the same latency, but twice the throughput.

Having said that, all of the benchmarks run were publicly available. There's nothing stopping you from configuring an AMD box yourself and seeing what numbers you get. In fact, I'm slightly surprised that the review site didn't do this.

Re:On Intel built and Intel controlled boxes.

[Anonymous Coward]

I couldn't disagree with your more. Actually if you read the article and looked at the test system specs and setup, the rigs were all configured very fairly and with identical drivers where possible. They even gave AMD an edge at 2.8GHz overclock for the FX-60 and didn't put the Conroe systems up against an AM2 DDR2 system, which is widely known to be on hardly faster and significantly less mature/stable currently than the nForce 4 SLI chipset on the socket 939 system that was used.

These numbers, like all others I've seen, scale accordingly and I'd say AMD is in for a bit of a beating this year. Not that I care either way, I'd actually rather see AMD keep their edge over ChipZilla but in the end we're all benfitting from Intel's new Core 2 Duo series. They're going to raise the bar, which is nothing but goodness.

Re:On Intel built and Intel controlled boxes.

[Anonymous Coward]

Who cares if intel built the boxes. If they have some magical method of eaking this much more performance out of their chips, well we'll just use the same technique on our boxes, no? It's just hardware dude, which anyone can put together.

Re:On Intel built and Intel controlled boxes.

[tomstdenis]

DDR2-800 is not generally available to the retail world. So the only benchmark that makes sense is this

AMD FX-62 sales volume: LOTS
Intel Core 2 Duo sales volume: Zero.

Not only that but how hard is it to go in the bios and make and AMD64 processor perform sub-optimally? Sure it's DDR400 but CL4-4-4-10, and you need the ECC scrubing turned on, disable the cache and ...

If Intel really wanted a benchmark they should ask AMD for engineering samples of next year's cores and they could pit them together.

Tom

Re:On Intel built and Intel controlled boxes.

[Jah-Wren Ryel]

DDR2-800 is not generally available to the retail world.

http://www.pricewatch.com/memory/845489-1.htm

If Intel really wanted a benchmark they should ask AMD for engineering samples of next year's cores and they could pit them together.

ring...ring...ring...
AMD: Hello?
Intel: Hello, AMD?
AMD: Yes?
Intel: Intel here. We've had to cut back on our industrial espionage budget this year, seems we've had an unexpected revenue shortfall and can't afford that group any more
AMD: Have you considered outsourcing it to India?
Intel: Well, no, not really. We were hoping you could just send us some samples of your lab prototypes.
AMD: Sure, sure, say no more. We'll send those over right away via courier. You'll have them on your desk first thing in the morning.

Somehow, I just don't see that happening...

Re:On Intel built and Intel controlled boxes.

[Anonymous Coward]

The same usual nonsense you're always talking (every single one of your posts I've seen was FUD or trolls).

I'll fix your post for ya:

AMD FX-62 sales volume: a few units (most people don't wanna spend 1500$ on a CPU)
Intel Core 2 Duo sales volume: lots starting next month (significantly faster than FX62 and at a FRACTION of the price)

Not only that but how hard is it to go in the bios and make and AMD64 processor perform sub-optimally? Sure it's DDR400 but CL4-4-4-10, and you need the ECC scrubing turned on, disable the cache and ...

And then it would be obvious that those AMD boxes were performing slower than in every other similar benchmark we've seen. And the results would differ from other 3rd party benchmarks we've seen before... Somehow, even if they have a better product (finally), then they *must* be blatantly cheating (by setting ridiculously slow settings that would make this PC slow as molasses) to appear faster? Right. It just *can't* be faster, AMD fanboys can't possibly admit "defeat"... (every other benchmark that had similar results must have been rigged too, and when we all have fast Core 2 Duo CPUs in our PCs and they perform that fast.... well, they've still must have cheated somehow! impossible!) Wait for the next benchmarks if you don't believe it. In a month's time there should be tons of them, and they'll somehow all concur (of course you'll say Intel paid them all for their reviews or something)

If Intel really wanted a benchmark they should ask AMD for engineering samples of next year's cores and they could pit them together.

Bzzt! Wrong again! More like next MONTH's cores. And if you take AM2 cores and put it against the Core 2 Duo (and at the clock speeds you can expect them to be), they'd still lose.

AMD fanboys will never admin Intel for once has the superior product (after years of netburst junk). Denial, FUD, conspiracy theories, anything will do to discredit it.

You keep buying 1500$ FX62's while I buy faster Core 2 Duos for 300$.

Re:On Intel built and Intel controlled boxes.

[(H)elix1]

As long as they are comparing CPU's that are a going to be launching, I was wondering why they did not use an AMD AM2 CPU that uses the same DDR2 RAM. "Typical" DDR2 is not very fast, thus the current real world comparisons of the 939/AM2 CPU's don't give a huge boost to the latest CPU. If DDR2-800 was common, the change over makes a bit more sense - possibly just what AMD was thinking as well.

And yes - you can dog a machine by tweaking the BIOS. Our kit was in a final bakeoff with a competitor - the customer was re-imaging the OS between installs, but we were last to use the server. The customer tried to drop a test file on the machine using a floppy disk and found the floppy missing. Went into the BIOS and we noticed 'someone' had set every possible BIOS setting to a worst-case condition. Turns a fairly fast machine into something not so quick. Won the gig, but I have to wonder how often this happens.

Re:On Intel built and Intel controlled boxes.

[Vexorian]

It's not like they 'll win anything from faking benchmarks they would eventually be proven wrong, and that would be a bad thing for them, so just leave the paranoia.

Re:On Intel built and Intel controlled boxes.

[Anonymous Coward]

The problem with your anology is that the industry funding of global warming research is presenting a real and alternative approach to the problem of global warming. It is not that global warming isn't happening it is a matter of finding the core problem. To blame global warming simply on economic output and not investigating other contributing factors is just bad science. It amounts to blaming recent volcanic activity on the production of green house gasses. Before such wild accusations are made yoou need to produce credible science to support the statements. The global warming crowd hasn't done that at all, rahter they make scarry statements to get the public to continue their funding.

As to Intel. well it has been pretty obvious to me that they have had issues now for some time. It sort of reminds me of Apple when they where promoting the PPC. Lots of hot air and nothing to back it up. In this case though I think Intel realizes they are screwed if there isn't any substance to their claims. Frankly Intel would do well to get a 20% lead on AMD on a performance per watt rating. So yeah Intel has likely optimized the benchmarking in their favor, though I doubt to the extreme as rebuttal benchmarks will come real fast. I don't however see them being successful with totally hollow performance "leaks". There has to be some meat with the potatoes or they will have trouble recovering, Intel knows this.

As far as independant tests that is all well and good but I want to see data on released hardware. Even at 10 to 15 percent I'm not to sure I'd want to switch back to Intel. AMD is the company that had the foresight to give me and the world a cheap 64 bit environment, that works rather well. So unless I see compelling reasons to change I'm likely to stay with AMD. The one thing that could change my mind is Floating point and the performance increase it might offer realworld CAD. Getting good benchmarks for that sort of thing is very difficult and slow, by the time those sorts of benchmarks hit the street I suspect AMD will be offering increased performance anyways.

What Intel may have is more head room in this processor than AMD has currently. That is they may be able to ramp speeds very quickly. Yes the Mega Hurtz race may very well be back for a short visit. But first Intel has to show that they can actually produce the processor.

Interesting times ahead.

Dave

Re:On Intel built and Intel controlled boxes.

[kscguru]

Is widely known to be hardly faster and significantly less mature/stable

Funny that I've been watching AM2 carefully for the past month, and only agree on "less mature". AM2 is not faster by itself, but it does open the door to DDR2 memory. Which means Intel went out of their way to compare an AMD on DDR memory with an Intel chip on DDR2, when Intel could very easily have set up the "equivalent" AMD system on DDR2. When they deliberately don't match memory technologies, I'm suddenly very suspicious of Intel's benchmark.

My socket-AM2 system has been stable - except for Tomb Raider, which does seem buggy (and I'm blaming graphics drivers for that). nForce4 is a buggy chipset period, I don't see how that is any advantage at all.

I expect Intel to jump ahead with their Core design, and then AMD to make up much of the difference with K8L later this year. But on the server side, AMD is going to eat Intel's lunch for a long while yet.

Re:On Intel built and Intel controlled boxes.

[Anonymous Coward]

However, now we all know that the upcoming intel cpu is faster than the 3-4 years old amd one.
Whoah! :D

Re:On Intel built and Intel controlled boxes.

[Anonymous Coward]

For someone being so vehemently anti-FUD, what's with the $1500 figure for the FX62?

Seriously, if you want to make a point, don't make up numbers. Sheesh.

Re:On Intel built and Intel controlled boxes.

[MooUK]

Whether true or not, your post comes across as very fanboyish too.

You should never take a manufacturer's claims about their product as fact. If separate, independent testing proves that the Intel offering is better than the AMD chips mentioned, then that's a different matter. In this case, the only information available is from Intel and those they chose, hence it is unlikely to be independent.

Re:On Intel built and Intel controlled boxes.

[Anonymous Coward]

Intel: (upon inspection of the base code) *gasp* this is over extended 486 code!

AMD: Heh heh heh...

Intel: Someone look that up don't we own the IP rights to that?

AMD: Well we had to start somewhere...build it cheap and they will come.

AMD is a hack, they basically got started stealing IP and have used price schemes to get a firm grip on the industry. Kinda reminds me of the people who think Linux will ever be something larger than a niche OS and will replace the Windows desktop...keep on smokin.

Re:On Intel built and Intel controlled boxes.

[Anonymous Coward]

The Pentium 4's design is superior in the ranks of a server chip, but I'm afraid they simply over-designed it to the point they're carrying too much baggage in the means of pipeline stages to keep encoders happy. And I really hate AMD users who validate everything by the benchmarks since almost any benchmark currently available is optimized either which way, and the only way to really tell is to use the damned hardware and get a feel for it.

Of course, it may seem trollish to bag at benchmarks, but really, I simply don't feel like they prove anything anymore. None really test for what they need to (multiple instructions? multiple threads? Fully qualified float, integer and vector math? Please, 3D mark used to be good, but they've been caught cheating, so let's not go there.

I really love and admire both AMD and Intel (and these days I'm really more on AMD's side, simply because I'm rooting for the underdog, and I'm a poor college boy), but I feel like neither of them are playing fair anymore. AMD gets a rush off copying every implementation Intel's ever made, and yet, when Intel does the same they cry foul? That's not sportsmanly at all, that's just being a whiner. AMD's still a multibillion dollar corporation, and they CAN catch Intel; they just need to do it and shut their upper lip about it.

Intel on the other hand, still employs a great deal of engineers that sit on their asses because of their current marketing team. There is so much they could be doing with Pentium M that they simply won't do, hopefully this deal with Apple will help them to. There's so much they could do with BIOS that they just won't do, hopefully Apple can alleviate this as well. There is simply so much innovation to be had, and it's so stifled by old technology, bad marketing and biggotry.

On a side note; if Intel went with the Pentium M now, that could really be a sign of monopolistic behavior that AMD could use. The Pentium M was designed by nature to run cooler, period, and thus, can outclock its competitors. I believe the IPC is still a bit lower than A64, but mainly due to A64's bitwidth. We'll see in about a year.

Re:On Intel built and Intel controlled boxes.

[saleenS281]

except that benchmarks have already shown ddr2 really doesn't do much of anything for the AMD chips because they aren't bandwidth starved as it is.

Re:On Intel built and Intel controlled boxes.

[tomstdenis]

[DISCLAIMER: NOT SPEAKING ON BEHALF OF AMD...]

AMD has challenged Intel to duels [go to their sunnyvale lab they [used to?] have posters up in the public space that say so. I wouldn't say AMD wants to take stuff out of the lab and make it public but I suspect if Intel said "let's set up a contest with current retail parts" they MAY [see above] answer the call.

It's no big secret that AMD wants to challenge Intel [and beat them] on every front.

Intel is comparing next gen stuff to current to say "look our /current/ stuff is so much better" and hope you don't notice. It's pretty lame because Conroe/etc doesn't look like it's bad.

I'd love to see Intel do MP benchmarks in public though. They'd get their asses handed to them.

Tom

Re:On Intel built and Intel controlled boxes.

[tomstdenis]

Hint: look up my CV and see where I work...

We deal with crooked benchmarks all the time. The fact that these are Intel maintained boxes comparing next-gen stuff against RETAIL cores is stupid. Even if they release the cores in July it will be months before every corner store picks it up. Those FX-62's they comparing against were designed years ago [though the process/design has changed since] and probably fab'ed many months ago.

It isn't as if AMD is saying "Ok we made Opteron, we're done, let's go make plush bunnies now". If Intel was actually serious they would challenge AMD to show their hand [*]

Tom

[*] I won't say what the "hand" is.

Re:On Intel built and Intel controlled boxes.

[jiushao]

I'd love to see Intel do MP benchmarks in public though. They'd get their asses handed to them.

Well, Anandtechs tests of a 4 core Woodcrest server against a Sun Niagara and 4 core Opteron sure seems to suggest otherwise: http://www.anandtech.com/IT/showdoc.aspx?i=2772

Granted you may be after 8-way or higher, but that is an interesting enough test. The Woodcrest makes an extremely good showing there.

Re:On Intel built and Intel controlled boxes.

[tomstdenis]

2.2Ghz Opteron against 3.0Ghz core of the future?

*I* have 885s in my 2-way Tyan board at my house. That's a 400Mhz or 16% boost in CPU performance. I'm sure anandtech could get themselves some 285s for a benchmark.

Tom

Re:On Intel built and Intel controlled boxes.

[Anonymous Coward]

Powerful flamebait.

Except you forgot that AMD had rights to produce 386 and 486 procesors alongside of Intel (Intel actually needed a second supplier to get in the IBM PC).

Also you forget that AMD [and Cyrix] were the ones to extend MMX and AMD was the one to come out with FP SIMD first. And AMD was the one to come out with a sensible 64-bit extension set to x86. And AMD was the one to bring point-to-point networking to the x86 world (they didn't invent it but they did embrace and extend it :-)).

Intel has been hyping Netburst for far too long and is only recently moving to "Core" as their flagship products. If you want to talk about "over extended 486 code" look at the Pentium 4. It was crap when it first came out, it's crap today.

Re:On Intel built and Intel controlled boxes.

[Anonymous Coward]

Sorry. They were 1500$ last I checked (last month). They've now dropped to 1400$ CDN...

So instead of being 500% as much for a slower CPU, You're "only" paying like 450% as much or whatever (still for a slower CPU).

The accuracy of the number I mentionned is somewhat irrelevant. At the same price, I still wouldn't buy a FX62 because it's slower (a matter of price/performance ratio, given power usage, architecture and all are both excellent)

Not made up numnbers, just outdated by a few days (I don't follow top-CPU daily pricing much - sue me?)

Re:On Intel built and Intel controlled boxes.

[jiushao]

Unless the Opteron magically scales better than linear when going from 2.2GHz and 2.4GHz (both tested) to 2.6 GHz it is still way behind, not only on raw performance but also on power consumption and price/performance. The woodcrest is a CPU of the future yes, the future being less than two weeks from now.

Re:On Intel built and Intel controlled boxes.

[tomstdenis]

It's hard to tell, they randomly jump from 275 to 880 and back. First off, why didn't they just choose 285s? Failing that, why did they jump between 2 or 3 different AMD cores? Who knows if their results are even accurate.

On the crypto side, the results are hard to read. The graph shows more signatures/sec for AMD but the table lists otherwise. Even still, I find it hard to believe Intel has any lead on that market. AMD has a 5 cycle multiplier and three ALU pipes for bignum math [hint: this is my passion]. Unless Intel has a 3 cycle multiplier or faster L1 (doesn't look like it) it should clock in at about the same pace. Doing bignum mults/sqrs I routinely get an IPC of nearly two on my 885 box.

It could also be that the code in OpenSSL [or whatever they used for SSL] is not tuned well. My TFM math library beats OpenSSL on x86-64 and matches it on x86-32 [both intel/amd] and PPC32 platforms.

Eitheway, I'm not saying it's impossible for Intel to win out on some marks. I'm just questioning the validity of the test because they seem to use random collection of boxes. If they want to make a point they could just pit some 285s against it running more open tests. I'd rather see Intel win by merits alone and not questionable testing practices. If they *are* faster it gives more incentive for AMD to catch up next year.

As for your comment about scaling linearly... If the task fits in the cache, generally it's true. At least for crypto work. AES takes 260 cycles @2.6Ghz ... it takes 260 cycles at 2.2Ghz as well ...

Where things skew is on the memory. It takes more cycles at a higher clock rate to access system memory. Which means that you may get the same walltime performance but the cycles/operation can go up.

Tom

Re:On Intel built and Intel controlled boxes.

[jiushao]

It's hard to tell, they randomly jump from 275 to 880 and back. First off, why didn't they just choose 285s? Failing that, why did they jump between 2 or 3 different AMD cores? Who knows if their results are even accurate.

It seems a bit odd that they use different machines yes, however, Anandtech is typically fairly thorough with these things so I see very little reason to doubt the accuracy. At any rate it does satisfy your requirement of tests independent of Intel.

On the crypto side, the results are hard to read. The graph shows more signatures/sec for AMD but the table lists otherwise. Even still, I find it hard to believe Intel has any lead on that market. AMD has a 5 cycle multiplier and three ALU pipes for bignum math [hint: this is my passion]. Unless Intel has a 3 cycle multiplier or faster L1 (doesn't look like it) it should clock in at about the same pace. Doing bignum mults/sqrs I routinely get an IPC of nearly two on my 885 box.

I see no problems reading the results for the crypto test, the Opteron and Woodcrest run more or less neck-to-neck when using 4 threads (the opteron crushes the Woodcrest running with 8 threads since the machine in question is apparently an 8-core deal), the Woodcrest wins on smaller key-lengths and the Opteron wins on larger, no big difference though. It seems fairly clear that a 2.6 GHz or higher Opteron will take that particular test for AMD, though with a very thin margin.

As for your comment about scaling linearly... If the task fits in the cache, generally it's true. At least for crypto work. AES takes 260 cycles @2.6Ghz ... it takes 260 cycles at 2.2Ghz as well ...

Right, I said "Unless the Opteron magically scales better than linear". Looking through the results the Opteron is quite a bit behind on most tests even if we assume that it somehow managed to run at 3GHz and scales there linearly. So the Woodcrest is not only higher-clocked, it most often performs better clock-by-clock.

Eitheway, I'm not saying it's impossible for Intel to win out on some marks. I'm just questioning the validity of the test because they seem to use random collection of boxes. If they want to make a point they could just pit some 285s against it running more open tests. I'd rather see Intel win by merits alone and not questionable testing practices.

Can't really fault you for being cautious. I would say that the evidence that the Conroe based chips are superior to the K8-based ones has piled up high enough to thoroughly convince me however. I were not too suspicious about the tests done at Intel-controlled facilities, sure they put the machines together, but they did not control the software, indendent testing on AMD machines suggest that the AMD set-up is indeed fair. This Anandtech article is then only confirmation for me, a Conroe brought into a non-Intel lab still seems to perform beautifully.

If they *are* faster it gives more incentive for AMD to catch up next year.

Sure is, there is no downside here for consumers; Faster, lower-power and cheaper processors ahead. I really think that AMD has been a bit pricier than I had exptected from them as of late, sure the X2's are nice and all but the entrypoint price was much too steep for me. The move to 65 nm and K8L redesign ought to make the situation a bit less skewed and set us up for a nice price-war.

Re:On Intel built and Intel controlled boxes.

[fitten]

Possibly... AnandTech has some interesting benchmarks for server based stuff running Linux (64-bit Gentoo) here that show Woodcrest doing pretty well. I think most of the nitpicking (DDR2 vs. DDR and the like) is just living in denial. AM2 has shown to be a non-issue for anything except the most memory bandwidth intensive benchmarks, and most of those are synthetics, which are pointless. There are a few HPC benches that show decent gains for AM2 but how many people run particle physics simulations and airflow solvers out there? Most just want more frames in F.E.A.R. or some other game. There have been *lots* of benchmarks of all kinds posted all over the place for Core2 and they are all consistent, even when using DDR2-667, which has both higher latency and only marginal bandwidth gains over DDR400 (basically equivalent).

Throw in the panic induced flailing about that is called 4x4 and you have a recipe that says AMD is sweating bullets, and they should be. I have four Athlon64s (one is an X2) and three AthlonXPs and right now, my next machines are looking very much like they are going to be Core2 based.

Re:On Intel built and Intel controlled boxes.

[kill-1]

I've heard that so often, and basically you're right. But all of the benchmarks show that the new Intel E6700 is 10-40% faster than the best Athlon FX. They can't fake that much with specially built boxes.

Core 2 Extreme

[mfh]

You have a point about the Intel thing, there. Just like the response time on a monitor -- if the benchmarks come from the manufacturer, how valid can they truly be? Where are the stips?

Point is -- Core 2 Extreme has great specs but the map and the landscape are wholly different. Time will tell.

Re:Core 2 Extreme

[cnettel]

The difference is that there is no trivial way for an end-user to dispute those performance numbers. It's even out of reach for several hardware sites (and there are real differences between different production sets of the same panel). Compare this to CPU performance: ANYONE will be able to run these benchmarks in two months. Anyone can run them today on the AMD and NetBurst side to get reference data. If the Intel results differ a lot at actual release, hell will break loose and Intel would be really out of touch to think that they can succeed. Mainstream users won't care anyway, and the technical users would certainly disprove of the methods.

So, Intel could do this, but they are probably quite aware of the consequences.

Osborne Effect

[Visaris]

Intel is suffering from the Osborne Effect. They have hyped their new products (which are comming in July/August of 2006) so much that no one wants their current parts. This has forced Intel to drop the prices of netburst (read: P4) parts through the floor to keep moving them. Intel is selling many parts at a loss, and they have more price cuts (up to 60%) planned for the 23rd of July. Conroe is a great chip, but it currently has bad yeilds and will not make up a significant portion of Intel's shipped CPUs until the end of this year. At that point, Conroe based chips will be 20% of production; you can only imagine how many will be available on launch, a whole 6 months earlier than that. Intel has a killer chip on their hands, but it will be along time before Intel is able to ship enough of these to do much to the market. In the mean time, Intel will continue to sell their old tech at a loss to clear out inventory and try to keep AMD from making more marketshare gains... I don't think it is going to work.

Re:Osborne Effect

[Anonymous Coward]

Conclusion? Microsoft is bad company.

Re:Osborne Effect

[AnotherDaveB]

Intel is selling many parts at a loss

What makes you think Intel is selling parts at a loss?

Re:Osborne Effect

[Visaris]

What makes you think Intel is selling parts at a loss?

In a sense, all CPU companies sell some chips at a loss. Processing a wafer costs almost the same amount no matter what chip is on it. A mess of CPUs come off of the wafer and are then binned based on their power and performance characteristics. CPUs from each bin sell for different amounts. The top bin with the best chips sell for a lot and make a nice margin. The lower bins sell for quite a bit less, and have less of a margin. Many CPU companies will even sell some of the lower binned CPUs for less than the cost to produce them because the higher-binned CPUs will offset this loss. Further, selling a chip for something is better than not selling the chip at all.

I say that Intel is selling many CPUs at a loss because the low end parts are the ones they are dropping their prices on the most. Further, Intel is dropping prices on their 90nm chips the most, and these are quite a bit larger than their 65nm parts. In the CPU industry, the larger chips cost more to make.

So, what I see is Intel dropping prices on the parts that already have low margins, and are the most expensive to produce. This is not a good thing for them.

Re:Osborne Effect

[Kjella]

Except Intel doesn't need the sales of PIV to get Conroe out the door. Their latest FUD campaign has been that the Conroe chips will be so superior to anything else, noone would want one from the current generation. They're trying to stall the whole market, and from the mainstream press I've seen I think it's working. The PIV price cut press is catching the "bargain" hunters, and the Conroe press is keeping "performance" hunters on the fence to see what's coming. And I don't see any major games released right now that mandate a new CPU purchase at this time. Of course, it's a temporary measure but as long as there's no delays in the Conroe launch I think it's seriously pissing on AMDs parade right now, because well... their chips are clearly superior right now, today.

Re:Osborne Effect

[zakath]

Intel is suffering from the Osborne Effect. They have hyped their new products (which are comming in July/August of 2006) so much that no one wants their current parts.

I don't think Intel is suffering from the Osborne Effect. People don't want their current products because the competition has a better offering. The only option Intel really has is to hype future products because it has become common knowledge that their current line up can't compete with AMD. The hype you're hearing is more of an effort to stop the exodus to AMD, it's yet to be seen if that will work.

Intel is selling many parts at a loss

...and you know this how? Are you privy to the details of Intel's cost/unit? Yes, they've cut prices but they may have had plenty of room to do so and still make money.

Re:Osborne Effect

[edxwelch]

Sorry, but I don't buy that. The price cuts are because Intel is desperate to win back marketshare from AMD, before it's too late.
The Conroes are also being offered at a huge discount
A Conroe 2.40GHz/4M costs $316, half the price of the equivalant Athlon (the Athlon 64 X2 4800+ cost $645)

Re:Osborne Effect

[Credible]

I'd love to know where you 'facts' that conroe yields are low and that there are supply problems come from. Intel has its issues - but last I heard - yields were good, and intel don't tend to have supply problems. What you say about the Osborne Effect is true - but it applies doubly to AMD. It's clear that Intel has changed gears, and as a large company is out to "smite" the smaller competition.

Re:Osborne Effect

[Anonymous Coward]

And Bush is evil. And the RIAA sucks. And DRM sucks. And Sony is evil. So is SCO. But Google is good. So is AMD. BTW, I'm typing this in Firefox on Linux. There, that should about cover it, now mod me up.

Re:Osborne Effect

[qbit23]

A lot of wishful thinking presented as fact.

1) "which are comming (sic) in July/August of 2006" Woodcrest launches June 26. But you already knew that.
2) "no one wants their current products" In which alternate universe?
3) "Intel is selling many parts at a loss" Dicounts don't imply loss.
4) "Conroe...has bad yields" Source: AMD message board?
5) "Conroe...will not make up a significant portion of Intel's shipped CPUs until the end of this year." Just plain false.
6) "Conroe based chips will be 20% of production;" You just made up that 20%, didn't you?

Re:Osborne Effect

[Anonymous Coward]

And by the time they're shipping in volume, the 4 core Opterons are due.

Their new motto is "Leap Ahead." They'd better shake a leg.

After several years of incremental improvements, it's starting to get fun again.

Re:Osborne Effect

[Overly Critical Guy]

Hardware analysts are saying AMD will lose its edge when Core 2 comes out, which beats AMD on speed and power consumption. It will be interesting to read opinions around here since there's a definite anti-Intel slant.

Re:Osborne Effect

[Anonymous Coward]

Well, I agree with the first three things anyway. Absolutely. The rest is debatable, but I guess I agree with most of those too, for the most part.

Re:Osborne Effect

[kimvette]

I don't think so.

There is ALWAYS something faster coming out. Anyone who has ever bought a computer can tell you this.

It's been happening for years.

Intel is in the lead, AMD leapfrogs Intel. Intel races past AMD. Mhz is everything. No, instructions-per-Mhz is everything. AMD is faster again. Oops, Intel just released the Super-extreme-hyper-overclocked-needs-a-2-ton-air- conditioner CPU to take the lead. Wait, AMD has a brand-new architecture that uses less power and benchmarks 50% faster at an equivalent clock speed. Oh-oh, Intel has a new chip coming out, AMD gonna die? No, AMD gonna release another CPU which will leapfrog Intel.

It's been going on ever since AMD came out with their 386DX40 and later their 486DX4. They're in a constant game of leapfrog, and I don't think that is going to change any time soon. The difference here is that not only is AMD able to produce their own designs, but they can bring them to market fairly quickly and maintain longer leads without resorting to massive overclocking to do it.

Buy Intel or buy AMD? Buy whatever the best bang-for-the-buck at the moment is, if you OS will run on it, or just flip a coin. In a month or two a faster processor will be out. It's not like 1991 where the CPU you bought was already out for four years and hasn't gotten any faster. New CPUs come out by the month, or at least the quarter, and they're only getting faster more rapidly. Personally, I'm partial to AMD, but frankly Linux support for bleeding edge AMD chipsets lags slightly behind bleeding edge Intel chipsets so I haven't bought AMD in a while. :(

One problem I do see with both architectures: they're turning to multiprocessing to increase performance. This is great because I love the responsiveness of SMP systems (my latest machine was a downgrade in responsiveness- went from a dual Pentium III to a Pentium 4) because it is less likely for a single CPU-intensive app to tie up the GUI, but if you need to run a single-threaded app faster, having 16 processor crammed into one box won't help one iota (aside from handling other tasks). They need to make absolute increases in performance in addition to bringing SMP into the mainstream desktop and notebook. Besides, while many apps are already multithreaded, you still need each thread to be executed as quickly as possible.

Re:Osborne Effect

[kill-1]

In this case the Osborne Effect only applies to enthusiasts. Do you think average John Doe engages in speculation about Intel's new processor architecture?

The reason why Intel is selling parts so cheap right now is simple: they have started to massively clear their P4/Netburst inventory. In the next months it's going to degrade to their "Celeron" line of processors. Noone will want to buy a P4 in half a year.

Re:Osborne Effect

[Anonymous Coward]

Bad yields??? Excuse me? The process that Conroe is being manufactured on is QUITE stable and I ASSURE you as somebody who works at Intel (though I cannot quote yield or ISO numbers for fear of getting fired) the yield on the part is well within Intel manufacturing standards--i.e. REALLY high yield....

I assure you, when conroe actually "launches" the channel will be filled and it will not take "6 months" to do something with them in the market. You are quite ignorant in your assumptions.

Re:Osborne Effect

[pantherace]

Everyone needs to remember that Intel's core chips are 65nm. AMD begins volume production of that next year? So Intel will have an advantage for a time. (Provided both aren't having issues like most people had when starting with 90nm.) It really isn't shocking that they would have lower power consumption than 90nm parts. AMD should worry over the more IPC, however, that may partially be due to the same thing. When this chip gets released, it will probably be top dog. It's not released so at the moment, AMD is leading...

Intel has a good thing going for them with the core chips. However, it's not just black magic to their compeditors, who pretty much started the whole processing power per watts. AMD can and will put 65nm processes into production, and they will tweak the core they use.

(And you will note that unless Intel or AMD has changed TDP measurements, they measure them differently: Intel's measurement is based on average usage (and it can go above that) while AMD's is worst case. (And even there, they do it for *classes* of processors. One of the earlier Athlon 64 revisions was 89W across the board. 1.8GHz to 2.4GHz or something like that... it's certainly going to be lower. Looked at it this way: AMD's value is highly pessimistic, while Intel's is highly optimistic.)

It's all really nice... except I, personally, am not taxing my current hardware that much, and I've got a single core processor (Granted, if I do need more speed, it runs very well on a bit more voltage to go to a 50% overclock, and get about 5C hotter). Faster processors are nice, but there's a point where you just go: It's fast enough. 95+% of the time it is clocked DOWN to it's lowest setting. Other things are a lot more important to performance now than the processor.

Re:Osborne Effect

[turgid]

AMD processors have a better I/O architecture (hypertransport) and scale properly in SMP systems.

Intel processors are only faster if you are running small, tight loops that fit in L1 cache. AMD processors are faster when doing more multitasking.

Re:When's this gonna stop?

[purpledinoz]

I beg to differ. Intel and AMD pretty much own 90% of the PC/laptop market, and 90% of those boxes have Windows installed. The PC world revolves around these companies. Also, what do you mean "growup"? Did I make a poop joke?

Re:Yawn

[purpledinoz]

What are you talking about? This is standard PR. Intel does it, AMD does it, ATI does it, nVidia does it...

Consequences

[Visaris]

While you have a decent point, I don't think you remember history well. A few years ago, Intel created the "NetBurst" architecture. They ended up with a 31-stage pipeline, and Intel said that they would take CPUs to 10GHz by 2005/2006. Intel even had press events and demonstrated a 5GHz NetBurst P4.

Where are our 10GHz Pentium 4s now? Does anyone really remember Intel's promise? Does anyone really care?

Intel can fudge benchmarks and make crazy promises all they want. In the end, everyone seems to still love them because they are Intel. I don't think Intel will suffer at all, whether they lie or not.

Re:Consequences

[NutscrapeSucks]

Where are our 10GHz Pentium 4s now? Does anyone really remember Intel's promise? Does anyone really care?

Intel did put itself through a whole round of public humiliation over this.

And are you actually suggesting that Intel was lying when they stated they could scale NetBurst to 10Ghz? It turned out to be huge engineering and strategic mistake that's cost them dearly (see recent financial news).

Re:Consequences

[cnettel]

It's one thing to say "in five years, this should scale to umph GHz". It's another to say "look, this is what we are going to release real soon now". If you think that Intel's credibility wasn't hurt in the minds of those (that is, geeks) that at all care about these things, I also think that you are wrong.

Anyway, we'll know in less than two months.

Re:Consequences

[TheRaven64]

I was at a talk a few months back by the chief architect on the NetBurst team, who has since left Intel. He had some quite interesting things to say. One was that the P3 and 4 were great fun to work on, because they were the last chips anyone was allowed to design where performance was the only constraint. Now, power usage is far more important, and it will continue to be so for quite some time (i.e. until we start using some as-yet-uninvented form of magic to make our chips). The other thing he said was that he expected the P4 to top out at about 5GHz. Management told everyone 10GHz, because they thought the engineers were being too cautious. Even he was surprised when it failed to even make 4GHz.

Past vs Future benchmarks?

[thedletterman]

I think AMD should respond by benchmarking their 65nm dual core chip due out 2007 against the Intel Pentium 4.

Re:Past vs Future benchmarks?

[archen]

AMD doesn't seem that quick to hype processors that look good on paper in the same way Intel does. I think this is a fair comparison for the current generation. I find it sort of strange that AMD has been sticking with 90nm for so long, but maybe their getting a better value out of their fabs this way. They've already been able to do some pretty amazing things on the power consumption front with the current tech.

Re:Past vs Future benchmarks?

[NutscrapeSucks]

BS. When AMD was leading up to the Hammer CPUs, they were doing the exact same benchmarketing that Intel is doing now. The truth is that if AMD had anything to show, they would be showing it.

Re:Past vs Future benchmarks?

[tomstdenis]

Intel: 8086, 286, 386, 486, p5, p6/pii, piii, piv
AMD : 8086, 286, 386, 486, Am586, K6, K7, K8/Hammer

Comparing K8 to P4 makes sense. They're both eight generation. The P4 was Intels answer to K7 and the Hammer was AMDs answer to the P3. Comparing Conroe to K8 doesn't make sense because Conroe is a 9th generation part. It'd also make more sense once Conroe is readily available. I can go to a store and buy a 285 today. I can't say the same about Conroe. Wait till the 9th gen AMD processors are out if you want to make this comparison on a technical level [or wait until Conroes are in ready supply].

Tom

Re:Past vs Future benchmarks?

[rthille]

You post makes sense, but I'd got a bit further or a litle sideways, depending on how you look at it.
Basically, I don't care what generation a chip is, I care about what its specs are, including price and power consumption.
Learning about future chips might get me to hold off a purchase, but I'm still going to compare shipping, pricegrabber-able, 'real' chips, not results on engineering samples.

Re:Past vs Future benchmarks?

[tomstdenis]

No rational computer shop will wholesale replace all of their machines for another product line because the competitor is 6 months ahead [or whatever the gap is]. Desktop users may flip/flop but by that logic they could just flip back in six months...

That said, I'll probably be picking up a Conroe part later this year to do benchmarking on. I'd be hard pressed to make the switch completely given that my desktop workstation is um ... a bit more powerful than a single Conroe :-)

Anyways, all the power [or lack of I guess] to Intel. Competition is good, specially if companies like Dell start opening their doors a bit more.

Tom

"..select members of the press" ??

[tkjtkj]

what kind of crap is that? Are they choosing the reporters who have been 'good lil boys and girls' and refrained from being critical of Intel? Do they believe we're idiots??

Re:"..select members of the press" ??

[2nd Post!]

Aren't we? Isn't the average IQ only 100? Isn't 50% of the population between 90 and 110, with 75% of the population below 110?

So it seems par for the course to expect everyone is stupid; you'll be right 3 out of 4 times!

Re:"..select members of the press" ??

[jcr]

Are they choosing the reporters who have been 'good lil boys and girls' and refrained from being critical of Intel?

Maybe they're choosing the reporters who have enough readers to be worth inviting.

-jcr

Re:"..select members of the press" ??

[buswolley]

Slashdot IQ Poll-check!

Re:"..select members of the press" ??

[tkjtkj]

yup.

Re:"..select members of the press" ??

[tkjtkj]

If that is so, how arrogant! They believe they know the readership statistics of every reporter?? They know which of them might be writing some important book, or lecturing at a major tech show?? No, i dont think so.. and neither do you,im sure. They want good press, period. And theirs is a self-serving approach to dealing with the public. But then, again, corporate america isnt famous for truthfulness.

Re:"..select members of the press" ??

[jcr]

how arrogant!

Excuse me? YOu have a press event and can only invite a limited number of people, who are you going to ask to come? If your answer is some random collection of bloggers who may or may not write a book that matters someday, then you'd be a completely incompetent PR agent.

-jcr

Not even a Current AMD System

[Anonymous Coward]

Why did they benchmark it against a DDR1 socket 939 FX-60? Of course the memory performance is going to be slower, as will be a few other things. What about the AM2 FX-62 with some DDR2, that would be a more compareable benchmark. Not to mention that you can buy an AM2 processor and mainboard and have it in your living room right now, whereas this new Core2duo stuff is still way off in the distance. Intel must be very frightened of AMD if they need to drum up this much business right at the launch of AMD's AM2.

Re:Not even a Current AMD System

[Lead Butthead]

Why did they benchmark it against a DDR1 socket 939 FX-60?

Standard marketing practice. I remember a RAID controller vendor had the gull to disable tag command queuing on their competitor's product before running benchmarks against their own, then declare their product superior.

Re:Not even a Current AMD System

[JorDan Clock]

Actually, if you were to do some research, you would realize AM2 processors have, at most, a couple of percentage points gain over their Socket 939 counterparts. Athlons simply don't benefit realistically from DDR2 yet. K8L may improve that, but for the moment using a 939 Athlon FX against a Conroe is just fine for benchmarking.

But don't take my word for it. Check out what Anandtech has to say.

Re:Not even a Current AMD System

[smash]

I think you mean "gall"

Do Intel choose which benchmarks are run?

[edxwelch]

With the previous Conroe benchmarks, Intel specified which benchmarks could be run. I wonder if this is also the case in this review, because noticable absent is the SYSmark benchmarks.
It is standard practice in biased tests to only include the benchmark where your product does well.

Unbiased review, isn't it?

[d.3.l.t.r.3.3]

Why to compare a new generation of CPUs against and overclocked setup of one of the current best chips? Why not taking the test against an AMD chip with a similar NATIVE clocking or, maybe, against a similar X2, since the core duo is a dual core chip? And once again, it's fair to compare two completely different architectures by the sound of their clocking? Nobody remembers the ruckus that Intel did when AMD introduced a better architecture that simply ate Pentiums at equal clocking? Just a pointless piece of hype.

Re:Unbiased review, isn't it?

[JorDan Clock]

The FX-62 is a dual-core processor. In fact, here is the details from AMD themselves.

Re:Unbiased review, isn't it?

[d.3.l.t.r.3.3]

Ops... My fault.

benchmarks are still good

[abigsmurf]

despite being on an intel optimised PC. It's the equilivant of being on a fresh install of a PC built by a talented system builder with a high budget. There may be comparissons between a lesser AMD system but most of the benchmarks so far have been ones directly comparable to other benchmarks performed by regular sites on systems they've built. I'm looking forward to the conroes because they seem to be bringing the price of CPUs down again which to me seem to offer less bang for buck than in the 2800xp days. Have they released a price list for every upcoming CPU yet?

I'm just waiting for Sharikou, Ph. D to show up.

[MSFanBoi2]

Soon this clown will show up spouting all his anti-Intel pro-AMD rhetoric.

He makes the worst of the Mac, Linux or Microsoft fanboi's look a touch out of the ordinary.

Both the Woodcrest and Conroe have shown time and time again in INDEPENDENT testing, to be quite a bit faster than any of AMD's options. I've been testing a Dell 2950 with Woodcrest and it simply smokes the HP DL385 dual core setups time and time again in both SQL 2005 (mixed size transactions) and anything else I throw at it, most of the time by 30-40% real world numbers. Other testers have seen much the same.

Intel just pulled a Microsoft. Microsoft was caught napping by Netscape. Intel was caught napping by AMD. It won't happen again.

Kinda Disagree, Kinda Agree

[mfh]

I'm on the fence with you on this one. Sometimes a vendor will try and bombard the market with misconceptions just to create enough confusion that the mainstream market will be hypnotized by long enough for the company to make budget. Is that the case with Intel? Not as much as it used to be, because people are onto the way they used to do business. ViiV really hurt Intel. Did you know that the media center ports on HP Media Center systems won't do audio alone? They need the video cable too! Half of my customers buy the AMD Media Center and the other half get the ViiV. How many of them do you think blame ViiV and not HP and not Microsoft?

Where do you point a finger?

It might not even be related to ViiV, but the customer doesn't know it. Their negative experience ties in and compounds the disatisfaction. Because they knew they bought a ViiV -- they remember that.

There are a lot of forces working against Intel... including bad design on systems running it. So anything else might tip the scales one way but certainly not the other...

I'd quit while you were ahead.

[MSFanBoi2]

In other words, you don't know that Intel is selling under cost...

For Intel to fully prodcue one Pentium 4 processor at 90um costs them about 24 bucks start to finish...

So now that AMD has annouced price cuts to compete with Intel's price cuts, are they selling under cost as well?

Re:I'd quit while you were ahead.

[fastgood]

to fully prodcue one Pentium 4 processor at 90um costs them about 24 bucks

If the new Intel chips in the $150-$300 range really were so monumentally better than the existing line, they'd drop the Celerons and continue making the 2.8-3.4GHz P4s as their entry level.

However, the new Intel is going to try increasing market share by focusing more on better marketing than just better engineering.

In related news

[Anonymous Coward]

Why Does Slashdot Have This Strange Typing Problem With Lower-/Uppercase Letters In The Titles All Of A Sudden????

Can't you write like a sane person any more?

Re:In related news

[thePowerOfGrayskull]

Because in a title, it's correct for all first letters to be capitalized (except for first letters of articles).

Isn't this still just hype?

[dtjohnson]

Used to be that Intel (and others) would give samples to reviewers, under a non-disclosure agreement, a short while before the launch date and then the reviews using real hardware would be posted on the day of the launch. Now, we are getting more and more 'reviews' of hardware that is not actually being sold or even shipping or even with a firm launch date. This just seems like reviews of vaporware hardware designed to hype the product rather than an actual measurement of real performance of a product that can actually be purchased. There have now been so many hypware reviews of 'Conroe' that the real reviews of the actual product that ships will probably be ignored as 'old stuff.' If the product is really good, it will be just as good on the day it ships and doesn't need a lot of pre-shipping hype. If the product is not really quite as good as the reviews of pre-shipping samples suggest, then the reviews are nothing but misleading. Either way, why not just wait until the product actual is shipped to start posting reviews of it?

Re:I'm just waiting for Sharikou, Ph. D to show up

[DeafDumbBlind]

I haven't seen any 'independent' tests yet where the machines contained identical hardware besides the MB and CPU. With that said, the Core Duo looks like a winner, and it looks like Intel will have the performance crown for the first time in a few years.
It will be interesting to see how AMD's K8L part on 65NM will do, but that thing won't ship until next year.

Still, IMHO, Intel will never recover to its pre P4 glory days. Before the K8 chips became smash hits in the enterprise, NO ONE seriousely considered AMD for a business app. That mind share has shifted. AMD is now viewed as at least Intel's Equal in terms of performance and stability.

Your analogy with Microsoft is a bit flawed. MSFT was able to leverage its OS monopoly to crowd Netscape out. CPUs are Intel's cash cow. Intel can't give them away for free to crush AMD.

It's great to see Intel back in the game, it should drive down the prices of the high end desktop CPUs and force AMD to innovate more quickly. They've been napping for the past couple of years as well...

Benchmarking

[Ice Wewe]

This is an unfair comparison. The benchmarkers are using a Socket 939 FX-60, which uses the older, slower DDR RAM. If they really wanted to compare the next generation CPUs, they'd get a FX-62 Socket AM2 (940 pin) CPU, and test it with the same DDR2 RAM that was used in the intel test system. I think they'd find AMD would be able to match the performence of the Conroe platform.

Thinking is...

400/667

0.599 = 60%

That means that the DDR RAM is only 60% as fast as the DDR2 RAM they're using, and will only pass 60% as much data. (Assuming that they're Dual Channeling)

Re:Benchmarking

[popsicle67]

Don't expect any relevant numbers out of Intel. If their shit could stand up in the real world everybody and his cousin would have the damn thing to run tests on so Intel could sit back and say "Yup we're back". Instead they are only giving out numbers from contrived benchmarks because AMD pwned 'em again.

Re:Benchmarking

[Anonymous Coward]

Why don't you AMD fanboys shut the fuck up and look at some inofficial numbers?

http://www.xtremesystems.org/forums/forumdisplay.p hp?f=59

There's a start where you could actually look for yourself instead of covering your ears and shouting "LALALA, Intel must be faking it". Stupid moron.

Re:Benchmarking

[Anonymous Coward]

I think they'd find AMD would be able to match the performence of the Conroe platform.

Geez, being a fanboy must be a real pain. Conroe does 270fps avg in FEAR vs. AMD's 190fps and you think this could be matched with just faster RAM? Then how come AM2 is such a dud with no real performance gain?

You need to get out of your cave, dude.

Re:Benchmarking

[Anonymous Coward]

Maybe the intel fanboy should take a chill pill. Calm down dude! That's like flamebait material! All I'm saying is that the review is biast towards intel. Actually, what I guess I'm trying to say is, AMD pwns intel, once again!

Re:Benchmarking

[JorDan Clock]

Actually, the performance benefit of DDR2 with Athlons is far less than 60%. Less than 6% in most cases. Check out this little comparison.

I'm not an Intel fanboy or anything. I just think people need to be more objective. Intel won this round. Maybe AMD can make it up with K8L? Until then, however, I'm going to be buying myself a Core2 system.

FUD? WTF?

[Anonymous Coward]

Why are 3/4ths of the stories on slashdot user tagged as "FUD"? There is no attempt to spread fear, uncertainty, or doubt in this article. It's merely stating that, with the hardware available to them, HotHardware's tests showed the Conroe as having a significant advantage over the FX-60. The article even repeatedly makes the caveat that they had limited access to the hardware, and were only able to run specific benchmark. In other words, HotHardware was remarkably honest.

P133

[x_MeRLiN_x]

Will these ourperform my P133?

Re:When's this gonna stop?

[Anonymous Coward]

Almost.

These might be your actual views, but I (and most) can read it as troll only.

Wait to the contest starts

[Enrique1218]

before you declare a winner. Media seems to have this fascination with projecting results before the contest has even started. Political election, sports, and technology. AMD or intels next generation are not shipping yet, everyone is projecting Conroe as a clear winner. Even if Intel Conroe will be the better processor, are they expecting to keep yields up next to AMD's. It would be a different story if Intels chip is 15% better but 30% more expensive. Lots of things can happen, companies don't talk about openly about all aspects of their product strategy. They tend to hide weaknesses and counter strategies.

Re:Wait to the contest starts

[Anonymous Coward]

Media seems to have this fascination with projecting results before the contest has even started.

I think that statement could be modified to read as follows and still be as accurate:

The typical human being lives his life in his mind. He does not see the world as it is but as he/she thinks it is. One way in which the typical human being carries out this "living life in his mind" phenomena is to spend a great deal of mental time and energy thinking about the future, to the detriment of his/her ability to cope with normal everyday events in an effective non-stress creating manner.

This article paid for by Intel.

[gd23ka]

Here, have a short, weird story instead:

It was dark when Madchen came home into her appartment. While she fumbled for the light switch
she heard the hoarse breathing of Goober. Dammit! Madchen thought, why can't he go out on his
own. With the lights on Madchen let out a sigh of dismay. "Oh dammit Goober! Look at what you
did to the couch!". Goober's brown eyes locked on to hers and he asked "What Goober Couch do?"
his voice a mixture of squealing tires and the dark bass sound of a v8 motor. "Why look! You
got all the gravy over the couch!". Goober turned and looked at the couch and without another
word began licking the gravy stains.
Goober preferred cold meals but Madchen had to have some hot food at least once a day. She headed
of the kitchen still voicing her upset but Goober chose not to listen even though he could hear
her as if she was sitting in the living room with her. He knew he had upset her again with
something trivial as food stains on the couch. When Madchen had her dinner fixed she came back
into the living room and flopped down next Goober on the sofa, immediately digging into her
chicken with steaming hot rice. Goober asked, "Goober have chicken?". Madchen silently glared
at him and replied "Goober have chicken bones: IF he asks for them in a full sentence".
"Oh Goody! Good! Good!" Goober exulted rowing happily rowing with his arms and making the odd sounds
he always made when he's happy. Madchen was wolving her meal down ravenously. Goober said
"Goober had rat today!". Madchen stopped. "That's I had a rat today, Gobber. Where did you find a
rat, goober?". Goober looked at her and said "Not here. When dark Goober had fun on street, found it,
hit it hard and then eat it. It is quick but very fun. You should do it too. Good for you too"
Madchen laughed quietly and went back to her food. Goober had his good sides and she didn't call him
Goober for nothing. He had introduced himself as "Good Bear" the day she first met him. The Army
research people didn't mind her calling him what she wanted as long as she shared the appartment with
him and didn't get eaten. Goober didn't mind either. He sometimes eyed Madchen's shapely thighs
speculatively but until now she always bought him off with steak.

Oh right so I ran afoul of FF's password feature

[gd23ka]

once more and posted this under my real name. Not the first time this happened and I'm not really upset about it because like I said this article is just another advertisement. Remember the many Windows Vista articles we had the past two weeks, more ads disguised as content. Anyway... enjoy my confusing little weird story, I already spotted two or three typos and grammar mistakes in it too but who cares. Just one thing for the real "geek" inside us all: Check out Sun's T1000/T2000 servers. Each of its SPARC CPU's has eight cores and they deliver a wallop of computing power.

For All You Nay-Sayers...

[vostok4]

Everyone here is constantly saying "Oh its an Intel system, built by an Intel team, vs. an AMD system, built by an Intel team... I'll trust the reviews when independant people get them."

If you looked a little you would see, that there are already lots of people with the Conroe in their hands. And it has shattered every PI, 3DMark, world record there is. We are talking about 10s 1M SuperPi runs, and if you know anything about that benchmark you will know that is absolutely crazy. Why not read some forums, like XtremeSystems or more specifically some benchmarking threads where the world record was broken on air w/ Conroe, but now its under LN2 for some other people (including coolaler) and is holding the world record.

non-Intel benchmarks...

[YesIAmAScript]

From independent equpiment:

http://anandtech.com/cpuchipsets/showdoc.aspx?i=27 71

You're right, these show lower numbers, more like 20% in gaming and still over 15% elsewhere.

Re:I'm just waiting for Sharikou, Ph. D to show up

[Anonymous Coward]

Where *isn't* he? All he's got these days is ego. Just ignore him.

DDR2 isn't higher latency...

[YesIAmAScript]

Well, actually it was in the beginning. But it hasn't been for quite some time.

You're getting messed up because DDR latencies are measured against the SDR clock, while DDR2 latencies are measured against the DDR clock. This is why you see things like 2.5 clocks on DDR measurements and you never do on DDR2.

Anyway, when you see a latency 5 clock DDR2, that's the same as a 2.5 clock DDR. Except that the DDR2 is very likely running at a higher clock rate and thus the absolute latency is lower.

I don't think the Intel tests actually cripple the AMD. That's somehting you can get sued for. Very likely they just control the test parameters very closely (selecting datasets, etc.) and used the settings that favored them the most. AMD also suggests tests to the hardware testing sites that make them look good. Some of them use them, too, so when you see one of those "independent" tests it still is making AMD look better than perhaps is representative.

ATI and NVidia do the same.

Re:DDR2 isn't higher latency...

[Agripa]

I posted asking about this before (http://slashdot.org/comments.pl?sid=173501&cid=14 439092) but never got a real answer. There was a discussion on RWT that touched the DDR2 versus DDR latency which other then the possibility of delaying the command cycle in DDR2 brought up that the DDR2 Intel controllers (and by extension the AM2 controller) were actually using 2 x 64 channels instead of 1 x 128 in order raise the efficiency. I never followed up on it completely as I have no DDR2 motherboards to play with. Going through the Intel north bridge technical documents should resolve it.

Re:DDR2 isn't higher latency...

[Agripa]

I was curious enough to dig through the Intel north bridge data sheets for the 875P DDR and 975X DDR2 memory controllers. AMD's documentation is not as easy to digest and I was not able to find any detailed data on the socket AM2 processors.

The Intel 975X interleaves two 64 bit channels while the 875P uses both DRAM banks as one 128 bit channel. Naive benchmarking will not show any advantage in fetching simultaneously from separate interleaved banks and Intel's FSB throughput is too low to take real advantage of it anyway. AMD should be in a better position as the AM2 memory controller should be able to use the full DDR2 bandwidth but it may be using the separate DDR2 banks as one with half the open pages and a poor total burst fill size. Adding a second core to a design already prepared for it is probably easier to justify then making major changes to the memory controller when you have a major derivative design coming out in the next year.

The FX-62 sales don't appear to be lots...

[YesIAmAScript]

Actually, from what I can tell, it still isn't available.

It's not on newegg. Pricegrabber only lists it at one place, and they say it comes June 30.

http://www.monarchcomputer.com/Merchant2/merchant. mv?Screen=PROD&Store_Code=M&Product_Code=120987&AF FIL=pricewatch&NR=1

It appears AMD preannounced the FX-62, at least in terms of availablity to other than 1st-tier vendors.

If FX-62 really ships June 30th, that'd only be 1 week ahead of Conroe, thus making your "ask AMD for next year's cores" comment extra stupid.

AMD's biggest efforts right now in making maximum hay from their AM2 releases before Intel makes their announcements, even to the point apparently of preannouncing their chips. And all this despite the approximately 0% speedup the AM2 chips have given clock-per-clock.

Re:I'm just waiting for Sharikou, Ph. D to show up

[Overly Critical Guy]

You're right; independent benchmarks show the Core 2 chips beating the pants off AMD's latest offerings. Even hardware analysts are now saying AMD will lose its edge in the server and desktop markets and will fall behind. However, don't expect anyone to listen around here--years of anti-Intel articles have shaped a certain viewpoint that only AMD is good, because they're the underdog. But that underdog is about to be beaten with, frankly, superior technology. The anti-Intel trolls can continue their rhetoric, but when the Core 2 chips are everywhere, AMD will just plain be behind in both performance and power usage. The Core 2 chips can even easily be overclocked to over 4Ghz, while AMD's latest struggles to reach 3Ghz. They're not even on 65nm yet. So it's pretty clear Intel will be ending the year with a definite lead.

Re:I'm just waiting for Sharikou, Ph. D to show up

[Anonymous Coward]

But that underdog is about to be beaten with, frankly, superior technology. The anti-Intel trolls can continue their rhetoric, but when the Core 2 chips are everywhere, AMD will just plain be behind in both performance and power usage. The Core 2 chips can even easily be overclocked to over 4Ghz, while AMD's latest struggles to reach 3Ghz. They're not even on 65nm yet.

Cue the apples and oranges. It's nice to see that the Intel Kool-Aid doesn't discriminate. Good luck with that.

This is all silly

[modernbob]

90% of the PC's bought are sold to people who don't know the difference between Ghz and dual core. The hardware so far outstrips the software's ability to use it that it makes these comparisons kind of lame. I think both Intel and AMD need to shift to a new way to market themselves . With the exception of gamers and people that live tech no one really cares if the machine does something several milliseconds faster than something else. If they can write a letter and send email and the machine doesn't run painfully slow they are happy. Lets face it, this is where the vast majority of users of PC's are!

Re:This is all silly

[Dark_MadMax666]

Company I work for for example asks that every new config we buy for research dept is benchmarked against set of in house application.Every % matters for statistical apps which may run several days -That was exactly reason we jumped intel's ship 3 years ago (they suck hardcore at FP) .

  There is market for performing CPU ,fact that there is billions of morons out there who waste/do not need computing power changes nothing.

Re:This is all silly

[Anonymous Coward]

I was under the impression that us, being the 10%, might notice a little. If you look at the title, this discussion isn't about Intel's market strategy, it's about the benchmarks.

Re:I'm just waiting for Sharikou, Ph. D to show up

[P3NIS_CLEAVER]

I think this is just the business cycle for the two companies... once the street catches up with Intel's newest offering AMD will have something newer and better on the horizon.

Re:I'm just waiting for Sharikou, Ph. D to show up

[Dark_MadMax666]

While if benches are true (but so far I saw anandtech woodcrest bench rigged in favor on intel - they used motherboard without NUMA support for AMD which crippled opteron's bandwidth ) it will be definitely a hit to AMD.

          Now as of 65 nm vs 90nm - it means that AMD has potential to increase their current chips stats,while intel already exhausted it (45 nm is nowhere near) ,combined with K8L it may as well bring them 40-60 % performance increase over their current 90nm K8 . Just jump to superior technology gives performance increase and less power consumption. Intel already jumped. AMD will in q4 2006. So I think Conroe will be superior either till q4 2006 ( 65 nm) either till q1 2007 (K8L on 65 nm).

"Benchmarks Strong"

[Anonymous Coward]

Shouldn't that be "strongly"?

Where are the real benchmarks?

[Anonymous Coward]

Enough with the toy, synthetic and vendor optimized benchmarks. Give us men real benchmarks and lots of them to ensure the results are consistent across the board and not due to specific vendor optimizations for we know certain vendors aren't very honest. Not to pick on Intel but check out the following url's as examples http://www.swallowtail.org/naughty-intel.html and http://www.technewsworld.com/story/49156.html.

Both 32 and 64-bit versions of both if available
Matlab, Mathematica, Maple
OpenSSL encyrption
Softimage, Maya, Modo
Oracle, MSSQL, Postgresql, Mysql

Feel free to add to the list. AMD looks like they have the architecture superiority while Intel's new Conroe will bandaid their architecture with a larger cache. Curious to see what happens to performance when cache is filled.

Re:I'm just waiting for Sharikou, Ph. D to show up

[malkavian]

Just a quick prediction: Intel will trump AMD by pushing the Conroe out. AMD will cut prices, so their chips fit nicely into the price/performance bracket that the market will allow them to sell at. Consumers win.
The speed freaks will buy Intel for performance, as it gives the absolute fastest. The people that want good performance at a price that doesn't bust the bank will probably buy AMD (and that's the way it used to be with AMD & Intel for quite a while).
Then, AMD will likely make modifications in their next architecture that speed their offering up past that available to Intel. And the speed freaks will buy AMD. Intel may, or may not put their chips in the 'bang for bucks' ladder, as they still have the name to leverage.
Then Intel will release another architecture, and so on. This is called competition. It's good. It makes sure that both sides don't do a Microsoft, and get the chance to turn the PC chip market into a monocultural wasteland, and have the market stagnate. Others entering the fray will also be good (how long until China get their own offerings out?).. A heterogenous system is more robust, and results in better product at the end (that's what happens when you get real competition).. That's what's happening in the chip market.
Really, the only thing to look at in the big picture is that the general public win. We get better prices for better tech. The speed crown of the day is a passing thing, for either side.
As long as I get good tech for a good price, I consider it a win. Whoever gives me that price, Intel or AMD.

No one seems to remember that AMD is 64bit

[Anonymous Coward]

No one seems to be pointing out the most glarily obvious fact: AMD's chips are 64bit and Intel's AREN'T! For me, if I was building a new system, then AMD is the way to go. Running Linux 64bit shows many performance inprovements and even 2003 Server 64 does as well. These new chips from Intel don't have their EM64T, so for a long term prospect for many, it is a dead end. ALL of AMD's chips are now 64bit. Even the $298 eMachine I saw at WalMart last night with the Sempron 3100.

Re:No one seems to remember that AMD is 64bit

[heinousjay]

This will sound like a troll, but I don't intend it that way.

Why would Intel worry about satisfying a tiny fraction of the market? Most people are perectly happy with 32 bit CPUs right now. 64 bit is useful in many cases, but it's more than the general market currently needs. People that need or want 64 bit CPUs have the option to buy them, they just can't have the performance benefits the Core 2 offers yet.

Re:No one seems to remember that AMD is 64bit

[Anonymous Coward]

Intel Core 2 processors will feature ***EM64T***, Virtualization Technology, and Execute Disable. The release will also introduce LaGrande Technology, SSE4, Enhanced SpeedStep Technology, and Active Management Technology (iAMT2).

Extended Memory 64-bit Technology (EM64T) is Intel's implementation of AMD64, a 64-bit extension to the IA-32 architecture. See the AMD64 article for architectural details.

RTFM

Re:No one seems to remember that AMD is 64bit

[Anonymous Coward]

That's exactly what everyone said about the move to 32bit. Windows v3.x and Dos were 16bit and only took advantage of some of the aspects of the 386. It took many YEARS before MS even shipped a full 32bit system for the desktop, and that was XP. So, almost 10 years went by with a 16/32bit hybrid system of v3.x/95/98/SE/ME. However, all the chips of that era were 32bit. There are many benefits of 64bit computing, and as usual, no one in the reviews is pointing this fact out. There have been 64 bit systems since 1992 for the microprocessor world when the Alpha was released. The Xeon based parts and most of the fastest new chips will go to 2 markets: Businesses and Gamers. And, a lot fo games have a 64bit client now as well. Sure it's going to take time for this market to mature, but people are holding on to their systems longer. That system that most people buy today will still be in use 3-4 years from now, and having a 64bit furture is a plus. Come one, it wouldn't have taken them that much to do it. The Core chips is based on the P3 and Athlon designs anyway. The P4 sucked and it's death is way overdue. When a 2.4Ghz P4 is only barely about 2-3 times as fast as my 500Mhz P3, you know it's a waste.

i havent "rtfm" but ...

[Anonymous Coward]

if intels conroe beats amds ass on a per hertz
NOW .. well that's just wierd. so a 2.8 gHz intel-NOW beats
a 2.8 gHz amd which after all beats a 3.8 ghz intel, oops-i.
seems the HZ-wars aren't over yet. (wondering
what a 3.8 Ghz a-m-d might do ...)

went out of their way?

[YesIAmAScript]

They compare their fastest against the fast AMD available. Did you try to buy an FX-62? It's just not out there. I'm not sure why. But either way, Intel probably just couldn't get hold of the FX-62, or at least didn't go the extra mile to do so.

But saying they went out of their way to not use an FX-62 seems a stretch.

do you have some references for this?

[YesIAmAScript]

Why you think Intel is selling at a loss?
Why you think that Intel is having bad yields?

Yes, Conroe will only be 20% of Intel's production. But that's very likely sufficient. Conroe is their high-end chip. They already have Core Duo out and the remaining P4s and P4Ds (even 65nm P4s).

High-end chips make up a small amount of the market. 20% seems like plenty.

I agree that finding Core 2 Duos might be difficult for a while. But then again, when I bought my AMD X2 4200+, it wasn't easy to find, and ultimately I had to get it because the 4400+ was impossible to find at retail prices. And this was a almost two months after it came out.

Nice to have competition...

[ponos]

The next year belongs to Intel, but AMD is likely to catch up very fast. First of all, the transition to 65nm is quite close and it will definitely allow a decent speed boost. Even if that isn't enough, a new "improved" Athlon64 rev. G core is just around the corner (and I'm not talking about K8L).

The advantage of Core 2 Duo is indeed spectacular (which is a good thing) but the competition from AMD will be fierce. I don't think AMD will soon surpass Core 2 (at least not before the K8L or K9 has matured) but the difference won't be as prononouced as it is expected to be right now. In the long run, Hypertransport and the headroom allowed by socket AM2 (current Athlon generation is definitely not memory speed limited, which has been a rare occurence in the processor world for as long as I can remember) will be an important strategical advantage. Other AMD technologies like "accelerator" CPUs linked via Hypertransport or Z-RAM may or may not prove to be significant in the real marketplace. The Core 2 processor definitely has higher IPC but at the cost of huge caches, to keep him running. Furthermore, the Core 2 is already at a 65nm technology, so don't expect power consumption gains in the immediate future. If the Core 2 is already at 2.93 GHz (P4 stuck at 4 GHz), I wonder how far it can scale with the current process technology.

P.

Re:I'm just waiting for Sharikou, Ph. D to show up

[qval]

oh, you mean like microsoft wasn't caught napping by real or google?

I don't think AM2 FX 60 is widely available yet

[MojoStan]

I was wondering why they did not use an AMD AM2 CPU that uses the same DDR2 RAM.

I also wondered, but my quick froogle search for available AM2 (DDR2 800) FX 60 CPUs brought up nothing. On the other hand, Socket 939 (DDR 400) FX 60 CPUs (which Intel used) are widely availablle. Therefore, the fastest available AMD Athlon platform that Intel could buy was Socket 939 FX 60, which uses DDR 400. It seems Intel made it even more fair by overclocking the FX 60 to approximate the performance of the unavailable FX 62.

Maybe Intel could have "borrowed" an AM2 FX 60 from an OEM, but I'm sure AMD wouldn't have allowed this.

Re:I don't think AM2 FX 60 is widely available yet

[(H)elix1]

I've already seen the AM2 x2 5000+ in a personal rig (not mine) last week, so I got to believe Intel could get their hands on one of the DDR2 using CPU's. I'm just wanting to find out if fast DDR2 makes a big impact. Not seeing a huge boost with the slower DDR2 over the AMD chips using DDR1.

Fanboy denial

[unassimilatible]

I strongly suggest you check out xtremesystems.org forums, where the benchmarks have even AMD fanboys like yourself are ready to switch to Intel. These conroes have already run SuperPi 1M @ 10.xx seconds. Out of the box they will do 1M 15 seconds. An FX couldn't do that on LN2.

A $316 Conroe can blast a $900 FX, and Conroes have been overclocking to 3.8+GHz on air.

AMD is doomed, unfortunately, which means Intel will rest on its laurels again.

Hey fanboy, get a clue!

[Anonymous Coward]

Check out xtremesystems.org and tell me that FX' aren't dead as Elvis. Not even close. Sell AMD short.

Wow, you know nothing about RAM

[unassimilatible]

No, the DDR2 is not faster than the DDR because it runs at much higher timings! Learn something before you post. Conroe destroys FX. Read some freaking enthusiast sites like xtremesystems.

Big whoop.

[fernandoh26]

So wait a minute, Intel's brand-spanking new processor can outperform AMD's 1 year + old processors? Big whoop. I would've been genuinely impressed if Intel had released these simultaneously to AMD's processors, now I simply feel that my intelligence has been insulted by Intel acting like it has to prove its new processor is all that and a bag of potato chips. Everyone's heard of Moore's Law, everyone expects a processor released over a year after a competitor to be better. What's Intel going to do when AMD releases a mega mack daddy processor that blows them out of the water in a few months?

SuperPi benchmark

[truth_revealed]

If you believe that SuperPi is representative of real-world CPU performance, then you will be very happy with the new Intel chip. I personally care more about multiprocess/multithreaded server loads using largely integer math with working datasets several orders of magnitude larger than the on-chip memory cache. Under such a scenario you want very fast memory transfer speeds between CPU and main memory. I have not seen any Conroe benchmarks related to that. I wonder why. We'll see how well the chip performs when it actually ships and is tested under conditions not dictated by Intel.

Against others?

[FluffyWithTeeth]

I don't suppose anywhere's got these chips compared to Pentium Es and such? It's nice seeing them in gaming performance, but I'd like to know how they compare against the current big chips in video encoding and such.

Re:Intel's a bit wierd now. CORPORATE AMERICA IS

[Anonymous Coward]

110% agreed, and this is the case with corporate america in general, especially in tech companies that actually produces goods based on hard sciences - you can't have a blindman at the wheel. How many of you work in MIS/IS/IT dept.'s here, and have "managers" who can't even do the job themselves, or understand it as well as you do? I would wager a LOT of you do... this is a problem. Too many "boardroom stooges" and not enough technically saavy mgt. exist out there, but that is what you get when 'fratboys' (rich boys) put their pals in jobs they do not belong in.

why the talk of bandwidth?

[YesIAmAScript]

What was this to address?

I was speaking of latency, and you speak of bandwidth and burst sizes.

Neither bandwidth nor burst sizes affect latency.

The indications of AM2 performance definitely go against your idea that AMD is more prepared to take advantage of DDR2. Their performance gains are nil, when the latency stayed about the same and bandwidth increased significantly.

Re:why the talk of bandwidth?

[Agripa]

My point is that Intel DDR and DDR2 chipsets use very different memory configurations (one bank of 128 bits versus two interleaved banks of 64 bits) and that the latency tests comparing them may not be accurately measuring what is really happening. Depending on the specifics of the test, they could be comparing one bank 128 bits wide versus one bank 64 bits wide running at twice the clock speed in which case fill rates and latency should be almost identical. If the test uses a block size larger then the burst size on one DDR2 bank, both interleaved banks will be used and the DDR2 fill time could be almost half of the DDR time except the Intel FSB does not have the transfer rate to support it. Were the latency tests for the first available word or for a cache line fill or what? In a lot of cases, it is not clear what was actually being measured and none of the tests I have seen made any attempt to simultaneously fetch from separate interleaved areas of memory where the 128 bit wide DDR system would perform relatively poorly.

Read what I said about AM2 performance again. They SHOULD be able to make better use of DDR2 then Intel because of lack of FSB limits BUT something is preventing it. I was not able to find any AM2 detailed specifications online and the Athlon 64 BIOS programming guide does not cover the AM2 yet. My guess is it is treating both 64 bit banks of DDR2 as one bank 128 bits wide (Like the Intel DDR controllers, the Athlon 64 DDR controller does not support interleaving two 64 bit banks that I could tell.) and using the smaller DDR2 burst size. This will give almost identical latencies for very small transfers but make the command overhead percentage larger then if it used the larger burst size and interleaved banks.

Uh, no, Core 2 is 64bit

[xevious1]

I don't know where you guys get your information. Core 2 most definitely is 64 bit. I won't even start on whether 64 bit is really even useful right now. Tony

I'm sure the interleave size is 64 bits

[YesIAmAScript]

And since a cache line is 32 bytes on an Intel (16? no matter, either way it's 128 bits), that means all cached accesses utilize both channels and will have the increased transfer rate. It's apparent in the benchmarks anyway.

I do agree that configs can alter first word latency more than full burst latency. But I honestly have trouble understanding what interleaved 64-bit banks means versus 128-bit banks. Classic interleaving meant simply pulling from both banks at once, which would produce the same results as a 128-bit bank in all but edge cases.

I don't really agree AMD's integrated memory controller means they can use DDR2 better. In fact, I'm not really sure what AMD's integrated memory controller is really good for. It might mean lower latency, or it might not. It definitely locked AMD into DDR longer than they should have been with it. And the problems with routing the traces all the way from the CPU past the Northbridge (due to board layout issues) contributed to the problems trying to get 4-DIMM DDR working (which would have been easier with DDR2 also). It also definitely means all DMAs have to go up to the CPU and back down. I'm not crapping on AMD for their memory architecture, their NUMA stuff is fantastic. But I'm not sure that putting the memory controller on CPU is inherently better than on the Northbridge. I think that in low-cost chipsets the video accelerator has to go on the Northbridge to be cost effective, and that means even more memory accesses that would have to be routed up to the CPU and back down if the memory controller isn't on the NB. Because of this, I see a tough road for AMD in the non-gaming and laptop part of the market.

I don't get why command overhead goes up when you change the configuration. Does the controller not issue commands in parallel to the two banks in one of the configs?

Re:I'm sure the interleave size is 64 bits

[Agripa]

Classic interleaving was used because the DRAM arrays were unable to supply data fast enough for the busses of the time. Interleaving allowed you to use one bus at twice or even four times the frequency that one DRAM array would support. Internally, DDR and DDR2 chips actually act very similar to this when they send busts of 4 or 8 transactions across one bus.

Each cache line fill is 64 bytes. The processor actually picks which of the 4 sets of 16 bytes it wants from the 128 bit DDR bank which then returns that one first and the other 3 sets of 16 bytes afterward in a total burst of 4 transactions. DDR controllers like the Intel 875P and the Athlon 64 DDR use two banks of 64 bits in lockstep. Each bank is presented with exactly the same commands and addresses and each transaction is 128 bits in size.

With the Intel DDR2 controllers, you have the option of interleaving both 64 bit banks such that the 64 byte cache fill lines come from alternate banks. Now the processor uses a burst size of 8 instead of 4 to get 64 bytes from one 64 bit bank. Since DDR2 can run at twice the clock speed of DDR, it can fulfill the entire request just as quickly as a DDR bank of twice the width. Since the other DDR2 bank is free, the DDR2 controller can execute 2 different memory requests simultaneously as long as they are both in different banks. This could be very handy on a multiprocessing system where 2 processors could alternately access each bank simultaneously. Now if someone would just make dual core processors . . .

AMD's advantage is that they are not limited by FSB throughput (Intel's fastest FSB is slower then 128 bit wide DDR2-800) and the benchmarks do show very high DDR2 memory transfer rates. Unfortunately, it appears they only modified the older Socket939/940 memory controller enough to support DDR2 and may be still using 1 bank of 128 bits with a burst size of 4 resulting in larger command overhead and fewer available pages. I do not have access to the AM2 design or programming specifications to verify this however.

You could think of AMD's integrated memory controller as just moving part of the north bridge onto the CPU. The result is better latency between the CPU and memory (one less off chip bus to deal with) while increasing the latency between any I/O devices and memory. As you point out, it was probably most important to them in facilitating their NUMA architecture.

None of this answers why latencies were so much worse with the initial DDR2 chipsets. Some of it was that the initial DDR2 cores were running at a lower frequency then the DDR cores. The cores themselves were different do to at least supply voltage changes making the established DDR cores more comparatively mature. There are some minor command differences as well but I have not seen an accurate enough description of them to judge.

okay, I'm getting it now...

[YesIAmAScript]

I see. You say that 875P draws from A and B at the same time: AB AB AB AB. That makes sense to me, since if you do the math with DDR400, it was the only way to maximize the "800 MHz" FSB on the P4-3GHz.

But that DDR2 chipsets might draw from A and B alternately, sharing the bus. A B A B A B A B.

The 2nd appears slower, since it takes 8 clocks, but the clocks are so fast that both fill the FSB bandwidth equally.

And you're saying that even though the 2nd is "fast enough", if they went to the first system again, they'd be able to double throughput on an Athlon, although an Intel would be FSB-limited.

Right?

So I have to ask, doesn't system 1 require twice as many data lines on the mobo as system 2, since system 2 timeshares them? If so, can't you just look at the socket or mobo to see which is being used where?

I do agree that what AMD did was move the NB on-chip. I just don't see the sense in that, given that the NB is not going to go away, in fact, it is going to become more important as the graphics accelerators move there. I just don't see AMD's solution making financial sense in the bulk of the market in the long term. It does make sense where latency is paramount (servers? render farms?) and where you're gonna have a graphics chip too complex to ride on the NB (gaming), but not in the mainstream "office" type system or laptop, which together make up well over half the market.

Re:okay, I'm getting it now...

[Agripa]

In the first case, the 875P and the socket 939/940 Athlons fetch from memory like AA AA AA. There is only one bank which is 128 bits wide.

In the second case, the 975X and other Intel DDR2 memory controllers fetch AB AB AB. They use both banks of 64 bits each simulaneously.

For the 875P and the socket 939/940 Athlons, the 64 bit DIMMs are arranged in pairs to make a single bank of RAM 128 bits wide. The memory controller fetches 4 lines of 128 bits at a time to fill the processor's cache. Note that there is nothing inherent about DDR memory design that requires them to use 1 bank of 128 bits. They could also have used 2 interleaved banks of 64 bits like the current DDR2 controllers but there was not enough advantage to doing so until DDR2 came out. Either Intel was planning ahead for dual core consumer processors or there were performance reasons to use 2 seperate banks with DDR2.

For the Intel DDR2 controllers, the 64 bit DIMMs can be accessed separately but simultaneously allowing the fetching of data from 2 non continuous blocks of memory as long as they are in alternate banks. Since each bank is half the size of the single 128 bit bank in the earlier DDR controllers, they fetch 8 lines instead of 4 but do it at twice the speed. It is like having two completely separate memory controllers with the odd addresses on one and the even addresses on the other except they alternate every 64 bytes. Both systems (1 bank of 128 bits and 2 banks of 64 bits) use the same number of data, address, and clock lines except for the interface differences between DDR and DDR2. There is actually an interesting comment about this in the Athlon 64 technical data about how the address and clock drivers for each 64 bit DIMM mirror each other allowing 4 total DIMMs without external buffering.

Intel would need a FSB-1600 to keep up with their DDR2-800 controllers. AMD does not have to worry about it since their memory controller presumably communicates with the crossbar on their processor at least that fast. There are actually some indications that a single core on the socket AM2 processors can not fetch data fast enough to saturate the memory controller. There may be an unresolved design issue or it could simply be that a single core can not issues loads fast enough. I have not seen any dual core tests to confirm anything though.

I just don't see AMD's solution making financial sense in the bulk of the market in the long term.

Hehe. Linus Torvalis was just involved in a discussion like this over at RWT. It makes PERFECT sense for AMD to have done this. It allows their fabs to produce a single basic design for multiple markets instead of dividing their effort into multiple designs. The mass production of one design lowered the cost enough to make up for any extra expense in the consumer market. This same effect has allowed x86 to overtake higher end markets from the bottom over time. Presumably Sony and IBM (or at least Sony) are looking to do the same thing to x86 with Cell starting with the console market but I think Cell is too specialized to pull it off not to mention Sony's past successes in introducing new standards and products. They have become expert in snatching defeat from he jaws of victory.

Incidently, I think one of AMDs first mistakes was not allowing cheap dual processor systems using socket 939 Athlons with ECC and non registered DIMMs. They could have sold twice as many processors to the high end consumer market but there are many reasons including fab capacity that this may not have been an option. Their upcoming 4x4 initiative looks like a step back to this.

I was ahead of the curve on this...

[YesIAmAScript]

I'm not piggybacking on Linus.

Hehe. Linus Torvalis was just involved in a discussion like this over at RWT. It makes PERFECT sense for AMD to have done this. It allows their fabs to produce a single basic design for multiple markets instead of dividing their effort into multiple designs. The mass production of one design lowered the cost enough to make up for any extra expense in the consumer market. This same effect has allowed x86 to overtake higher end markets from the bottom over time. Presumably Sony and IBM (or at least Sony) are looking to do the same thing to x86 with Cell starting with the console market but I think Cell is too specialized to pull it off not to mention Sony's past successes in introducing new standards and products. They have become expert in snatching defeat from he jaws of victory.

Long term, that doesn't work. I do see what it did for them now, I don't see it continuing to work. Because all the "single solution" stuff won't reduce the number of chips on the board. An Athlon solution has CPU, NB, SB and graphics accelerator (possible NB and SB together). An Intel solution now has just CPU, NB and SB (SB is almost vestigal). AMD simply cannot match on price when they have to make more chips, no matter how broadly they apply the chips to the market. I don't see the GPU going onto the CPU, so they can't ditch a chip that way. AMD seems like they think they can put the GPU next to the CPU (opening up HT on the board), but that doesn't save chips either.

Let me ask you this, could a Mac Mini have been made with an AMD solution?

I've heard rumors Intel wants to go to FB-DIMMs with Conroe. We'll see what happens if they do that. That could really hurt them in the same way you mention with Athlon non-registered ECC.

Re:I was ahead of the curve on this...

[Agripa]

If anything I am but it was more a case that I agreed with his position in the debate about the costs of AMD's integrated memory controller and whether is was appropriate for single processor consumer systems.

It is not a matter of reducing the number of chips on a motherboard for AMD or necessarily matching Intel's price. It was a matter of having a solution that would work were Intel was weak (2 to 8 cpu workstations and servers), using one design to cover both that market and the single cpu consumer market, and using greater economy of scale to cover the extra research, development, and production expenses. I agree with the notion that the mass production of one basic design paid for the added cost of the integrated memory controller which in turn allowed them to best Intel in the 2 to 8 processor server market.

    If anything, the AMD design actually uses fewer chips then the Intel one since the north bridge as such no longer really exists. AMD chipsets still have one but it no longer has a memory controller and serves mostly as an I/O bridge for the graphics subsystem. If it was not for SLI motherboard configurations, there would only be a south bridge with all of the I/O hanging off of it. I am not sure if they are thinking of this with their upcoming 4x4 systems but the possibility exists to place the graphics processor into one of the HT connected external slots making their current north bridge completely vestigial. I do not see the graphics chip vendors wanting to split their wares between PCIe and HT voluntarily but the rumors about AMD and ATI have been suggestive. These kinds of things look to be at least a year away making them at the horizon of speculation.

As for Apple, it was more important that they switch away from PowerPC to x86 while they are strong then the choice between Intel and AMD. Once that was decided, Intel was the obvious choice just based on mitigating the risk as much as possible. I do not know any technical reason a Mac Mini could not be made using an AMD processor and chipset but why make things more complicated then necessary when you are transitioning to a new architecture? I am sure the Apple skunk works has some AMD systems running Mac OSX as a plan B just in case. Apple with Steve Jobs at the helm has been nothing if not prepared for the unexpected future. I am reminded somewhat of Intel when they made the life or death switch from memory production to microprocessors.

I've heard rumors Intel wants to go to FB-DIMMs with Conroe. We'll see what happens if they do that. That could really hurt them in the same way you mention with Athlon non-registered ECC.

They have working north bridge memory controllers supporting 4 banks of FB-DIMMS already but the RAM is currently too expensive when compared to DDR2 for the consumer space. That design is really for systems requiring more then 4 to 8 GBytes of RAM where DDR or DDR2 become very difficult to layout when your memory controller is on a single chip. FB-DIMMS are also poorly suited for the portable market do to the extra power required.

I do not know why AMD segmented their market between the single and dual processor system instead of between the dual and quad processor systems. If they had done it the other way, it could have been like having dual core systems 2 years earlier. Presumably their bean counters knew what they were doing but the 4x4 design they recently showed makes it look like either they have reconsidered that initial decision or things have changed enough leading to a different outcome.

since you don't cover it, I will

[YesIAmAScript]

No, the Mac Mini as-is could not be done on an AMD solution.

It would simply cost too much because you would need to purchase an additional graphics solution chip. Apple could do that before because they were using an outdated processor (G4). Buying an up to date, dual core CPU plus GPU plus NB plus SB just costs too much (even if you can fit it in there) to make the margins Apple demands.

The complexity of chips does affect their cost. But even a simple chip with a lot of pins (an AMD NB) costs almost as much as a complex chip with a lot of pins (an Intel NB). the difference is here that the Intel NB also is a GPU in mainstream solutions.

So, I say again, just count the chips. AMD is looking at a 4-chip solution, Intel has it down to 3, and the extra chip isn't even a chip one, it's a graphics chip, which even in basic form costs $20. You also have to add the cost of graphics RAM (unless you get a chip with embedded GRAM, which costs more but saves money overall compared to external GRAM). Intel's solution leverages the onboard RAM, which it can do easily since it has the RAM directly attached to the NB.

Putting the GPU in an HTX slot or HT socket doesn't save money in the same way that not having a separate GPU does. And you can't remove the NB as long as the machine has any slots or PCI or PCIe peripherals (like Firewire interfaces). The NB in an AMD system acts as an HT to PCIe/PCI bridge.

And that is why I doubt Apple is seriously looking at AMD right now. I'm sure they got Mac OS X running on an AMD, but I doubt they even built any AMD hardware prototypes. AMD can't meet Intel on cost structures on the Mac Mini or Mac Book (non-Pro), and these make up a lot of Apple's sales. They also can't meet it on XServes. All of these are because all of these machines can usefully use Intel's on-chip (950) GPU. Apple could probably not use AMD for their other laptops either given how hot they already run with Core Duos, which provide Athlon X2 4200+ performance at just over half the power/heat, they wouldn't make it with an Athlon X2.

So Apple could use AMD in the iMac and in the towers, but nowhere else.

And that is almost certainly why Apple went with Intel. And it's why the AMD+ATI rumor made a lot of sense to me. If AMD wants to sell to Apple and other cost-conscious manufacturers (presumably Dell amongst them) they need to have a solution that matches Intel on cost on the low end. And licensing a graphics engine from ATI (probably one of the ones they got with the BitBoys acquision) would do it. From a competition standpoint, I don't really want AMD and ATI to merge, but companies see these things in a different light. They want to maximize their profits, not my choice.

Mentioning 4x4: there's another thing AMD's onboard NB locked them out of. Since the CPU interfaces directly to the RAM, that means you have to have a RAM subsystem for each CPU chip. That means you have to install at least 4 DIMMs in a machine. In a world where a 1GB (DDR400) DIMM costs only 30% more than a 512MB DIMM, making the user install 4 512MB DIMMs instead of two 1GB DIMMs is a killer. That means they couldn't do affordable MP unless they got both cores on one chip (which they did very well).

Additionally, if you have seen what Intel showed at their conference (IDC) this year, they showed a Core Duo plus North bridge plus South bridge plus CPU power controller on a single chip (obviously multiple chips on one package). That included a GPU of course, since there is on in Intel's NB. Apple could buy that from Intel and make a smaller and cheaper (although multiple chips on one package don't save nearly as much money as putting more on one die) system than they can with AMD. Why? Because AMD doesn't make a GPU or a NB (they do make an SB, Geode). That means anyone who wants them together has to work with manufacturers to get them all put together themself. That's difficult to do if you don't have very high volumes.

All of this just underscores AMDs difficult position in the laptop and mainstream markets in the next couple years. They're not stupid, I'm sure they see it. But they'll need to not just understand it, but also fix it before they can compete with Intel in these markets.

Re:since you don't cover it, I will

[Agripa]

I am not convinced it is possible to resolve Apple's use of Intel versus AMD processors except possibly in the case of laptops which would naturally lead to them using Intel over their whole product line (at least to start) even if they determined AMD would be a better choice for some configurations. Even including all of the public pricing information available and different configurations, Intel could have made an offer too good for Apple to refuse and there is enough evidence they have done such things is the past not to factor that uncertainty in. Apple made the safe bet even without technical considerations.

there's another thing AMD's onboard NB locked them out of. Since the CPU interfaces directly to the RAM, that means you have to have a RAM subsystem for each CPU chip.

There is no requirement for dual Opteron systems and by extension the proposed 4x4 system to have memory on each CPU. At least one of the early dual processor boards only attached DIMM sockets to one CPU and the other CPU had no local memory. Benchmark tests on that system as well as others configured with memory on only one processor actually showed very little performance degradation. I tend to look at RAM expansion first so would naturally never advocate buying a motherboard restricted in that way but it does work.

The only difficulty I find in AMD's position is their chipset support. Any of my Intel systems have excellent driver support for everything included in the Intel chipset. I have several nVidia test systems and I find between poor driver support, inadequate documentation, and flaky IDE/ethernet/PCI controllers that they are difficult to justify.

AMD Fanboys Settle Down

[moogle10000]

Okay all you AMD Fanboys - Settle down. I'm one too... I do really like AMD's DESKTOP processors. However, for the portable/low power market the Intel Dothan/Yonah just wins hands down. I never thought I'd like Intel processors again until I got my Dothan 1.8gHz laptop. Amazing. If the Intel Core 2/Conroe/Merom/Woodcrest is anything like the Dothan/Yonah processors in terms of speed and responsiveness, AMD will be in serious trouble. Furthermore, I've noticed that the C2 (Core 2) has a lower TDP than the comparable AMD processors (maximum of 65W) - This would deal another serious blow to AMD. On a closing note: The difference between DDR2 and DDR on AMD systems is inconsequential. Intel won this one - *I* don't particularly like it, but they did. And if they can actually make a product that can compete with AMD, good for them. (Netburst -- Barf!)