Nice one Norde - why don't you go back to the hole you came from. That's the whittiset retort I have *ever* read in a comments section, have you just come here from Digg? Please don't take that as a complement, it was 110% sarcasm.

I'm surprised you have even evolved fingers to type, but it seems your grammar still needs work. Maybe that's just more a sign of your limited grey matter though.

Given your single lined response, I bet you gave up reading after the first line - she did use many big. complicated words like "Bastion" for example. I doubt you could have kept up.

if Sylvie was in fact talking about stoves instead, I mean being a woman and all.

"The reason is that the benchmark really loves cache and Intel has always had larger, faster caches than AMD."

Did you read this someplace, or is it just your humble opinion. I think you should consider another career, a real one, instead of supporting the despicable Intel monopoly. They have become lately more honorable and fair as the Nazis, they truly deserve our support.

Bollocks.

When intel changed its TDP specification from maximum power a chip could theoretically pull, to something it "typically" pulls, it's a "standard".

When AMD does the same thing several years after intel did it, it's called "conjured up".

I agree with other comments that SPECjbb isn't a very good server benchmark; 2008 spec seems more promising though (need a more thorough read).

Bleh this brings back memory... Didn't bother posting before but in the "AMD vs Intel - How Shanghai shapes up" article, an Intel setup without an optical drive and less ram modules was being compared against and AMD setup with optical drive and more modules (same total memory???). I think Intel setup still would win but margin is not nearly as large.

That is my main beef with SPEC, the components are not isolated and measured individually. Or at least have a component set of components (RAM, fans, HDD, optical drives, etc). Is maybe okay for server space but make it useless for consumers unless they have numbers for every system major vendors sell.

I rather just have individual numbers and while ACP isn't perfect, it is certainly more usable. Besides ACP and TDP aren't the same thing! TDP is for thermal requirements while ACP measures energy usage (in a way AMD decides, for now).

SPECjbb is *NOT* a good server benchmark. Those of us who work on JVM internals laugh about how bad SPECjbb is, and how the optimizations we apply to get good SPECjbb numbers often don't apply to real server workloads.

For years AMD compared poorly to Intel, as AMD would supply TDPmax as the TDP, whereas Intel would provide TDPtyp (75% of TDPmax or thereabouts). AMD took a long time to switch to something that was actually a fairer comparison.

Intel might have started incorporating thermal diodes into their CPUs before AMD, but it was only a year or two earlier, and it was so long ago compared to the difference that it is a stupid point to even make. You can argue strongly that the Pentium 4 required these diodes because it was such a heatbeast anyway whereas the AMD chips of the time were only starting on the room heater path.

Oddly enough now Intel has a core architecture that generally runs far cooler than the TDPs they publish, whereas AMD hit their limit mostly - a complete turn around from 3 or 4 years ago.

The SPECjbb benchmarks are not "one of the better SPEC benchmarks". They purport to process transactions but they
deliver transactions in batches (like 500 at a time), do not have high network traffic, they do not create significant amounts of physical disk I/O, they do not generate a high volume of context switches. In short they are the opposite of anything like a real web server or ecommerce workload.

If you think 8 years ago is recent, then sure, AMD has just "recently" added a thermal diode.

uninformed comments don't change it either.

SOS, by Sylvie.

Sour grapes does not a nice Sylvie make.

Bashing AMD does not change reality. so get over it.

The pentium DC, which is basically a Core2 w/ less cache, has a rated TDP of 65W- the same as the regular Core2 with twice as much cache. The actual TDP for the DC is < 55W @ max load (at least from the outlet power draw of my system which comes in @ 80W @ max load).