Slashdot Mirror
New Xeon CPU Hot and Underpowered
Kasracer writes "Web Sites The Register and GamePC received several of the new dual-core processors from Intel, dubbed 'Paxville', and ran a battery of tests on them. What did they find? From the article: 'There's no doubt about it, Intel's dual-core Xeons are their most power hungry Xeons to date ... Even when idling, two dual-core Xeons consume nearly 400W of power at any given time, which is amazingly high, even by Intel's standards ... their new dual-core chips (while powerful in their own right) simply are bested across the board by AMD's dual-core Opteron processors. Even worse, the Opterons typically perform much better while running at slower clock speeds and only having half the amount of on-die L2 cache to utilize.'"
The Register didn't test anything. They're just providing a short and more to the point commentary on the GamePC review, who appear to have adopted a more diplomatic tone so as to receive Free Crap To Review in the future as well.
These things are apparent when you Read The Fucking Article, you know? The thing you're supposed to do before submitting? Gah.
The situation (Intel's, not necessarily the difference) should change RSN when Intel's 65nm process comes online. Looking at the huge lead AMD has right now, I don't see how Intel can beat them in both power and performance anytime soon unless they're willing to add a few hundred more pins to their sockets to accomodate on-die memory controller(s). I doubt Intel will do that. I also doubt they'll come back a bit from the 150W/CPU these Paxville chips crank out, so they'll be using the process headroom for higher clocks and/or larger caches.
Ultimately an on-die memory controller is the only way to bridge the increasingly large gap between the CPU and RAM. Intel's managed amazingly low latencies to RAM given that they've got an entire extra bus and chip to run through, but they're still ~50% higher than AMD's. The netburst architecture was supposed to be insensitive to RAM latencies but Intel is not keeping up in the bandwidth department either, and it's clear that these CPUs are suffering from a lack of RAM bandwidth (twice the processing power per chip, but no increase in bandwidth).
High-speed Road Trip (18.000KPH)
(Disclaimer: I haven't read the entire article in depth yet, and I don't know **** about chip design)
/.) promoting the Pentium M as a desktop chip (IIRC passive cooling was possible with a reasonably-sized desktop heatsink).
I assume that this chip is basically derived from the Prescott P4, which (in spite of all the hype) has been considered by some commentators to be a dead-end in chip design and a mistake on the part of Intel.
There are rumours that Intel are now using the Pentium M (ironically, a chip which supposedly owes at least as much to the Pentium III design as to the Pentium 4(*)) as their "reference" design instead of Prescott. Bearing in mind that the Pentium M has been praised for providing performance approaching that of the P4s with *much* smaller power requirements and minimal cooling needed, this wouldn't surprise me. In fact, I've read several articles (including one via
(*) And I don't know if that was the previous-gen P4; it's been said by some that Prescott was different enough to warrant the "Pentium 5" name. At any rate, the Pentium M isn't simply a power-efficient Pentium 4.
"Slashdot - News and Chat Sites Deviant". (Click "homepage" link above for details).
Yup, power consumption figures are for the whole system.
But using roughtly equivalent systems (same power supply, an Antec TruePower so you're looking at an "at or above" 75% efficiency power supply when drawing 200+W from it, a single Raptor 74Gb and a Plextor DVD-RW drive), AMD's Opteron system top at 235W idle (for the 2.8GHz Opteron box) without using PowerNow's power management system (GamePC reports that the total power consumption @idle fell to around 170W using PowerNow) while Intel's 2.8GHz Xeon system chews through 390W idle...
"The way we can tell it's C# instead of Haskell is because it's nine lines instead of two." -- wadler
Can you tow a camper, boat or trailer with your little 1.6L engine? How about hauling 2,000 lbs worth of gear? Can you fit 8 people comfortably in your VW? Believe it or not some of us actually use a full size SUV for something other than commuting in on a regular basis.
Just because you don't use your car for anything other than hauling yourself to work and back doesn't mean everyone else uses their vehicles the same way.
The max speed limit on the interstates in the US is 75 mph, generally 60 mph on highways/interstates that run through cities.
Read this article:
p eak_at_intels_65nm_pentium_4-11.html#power_consump tion
http://www.tomshardware.com/cpu/20051007/a_sneak_
In short, decreasing the die size has allowed Intel to reduce power consumption a bit, but it's also increased thermal density to such a degree that they can't clock them as fast as even the current crop of 90nm processors. Why upgrade to these chips?
It's time to drop the gamepc link. Apparently they don't like the extra page views that /. is sending them.
Federal govt receipts:
1998 : 1.747 B
1999 : 1.857 B
2000 : 2.043 B (Last year of Clinton's presidency)
2001 : 1.994 B
2002 : 1.814 B
2003 : 1.795 B
2004 : 1.926 B
IANAE, but it looks to me that if the tax cuts had not been implemented, there wouldn't have been such a dramatic fall in revenues.... Laffer curve? I'm laughing so hard, I'm begining to cry....
Most Athlon64 motherboards have a feature called "Cool and Quiet" which if you install the proper software, will on the fly underclock your PC. To give an example my cpu is an AMD 3000 which runs at 1.8ghz. Most of the time it runs at only 1 ghz using Cool and Quiet and only steps up the clock speed when needed.
AMD is so far ahead of Intel in so many areas: Cool and Quiet, perforamce per watt, on die memory access, dual core, etc. that is not funny any more. If marketing did not matter so much Intel would be in deep.