Slashdot Mirror
SpeedStep On Your Desktop - Intel's Prescott-2M
Kez writes "Intel's Prescott core has undergone a few changes, and the latest version - Prescott-2M - includes new features, one of which is Enhanced SpeedStep technology. Given the jokes about the heat that the Prescott gives out, Intel had to act. It was inevitable that a power (and heat) saving technology such as SpeedStep would find its way into desktop PCs. HEXUS.net has an article looking at the new Prescott-2M based Pentium 4 660 and Extreme Edition 3.74Ghz CPUs, examining their new features and performance."
only fry one egg at a time whilr doing anything other than staring at the login screen?
The cesspool just got a check and balance.
Slashdot
Dupes on your desktop
I look forward to heating my house with my new Intel processor!
I'm a big tall mofo.
...why should it be? AMD used to have heat issues, and they managed to find a way around them. To me, this almost seems like cheating on Intel's part. And what are they going to use as the excuse to "step down" a processor I paid to have run at a certain speed? With a laptop, they have the "save battery" excuse, which is a valid one (but still over-ridable by the user) - what's the desktop equivalent? The fact that they can't cool their processors is definitely not a good "excuse".
Dear Intel,
Thanks for taking such a GREAT APPROACH to your heat problems. I can't WAIT to use one of these new processors in my desktop, only to watch my whole computer DROP IN SPEED as I am an hour into Doom 3. I don't know that I can speak for everyone, but the whole design efficiency thing is overrated anyway. I simply can't live without the noise of a jet engine in my case. Keep cranking up those Mhz and I will continue to have my cpu throttled everytime I do something useful.
You're the best,
Sarcastic Consumer
"When life gives you lemons, don't make lemonade. Make life take the lemons back!" -- Cave Johnson
These look like some nice enhancements to the current Prescott core, especially the 64-bit part. My question is whether or not these will work with current motherboards, such as my Abit AG8 with a 915P chipset. If so, that would offer a nice upgrade path.
Now my CPU can slow down if I'm working it too hard.
This has got to be the best idea since hoola-hoops!
If you don't know what AltaVista is (was), get off my lawn.
No more mucking around trying to get that gas stove to just the right temperature!
not bother to upgrade.
I mean the whole idea of a faster CPU is to get more work done. So, why buy one and then let it idle most of the time?
Hurry up and post the new Paris Hilton story so we can chat about it! Thanks
Will this new processor work with current 915P and 925 chipset motherboards?
Hey, if our brains do a nice enough job using lots of parallel instruction "dumb" processors, why are we so obssessed with ultra fast only-a-few-instructions-at-a-time single processors? I think the whole approach of the cell architechture is the right way to move forward.
Please spare me the "the brain can't multiply 100000*1234555 fast enough" argument. We can have the best of both worlds: complex single "cells" (unlike brain cells) repeated many many times for parallelism.
There are two kinds of people in the world: Those with good memory.
Anandtech has another article up, but it emphasizes the increase in L2 cache and the effect this has on performance.
There's no live stream of the Article, due to interfearnce from fans in the background. Can't think where they were coming from ?
"Sweet llamas of the Bahamas !"
Intel was wise in renaming the throttleable version.
The original P4 is the SUV of CPU's.
This is a free program that lets you control speedstep in XP (something you could do with windows 2000). I have my laptop set for full performance when on AC and Max Battery when unplugged.
The interesting this about this architecture is that it is near-identical to what was implemented on the 8255 using PicoJava. Sadly, it never took off and at best is just a footnote in history.
Can I interest you in an article from someone who knows WTF they are talking about?
AnandTech
I don't know what the Hexus kid was on - but I feel safe trusting my reviews to people who have trouble writing big words!
From Page 2: Being LGA775 CPUs, the new processors all look the same. Being press samples that I get the privilege of testing, they're also unmarked with any meaningful information bar the slightly exciting Intel Confidential. So I draw on them. Not quite the Mona Lisa in miniature, mind you, rather an idea of what it is. Any retail example you purchase will be umblemished with my scriblings.
See this posting.
IRC: Grounded0 @ IRCnet. "I was lucky get into computers when it was very young & idealistic industry" -Steve Jobs
"Given the jokes about the heat that the Prescott gives out, Intel had to act" Geek Jokes are such an important driving factor in improving technology- jokes and Slashdot posts, of course! ;-l
Is it anything like AMD's Cool N' Quiet? Which is great. My Athlon 64 3400+ Clawhammer is currently.. 95~ degreesat 1.0ghz and when I launch a game or something that needs the speed, it clocks back up to 2.2ghz.
Not a Twitter sockpuppet... but I wish I was.
The poster is apparently clueless as hell - AMD is already shipping CPUs with variable frequency for desktops. Google for Cool'n'Quiet.
It's just unfortunate that while the advantage of having a stepping processor on the desktop will cut down on the heat in my building's rooms, it will also cut down on the processing power at the same time.
;)
Honestly, most people won't ever notice the difference since what they'll use it for is word processing and spreadsheets. They don't need the gawdoffal (tm) power these computers have now. In fact, the only thing driving the continual 3-4 year upgrade cycle is poorly written code and programs so huge that we'll never use all the features.
Which makes me think that maybe we should call some moritorium on "new" software, perfect what we do have, use lower power processors that are simply more efficient, and stop buying pcs altogether.
NAH!!! Just kidding! I mean, who really wants to stop this computer-centric commerce and put intel and microsoft out of business anyway... (looks around and sees slashdot's mac, linux and other os users...)
Uh, easy guys. It's only a joke...
Who's worried about their CPU heat at idle or whilst word processing? It's the heavy processing that needs full porcessing power that manefests heat problems.
Free Mac Mini Yeah, it's
Imagine if every PC currently in use had such a CPU, that could reduce clock speed (and thus power consumption) when load was low. The power savings globally would have to be massive (assuming everyone wasn't running SETI@home).
Dan East
Better known as 318230.
I have to admit that I am kindof torn about this. I have an older Athlon chip that use to get QUITE hot (more recent heatsinks solved this). However, one of the things I use to prefer about Intel chips is that they'd stop working before they'd fry themselves. This may not seem important, but it is to me... I had an Athlon 1800XP that burnt out itself and the motherboard within 2 seconds of startup because the heatsink/fan were underrated for the job (despite it's claims to the contrary). Now I'm a little more willing to pay for something that will "step" itself down instead of dieing.
Jeremy Logan's Website.
Another Intel 600 series Pentium 4 review from Hardware Analysis.
I'd guess that its mostly a problem with the English language, rather than technology knowledge. Here's another better written article http://arstechnica.com/news.ars/post/20050220-4632 .html
A deamon that automagically sends support@intel.com a standard mail when CPU is 100% at low speed.
This mail looks as follows:
"Dear support team,
If you received this mail, it's because of a malfuction in your Prescott CPU #22354432, which reached a idle state of 0% during 68 ms while still being in low frequency mode. It had a temperature of 56C
feel free to ignore this mail as you ignored the 122563 previously sent by this deamon, and just as you ignored the need of a solution for CPU heat problems.
Yours sincerely,
The Intel bugger deamon."
*squeak*
AMD's version called PowerNow is similar to Speedstep but I hate it. My laptop (Compaq R3000) There is no supported AMD app that allows me to set the CPU (AMD Athlon XP 2800+) to full. It always stays at 700MHz even if I try to crank it up. I have it dual booting XP and Gentoo. Even doing an emerge world doesn't see it go up. Doing a google on it seems to show others are having the same problem with the AMD64s as well.
The only way I was able to adjust it was a third party tool that adjusted the clock multiplier and that was only in XP. CPUFreq and powernowd don't seem to work with it.
there's nothing new about Jokes on Intel's power consumption. I remember back in 1994, a cartoon on PCMagazine, which portraited a chef. The title: "Cooking with the 586".
I still wonder why it took Intel 10 years to realize there was a problem.
Remember in the early 90's when many of you were still in grade school and the TCM based mainframes with their 400psi water chiller pumps were beginning to make way for the CMOS era? And we heard that a 10 CMOS CEC could easily replace a 2-3 TCM CEC because even though each one was rather slow and low powered they could gang them together and heat would not be a problem? We all chucked our TCM mainframes, got rid of all the chiller machinery with hacksaws and went on our merry ways.
Well it looks like the prognostication for a Brave New World was a little premature. It looks like we'll start to see the return of complex and expensive water chillers yet. Not the homemade black tee shirt and Krispy Kreme version but real, large, complicated chiller pipes that are built right into the CPU chip.
From TFA: ... that rides the same 266MHz bus (1066MHz affective)...
Sounds like a lot of affective vibrations while riding.
CC.
TaijiQuan (Huang, 5 loosenings)
How about reducing the frikkin' power dissipation? Not to Pentium-M levels, but to Athlon 64 levels at least. People used to joke about Athlons, and now look at them, AMD fixed the issues without running them at 300MHz (that's the speed of Pentium-M processor in my notebook as I write this).
An example: multitrack audio with native plug-ins; I could imagine getting part of the way into a session, then heat build up makes the CPU throttle back just in time to spike the processor on a recording pass.
And what about render farms, or any kind of parallel processing? Imagine having to allow for wildly varying thread execution times because of temperature ("node 42 runs 5% faster than the others because it's near an open window")...
It strikes me that this processor is set to be the replacement for the Celeron rather than the P4ProMkIIRevCExtremeHT* (or whatever the hell Intel's flagship is called these days...Intel's naming system leaves me Plentinumb). Think about it: it's faster on paper than the current Celeron line and power managment rather than heatsinking allows lower cost construction, which makes it ideal as a cheap replacement office (or low-end home) machine where full-time processing power isn't critical. Since its unlikely anyone would recommend such a processor for big iron, that leaves a nice marketing niche for the Xeon. It could also be an attempt to make people use more efficient processors, one way or the other: less power when power isn't needed, better processor architecture where it is.
"...that needs full porcessing power"
Porcessing (v): The conversion of waste materials (notably sawdust, cotton waste and chicken-shed sweepings) into pork products. See also Hormel Foods, SPAM. (Some typos are too good to resist making fun of).
*This "HT" thing in all the Intel ads...that does stand for "High Temperature", right?
I am running Gentoo now and I have had a lot of problems with it, moreso than any other Linux distribution I have used. Try another distro and perhaps your CPU will run at the right speed.
Intel has apparently not posted SPEC numbers for these processors, and in fact seems to avoid publishing official SPEC numbers for non-Xeon processors. By contrast, AMD does post SPEC numbers for the FX-55, and the Opteron 252 results were available the day the chip was announced. The comparison between the latest Opterons and Xeons is none too flattering for Xeon although the 2MB cache should help the SPEC FP numbers quite a bit. The problem for Intel is that P4 still consumes gobs of power and produces a lot of heat even when it isn't doing anything. By contrast, the Athlon 64 3000+ (90NM) that I'm typing this on maxes out at about 65W, which is roughly the P4's idle power consumption. This machine torches the 2.6GHz P4 machine I have at work at compiling and running Java programs (of course, I'm running Fedora Core 3 here, and Win2K there so it isn't apples to apples). It is hard to see how Intel is going to cool 2 such cores on a single die whereas AMD shouldn't really have a problem. Note, that I'm not particularly an AMD fanboy. I have a couple of Dual Celeron boxes and a Dual PIII box, but Intel took a very wrong turn when they went the P4 route, and I don't see anything that indicates that they are getting back on track. The multimedia performance is nice, I gues, but realistically, how many users spend the bulk of their time encoding video?
It is? I have a first generation Prescott 2.8E socket 478, and a Gigabyte motherboard. It's been running at 50C lately...how would I go about turning this on?
From reading the articles mentioned by previous posters it seems pretty clear that the best desktop cpu's are the AMD64 90nm CPU's. Assuming you care about power/noise/heat, this is.
The articles I'm referring to are:
From AnandTech
http://www.anandtech.com/cpuchipsets/s
From the Tech Report
http://www.techreport.com/reviews/2005q1/
Actually what's more interesting than the SpeedStep thing, is the fact that the 600 series uses 25W less power than the 500 series on full load ( that's something when you consider that there is 1M more memory on the new chip).
Up til now Intel's 90nm process was a huge failure because of the heating problems and forced Intel to abandon their plans to hike speeds above 3.8GHz
I bet these processors will be used in the Media Center PC's where noise and heat are a big issue, and performance is not. Nobody wants a constant hum coming from their av rack.
Do any current, on sale Intel Motherboards support this chip?
To lock down (or increase) the cpu frequency in XP, I got clockgen at cpuid.com. You can adjust the voltage and frequency in there plus you can create shortcuts using command line options that will change accordingly.
I go into cpu0 and the folder is empty. I suspect I have something wrong in the kernel but I have enabled Athlon/Duron powernow as well as cpufreq in the kernel and nothing seems to be working.
However I suspect that since the laptop chipset is NForce3 I may want to do the AMD Athlon64 Powernow as well. Some of the R3000 laptops are using the Athlon64 but mine has the Athlon XP-M in a 754 pin
package
Let me know if you can help me offline and I can get you my email address.
Is there any way to throttle down a P4 without speedstep? I've got a 3.2ghz P4 which requires a screaming fan to keep cool despite the processor claiming it is using 1% capacity. If I'm just browsing the web is there any program out there that can slow down a regular P4 chip?
Introducing Microsoft Vacuum 1.0 The first Microsoft product that doesn't suck.
Regretably, as my machine is a Duron, I doubt I can go much further to help. I don't have one of the Athlon 64 machines, I merely lust for them.
I would just suggest slapping every PowerNow option to "Yes" in the kernel and double-checking that you are rebooting to the correct one (no offense intended, I screw this up all the time). If that doesn't work, I'm out of ideas, as that Worked For Me.
Although, come to think of it, you might also try a kernel version or two back; compiling a kernel on your XP-M 2.xGHz machine is probably just a wee bit less painful than compiling a new kernel on my 500MHz-locked Duron. I try to avoid that when possible. I'm probably nuts for using Gentoo on this thing, but I so love the rest of the distro...
I will try it (that was my next step but wanted to see if there was anything else)
I wouldn't have got this laptop except for the screen (1280x800) and the CPU combo plus it was only $1000 CDN as it is a refurb. The machine is in immaculate condition.
The big stinkers for this laptop though are:
1. battery time (maybe an hour and a half and the battery is definitely brand new)
2. need to use NDISWrapper for the wireless driver but it works
3. PCMCIA is giving me grief as it sees a card in the slot but no info comes back when queried
4. Setting up the video took a while due to it isn't a normal screen resolution but google and linuxquestions.org had a lot of info as did the R3000 mailing list archive.
IIRC, they won't throttle automatically. I think they just shut off. But that's enough to prevent damage.
I rarely criticize things I don't care about.
Strangely enough, although it recognizes my cpu as "AMD Athlon(tm)XP Processor 2800+" Installing CpuFreq support for the Athlon64 rather than the Athlon XP seems to have got Powernowd working.
I suspect the new Athlon M processors are like Athlon64 lobotomized. CPUZ recognizes it as a Sledgehammer but Gentoo AMD64 won't install as it says it doesn't have long support
If it's a 754 pin, AthlonXP-M then I guess it must be a Sempron. There is only one 754 pin Sempron I think (not sure - may be more by now), and it's an Athlon 64 core without the 64 bit extensions. Which would fit what you're saying. Most of the Semprons are old (Thunderbird?) 32 bit cores, but there is at least one which is the new core, but crippled. See what an up to date version of Sandra calls it.
Hope that helps, and isn't completely wrong.
I am giving that a shot right now but, I had looked into it before thinking the same thing but the only Sempron I found with a 754 package had more cache built in than this one.
Very strange
Sandra recognized it as a "M4 Athlon 64 (K8 Clawhammer)" but it said it had NX but no AMD64 Technology
Although CPU-Z says it was 754, Sandra said it was a Socket A which sounds more believable
Somebody elsewhere posted a link to a comparison showing Athlon64's temperatures were somewhat lower or comparable, but for desktop systems, once you get a system that automatically adjusts its CPU speed based on load, the "average" temperature becomes meaningless, and concerns over overheating virtually disappear, because the processor spends a large amount of time throttled down anyway. It'll run full tilt for only as long as you play that 3D intensive game, or for as long as it takes gcc to compile your kernel.
Consider my situation (A64 3200 with the CPU's PowerNow active and the MSI motherboard's "Core Cell" logic controlling the CPU): running KDE with all the fancy graphics on and some cute animations going on in the taskbar, 3 browser windows with most of them having multiple tabs open, kmail open, amarok playing some music, editing in a browser window or in the other likely scenario, editing in Kdevelop or a Kate editor window, or running MC doing file/system maintenance, and yet my CPU load doesn't crack 6%. You really have to make your CPU *work* before it'll even go to full speed.
Temperature is simply not an issue anymore for me, and I really feel sorry for those folks sticking with Intel because they think AMD chips are inferior, or just because AMD chips don't have "Intel" printed on them or whatever.
Haven't heard of the latest models. I won't seriously pay attention till their Athlon64 version that supports PCI Express comes out and will then wait for MSI to make a motherboard to support that, but I'm in no hurry to replace my existing system anyway since, a) It "just works", and, b) Its "fast enough" (for now).