P4 winning on all fronts
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
-
- Site Admin
- Posts: 12285
- Joined: Sun Aug 11, 2002 3:26 pm
- Location: Vancouver, BC, Canada
- Contact:
a sign of the times... from hardocp---
<br>
<br>--------------
<br>Looking for 3GHz on the cheap?? Want that 533MHz + Quad Pumped FSB for a little hardcore gaming action without breaking the bank, well then, you might want to look at this review here from ***AMDWorld.*** The AMDWorld guys had some pretty good stuff to say about Intel’s 2.26GHz CPU, check it:<!-- BBCode Start --><A HREF="http://www.amdworld.co.uk/p422.htm" TARGET="_blank">http://www.amdworld.co.uk/p422.htm</A><!-- BBCode End -->
<br>----------------
<br>
<br>So a website called <!-- BBCode Start --><B>AMDWorld</B><!-- BBCode End --> reviews an Intel P4 and declares it a clear winner. Hmmm. The other amazing thing: using the stock fan/cooler, they got an overclock to 2.89G w/ just 0.1V higher CPU voltage and max CPU temp of 41C. Given that, it would be easy to make this thing run silent at 3G!!! <IMG SRC="modules/phpBB_14/images/smiles/icon_eek.gif">
<br>
<br>I've been a AMD fan since the early K7 days, but this is not possible to ignore, especially given the lack of cooling progress with the 0.13 micron XP2200+. <IMG SRC="modules/phpBB_14/images/smiles/icon_rolleyes.gif">
<br>
<br>
<br>--------------
<br>Looking for 3GHz on the cheap?? Want that 533MHz + Quad Pumped FSB for a little hardcore gaming action without breaking the bank, well then, you might want to look at this review here from ***AMDWorld.*** The AMDWorld guys had some pretty good stuff to say about Intel’s 2.26GHz CPU, check it:<!-- BBCode Start --><A HREF="http://www.amdworld.co.uk/p422.htm" TARGET="_blank">http://www.amdworld.co.uk/p422.htm</A><!-- BBCode End -->
<br>----------------
<br>
<br>So a website called <!-- BBCode Start --><B>AMDWorld</B><!-- BBCode End --> reviews an Intel P4 and declares it a clear winner. Hmmm. The other amazing thing: using the stock fan/cooler, they got an overclock to 2.89G w/ just 0.1V higher CPU voltage and max CPU temp of 41C. Given that, it would be easy to make this thing run silent at 3G!!! <IMG SRC="modules/phpBB_14/images/smiles/icon_eek.gif">
<br>
<br>I've been a AMD fan since the early K7 days, but this is not possible to ignore, especially given the lack of cooling progress with the 0.13 micron XP2200+. <IMG SRC="modules/phpBB_14/images/smiles/icon_rolleyes.gif">
<br>
-
- Site Admin
- Posts: 12285
- Joined: Sun Aug 11, 2002 3:26 pm
- Location: Vancouver, BC, Canada
- Contact:
yeah, I agree with you that AMD has been good for all. They've got to catch up quick, though. Right now, methinks they've got to improve their 0.13 micron process. Otherwise, t-breds will be stillborns.
<br>
<br>When you consider that the CPU represents only a small portion of the system price, and add the costs of more effective than stock HSF that you might (will) want for quiet cooling, then even the price is hardly a win for AMD.
<br>
<br>
<br>
<br>
<br>When you consider that the CPU represents only a small portion of the system price, and add the costs of more effective than stock HSF that you might (will) want for quiet cooling, then even the price is hardly a win for AMD.
<br>
<br>
<br>
-
- Posts: 7
- Joined: Mon Aug 26, 2002 12:40 pm
- Location: UK
-
- Site Admin
- Posts: 12285
- Joined: Sun Aug 11, 2002 3:26 pm
- Location: Vancouver, BC, Canada
- Contact:
Here's a sobering alternative view... http://www.overclockers.com/tips084/
-
- Site Admin
- Posts: 12285
- Joined: Sun Aug 11, 2002 3:26 pm
- Location: Vancouver, BC, Canada
- Contact:
Beat the Intel cartel that keeps all the majors like HP, Compaq & Dell from using AMD. The big $ is NOT in retail, it is in OEM. How? Ummm... Get Bill Gates to advance you about $5 billion or so for a 5 year war of attrition against Intel & arm yourself with a team of sales/marketing ninjas who play just as dirty as intel while beating P4s, P5s and whatever else in the test labs?
Intel's pockets are just too deep.
Intel's pockets are just too deep.
Well, I have just finished torture testing my 2.26B chip. At first I thought I could get 2.72GHz stable @1.50V, then I ran tests known to stress Pentium4B CPU's with high FSB the worst. AMDWorld admittedly did not run these benchmarks, and probably had a false impression of stability at 2.89ghz, just like many people who contribute to Overclockers.com database:
1) Prime95 and 3dmark2000 (not 2001se) simultaneously, people on Anandtech mentioned how it brought down seemingly stable p4 cpu's (took me down some 50mhz, because of Windows memory corruption after the 3dmark run--Windows became very slow). This happened all the way down to 2.67ghz.
2) memtest86, a self-booting utility--this one really takes the cake... since it boots from a floppy, it gets exclusive control to your system and memory--and boy does it use it! I got errors all the way down to 2.64ghz, and these were purely CPU-related, because memory was set to slow and increasing VCore fixed the errors.
So my highest stable speed is 17.0x154 = 2.62ghz @1.50V. For safety, I'm currently running at 17.0x153 = 2.6ghz, which at default voltage creates about 60.2 watts max heat. I am running a 512mb Corsair XMS3200c2 stick at 408MHz @cas2, which helps get blistering 5500+ Sandra Drhystone scores.
1) Prime95 and 3dmark2000 (not 2001se) simultaneously, people on Anandtech mentioned how it brought down seemingly stable p4 cpu's (took me down some 50mhz, because of Windows memory corruption after the 3dmark run--Windows became very slow). This happened all the way down to 2.67ghz.
2) memtest86, a self-booting utility--this one really takes the cake... since it boots from a floppy, it gets exclusive control to your system and memory--and boy does it use it! I got errors all the way down to 2.64ghz, and these were purely CPU-related, because memory was set to slow and increasing VCore fixed the errors.
So my highest stable speed is 17.0x154 = 2.62ghz @1.50V. For safety, I'm currently running at 17.0x153 = 2.6ghz, which at default voltage creates about 60.2 watts max heat. I am running a 512mb Corsair XMS3200c2 stick at 408MHz @cas2, which helps get blistering 5500+ Sandra Drhystone scores.
-
- Site Admin
- Posts: 12285
- Joined: Sun Aug 11, 2002 3:26 pm
- Location: Vancouver, BC, Canada
- Contact:
Writing this from the Bay area, down here attending the Intel Developers Forum in San Jose this week, trying to learn interesting new stuff to report about, network with engineers in likely companies, get the word out about SPCR etc. Interesting & exhausting couple days so far, lots to write about later.
Talking with one thermal engineer, I learned that different benchmarks and stress tests run different portions of processors. Mad Onion's 3dmark2000 was identified as the program that stresses the P4 the most, so you seem to be on to something, Leo. That memtest86 sounds interesting, will mention it to this engineer to see what he has to say.
Another pertinent issue is the actual physical location of the thermal diode in the die. On the P4, it is located in one corner; the hottest portion is slightly past center over on the other side, and the temp difference betw. these 2 points, I was told, can be as much as 20C!!
More on this later.
Talking with one thermal engineer, I learned that different benchmarks and stress tests run different portions of processors. Mad Onion's 3dmark2000 was identified as the program that stresses the P4 the most, so you seem to be on to something, Leo. That memtest86 sounds interesting, will mention it to this engineer to see what he has to say.
Another pertinent issue is the actual physical location of the thermal diode in the die. On the P4, it is located in one corner; the hottest portion is slightly past center over on the other side, and the temp difference betw. these 2 points, I was told, can be as much as 20C!!
Last edited by MikeC on Wed Sep 11, 2002 5:55 am, edited 1 time in total.
Mike, a major question you may want to ask them: does P4 make thermal throttling decisions based its thermal diode? If yes, at least we don't have to worry about unexpected throttling (even if the central part gets hot).
BTW it's great that you're pulling for us quiet people! Eagerly awating your writeup.
As for P4 cooling: Arctic Silver 3 should drop CPU temperatures another 1.3C over the AS2 that I'm currently using.
BTW it's great that you're pulling for us quiet people! Eagerly awating your writeup.
As for P4 cooling: Arctic Silver 3 should drop CPU temperatures another 1.3C over the AS2 that I'm currently using.
-
- Posts: 262
- Joined: Sun Aug 11, 2002 3:26 pm
- Location: Worcester, UK
-
- Site Admin
- Posts: 12285
- Joined: Sun Aug 11, 2002 3:26 pm
- Location: Vancouver, BC, Canada
- Contact:
It was only a remote screen of a speed counter showing the Hz as they used a software to oc a system behind the stage. No info on what kind of cooling, they were just doing their usual tease -- a bit of cleavage and a coy smile .ChiefWeasel wrote:I hear Intel have a P4 4.7GHz on display there? Have you seen that Mike? What kinda cooling are they using?
actually, the p4 thermal monitor sensor is independant of the thermal diode as stated in the p4 datasheet section 6.3.1, "[the] thermal diode is separate from the Thermal Monitor’s thermal sensor and cannot be
used to predict the behavior of the Thermal Monitor."
p3 datasheets used to publish the maximum Tdelta from the thermal diode to the worst case hotest point on the processor. looks like that information is no longer publish in the p4 datasheet. i can think of several reasons for this, but the dual catastrophic thermal protection mechanism built into the p4 should be sufficient in any case.
used to predict the behavior of the Thermal Monitor."
p3 datasheets used to publish the maximum Tdelta from the thermal diode to the worst case hotest point on the processor. looks like that information is no longer publish in the p4 datasheet. i can think of several reasons for this, but the dual catastrophic thermal protection mechanism built into the p4 should be sufficient in any case.
I have just run BURNP6.EXE on my computer in High priority. This site mentions how this program squeezes the most power out of a CPU--around 88% of max theoretical power!
My results speak for themselves: CPU temperature rose to 61C, a whole 7C higher than HotCPU Tester was able to manage! If the 88% figure holds, my CPU was pumping out an unholy 69 watts during this time! (This is running 2.26B @2550mhz 1.50vcore). The worst for a CPU like mine, according to Intel, should realistically be ~60 watts.
Should these results be taken as realistic, even for a worst-case scenario? Maybe not, but it's best to be able to accomodate it anyway.
My results speak for themselves: CPU temperature rose to 61C, a whole 7C higher than HotCPU Tester was able to manage! If the 88% figure holds, my CPU was pumping out an unholy 69 watts during this time! (This is running 2.26B @2550mhz 1.50vcore). The worst for a CPU like mine, according to Intel, should realistically be ~60 watts.
Should these results be taken as realistic, even for a worst-case scenario? Maybe not, but it's best to be able to accomodate it anyway.