Message boards :
Number crunching :
Whetstone/Dhrystone problem
Message board moderation
Previous · 1 · 2 · 3 · Next
Author | Message |
---|---|
htrae Send message Joined: 3 Apr 99 Posts: 241 Credit: 768,379 RAC: 0 |
The AMD K7 line of mobile processors uses "PowerNow" technology to throttle the processor and doesn't use "CoolnQuiet" as the K8(A64) uses. I believe there may be a "PowerNow" utility in the control panel of WindozeXP that you can access for this. ControlPanel>PowerOptions>PowerNow Here's a link that explains it a bit better. It also has some download utilities (DashBoard)that may help you diagnose the problem. I don't know if this applies to your situation as I've never owned a laptop before. http://www.amdboard.com/powernow.html Thanks for the Bios version, unfortunately I can't find any info on whether it has any sort of thermal throttling. |
Astro Send message Joined: 16 Apr 02 Posts: 8026 Credit: 600,015 RAC: 0 |
Do you know CPU-Z? Morning, I "force cooled" the laptop, booted it up, ran Boinc and CPUZ. Here's the data: Time Freq Whet Dhry 8:49 1788 1649 2766 8:52 1788 1653 2781 8:54 1789 1545 2387 8:56 1788 1571 2336 8:59 1788 1255 2430 9:03 1789 1120 1951 9:06 1788 1102 1793 9:09 1788 1106 1872 9:14 1788 922 1935 Using task manager, peformance tab, it shows 100% cpu throughout. I wonder what Windows is monitoring to get it's reading? Maybe the core freq? As we can see, the Frequency doesn't change much. does this help? tony |
Astro Send message Joined: 16 Apr 02 Posts: 8026 Credit: 600,015 RAC: 0 |
One HP/Compaq update page for you. :) Thanks Jord. I have the latest Bios downloaded, but need to wait till I'm in South Carolina so I can use my floppy drive(it's not up here). |
Astro Send message Joined: 16 Apr 02 Posts: 8026 Credit: 600,015 RAC: 0 |
Thanks for the Bios version, unfortunately I can't find any info on whether it has any sort of thermal throttling. PowerNow doesn't exist on this laptop. Any other ideas? tony |
W-K 666 Send message Joined: 18 May 99 Posts: 19091 Credit: 40,757,560 RAC: 67 |
If I rememeber correctly, AMD fitted a thermal diode to their CPU's to interrupt the clock, which effectively slows down the CPU. I don't think we are talking about the same kind of thottling. Several years ago it was reported that some CPU's when overclocked showed a decrease in performance even though utilities such as CPUZ etc. showed the clock frequency as steady. What was happening inside the CPU the thermal diode was being triggered and this actually interrupted the clock internally, putting the CPU to sleep for a few clock cycles to allow the CPU to cool slightly. When the CPU had cooled the diode then allowed the CPU to operate again, if it again overheated the cycle started again. The throttling you are talking about actually switches the clock to a lower frequency, either when the CPU is hot or when the CPU is not doing any work, this saves wasted energy. Shortly after the introduction of the 2nd generation of AMD AthlonXP 2200+, AMD forced the MoBo makers to introduce a overheat circuit that shut down the PSU if they wanted their MoBo's to be endorsed. Andy Edit/ I am assuming this is might be what is happening to Tony's sick laptop, as the reported clock freq is staying steady. It might be worth trying underclocking if the BIOS allows this. /Edit |
Kokie Send message Joined: 18 Nov 03 Posts: 12 Credit: 8,294,384 RAC: 0 |
We are talking about the same kind of throttling. If you want to protect a CPU against high temperatures you have to reduce it's heat dissipation, and the only way to do this is to lower the core frequency (and, of course, the supply voltage, see below). The first CPU that was able to do this without BIOS or mainboard support was the P4. It used a thermal diode to measure the die temperature and at a certain threshold, a built-in circuit forced the built-in clock generators to lower the core clock. We can say that this CPU was self-protecting. When AMD then came out with the XP it had only such a thermal diode (which cannot change any frequencies but only deliver the value 'temperature' through a voltage) and thus being not able to protect itself. The XP definitely needs some extra circuit on the mainbord. If you want to save energy, it a bit more complicated since the user should not notice what happens behind his back. All the current power saving (like SpeedStep or Cool'n'Quiet) is done by reducing the core frequency and the supply voltage. To have this working smoothly you need a CPU that supports it, a BIOS that supports it and a driver that supports it (and an OS that supports the driver...). All this left me completely confused when I saw Tony's table with the frequencies and the benchmark values. I must admit that I'm about to surrender. I'd expect a coherence between the core clock and the benchmark results. Either the measured clock is wrong or the CPU has an increasing amount of other things to do. But these values simply don't fit. I'm afraid I can't help, sorry. Kokie |
Astro Send message Joined: 16 Apr 02 Posts: 8026 Credit: 600,015 RAC: 0 |
All this left me completely confused when I saw Tony's table with the frequencies and the benchmark values. I must admit that I'm about to surrender. I'd expect a coherence between the core clock and the benchmark results. Either the measured clock is wrong or the CPU has an increasing amount of other things to do. But these values simply don't fit. That's OK. It's got me confused as well. I don't know at what point in the circuit that is measured to display frequency. I have had taskmanager open and can say that NO other processes are or were running. Edit, it's been running most the day. Freq 1788, whet 740, dhry 1216 |
j2satx Send message Joined: 2 Oct 02 Posts: 404 Credit: 196,758 RAC: 0 |
Before you throw it in the dumpster, let some techno geek have a go at finding the problem. It is an interesting problem. |
Astro Send message Joined: 16 Apr 02 Posts: 8026 Credit: 600,015 RAC: 0 |
Before you throw it in the dumpster, let some techno geek have a go at finding the problem. It is an interesting problem. I have had the thought about putting Linux on it or just leaving it as is and just letting it set in the corner crunching at a slow pace until death. I don't know any geeks that are by me and close enough to do any more than I can. I'll be flashing the new bios upgrade in about a week. I'm leaving Michigan for South Carolina on/about the weekend. I know a bit, but am usually amazed at how much knowledge can be found in these fora. Not everyone know everything, but with a group this size, someone is bound to have the knowledge. thanks for your time tony |
htrae Send message Joined: 3 Apr 99 Posts: 241 Credit: 768,379 RAC: 0 |
" I'm afraid I can't help, sorry. " I'm baffled here also. Core frequency remains constant while the benchmarks drop off, go figure...!!!. The only other thing I can think of is that other concurrently running processes are starting up, hogging resources and affecting the benchmarks. Try using this utility to shut down as many processes as possible and try running the benchmarks. http://www.the-old-sea-dog.net/enditall.html Sorry, but I'm running out of ideas. Edit: Sorry, I was typing this reply up as you were yours and didn't see that you remarked that no processes were running. |
j2satx Send message Joined: 2 Oct 02 Posts: 404 Credit: 196,758 RAC: 0 |
All this left me completely confused when I saw Tony's table with the frequencies and the benchmark values. I must admit that I'm about to surrender. I'd expect a coherence between the core clock and the benchmark results. Either the measured clock is wrong or the CPU has an increasing amount of other things to do. But these values simply don't fit. No kidding, my two 2200+ have been running 1684 / 2848 @ 1.800MHz and 1683 / 2841 @ 1.799MHz..........basically twice the FP and Integer at the same FSB. |
j2satx Send message Joined: 2 Oct 02 Posts: 404 Credit: 196,758 RAC: 0 |
Before you throw it in the dumpster, let some techno geek have a go at finding the problem. It is an interesting problem. Have you run SANDRA and let it suggest performance improvements? |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
Tony, probably a weird question, but have you tried to run a game on that laptop? One that really takes much of the CPU, like Half Life 2, NFS: Underground 2 or Doom 3? If any of those games can hack it on your laptop without much slow down, with the embedded video chip & sound chip, I'd be turning to look it's something other than heat in its totality. |
Bronwyn & Grym Send message Joined: 12 Apr 00 Posts: 42 Credit: 1,163,872 RAC: 0 |
|
Astro Send message Joined: 16 Apr 02 Posts: 8026 Credit: 600,015 RAC: 0 |
Tony, probably a weird question, but have you tried to run a game on that laptop? In short NO. Believe it or not, I don't own any games. I used to play Colonization, Civ, Civ II, Civ call to power, and Civ III. My Civ III cd is damaged, so I don't have any available. Not up here, and not in South Carolina. |
Kokie Send message Joined: 18 Nov 03 Posts: 12 Credit: 8,294,384 RAC: 0 |
Has anyone thought about memory impact? Could the memory be a little flaky? Yes, sure, but memory issues usually sound more like 'BSOD', 'sudden crash' or something like that, but not like 'slowly decreasing performance'. |
Astro Send message Joined: 16 Apr 02 Posts: 8026 Credit: 600,015 RAC: 0 |
Has anyone thought about memory impact? Could the memory be a little flaky? I ran prime 95 for 30 minutes without error. As stated in the other thread, Memtest, configuration builder, and DFT (hitachis drive fitness test) CD causes the puter to shutdown. I even booted to Bios screen and as I was scrolling across, it shut down. |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
In short NO. Pssshh. As if anyone in your position can't download games from IRC. ;) |
Don Erway Send message Joined: 18 May 99 Posts: 305 Credit: 471,946 RAC: 0 |
Hey Tony. Wow, weird one. Have you tried examining the individual processes that are running, as this occurs? Just looking for 100% does not tell you if something else is coming along, and grabbing up CPU. Could be a virus, or trojan... Or, it could be an antivirus kicking in, or something like that.. I bet, if you watch the processes in task manager, (and sort them by cpu %), you'll see something. Don P.S. I've got mycroft running a mobile Athlon XP 3000 - just a bit overclocked from 2.2 to 2.4 gig. Running prime95 just fine. Lots of fun! |
Professor Desty Nova Send message Joined: 17 May 99 Posts: 59 Credit: 579,918 RAC: 0 |
You could run a live-CD Linux (like Knoppix, etc.; it runs of the CD, no need formating things) and then run BOINC on it for a few hours, and see how it goes. SETI@home classic workunits: 1,985 CPU time: 24,567 hours Professor Desty Nova Researching Karma the Hard Way |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.