Pentium(R) D CPU 3.20GHz question
5 Aug 2010 21:01:01 UTC
Topic 195223
(moderation:
Do any of you here run one of these?
It is one of my collection (I am on it right now)
I am wondering why it appears to be so slow compared to any of my other machines.
I have forgotten what its WU time was in the past but this dual core is taking around 30 hours and my other machines do the Global's between 6 and 13hrs 24/7
The heat sink looks clean and I have the panel off and an extra fan on it just like the others.
My AMD 3200 is the one running singles out just over 12hrs so this P4 3.2 seems real slow doing the same WU's @ 30+ hours.
I tried a couple links with temp. tests and they didn't actually seem to be doing what they say or they were just trying to get you to buy something.
(which I will not do since I spend enough on these pc's over the last 10 years)
Any tips or suggestions are appreciated.
Pentium(R) D CPU 3.20GHz question
)
I run a comparable Pentium 4 3.0GHz. It runs the GWC1's in about 10 hours.
Checking the benchmark values between our two computers, it almost seems as if yours is damaged. Of course, it could just have been a bad benchmark, that it was done while the CPU was busy with something else or something 'disruptive' was loaded while the benchmarks took place. (I notice when Winamp is loaded, doesn't even need to be on, just in memory, that my benchmarks go down by 500 MIPs).
Still, your Measured integer speed 1581.92 million ops/sec vs my Measured integer speed 2951.81 million ops/sec? Sounds like something wrong with the CPU. Or missing/corrupt motherboard drivers.
(yes, I know, the benchmarks aren't used here at Einstein. But it does give a good indication whether something is wrong or not!)
RE: Do any of you here run
)
I'm sorry, but I've never owned a Pentium D. The nearest thing I've got is a 2.8GHz Northwood P4 with HT.
I would guess that it's probably one or more of several possibilities
* It's being frequency throttled
* It's being thermally throttled
* There are other compute and/or memory intensive apps running in competition
When you take a look at the tasks list for that machine there is unusually large variability (vertically) in both the CPU time column and the total run time column. My experience is that task times are reasonably constant so this variation is unusual. It sort of implies different amounts of throttling for different tasks. Also, the difference between CPU time and run time is quite large. Something else is using a significant amount of CPU.
The BOINC benchmarks for your machine are listed as 1058/1581. My Northwood P4 shows 1314/2834. I would imagine your machine should get markedly better scores than mine.
Have you tried running any CPU testing utilities? For starters, use CPUZ to check frequencies and memory settings, etc. Use something like CoreTemp to check idle and load temperatures. If frequency were changing under load, I imagine you would see this on CPUZ.
Cheers,
Gary.
I did try CPUZ and it seens
)
I did try CPUZ and it seens ok and I tried CoreTemp the other day but after I downloaded it it said it does not support this particular processor so I don't know if it is a temp. problem.
This machine doesn't run any other programs and it just gets used for the other basics like the other ones (email and such)
Task manager says it is running only Einsteins.
I even just tried running a extra fan and low temp in the room and it made no difference.
try speedfan
)
try speedfan http://www.almico.com/speedfan.php for temps
seeing without seeing is something the blind learn to do, and seeing beyond vision can be a gift.
Thanks Paul.....speedfan
)
Thanks Paul.....speedfan works great!
So it tells me that
Remote temp. 42C
Internal temp. 55C
CPU temp. 97C ......ok that sounds a bit warm.
Maybe I will shut it down for a couple minutes and take a closer look at the heat sink when the fan isn't spinning.
Seems odd that this one is hot and the others aren't since this one is in the coolest part of the room.
Thanks again Paul.....(of course I will take any suggestions)
Well your CPU ain't cooling
)
Well your CPU ain't cooling well despite an otherwise rather cooler system. It fact it's probably close to irreparable damage. Read and print out this post beforehand if you want to follow my advice .... :-)
With the system off ( inspect for dust too ), slightly jiggle the heatsink/fan over the CPU and feel if there's any 'play'.
If so : those four 'click and turn' plastic retaining pins aren't holding the heatsink/fan array firmly against the mobo, thus also not well enough against the upper CPU surface. Try and fix that if you can, mind you I've sometimes found that rather difficult on occasions as the little detent/notch/ledge on the surround of each retaining pins can wear and/or break ( it's the bit that grabs the head of the pin as it thrusts through the mobo hole thus holding it firmly in situ ). You may be up for a new heatsink altogether, in which case my following advice is also relevant.
Otherwise : most likely the thermal paste b/w the CPU and heatsink is 'overdue'. It can degrade with time and heat. You'll need to get a new tube of paste before proceeding with any of my advice :
- shut everything down and remove the heatsink and inspect the paste - odds are it is discoloured ( depending on what it was de novo ), dry and not covering the upper surface of the CPU adequately.
- clean the upper surface with or without removing the CPU from it's seat ( depending on how messy things are, preferably leaving it in ). There are a number of very handy spray can products for this type of work, but the main worry would be damage to CPU pins, debris clogging the socket's holes for the pins, and debris causing shorts b/w pins. The interface b/w CPU and it's underside mount ( ie. the entire pin side of the CPU and the entire socket ) should be utterly pristine. Close and careful inspection with some magnifying gadget is very rewarding here. A static grounding strap is not a silly idea either.
- remount the CPU in the socket if removed.
- clean the underside of the heatsink.
- clean the surrounding mobo.
- new paste applied. Be careful if the heat paste is metal based ( eg. silver, in which case it looks greyish ) as that means it's also electrically conductive in addition to being thermally conductive. That means any excess squeezing out when you put the heatsink on can short out adjacent stuff. If it's that white silicone based type of paste then it doesn't conduct current.
- remount the heatsink ( plug in the fan etc ... ), check for wobbles as discussed above.
- clean the surrounding mobo, again.
- fire it up and see how it goes. If you go into the BIOS screen first up ( hit 'del' early on ) then go to the 'PC Health' section which usually displays the CPU temp. Leave it at that BIOS screen for a good half hour, get a cuppa for yourself etc, and watch the temperature. While you're at it set the CPU temp warning ( if there is a setting for that ) at the lowest, say 60 degrees C.
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
97c is a bit over 200f so
)
97c is a bit over 200f so yeah id say thats a bit hot :|
while your checking the heatsink be sure to check the fins behind the fan, dust bunnies just LOVE to hide there. also check any other vents and or fans. what i do when im diagnosing a heat issue is while i have the case open i unhook the power cable from the hard drive so the system wont boot. then i start it with the side panel off for a few minutes to check on the fans.
the fan on the heatsink should still be working otherwise i imagen youd have gotten some sort of alarm. but it could be running slow.
if your motherboard supports it speedfan can show the fan RPM's as well. tho on my vista system i have to right click then "run as administrator" to get access to the fan rpm's.
edit: mike beat me to it! :)
seeing without seeing is something the blind learn to do, and seeing beyond vision can be a gift.
Yeah I was just hoping for a
)
Yeah I was just hoping for a temp check like "speedfan" to use instead of the bios version so I could watch the temp changes if this was going to be the problem.
And it is and after just cooling it down 5 degrees it cut the wu time from 30hrs back to between 12hrs and 13
This is one of my machines that I put a better heat sink/fan cooler in to replace the cheap stock version a couple years ago.
And it doesn't have a removable fan otherwise I would pull it off and try another fan (and of course the sink is easier to clean)
And of course I hate even shutting one of these off but I think I better do that for a few minutes and see what I can do to bring the temp down farther but I think I will just "tigerdirect" another one today so it will arrive here this week and I can replace the one I have in there now.
But now I have used "speedfan" on the other ones too....just to make it a easier way to check that and a few other things.
Thanks again.
tinker with it. speedfan can
)
tinker with it. speedfan can be set to alert you if the cpu goes over a certain temp. i think that can be email as well. or to play an audio file, etc.
seeing without seeing is something the blind learn to do, and seeing beyond vision can be a gift.
I believe the temperature of
)
I believe the temperature of a Pentium D may be around 70 degrees Celsius maximum, at least it may with a Pentium 4. 97 degrees is way too hot for such a processor. Newer Intel 'Core Duo' and such CPUs allow higher temperatures, up to 100 degrees or so.
Mike Hewson's advice is good: remove the heatsink and blow it clean, put (a little) new thermal paste between it and the processor after cleaning off the old paste, and then make sure it is tightly fitted to the processor when you put it back on. I did that with my Pentium 4 system, which was hitting 67 degrees or so while crunching, and the temperature dropped to 43 degrees while crunching. A large difference.
You're lucky it's an Intel, because that means it throttles itself down to prevent damage from overheating. That's why it was so slow: it's supposed to slow down when it's overheating to prevent even more overheating. This is an interesting (classic) little movie about that: http://www.youtube.com/watch?v=NxNUK3U73SI
I don't know whether AMD has the same mechanism built into their CPUs nowadays. I hope so, but according to that little movie their older CPUs surely did not have it.
Good luck.