Message boards :
Number crunching :
Version 4.05 Warmer CPU
Message board moderation
Author | Message |
---|---|
Petit Soleil Send message Joined: 17 Feb 03 Posts: 1497 Credit: 70,934 RAC: 0 |
I have just finished my first WU using SETI 4.05. Not only it took 40% longer to process but the CPU ran 4°C warmer then before. That is a fact. I have proved before on another thread for those who remember, that I know my machine and I monitor it's temperature very precisely. It proves that there is defenitely something wrong with that version. It seem that the CPU is actually wasting his ressources on something that has nothing to do with "pure SETI processing" like searching for Gaussians, spike, triplets, etc. Logicaly, an increase of CPU temperature means more processing works, and therefore slower processing times. Something is wrong here. Marc |
JAF Send message Joined: 9 Aug 00 Posts: 289 Credit: 168,721 RAC: 0 |
I'm now using 4.05 and my machine crunches at 97 - 98 percent (winXP with the task manager active). My CPU temperature is the same as with 4.03. I can't see how your temperature would increase 4 degrees C unless you were crunching a less than 97 - 98 percent CPU usage on 4.03. CPU activity can't go above 100%, so you were either running under 97 -100 percent under 4.03 or you need to check your computer cooling. Any other opinions? |
StanSuarez Send message Joined: 17 Mar 04 Posts: 5 Credit: 72,403 RAC: 0 |
My WUs take about 30-40% longger to finish too. Unfortunately, I didn't monitor temps closely on my BOINC boxes. I doubt that the client is more "efficient" if it takes longger to finish a WU. :( |
Petit Soleil Send message Joined: 17 Feb 03 Posts: 1497 Credit: 70,934 RAC: 0 |
> I'm now using 4.05 and my machine crunches at 97 - 98 percent (winXP with the > task manager active). My CPU temperature is the same as with 4.03. I can't see > how your temperature would increase 4 degrees C unless you were crunching a > less than 97 - 98 percent CPU usage on 4.03. > > CPU activity can't go above 100%, so you were either running under 97 -100 > percent under 4.03 or you need to check your computer cooling. > > Any other opinions? Thanks but I know that it can't go above 100% But it does not everything. Threads are just one aspect of CPU usage. You can run program X at 100% and having temperature higher or lower then program Z also using 100% threads. Here SETI 4.03 was 99% and 4.05 is also 99%. To prove that try different game for example. All the games I play, from RISK II to Unreal tournement they all uses 99% but temperature varies from +/- 10°C My computer cooling is working very well. I have cleaned it (laptop) last week. I have proved on this thread that the non linear processing time were also non linear as for processing power usage. I will monitor the next units very closely. -.-. --.- -.. -..- . - --... ...-- .-.-. -.- |
CyberGoyle Send message Joined: 2 Jun 99 Posts: 160 Credit: 3,622,756 RAC: 26 |
> My WUs take about 30-40% longger to finish too. Unfortunately, I didn't > monitor temps closely on my BOINC boxes. I doubt that the client is more > "efficient" if it takes longger to finish a WU. :( > > Quite likely the 4.05 client is performing more calculations that the 4.03 client. Back in the day on Classic Seti, the 3.x clients took about 75% longer to complete WU's than the 2.x clients - specifically because 3.x performed more science. <a> |
JAF Send message Joined: 9 Aug 00 Posts: 289 Credit: 168,721 RAC: 0 |
Well, if you are running a process just under 100 percent and the room ambient air temperature is constant, your CPU/computer should stabilize to a certain temperature. If you quit the processes you are running and run different one's (like 4.05 instead of 4.03 Seti) your CPU/computer temperature should be the same (as long as the room ambient temperature hasn't changed) and the processes run at the same percentile. From what I've seen, 40.3 and 4.05 run at the same percentile. I'm no expert on this, so maybe someone else can help, but I think I'm correct. |
Petit Soleil Send message Joined: 17 Feb 03 Posts: 1497 Credit: 70,934 RAC: 0 |
> Well, if you are running a process just under 100 percent and the room ambient > air temperature is constant, your CPU/computer should stabilize to a certain > temperature. If you quit the processes you are running and run different one's > (like 4.05 instead of 4.03 Seti) your CPU/computer temperature should be the > same (as long as the room ambient temperature hasn't changed) and the > processes run at the same percentile. From what I've seen, 40.3 and 4.05 run > at the same percentile. > > I'm no expert on this, so maybe someone else can help, but I think I'm > correct. Well again one the point is missing. Different programs all running at 100% does not necessarely run at same temperature. Threads do not represents all of what your CPU is doing. RISK II 100% 68°C SETI CALSSIC 100% 72°C SETI BOINC 4.03 100% 62-64°C SETI BOINC 4.05 100% 70-72°C TRAIN SIM 100% 82°C etc. I was just finished with me last SETI 4.03 WU and temperature was 62°C BOINC started the next one using SETI 4.05 and temperature rised up to 68 and increasing up to 72°C for the complete process. Now tell me what changed in between these two WU ? Ambiant temperature was the same, My cooling system didn't change, etc. OK so we all now that WU takes longer to process with 4.05 and From what I can see from here it also uses more CPU ressources and this resulting in higher CPU temperature. Does the 4.05 version is actually doing deeper math analysis ? I don't know. Doest it have some debugging code that slows the process but also makes it harder on the CPU ? I don't know. Did I've just happen to download some funny WU that no others have had so far ? I don't know. I am just relating facts that I can see here on my screen and that is my CPU is working harder since it runs SETI 4.05 -.-. --.- -.. -..- . - --... ...-- .-.-. -.- |
MarkRH Send message Joined: 7 Dec 03 Posts: 32 Credit: 211,105 RAC: 0 |
I got to thinking about this and while I haven't really noticed any temperature increases (I'll have to check when my CPDN gets pre-empted and SETI starts again), it does make sense to me how two processes that use 100% of the CPU could have different temperature results. The heat is generated by electrons whizzing around the various pathways in the chip. If one process uses more areas of the chip to do its thing, more heat will be generated. I could see how a program that contains an infinite loop of adding and subtracting 1 from a value would generate less heat than a program that did more complicated mathematics utilizing more areas of the chip. Another thing that could cause a difference is how much interaction one program has with others areas such as RAM and Disks vs another. Anyway, with my system, anything that runs the CPU at 100% seems to run it around the same temperature. Mark H. |
StokeyBob Send message Joined: 31 Aug 03 Posts: 848 Credit: 2,218,691 RAC: 0 |
Wow. Mine is about 8 degrees hotter with setiathome 4.05. Processing a work unit went from 3 hours 20 minutes for two (Hyperthreading) to 4 hours. |
Misfit Send message Joined: 21 Jun 01 Posts: 21804 Credit: 2,815,091 RAC: 0 |
According to my Asus Probe utility: Current CPU temp: 57C/134F Current MoBo temp: 37C/98F CPU fan speed: 4141-4166 Right now my ASUS bios is set to run the fan at 11/16 ratio. I'll keep a watch on the temp and see if I need to turn up the fan. Thanks for the heads-up, Marc. |
StokeyBob Send message Joined: 31 Aug 03 Posts: 848 Credit: 2,218,691 RAC: 0 |
I may have spoke to soon about my CPU temp. I looked into things further and found my CPU fan runing at a lower speed than it should be. I'll have to check it out tomorrow. |
Lord Comage Send message Joined: 2 May 02 Posts: 4 Credit: 2,052,032 RAC: 0 |
I haven't noticed any increase in Mnementh's cpu temps between clients. Iv got a P4 3.06 GHZ overclocked to 3.4 GHZ on a 1/2 watercooling setup. My temps run; CPU 39C/100F MB 28C/82F A really important factor to consider is the ambient room temp, for me, running at a constant 68-70F. With CPU at no usage, temps are; CPU 33C/91F MB 26C/78F |
Petit Soleil Send message Joined: 17 Feb 03 Posts: 1497 Credit: 70,934 RAC: 0 |
> I haven't noticed any increase in Mnementh's cpu temps between clients. Iv > got a P4 3.06 GHZ overclocked to 3.4 GHZ on a 1/2 watercooling setup. My > temps run; > CPU 39C/100F > MB 28C/82F > > A really important factor to consider is the ambient room temp, for me, > running at a constant 68-70F. > > With CPU at no usage, temps are; > CPU 33C/91F > MB 26C/78F Thanks Lord but once again you have just proved to me that people just don't read the threads before replying. Regarding ambiant room temp, please read again what I have posted. ""I was just finished with me last SETI 4.03 WU and temperature was 62°C BOINC started the next one using SETI 4.05 and temperature rised up to 68 and increasing up to 72°C for the complete process. Now tell me what changed in between these two WU ? Ambiant temperature was the same, My cooling system didn't change, etc."" It doesn't seem to me that the ambiant temperature had much time to change in the half a second it takes to start a new WU !!! You seem to have a very good cooling system there, I wish I could have have the same in my laptop. OH ! Did I've mentionned it was a laptop... Yes I did. |
Petit Soleil Send message Joined: 17 Feb 03 Posts: 1497 Credit: 70,934 RAC: 0 |
Anyway since It seems to be that I am the only one who have notice that I must be crunching some weird WU. I have also notice that the game I use to play with BOINC running at the same time started to act funny. I will reset and reinstall boinc and see what happen. Thanks to all Marc -.-. --.- -.. -..- . - --... ...-- .-.-. -.- |
Siran d'Vel'nahr Send message Joined: 23 May 99 Posts: 7379 Credit: 44,181,323 RAC: 238 |
> > My WUs take about 30-40% longger to finish too. Unfortunately, I didn't > > monitor temps closely on my BOINC boxes. I doubt that the client is more > > "efficient" if it takes longger to finish a WU. :( > > > > > > Quite likely the 4.05 client is performing more calculations that the 4.03 > client. Back in the day on Classic Seti, the 3.x clients took about 75% longer > to complete WU's than the 2.x clients - specifically because 3.x performed > more science. > > SETI "Classic" was blazing the trail for this type of scientific work though. You would think that the BOINC SETI Client would have already been coded for the same processing as the v3.x "Classic" client was. All I can figure is that the processing may have evolved beyond v3.x "Classic", causing still longer processing times in BOINC. My WUs are now running longer than when I was using my P4 1.6 Ghz CPU with 256 KB L2 cache and 400 Mhz FSB. My current CPU is a P4 2.8 Ghz with 1 MB L2 cache and 800 Mhz FSB. We just need something, anything from the dev team to briefly explain the longer processing time. Is it really so much to ask for? EDIT: To follow this thread, my CPU is running 100%, 100% of the time when running BOINC, which happens to be 24/7. I haven't really observed the temps, but mine seems to be running normal temp wise. Also, BOINC is always running even when I'm using my PC for, this for example or for printing from my office suite or anything else I do with it. OFF TOPIC: I have seen this since moving to v4.09 CC and v4.05 SC. I can run 3 WU at once now and when up/downloading, I can do 3 at once also. I watched while I was downloading more WUs, one had finished and uploaded at the same time as 2 were downloading. Must be SUPER Hyper-Threading.... ;-) Has anyone else noticed this? L8R.... --- Rick A. - BOINCing right along now.... It can only get better! "There is no fate except that which we create for ourselves." Live Long and Prosper.... |
Petit Soleil Send message Joined: 17 Feb 03 Posts: 1497 Credit: 70,934 RAC: 0 |
God knows what was wrong and what I was crunching but I have reinstall everything from scratch and it is now running at normal CPU temperature. Thanks for your participation. Problem is fixed now. (but still take longer to process) marc -.-. --.- -.. -..- . - --... ...-- .-.-. -.- |
William Smith Send message Joined: 12 Jul 03 Posts: 10 Credit: 11,645 RAC: 0 |
When SETI Classic was in use CPU speeds where lower. To make it realsitc time for wu to complete not every thing was proberley done in anlysing of the wu. But now that CPU speeds are getting faster more things can be analysied, in a reason amount of time. It's Out there........ |
Heffed Send message Joined: 19 Mar 02 Posts: 1856 Credit: 40,736 RAC: 0 |
|
JAF Send message Joined: 9 Aug 00 Posts: 289 Credit: 168,721 RAC: 0 |
> God knows what was wrong and what I was crunching but I have reinstall > everything from scratch and it is now running at normal CPU temperature. > > Thanks for your participation. Problem is fixed now. > (but still take longer to process) > > marc Still it brings up an interesting question. I tight loop of ten minutes worth of a simple integer count verses a more intense calculation for the same amount of time and compare temperatures. Or run Seti Classic and then switch to Boinc Seti 4.05. I downloaded motherboard monitor because I wanted access to the cpu temperatures without rebooting and going into the bios. But then found out I didn't know the brand of motherboard in my Dell 600m notebook nd couldn't find out anything in the Dell documentation or a google search. I'll have to wait until I can get to my desktop and then breakout my c++ and do a bit of experimenting. I would still be surprised to get a 4 degree difference from two "near 100 percent" processes, but it does make sense the a complex CPU instruction(s) would generate more heat than a simple instructions. |
Petit Soleil Send message Joined: 17 Feb 03 Posts: 1497 Credit: 70,934 RAC: 0 |
> > TRAIN SIM 100% 82°C > > Which Train sim? (I'm also a TS fan :) ) > MS Train simulator (2002 I think) with a few patchs (TGV, Sandpatch) I have uninstalled it a few weeks ago because It starded to act funny. I am still in the process of learning how to creates tracks, very complicated... I run it on my Laptop Presario 2800 P4m 1.4 GHz / 512 DDR 266 / Radeon 7500 32 Mb RAM Results are good but machine is "working hard" !!! Regards Marc |
©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.