Thanks for your answer : )

i am sorry that my message wasnt too clear on some points, i was retyping after firefox ate my precedent post and sent it to another universe (frustration doesnt go well with concentration)


The cpu of my wife computer is cooled via thermo-electric system that provides, in exchange of some extra Watts (peltier effect) real low temperature to cpu known to spread a LOT of heat. 3.4Ghz is known to behave this way, so i bought this fancy little 'fridge' that makes it around 20-25 degrees under the 'normal' temperature.

i am going to make different tries with other agents on this computer (my wife wont like that, i am not supposed to mess with her best friend) and see if it comes from WCGA, HCMD itself. (a good old boinc vs wcga/xxxx will keep me busy for some time :)

i ll keep you informed about that.

Also, related to my post 'WCGA uses only 33% of cpu" it seems that on the 3.4Ghz it takes all that it can.

maybe a cpu optimization inside the program

just a lucky guess based on fuzzy memory....

I will be back with more material

Edit:

There i'm back and i have some answers.

i tried different projects and found out that wcgridHCMD _seems_ to be requesting more cpu% than other projects.

let me explain : most projects make let pause appears on cpu graph, or have a lessened cpu average use (either spikes between 20 and 50 or kind of constant line around 35%)

HCMD though, runs 'stronger' than others per default

so i played a bit with WCGridMaxCPU

i ran a test of genome at default, then at "100% setting" using wcgridmaxcpu.

temperature started to climb over 35° instead of 25° with default setting.

then i reinstalled with hcmd and ran hcmd at 60n then 40% setting, and at the moment i write those lines, cpu temp goes from 33 to 37 and TEC temp is stable at 38 but increasing slowly.

my guess is that the peltier is slightly insufficient in power, which will be solved next week (going 110 instead of 90W)

as it has a very fast response cooling effect, if the software makes regular pauses the temp stays ok, that s the case in most apps and games and stuff my wife runs on this machine. benchmarks tend to get temp a bit higher, around 45-47 after a 2 hours loop.

but HCMD is the only one that almost killed my TEC in less than 10 minutes, i dont know how much it can handle, but 60 seems a bit too much to my taste (if TEC dies, PC stays off until i get a new peltier component)

right now TEC is taking one extra degree per 2 minutes round, running at 42 and might be higher later.

so 40% is still too strong :/

i have to say that s the best coded program, at least the one that takes all that my cpu can give.

my other computer at 2.6Ghz doesnt have problem, temp is stable with stock intel heatsink, but a 2.6Ghz doesnt produce the heat a 3.4Ghz does.

also this is old fat pentium northwoods, i guess that on dualcores, the cooldown is so fast that average temp must stay very reasonable.

my 2 cents.... at 30% it seems the TEC can hold it.... will let it go for an hour and see....

thanks to all of you who red this (not very) useful post :)

Florent