Got the new case and as many of you were familar with my past post early on with my new 2.4c I was getting extremely high temps even though I had reapplied AS2, got New AS5, went from an alpha w/ 80mm ystech fan to a swiftech with the same 80mm fan to the swiftech with an 90mm
I had 3 scenarios
1) case side off outside of desk enclosure
sys temps (idle/load) = 25-28
cpu temp (idle/load) = 44-61
pwm temp (idle/load) = 30-42
NO issues in any stress testing or encoding....
2) case side on outside of desk
sys temps (idle/load) = 27-31
cpu temp (idle/load) = 45-63.5
pwm temp (idle/load) = 32-44
NO issues in any stress testing over the last few weeks after extensive testing....
3) case side on and inside desk
sys temps (idle/load) = 27-32
cpu temp (idle/load) = 45-65
pwm temp (idle/load) = 32-45
Had a few issues like read errors in tmpgenc...Ruled out error in source file and reran with case outside of desk or case door open and no errors and 6 hour encode...
So naturally I am thinking cpu heat. Then at the time I through back on the 80mm ystech fan that did about as good (within 1c) of the 90mm fan since it pushed 50cfm and this 90mm is pushing 58cfm max....I turned the dial down to 2200rpm fan on the 80mm dropping the cfm to likely 30cfm area and the temp on the cpu got to 68c with case side open and outside of desk and it made it through the encoding of 6 hour test with prime95 running in background....OK cpu heat is not issue and the chip is not throttling...
My theory then looked at that I would likely get a read error or error in prime95 when the PWM reached +44c range...What is the pWM you ask?? Many have speculated that it was the NB temp but that in fact is untrue...the temp is taken in an area of mosfetts (power do-thingies
)....It appears my fan duping the hot heat from the heatsink fins was blasting the area and heating it up and then the fact poor case cooling or at least inadequate for this endeavor was then cycling warmer air back over fin and thus not expelling the heated air quickly enough to cool down areas on the board as well as the ram heatspreader would get quite hot.....
Step to new case...
Antec tower with 2 80mm exhaust fans (ultra quiet 80mm) in back...1 intake front across the HDD and one side intafe fan aiming around the vid card/ NB chipset region...POwer supply has a dual fan system as well...
New temps with case closed in desk full dual POV-Ray encodings going simultaneously....Also testing TMPGENC with prime95 in the background...noted no errors yet...
sys temps (idle/load) = 25-28c ***obvious sign of better cooling as case temp stay close to room temp of 24c (75f)***
cpu temps (idle/load) = 43-61c ***little better likely do to better case temp by about 3-4c...seems to be limitation of HSF***
PWM temp (idle/load) = 28-40c ***much better here as obvious cooler case temps, better ventilation on that area, and cooler cpu exhaust is helping***
note: that with case outside of desk maybe 1c better in cpu only....with side door open 2c better cpu/1c better sys temp and 1c better pwm....
Conclusion the case has given me max case ventilation...Now I am left with Abit temp readings and likely limtations of the swiftech MCX400 heatsink....
Maty get the Zalman 7000al-cu for testing of cpu temps....
I had 3 scenarios
1) case side off outside of desk enclosure
sys temps (idle/load) = 25-28
cpu temp (idle/load) = 44-61
pwm temp (idle/load) = 30-42
NO issues in any stress testing or encoding....
2) case side on outside of desk
sys temps (idle/load) = 27-31
cpu temp (idle/load) = 45-63.5
pwm temp (idle/load) = 32-44
NO issues in any stress testing over the last few weeks after extensive testing....
3) case side on and inside desk
sys temps (idle/load) = 27-32
cpu temp (idle/load) = 45-65
pwm temp (idle/load) = 32-45
Had a few issues like read errors in tmpgenc...Ruled out error in source file and reran with case outside of desk or case door open and no errors and 6 hour encode...
So naturally I am thinking cpu heat. Then at the time I through back on the 80mm ystech fan that did about as good (within 1c) of the 90mm fan since it pushed 50cfm and this 90mm is pushing 58cfm max....I turned the dial down to 2200rpm fan on the 80mm dropping the cfm to likely 30cfm area and the temp on the cpu got to 68c with case side open and outside of desk and it made it through the encoding of 6 hour test with prime95 running in background....OK cpu heat is not issue and the chip is not throttling...
My theory then looked at that I would likely get a read error or error in prime95 when the PWM reached +44c range...What is the pWM you ask?? Many have speculated that it was the NB temp but that in fact is untrue...the temp is taken in an area of mosfetts (power do-thingies
)....It appears my fan duping the hot heat from the heatsink fins was blasting the area and heating it up and then the fact poor case cooling or at least inadequate for this endeavor was then cycling warmer air back over fin and thus not expelling the heated air quickly enough to cool down areas on the board as well as the ram heatspreader would get quite hot.....Step to new case...
Antec tower with 2 80mm exhaust fans (ultra quiet 80mm) in back...1 intake front across the HDD and one side intafe fan aiming around the vid card/ NB chipset region...POwer supply has a dual fan system as well...
New temps with case closed in desk full dual POV-Ray encodings going simultaneously....Also testing TMPGENC with prime95 in the background...noted no errors yet...
sys temps (idle/load) = 25-28c ***obvious sign of better cooling as case temp stay close to room temp of 24c (75f)***
cpu temps (idle/load) = 43-61c ***little better likely do to better case temp by about 3-4c...seems to be limitation of HSF***
PWM temp (idle/load) = 28-40c ***much better here as obvious cooler case temps, better ventilation on that area, and cooler cpu exhaust is helping***
note: that with case outside of desk maybe 1c better in cpu only....with side door open 2c better cpu/1c better sys temp and 1c better pwm....
Conclusion the case has given me max case ventilation...Now I am left with Abit temp readings and likely limtations of the swiftech MCX400 heatsink....
Maty get the Zalman 7000al-cu for testing of cpu temps....
Facebook
Twitter