I built a new system a couple of weeks ago using a GA-P35-DS4 (could not wait for the X38) and have been loading and testing applications on it. Currently runs XP Pro but will dual boot with Vista 64 once I have XP working well for legacy software and hardware. The system runs well, is fast, stable and quite but when trying to monitor it, CPU loading seems low < 40% during video rendering. Systems is built in an Antec P182 case Gigabyte GA-P35-DS4 motherboard F7 Bios Intel Q6600 clocked at 320 x 9 so 2.88GHz Scythe Ninja plus CPU cooler with supplied 120mm fan Patriot 4GB (2 x 2 Gbyte) PC2-6400 running at 800 MHz PowerColor 256MB HD2600XT Case fans – rear extract, intake in lower HD cage. Top fan fitted but not powered. 2 x Samsung 500Gbytes, 1 x Samsung 500Gbytes in lower cage 1 WD 300 Gbyte System drive in upper cage System is used for image and video editing/playing so does not require heavy graphics use, but needs to be quite. With the above settings, the system idle or browsing the web, the system runs with CPU fan turned off by the motherboard. Temperatures reported by Gigabyte Easytune are System 39C CPU 33 when idle. Room temperature is 21C That’s the first issue – those 2 temps should be the other way round, can the CPU be cooler than the case when CPU fan is not running. Core temp reports the cores as being 40, 37, 38, 42. Hard drive temps are 38 to 41C Under load – when rendering video - Gigabyte Easytune are System 39C CPU 43 and CPU fan ~ 850 RPM Core temp shows temperatures upto 58C Speed fan shows similar values to Gigabyte Easytune, but is not able to say which is which. Speed fan also shows CPU efficiency to be 34% to 38% whilst rendering. I have the latest versions of the above utilities, but which is believable? The low CPU loading indicates the system is IO bound, so does anyone know of any tools that will show the load on each hard drive?
IIRC Coretemp often reads wrong, I remember reading a thread about it (though I forget where). Something to do with the tjunction sensor or something, certain CPU's give an incorrect reading up to 20C out and it was determined that speedfan was most accurate. Something about some t-junctions being set at 80C and some set at 100C threshold would give incorrect readings. I would say if you are worried stick a thermometer in there and check it the old fashioned way. ::edit:: http://www.tomshardware.co.uk/forum/221745-11-core-quad-temperature-guide seems really similar to the one I remember reading.
Thanks for the link, needs a bit of study and provides a lot of useful information including how to calibrate Speedfan. Will have a detailed look this evening It explains the difference between Coretemp and the others is quite simple - Coretemp is reporting the temperature of the sensor built into the core, the others are reporting the temp of the CPU casing. So Coretemp reports a higher value since thermal energy is flowing from the sensor it uses to the CPU enclosure which is being cooled and whose temperature is reported by Speed fan. Also good explanation of t-junction, when the temp reported by coretemp reaches that value - 100C in this instance, then the CPU will shut down. Now I understand that and the differences, I can concentrate on finding the bottle neck in the system
What revision DS4 do you have? I only ask as I have the rev 1, with the towerblock of heatsink on the northbridge chip as well as large blocks on the southbridge and mosfets/power circuitry. It's the latter that fouled the installation of my Scythe Ninja, so I gave up and bunged the retail hsf on.
I have the latest version which is Rev 2.0 I checked this out before buying, on this forum and elsewhere - the heatsinks on the Rev2 version of the board are smaller and leave plenty of space around the CPU for the Scythe Ninja. I checked how well it might fit before screwing the motherboard into the case - there was plenty of room so I installed the heatsink whilst the motherboard was fitted in the case. The only problem I had was one of the prongs on the heatsink feet was slightly open and would not fit in the hole on the motherboard - so had to squeeze it closed to get it to fit - took about 4 attempts because I did not want to put too much pressure on heatsink/motherboard. The Ninja works really well, I had planned to leave the fan off but most of the time the CPU is reported by Speedfan as being in the low 30's C and the fan is turned off by the system. With the Ninja (CPU) fan pointing to the back of the case, there is not much air flowing across the motherboard so I am thinking of moving the case fan from the top of the case - where it seems to have no effect on cooling - to the front of the P182. If that has a significant impact, then I might try and find a quite fan to replace the Antec one which makes more noise than I want.
