Hi folks - I could use some advice on where to go with a troublesome workstation I just built. This is a light CAD workstations for a friend of the family, and it's failing stability testing. NO INTENTIONS of overclocking or tweaking this machine, everything must be stock/stable. FYI customer insisted on 3570k despite also having an (entry leveL) quadro card being installed.
GA-B75M-D3H mobo,
i5-3570k, stock HSF (see below for compounds)
Video (HD4000 for purposes of this testing)
Mushkin Enhanced Blackline 2x4Gb: DDR3 1600, 9-9-9-24, 1.5V
Silverstone PS07B mATX, 2x120mm intakes
Builder CX430 PSU
Crucial M4 SSD
Fresh Win7 x64 Pro, all Gigabyte drivers, latest bios.
Initial build, Win7 install went great - used the stock HSF along with Intel's factory compound. I ran IntelBurn several times (10x each), and NO stability problems AT ALL. Idled in the 20's. However temps would climb into the high 70's, so I decided to remove/replace heatsink compound. I should have left well enough alone!
Initially replaced with rice/bb of ICDiamond. Load temps dropped to mid-70's, but with a higher idle temp, and higher difference between cores (2-5°+ difference) under load. Began BSODing under IntelBurn. Also would not sustain Prime95 mesh for more than 10minutes or so. However despite temps under load never going above 75°C, still BSOD!
Removed and inspected - compound looked fine. Decided to use ArticSilver5 instead, light smear coat (which I usually do for polished/smooth coolers). Core temps were *much* tighter now - idle around 25-30°C, load still under 75°C, but STILL BSOD UNDER LOAD.
Ran MemTest86+, two passes no problems. Ran HCI MemTest, 0x124 BSOD withing 15minutes (it was loading all cores 100%, but STILL maintaining under 60-65°C during testing).
Cleared CMOS, (did NOT load optimum settings) and ONLY changed IDE/AHCI mode. Able to run HCI mem test for an hour, no problems. Still no stability with IntelBurn or Prime95.
Look, I'm not looking to go for some crazy record on either stress test, but I think any STOCK system should have no problem maintaining 100% load for a few hours at least. The way I see it, even though this stress testing isn't a realistic representation of every day load, there ARE software bugs/problems/hangs that can throttle up a CPU and keep it there. I feel a system should be able to handle this, temperature wise without self destructing or BSODing.
Any suggestions? THANK YOU for your time!
GA-B75M-D3H mobo,
i5-3570k, stock HSF (see below for compounds)
Video (HD4000 for purposes of this testing)
Mushkin Enhanced Blackline 2x4Gb: DDR3 1600, 9-9-9-24, 1.5V
Silverstone PS07B mATX, 2x120mm intakes
Builder CX430 PSU
Crucial M4 SSD
Fresh Win7 x64 Pro, all Gigabyte drivers, latest bios.
Initial build, Win7 install went great - used the stock HSF along with Intel's factory compound. I ran IntelBurn several times (10x each), and NO stability problems AT ALL. Idled in the 20's. However temps would climb into the high 70's, so I decided to remove/replace heatsink compound. I should have left well enough alone!
Initially replaced with rice/bb of ICDiamond. Load temps dropped to mid-70's, but with a higher idle temp, and higher difference between cores (2-5°+ difference) under load. Began BSODing under IntelBurn. Also would not sustain Prime95 mesh for more than 10minutes or so. However despite temps under load never going above 75°C, still BSOD!
Removed and inspected - compound looked fine. Decided to use ArticSilver5 instead, light smear coat (which I usually do for polished/smooth coolers). Core temps were *much* tighter now - idle around 25-30°C, load still under 75°C, but STILL BSOD UNDER LOAD.
Ran MemTest86+, two passes no problems. Ran HCI MemTest, 0x124 BSOD withing 15minutes (it was loading all cores 100%, but STILL maintaining under 60-65°C during testing).
Cleared CMOS, (did NOT load optimum settings) and ONLY changed IDE/AHCI mode. Able to run HCI mem test for an hour, no problems. Still no stability with IntelBurn or Prime95.
Look, I'm not looking to go for some crazy record on either stress test, but I think any STOCK system should have no problem maintaining 100% load for a few hours at least. The way I see it, even though this stress testing isn't a realistic representation of every day load, there ARE software bugs/problems/hangs that can throttle up a CPU and keep it there. I feel a system should be able to handle this, temperature wise without self destructing or BSODing.
Any suggestions? THANK YOU for your time!