- Joined
- Jul 23, 2010
Based on these quotes:
So here is the whole story:
My Mobo: Gigabyte H55-USB3
My CPU: Intel i5-750
My DDR3: Corsair Dominator TR3x6G-1600C8D >> I mistakenly buy these triple channel kits... I know it would be only dual channel in H55... This are where the problem starts...
- i5-750 default memory controller speed is 1333mhz...
- when I enable XtremeMemoryProfile (XMP) (1600 mhz 8-8-8-24, dram 1.65v, qpi 1.35v), my CPU VCore is 1.28V, temp is 56C (idle) to 99C (throttled because reached TJMax) (100% load)... It also OC'ed my CPU to 3.2Ghz...
- when I disable XMP (1333 mhz 9-9-9-24, dram 1.5v, qpi 1.1v), my CPU VCore is 1.184v, temp is 40C (idle) to 78C (100% load)... My CPU stays at 2.66Ghz...
Because of that 99C throttled, I pulled up some facts:
- I already reseated my CPU
- 6 fan cooled case. No problem with the case
- Fan maxed out at 2500RPM++. No problem with the fan
- Stock fan, stock paste, stock frequency no OC
So I asks the shop to replace my CPU with the new one... Still the same temp even with this new CPU...
Then I thought maybe 1.28v VCore is too high... So I am thinking to undervolts VCore to makes my CPU temp never goes above 72.7C (TCase spec for i5-750 based on intel websites)
Now there are three reasons why I thought I might hurt my i5-750:
1. It reaches TJMax and therefore the CPU is throttled... Happens about 10 minutes...
2. I realized XMP (which is supposed to X58 that can handle more qpi voltage) gives a wrong qpi voltage to H55. It puts 1.35v while intel says H55 max is 1.21v... It can hurt my i5 IMC... I already put XMP on for 1 ~ 2 hours...
3. The next stupid thing is, I modified the BIOS to makes DRAM voltages at 1.66v and qpi voltages at 1.1v (stock default)... 1.66v - 1.1v means 0.56v delta... More than 0.5v allowed... Look at the quoted posts above... I put this mode on for 1 hours...
After these three things happened, I realized my mistake (because I read some forums by googling)... Then while I'm regretting this, I continued some testing...
And here is my final config:
- 1.1v >> 1.21v qpi voltage
- 133 >> 160Mhz blck (OC 'ing the CPU from 2.66 > 3.3Ghz and my DDR3 from 1333 to 1600Mhz)
- 1.5v >> 1.64v DRAM voltage
- Makes all memory timings and settings from default to the same as XMP
- 1.28v >> 1.0375v VCore CPU voltage (undervolted)
Because of the undervolted VCore, the temp went down from 99C to 72C... Prime95 CPUStress test gives no error... Mission accomplished!
The Question:
1. As the title of this post says... Based on the three occurences above, did you think that my CPU is already damaged? For now it's still doing fine, though... Do I need to replace it again? But this time it's clearly my mistakes because of experimenting without read further
2. Is it normal if I enable XMP the temp at stock VCore reaches 99C?
Wrong. This is not the case for current Intel Platform. For LGA1156(P55/H55/H57), the DRAM voltage (VDIMM) and CPU VTT voltage (QPI/VTT) have to be within a 0.5V delta (If VTT is 1.2V, VDIMM cannot be more than 1.2V+0.5V=1.7V), otherwise you may end up with hardware malfunction and/or damaged IMC of the CPU.
http://www.intel.com/design/corei7/documentation.htm
Intel® Core™ i7-800 and i5-700 Processor Series Datasheet - Volume 1
p.67 and p.69 mentioned the absolute and normal operation specifications of VTT/DDQ(VDIMM).
So here is the whole story:
My Mobo: Gigabyte H55-USB3
My CPU: Intel i5-750
My DDR3: Corsair Dominator TR3x6G-1600C8D >> I mistakenly buy these triple channel kits... I know it would be only dual channel in H55... This are where the problem starts...
- i5-750 default memory controller speed is 1333mhz...
- when I enable XtremeMemoryProfile (XMP) (1600 mhz 8-8-8-24, dram 1.65v, qpi 1.35v), my CPU VCore is 1.28V, temp is 56C (idle) to 99C (throttled because reached TJMax) (100% load)... It also OC'ed my CPU to 3.2Ghz...
- when I disable XMP (1333 mhz 9-9-9-24, dram 1.5v, qpi 1.1v), my CPU VCore is 1.184v, temp is 40C (idle) to 78C (100% load)... My CPU stays at 2.66Ghz...
Because of that 99C throttled, I pulled up some facts:
- I already reseated my CPU
- 6 fan cooled case. No problem with the case
- Fan maxed out at 2500RPM++. No problem with the fan
- Stock fan, stock paste, stock frequency no OC
So I asks the shop to replace my CPU with the new one... Still the same temp even with this new CPU...
Then I thought maybe 1.28v VCore is too high... So I am thinking to undervolts VCore to makes my CPU temp never goes above 72.7C (TCase spec for i5-750 based on intel websites)
Now there are three reasons why I thought I might hurt my i5-750:
1. It reaches TJMax and therefore the CPU is throttled... Happens about 10 minutes...
2. I realized XMP (which is supposed to X58 that can handle more qpi voltage) gives a wrong qpi voltage to H55. It puts 1.35v while intel says H55 max is 1.21v... It can hurt my i5 IMC... I already put XMP on for 1 ~ 2 hours...
3. The next stupid thing is, I modified the BIOS to makes DRAM voltages at 1.66v and qpi voltages at 1.1v (stock default)... 1.66v - 1.1v means 0.56v delta... More than 0.5v allowed... Look at the quoted posts above... I put this mode on for 1 hours...
After these three things happened, I realized my mistake (because I read some forums by googling)... Then while I'm regretting this, I continued some testing...
And here is my final config:
- 1.1v >> 1.21v qpi voltage
- 133 >> 160Mhz blck (OC 'ing the CPU from 2.66 > 3.3Ghz and my DDR3 from 1333 to 1600Mhz)
- 1.5v >> 1.64v DRAM voltage
- Makes all memory timings and settings from default to the same as XMP
- 1.28v >> 1.0375v VCore CPU voltage (undervolted)
Because of the undervolted VCore, the temp went down from 99C to 72C... Prime95 CPUStress test gives no error... Mission accomplished!
The Question:
1. As the title of this post says... Based on the three occurences above, did you think that my CPU is already damaged? For now it's still doing fine, though... Do I need to replace it again? But this time it's clearly my mistakes because of experimenting without read further
2. Is it normal if I enable XMP the temp at stock VCore reaches 99C?
Last edited: