Gigabyte P67A-UD7 + i7 2600K

[eva2000]

First Sandy Bridge cpu run with Intel Core i7 2600K L038A660 with Gigabyte P67A-UD7. The Gigabyte P67A-UD7 was flashed from shipping bios F3 to F7f and immediately clocked it up on both cpu and memory end. 2x2GB A-Data 2000X - Elpida BBSE modules left on auto main and sub timings and set to 21.33 mem multi. Later tried Elpida Hyper based memory modules with F8x bios.

System:

• Intel Core i7 2600K L038A660
• Corsair H50-1 + 120x25mm Thermaltake 81cfm fan
• Gigabyte P67A-UD7 F7f bios
• 2x2GB A-Data 2000X Elpida BBSE
• 256GB Crucial RealSSD C300 Intel SATA 6G port IDE mode
• Sapphire HD5870 Cat 11.1
• 1200W Silverstone OP1200
• WinXP SP3 nlite fully updated

Quick Super Pi 32M runs untweaked

• 4.8Ghz 1.335v vcore LLC 1
• 5.0Ghz 1.390v vcore LLC 1
• 5.2Ghz 1.500v vcore LLC 2

4.80Ghz 1.335v​

5.0Ghz 1.390v​

5.2Ghz 1.500v​

 

[eva2000]

Elpida Hyper Testing on Gigabyte P67A-UD7​

The A-Data 2000X Elpida BBSE didn't like 7-8-7x at 2133Mhz as it's a low vdimm 1.58-1.60v type of modules so higher vdimm needed would error out. So swapping over to some Elpida Hyper testing. Tried one 2x2GB SuperTalent 2200C8 Elpida Hyper kit and while it did 32M pi at 2133mhz 8-8-8-24 at 1.66v vdimm but it only did it at 2T and didn't like 1T. So swapped that out for some Corsair Dominator GT 2000C7 7-8-7-24 Elpida Hyper testing which was much better once figured out a few things for the Gigabyte P67A-UD7.

Did notice the subtimings default in SPD on Super Talent 2200C8 were tighter in some areas than Corsair Dominator GT 2000C7. For instance tFAW = 32 on Super Talent vs 40 on Corsair.

While testing Corsair Dominator GT 2000C7, I think I have found a bclk hole somewhere. 100.2bclk which is what I have been testing with, is causing failed oc'd messages at 2133mhz 7-7-7-24, but changing to 100.1bclk or 100.6bclk or 100.7bclk and even up to 102.7bclk can memtest fine!

First run at 2131Mhz 7-7-7-24 leaving subtimings on auto except tRFC 74 and manually setting 1T and in memtest86+ v4.20 quicky test was capable at 1.70v vdimm, but needed 1.74v vdimm for 32M Pi from 2131-2139Mhz. Did a quick bclk bumping via ET6 in windows to validate at 2202Mhz 7-7-7-24 at 1.74v.

System:

• Intel Core i7 2600K L038A660
• Corsair H50-1 + 120x25mm Thermaltake 81cfm fan
• Gigabyte P67A-UD7 F7f bios
• 2x2GB Corsair Dominator GT 2000C7 - Elpida Hyper
• 256GB Crucial RealSSD C300 Intel SATA 6G port IDE mode
• Sapphire HD5870 Cat 11.1
• 1200W Silverstone OP1200
• WinXP SP3 nlite fully updated

 

[eva2000]

5001Mhz at 1.380v
2x2GB Corsair Dominator GT 2000C7
@ DDR3-2134Mhz 7-7-7-24 at 1.74v​

Only tightened tRFC 88 and set 1T CMD rate manually for first run @5001Mhz. Strange as I mentioned before with 21.33Mhz memory multiplier at 7-7-7-24 timings with 34x cpu multi 100.2bclk had issues, but now at 50x multi 100.1bclk had issues but 100.2bclk runs fine

 

[eva2000]

As explained here, after remounting Corsair H50-1 with proper mount and thermal paste contact, I managed to a bit more stability when it comes to all core benchmarking i.e. wprime 1024M at 5.2Ghz 1.490v bios (1.512v idle while ET6 reports 1.50v) windows was passable. But stil this Core i7 2600K L038A660 is hitting a ceiling at 5202Mhz on Gigabyte P67A-UD7 F8x bios.

5215Mhz was max validation and benchmarks below at 5202Mhz 1.512v idle.

5215Mhz 1.512v idle
4GB Corsair Dominator GT 2000C7
@2139Mhz 7-7-7-24 1.74v​

5202Mhz 1.512v idle
4GB Corsair Dominator GT 2000C7
@2134Mhz 7-7-7-24 1.74v​

Super Pi 32M

Aida64

MaxMem

Hexus Pifast

WPrime v1.55 32M & 1024M

System:

• Intel Core i7 2600K L038A660
• Corsair H50-1 + 120x25mm Thermaltake 81cfm fan
• Gigabyte P67A-UD7 F8x bios
• 2x2GB Corsair Dominator GT 2000C7 - Elpida Hyper
• 256GB Crucial RealSSD C300 Intel SATA 6G port IDE mode
• Sapphire HD5870 Cat 11.1
• 1200W Silverstone OP1200
• WinXP SP3 nlite fully updated

 

[dinos22]

These rigs hit the ground running with you don't they hehehe

 

[eva2000]

They certainly try to

Some load temp testing just to see where Corsair H50-1 cooling is at for Core i7 2600K L038A660. I did quick 3 pass Linx run and took screenshots at exact same time in the benchmark run.

Corsair H50-1 + Thermaltake 120x25mm 81cfm fan push
vs
Corsair H50-1 + Thermaltake 120x25mm 81cfm fan push + Spire 120x25mm 75cfm fan pull​

Only resulted in 0.5C to 1.5C difference which evened out to around 0.5C the longer the test ran.


Corsair H50-1 + Thermaltake 120x25mm 81cfm fan push

Linx

Corsair H50-1 + Thermaltake 120x25mm 81cfm fan push + Spire 120x25mm 75cfm fan pull

Linx

then ran 2hr Prime95 which stresses cpu more than Linx for Sandy Bridge cpus.