6x1GB Corsair + GSkill + Cellshock (HCF0 + D9GTR) CAS8 overclocks!

eva2000

Gawd
Joined
Apr 14, 2002
Messages
539
First time mismatched 6x1GB triple channel. Still manages to accomodate my push 120x38mm thick Scythe UltraKaze 2000rpm fan with memory module installed in dimm slot 1 :D

System specs
  • Intel Core i7 920 3836A756
  • Thermalright Ultra 120 Extreme + LGA1366 Bolt Thru kit + 2x 120x38mm Scythe UltraKaze 2000rpm 87cfm fans in push/pull
  • DFI LP UT X58-T3EH8 R.A51 - 01/23 beta bios
  • 128MB Gainward FX5200 PCI
  • Ram cooling: 3x 60x25mm Sunon 23.5cfm Maglev fans
  • 80GB Hitachi 7K80 SATAII
  • Pioneer DVD-RW
  • 1KW Corsair HX1000 psu
  • WinXP Pro SP3 full nLite updated hotfixes / Windows 7 beta dual boot

Memory configuration:
  • Dimm slot #1: 1GB G.Skill PC3-12800HZ (Samsung HCF0)
  • Dimm slot #2: 1GB Corsair DDR3-2133C9DF Dominator (Samsung HCF0)
  • Dimm slot #3: 1GB Cellshock PC3-14400 (Micron D9GTR)
  • Dimm slot #4: 1GB Corsair DDR3-2133C9DF Dominator (Samsung HCF0)
  • Dimm slot #5: 1GB Cellshock PC3-14400 (Micron D9GTR)
  • Dimm slot #6: 1GB G.Skill PC3-12800HZ (Samsung HCF0)

IMGP9614.jpg
IMGP9625.jpg


IMGP9629.jpg
IMGP9631.jpg


Seems with 6x1GB AUTO CMD rate = 2T, so for 1T you need to change CMD rate from AUTO to 1T (1N).

Bumped vdimm straight to 1.830v in bios and clocked 6x1GB straight up to DDR3-1606Mhz 8-8-8-24 1T to see how it runs. Everest latency speeds seem to get slower as you add more modules. So in terms of fastest everest latency the order goes = 2x1GB > 3x1GB > 6x1GB. Still 32M time with 6x1GB DDR3-1606Mhz 8-8-8-24 1T is pretty decent.

6x1GB @DDR3-1606Mhz 8-8-8-24 1T at 1.830v bios (DMM measured 1.835v real)





spi32m_9m31s641ms_sg_tn.png


everest.png
 
6x1GB @DDR3-1688Mhz 8-8-8-24 1T at 1.815v bios (DMM measured 1.820v real)
Bumped Bclk to 211 and reduced vdimm from 1.830v bios to 1.815v bios (DMM = 1.820v real) for DDR3-1688Mhz 8-8-8-24 1T.



cpuz_valid.png


Slightly faster 32M Pi times than 3x1GB D9GTR at same main timings even though below 6x1GB was done with considerably looser subtimings for tRFC, tWTP, tRRD and tFAW in bios. Will try tighter timings later.

spi32m_9m07s750ms.png


Everest SPD and bandwidth

everest.png


everest2.png


everest3.png
 
Best I can do keeping voltages the same @879Mhz 8-8-8-24 1T at 1.815v bios (1.820v real)

spi32m_8m41s968ms.png
 
Now that is an interesting mix of memory there! Nice results but, aren't you worried about your CPU warranty? ;)
 
Thanks Mike, why worry Uncore is within intel spec'd 1.35v (CPU VTT) and vdimm is within 0.5v of Uncore (CPU VTT). 1.82v real vdimm and 1.335v real CPU VTT - well pretty close at 0.485v according to DMM measurements :)

Now beyond...

6x1GB 7-7-7-18 1T
Seems this mismatched combo does pretty well tightening both main and subtimings managing up to DDR3-1718Mhz 7-7-7-18 1T at 1.875v bios. Needed a notch or two higher vdimm when subtimings were tightened otherwise I could of lowered vdimm by ~0.25-0.30v.

System specs
  • Intel Core i7 920 3836A756
  • Thermalright Ultra 120 Extreme + LGA1366 Bolt Thru kit + 2x 120x38mm Scythe UltraKaze 2000rpm 87cfm fans in push/pull
  • DFI LP UT X58-T3EH8 R.A51 - 01/23 beta bios
  • 128MB Gainward FX5200 PCI
  • Ram cooling: 3x 60x25mm Sunon 23.5cfm Maglev fans
  • 80GB Hitachi 7K80 SATAII
  • Pioneer DVD-RW
  • 1KW Corsair HX1000 psu
  • WinXP Pro SP3 full nLite updated hotfixes / Windows 7 beta dual boot

Memory configuration:

Super Pi 32M stable
  • DDR3-1606Mhz 7-7-7-18 1T 1.740v bios (1.745v DMM)
  • DDR3-1702Mhz 7-7-7-18 1T 1.830v bios (1.835v DMM)
  • DDR3-1718Mhz 7-7-7-18 1T 1.875v bios (1.880v DMM)

DDR3-1606Mhz 7-7-7-18 1T 1.740v bios (1.745v DMM)​

spi32m_9m02s203ms.png


DDR3-1702Mhz 7-7-7-18 1T 1.830v bios (1.835v DMM)​

spi32m_8m32s110ms.png


DDR3-1718Mhz 7-7-7-18 1T 1.875v bios (1.880v DMM)





spi1m_9s500ms.png


spi32m_8m26s781ms.png
 
Max 6x1GB 8-8-8-24 1T Clocks
Changing to 10x memory multiplier, tried to find max 6x1GB 8-8-8-24 1T clocks and ended up with DDR3-2000Mhz 8-8-8-24 1T at 2.04v bios (2.045v real) only for tried 1M Super Pi as 32M Pi test failed. I dropped bclk and mem clocks to DDR3-1960Mhz 8-8-8-24 1T at same vdimm but Super Pi 32M failed at 22nd iteration. Similar trend occured at DDR3-1940mhz as well.

I recall it's probably the Cellshock PC3-14400 Micron D9GTR holding the 6x1GB modules back as it needed more vdimm for do such clocks even on P35/X48 DDR3 boards closer to 2.1+ volts for around DDR3-1920 8-8-8-x. Decided to stop there as unnecessary stress to i7 IMC ain't needed when I shall have some nice DDR3-2000 CAS7 rated triple channel memory due my way this coming week :D ;)

System specs
  • Intel Core i7 920 3836A756
  • Thermalright Ultra 120 Extreme + LGA1366 Bolt Thru kit + 2x 120x38mm Scythe UltraKaze 2000rpm 87cfm fans in push/pull
  • DFI LP UT X58-T3EH8 R.A51 - 01/23 beta bios
  • 128MB Gainward FX5200 PCI
  • Ram cooling: 3x 60x25mm Sunon 23.5cfm Maglev fans
  • 80GB Hitachi 7K80 SATAII
  • Pioneer DVD-RW
  • 1KW Corsair HX1000 psu
  • WinXP Pro SP3 full nLite updated hotfixes / Windows 7 beta dual boot

Memory configuration:

6x1GB Samsung HCF0 + Micron D9GTR @8-8-8-24 1T at 2.04v bios (2.045v DMM real)





Super Pi 1M = 10.140s

spi1m_10s140ms.png


Super Pi 32M failed at 22nd loop @DDR3-1960Mhz 8-8-8-24 1T

spi32m_err.png
 
damn so 4.0 cpu speed is 'common' in the 920 ?

I'd say 3.6GHZ+ is common, even on air. I don't know about 4.0 but then again I am not monitoring it.

not on air, air 3.6-3.8ghz average, good ones on air like mine 3.8-4.0ghz 24/7. H20 add 200mhz to that.

isnt i7 like 20% faster than core 2 @ same speeds

Depends what you're comparing, if video/auto encoding/transcoding is your thing the difference is quite a large for core i7 aware applications!. Gaming not as much improvement.
 
Back
Top