You are not registered yet. Please click here to register!
First i7 920 D0 Step = mixed bag of results - i4memory.com - different look at memory
Tags Register Blogs FAQ Members List Calendar Mark Forums Read

Notices

Reply
 
Thread Tools
Old 19-05-2009, 04:15 AM   #1
eva2000
Administrator
eva2000's PC Specs
 
Join Date: Jul 22 2004
Location: Brisbane, Australia
Posts: 23,008
twitter.com/i4memorycom Facebook Page livestream.com/i4memorycom
Blog Entries: 42
Smile First i7 920 D0 Step = mixed bag of results

System:
  • i7 920 3851A368
  • EK Supreme LT 1366 / FESER 480 / MCP655
  • DFI UT X58-T3EH8 04/28 beta
  • 2x HD4870x2 Crossfire (Powercolor + Gigabyte flashed to Asus Top bios) with ATI Win7 9.4 (8.612 WHQL)
  • 6GB Corsair Dominator GT DDR3-2000Mhz 7-8-7-20
  • 750GB Samsung SATAII
  • Silverstone OP1200 PSU
  • Triple boot - WinXP Pro SP3 / Vista Ultimate 64Bit SP2 / Win7 Ultimate 64bit RC

Got my first D0 stepping Core i7 920 cpu hoping it will be better than my gem of a C0 step i7 920 3836A756 which managed 4769Mhz on TRUE120 AIR.

Stock idle and load temps look pretty normal to me with 28-30C room ambients. The VCORE = AUTO resulting within windows default vcore voltage is slightly higher on this D0 at 1.200v idle / 1.184v load compared to both my i7 920 3836A756 C0 cpus which both have vcore AUTO at 1.168v idle / 1.152v load.

eva2000 is offline   Reply With Quote
Old 19-05-2009, 04:15 AM   #2
eva2000
Administrator
eva2000's PC Specs
 
Join Date: Jul 22 2004
Location: Brisbane, Australia
Posts: 23,008
twitter.com/i4memorycom Facebook Page livestream.com/i4memorycom
Blog Entries: 42
Update: Can confirm see further below for stability tests at 1.312v 4000Mhz
Interesting the vcore required for HyperPi 8x 32M PI test might be just for that, as I started testing Prime95 small FFTs 1hr stable at 20x200 = 4000Mhz at 1.312v and Prime95 Blend custom 5.4GB test at same 1.312v and it's running without problems. So HyperPi 8x32M Pi test might come down to memory/voltage related settings. Will try lower vcore after the custom Blend test and then LinX.
Unfortunately, this D0 stepping i7 920 3851A368 is a mixed bag of sorts with 1 of the 4 cores being weaker than the rest. In Hyper Pi 32M x 8 instance pi test, one core errors out while rest pass at 1.296v vcore. But seems it only for HyperPI 8x32M as Prime95, LinX have no problems at 1.312v so far - see below here.

20x200bclk = 4000Mhz
  • Super Pi 8x 32M HyperPi instances needed 1.344v vcore to pass BUT
  • Super Pi 1x 32M single instance needed much less 1.296v vcore to pass being able to keep same 1.296v vcore all the way to 4104Mhz single Super Pi 32M without problems.

Boots easily at 1.248v



Single Super Pi 32M

@4000Mhz & @4104Mhz





Super Pi 32M x 8 instances (Hyper Pi)

@4000Mhz



And want to see the fastest i7 920 on the planet courtesy of Windows 7 and CPUZ flakiness during Hyper Pi 8x 32M Pi test ? How about 12200Mhz - 61x200bclk LOL


Last edited by eva2000; 19-05-2009 at 11:58 PM.
eva2000 is offline   Reply With Quote
Old 19-05-2009, 04:16 AM   #3
eva2000
Administrator
eva2000's PC Specs
 
Join Date: Jul 22 2004
Location: Brisbane, Australia
Posts: 23,008
twitter.com/i4memorycom Facebook Page livestream.com/i4memorycom
Blog Entries: 42
Max BCLK Testing - first look


Now the good news is bclk to CPU VTT volts required is much better! Initial memtest86+ v2.11 has me scaling as follows with default IOH/ICH volts of 1.11v and PCI-E frequency at 100mhz and 104mhz the same results:
  • 223bclk - 1.39v CPU VTT
  • 228bclk - 1.44v CPU VTT
  • 230bclk - 1.47v CPU VTT

Getting into memtest86+ v2.11 @230bclk was easier than with both my twin identical batch i7 920 3836A756 C0 stepping cpus and using less CPU VTT on average. For getting into windows needed more CPU VTT of course - 1.52v CPU VTT for 230bclk and Super Pi 1M so far until I figure out the right combination of voltages and settings for more





All above was done within 1st hour of installing this i7 920 3851A368 and with 2x HD4870x2 Crossfire setup, so shall see what I can squeeze out of this cpu later. Vcore wise doesn't look that great compared to other i7 920 D0 steppings, but bclk to CPU VTT required relationship looks to have alot of potential

Next is to see what the ceilings are for max HT enabled raw cpu mhz frequency and max triple channel memory frequency (how good the memory controller will be).
eva2000 is offline   Reply With Quote
Old 19-05-2009, 04:16 AM   #4
eva2000
Administrator
eva2000's PC Specs
 
Join Date: Jul 22 2004
Location: Brisbane, Australia
Posts: 23,008
twitter.com/i4memorycom Facebook Page livestream.com/i4memorycom
Blog Entries: 42
Looks like indeed vcore was a HyperPi 8x 32MB test related as I ran the following 4000Mhz at 1.312v (haven't tried lower vcore yet) without problems

Powermate wattage meter system power draw from wall in bold.
  • 1hr Prime95 Small FFTs (~533W - 544W)
  • 1hr Prime95 Custom Blend 5400MB memory allocated (~550W - 562W)
  • LinX 0.5.8.2 10 pass with 768MB allocated (what most folks incorrectly use instead of all memory test) (~562W - 570W)
  • LinX 0.5.8.2 10 pass with ALL memory allocated (~565W - 572W)

1hr Prime95 Small FFTs






1hr Prime95 Custom Blend 5400MB memory allocated








LinX 0.5.8.2 10 pass with 768MB allocated






LinX 0.5.8.2 10 pass with ALL memory allocated






System:
  • i7 920 3851A368
  • EK Supreme LT 1366 / FESER 480 / MCP655
  • DFI UT X58-T3EH8 04/28 beta
  • 2x HD4870x2 Crossfire (Powercolor + Gigabyte flashed to Asus Top bios) with ATI Win7 9.4 (8.612 WHQL)
  • 6GB Corsair Dominator GT DDR3-2000Mhz 7-8-7-20
  • 750GB Samsung SATAII
  • Silverstone OP1200 PSU
  • Triple boot - WinXP Pro SP3 / Vista Ultimate 64Bit SP2 / Win7 Ultimate 64bit RC

Last edited by eva2000; 19-05-2009 at 11:57 PM.
eva2000 is offline   Reply With Quote
Old 19-05-2009, 04:16 AM   #5
eva2000
Administrator
eva2000's PC Specs
 
Join Date: Jul 22 2004
Location: Brisbane, Australia
Posts: 23,008
twitter.com/i4memorycom Facebook Page livestream.com/i4memorycom
Blog Entries: 42
First run with Corsair Dominator GT 2000C7 kit at DDR3-2000Mhz 7-8-7-20 1T with i7 920 3851A368 D0 stepping and no problems with HyperPi 8x 32M Pi other than needing more vcore for just HyperPi compared with Prime95/LinX - 1.344v vs 1.312v. Ran 3dmark06 after HyperPi finished


Last edited by eva2000; 20-05-2009 at 02:09 AM.
eva2000 is offline   Reply With Quote
Old 19-05-2009, 04:16 AM   #6
eva2000
Administrator
eva2000's PC Specs
 
Join Date: Jul 22 2004
Location: Brisbane, Australia
Posts: 23,008
twitter.com/i4memorycom Facebook Page livestream.com/i4memorycom
Blog Entries: 42
So far every indication is this i7 920 3851A368 D0 is a slightly better version of my i7 920 3836A756 C0 twins, even bclk'ing on 20x is slightly easier - quick 32M run - yeah not that great on vcore compared to other D0s.


Last edited by eva2000; 20-05-2009 at 04:23 AM.
eva2000 is offline   Reply With Quote
Old 19-05-2009, 04:16 AM   #7
eva2000
Administrator
eva2000's PC Specs
 
Join Date: Jul 22 2004
Location: Brisbane, Australia
Posts: 23,008
twitter.com/i4memorycom Facebook Page livestream.com/i4memorycom
Blog Entries: 42
First overclock for GT memory on this i7 920 D0 and looking pretty good so far @DDR3-2148Mhz 7-8-7-20 1T at 1.875v bios set vdimm. Ignore the vcore used, I set it higher than needed to focus on memory clock first.

HyperPi 32M x8


Single Super Pi 32M


Everest Ultimate 5.0.1.1732


System:
  • i7 920 3851A368
  • EK Supreme LT 1366 / FESER 480 / MCP655
  • DFI UT X58-T3EH8 04/28 beta
  • 2x HD4870x2 Crossfire (Powercolor + Gigabyte flashed to Asus Top bios) with ATI Win7 9.4 (8.612 WHQL)
  • 6GB Corsair Dominator GT DDR3-2000Mhz 7-8-7-20
  • 750GB Samsung SATAII
  • Silverstone OP1200 PSU
  • Triple boot - WinXP Pro SP3 / Vista Ultimate 64Bit SP2 / Win7 Ultimate 64bit RC

Last edited by eva2000; 20-05-2009 at 03:38 PM.
eva2000 is offline   Reply With Quote
Old 19-05-2009, 04:16 AM   #8
eva2000
Administrator
eva2000's PC Specs
 
Join Date: Jul 22 2004
Location: Brisbane, Australia
Posts: 23,008
twitter.com/i4memorycom Facebook Page livestream.com/i4memorycom
Blog Entries: 42
Pushing my i7 920 3851A368 D0 to the max with HT enabled now. Seems cpuz validation is higher than my i7 920 3836A756 C0 when HT enabled - 4710mhz vs 4669mhz and D0 also has slightly higher max HT Enabled Super Pi 32M cpuz speed - 4685mhz vs 4618mhz. Only difference is for C0 tests I used single 8800GT while D0 was using 2x HD4870x2 Crossfire installed.

Like my C0 stepping, this D0 seems to handle high vcore with relatively low temps !!!

System:
From http://i4memory.com/111898-post6.html

i7 920 3836A756 C0 TRUE120 Air
  • Max 32M Super Pi = 4473Mhz HT enabled / 4652Mhz HT disabled
  • Max Validation = 4473Mhz HT enabled / 4769Mhz HT disabled

i7 920 3836A756 C0 EK Supreme LT / Feser480 H20
  • Max 32M Super Pi = 4618Mhz HT enabled / 4680Mhz HT disabled
  • Max Validation = 4669Mhz HT enabled / 4744Mhz HT disabled

VS

i7 920 3851A368 D0 EK Supreme LT / Feser480 H20
  • Max 32M Super Pi = 4685Mhz HT enabled / TBD HT disabled
  • Max Validation = 4710Mhz HT enabled / TBD HT disabled

All benchmarks done in Win 7 Ultimate 64bit RC.

@4710Mhz at 1.64v





@4685Mhz at 1.616v



@4643Mhz at 1.584v



@4560Mhz at 1.536v


Last edited by eva2000; 22-05-2009 at 01:30 AM.
eva2000 is offline   Reply With Quote
Reply

Bookmarks

Tags
230bclk, 3851a368, corsair dominator gt, dfi x58, i7 920 d0


Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools


Similar Threads for: First i7 920 D0 Step = mixed bag of results
Thread Thread Starter Forum Replies Last Post
i7 920 batch and oc'ing results - please share eva2000 General Intel Motherboards / CPU 102 10-06-2010 04:21 AM
Second i7 920 D0 Stepping 3851A368 #2 results eva2000 DFI Intel Motherboard / CPU 13 29-05-2009 05:11 AM


All times are GMT +11. The time now is 05:16 AM.

no new posts