Lynnfiled LGA1156 i5/i7 Overclocking guide request

Lynnfield CPU overclocking guide required?

  • Give me! Give me! Give me!

    Votes: 5 62.5%
  • Why would a pro like me need it?!

    Votes: 3 37.5%

  • Total voters
    8

andy5174

Distinguished
Mar 3, 2009
2,452
0
19,860
Is an overclocking guide for Lynnfield CPU required? Please vote!

Basic Lynnfield overclocking guide
(source: Computer Lounge Forum)

Settings for 3.6 GHz or 3600 MHz – possible on nearly every chip.
Ai Overclock Tuner - Manual
CPU Ratio Setting – 20
DRAM Frequency – either 1440 or 1800, whichever is closest to the original operating frequency of your RAM
Intel Speedstep Tech - Disabled
Intel Turbo Mode Tech - Disabled
QPI Frequency – 180
PCIE Frequency - 100
DRAM Timing Control – Set this to be the Timings of your specific RAM, i.e. 9-9-9-26. This setting is done in order, first setting 9, second 9, third 9, fourth 26.
DRAM Command Rate – 2T
CPU Voltage – 1.0v – 1.3v, start at 1.3v and stress test, then decrease and retest stability.
CPU PLL Voltage - Auto
IMC Voltage - Auto
PCH Voltage - Auto
DRAM Bus Voltage – Set this to the Voltage of your Ram
Load-Line Calibration – Enabled
CPU Clock Skew – Auto
PCI-E Clock Skew - Auto

Settings for 3.8 GHz or 3800 MHz – please note, this may not be possible on every chip.
Ai Overclock Tuner - Manual
CPU Ratio Setting – 20
DRAM Frequency – either 1440 or 1800, whichever is closest to the original operating frequency of your RAM
Intel Speedstep Tech - Disabled
Intel Turbo Mode Tech - Disabled
QPI Frequency – 190
PCIE Frequency - 100
DRAM Timing Control – Set this to be the Timings of your specific RAM, i.e. 9-9-9-26. This setting is done in order, first setting 9, second 9, third 9, fourth 26.
DRAM Command Rate – 2T
CPU Voltage – 1.0v – 1.3v, start at 1.3v and stress test, then decrease and retest stability.
CPU PLL Voltage - Auto
IMC Voltage - Auto
PCH Voltage - Auto
DRAM Bus Voltage – Set this to the Voltage of your Ram
Load-Line Calibration – Enabled
CPU Clock Skew – Auto
PCI-E Clock Skew - Auto
 

andy5174

Distinguished
Mar 3, 2009
2,452
0
19,860

You'd better pull out your CPU and check their connectors. Replace the m/b with P55A series which are with the LOTES socket asap if there's any sign of burnout.

BTW, I personally will get at least a UD4 for OCing like 4.0GHz.
 

rodney_ws

Splendid
Dec 29, 2005
3,819
0
22,810


I'm not really sold on the idea/theory that a 3.6 GHz i5 is going to have issues on a Foxconn socket... even long term.
 

axffox

Distinguished
Dec 14, 2009
3
0
18,510
you should put stuff about stress testing... I'm the owner of that OC guide @ overclock.net, I wondered why this got the first google result :p


I've bought a Xeon X3440 to see how it compares to my i7 860 in regards to overclocking. Too bad the i7 860 doesn't overclock as good as the i5s.
 

bob5568

Distinguished
Jan 28, 2006
147
0
18,690
It may be a function of motherboard, but I'm finding my build is happier at 3.6 (18x200), than 3.6 (20x180). running the bclk at 200mhz also lets me stock clock memory at 8x200.

Only thing I'm missing is good case cooling. This is causing my 100% load temps to creep higher than I want. That's next on my list of things to do.

Cheers,
Bob

i7 860/ga-p55A-UD4p/G.Skill (4x2gb) 1600 ddr3/8800 gts (g92)/Corsaid modular 520W/win7 64
 

andy5174

Distinguished
Mar 3, 2009
2,452
0
19,860

Your Vtt/IMC and PLL will need to be higher with higher BCLK. What Vtt and PLL voltage do you use?
 

RJR

Distinguished
Jan 7, 2009
1,065
0
19,360
That is the poorest excuse for an OCing guide I've ever seen, LOL!!!

Yeah, set your blck & ram and leave everything else on AUTO, thats OCing, HAHAHA.

Watch out you might not be able to hit 3.8 GHz., LOL !!!!!!!!

This is just pitiful.

 

bob5568

Distinguished
Jan 28, 2006
147
0
18,690


Yes, you are right. I had to raise Vtt to 1.24v and VPLL to 1.86v.

My Vcore is at 1.225v.

I don't consider these loads excessive, but I'm open to discussion.

Best,
Bob
 

andy5174

Distinguished
Mar 3, 2009
2,452
0
19,860

Are you aware of the absolute maximum Vtt for LGA1156 being 1.21V instead of 1.35V?

What's your 100% load Vcore shown in CPU-Z? The BIOS Vcore is insignificant due to it being different for different BIOS version and M/B.
 

bob5568

Distinguished
Jan 28, 2006
147
0
18,690
Andy, what is your source for the Vtt data? I know that testers at Anandtech were running Vtt way higher than I am. I'll look up my 100% load Vcore when I get home from work. I don't have that at the moment.

Bob
 

bob5568

Distinguished
Jan 28, 2006
147
0
18,690
Disregard....I found your source.

I also found significant discussion on the topic. Interesting. Thanks for the heads up, Andy. I haven't decided if I'll back Vtt down yet, but I might.

Bob
 

notty22

Distinguished
I don't think the qpi/vtt voltage will hurt anything. On my gigabyte ud4p bios it goes to pink at around 1.21. Thats all I need. Of note OCZ tech support regularly tells customers to up to 1.35 attempting to get memory stable. I'm not advocating thats "good" just noting it. This setting is effecting a controller inside the cpu now instead of the chipset , correct ?
http://www.ocztechnologyforum.com/forum/showthread.php?t=66325
 

bob5568

Distinguished
Jan 28, 2006
147
0
18,690
I'm not too worried, but having learned about the intel spec for the first time today, I do plan to explore what I can acheive with a 1.21 vtt. There is so much headroom between my current vcore and max vcore that I may be able to stabilize with a different balance.

 

bob5568

Distinguished
Jan 28, 2006
147
0
18,690
Based on the intel spec for vtt (thanks Andy!), I've reset my oc to 3.6 (20x 180). Vtt returned to 1.21v, and Vcore from cpu-z at 1.248v at 100% load induced from prime95. Stable thru 8 cycles of prime 95, but temps are higher than I like. My case is a problem, its a 10 year old enlight. Classy for its day, its day is way in the past. I have a case supposed to arrive on Monday with 2 fans in the top, one in the back and 2 in the front...a Lian Li. Once I rebuild, it will be fun to see how it improves.
 

andy5174

Distinguished
Mar 3, 2009
2,452
0
19,860


FYI:
My 100% load CPU-Z Vcore is one notch above 1.20V which is also the case for the i5 OC done by TOM!

Hence, I think 100% load CPU-Z Vcore around 1.20V should be sufficient for i5 at 3.6GHz.

 

andy5174

Distinguished
Mar 3, 2009
2,452
0
19,860
BTW, my next notch is 1.216V which is higher than TOM's 1.208V.

Not sure how TOM determine the stability, but I determine my stability with 24hrs Prime95 Small FFTs test and 25hrs-1mins prime95 Large FFTs test stable!

Note: Although small FFTs is well known for testing CPU stability, your cpu CAN BE unstable even if you get 24hrs stability on this test!

For instance, I get small FFTs stable for 24hrs with 1.20V, but fail Large FFTs within 5 hours in most of my SEVERAL Large FFTs tests!

Firstly, I thought it was Vtt, PLL or/and PCH voltage being too low instead of Vcore issue, because I passed 24hrs Small FFTs(CPU) test and 14hrs Memtest86+(RAM) test.

However, my system was not stable even I changed all of those values to their absolute maximum. Thus, I started tweaking the Vcore once again by making it 1.216V and
23hrs Large FFTs stable after this but do failed on 23hrs-11mins.

Finally, I took my PCH(responsible for I/O interface) one notch above the stock(1.05V) and get 25hrs-1mins stable!