Core i7/X58 Overclocking Thread

[jcool]

Running 4.45GHZ (165 * 27) with 1,45v with HT enabled 24/7 but has never tried BOINC or anything similar in paralell with gaming...

Wow, nice chip
Anyway you should try it some time, makes the SS really work hard

[eva2000]

maybe HT isn't really needed ?

Right now running TmpgEnc Express 4 and Authoring Works 4 on 15.7GB DVR-MS raw video record file converting it to DVD-video @19x211 = 4009Mhz at 1.328v with no probs. 5hr raw video takes 71mins to convert to dvd-video and 53mins to convert to DIVX with HT enabled and source raw file on 750GB Samsung SATAII hdd. With HT disabled and source file on single 32GB OCZ Core V1 SSD, DIVX conversion took 43mins - nearly 10mins faster

Recorded Obama's Inaugaration from my HTPC Vista Media Center.

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 - 01/06 beta bios
• HD4870x2 stock HSF
• 2x1GB Corsair DDR3-2133C9DF Dominator + 1GB G.Skill PC3-12800HZ (3x Samsung HCF0)
• Ram cooling: 3x 60x25mm Sunon 23.5cfm Maglev fans
• 750GB Samsung SATAII on ICH10R
• 32GB OCZ Core V1 SSD on ICH10R
• Pioneer DVD-RW
• 1KW Corsair HX1000 psu
• Vista Ultimate SP2 Beta 64bit / WinXP Pro SP3 full nLite updated hotfixes dual boot

4009Mhz @19x211Bclk with 8x mem multiplier 3x1GB HCF0 @844Mhz 9-9-9-24

• Record Time = ~5hrs 50mins
• TmpgEnc Authoring Works 4 edited commercials out = ~4hrs 57min DVR-MS file
• TmpgEnc AW4 DVD-Video Conversion with HT/750GB Samsung HDD = ~71min
• TmpgEnc Express4 DIVX Conversion with HT/750GB Samsung HDD @780Kb/S = ~53min
• TmpgEnc Express4 DIVX Conversion without HT/32GB OCZ Core V1 SSD @1000Kb/s = ~43min

without HT/32GB OCZ Core V1 SSD



Full screenshots here.

[day187]

the test isn't a very good comparison, can you run it again , without HT on the 750GB hdd please, then we can get an idea of the gains from the SSD and also the benefit (or lack of) HT

Thanks
Day

[Brahmzy]

I use HT all the time. It speeds up my rendering times immensly. For instance....

My 4.4GHz E8600 rendered a scene I created in 6m 6s.
My 4.0GHz i7 920 renders the same scene in 3m 9s with HT off.
My 4.0GHz i7 920 renders the same scene in 2m 2s with HT on.

Real world proof in VUE 7 - no synthetic benchmark crap. Now just think of what it does when I have 2-3 day render jobs!

THAT is precisely why I bought the i7 vs. a Q9550 or something...

[eva2000]

yeah could of been the SSD speeding things up (which is what i got SSD for hehe)

will retest without HT on 750GB SATA as well as with and without HT on single 32GB SSD, 2x32GB SSD raid 0 and 4x32GB SSD raid 0 on Highpoint 3520LF tomorrow!

[day187]

yeah could of been the SSD speeding things up (which is what i got SSD for hehe)

will retest without HT on 750GB SATA as well as with and without HT on single 32GB SSD, 2x32GB SSD raid 0 and 4x32GB SSD raid 0 on Highpoint 3520LF tomorrow!

nice one Eva

[vikingboy]

What is the literal name of this value in MemSet or the BIOS?

DRAM write recovery time

I'm having a lot of success and speed optimisations playing with memory timings....havent found optimal settings but i think the integrated memory controller is key to optimising these I7s. It certainly the limiting factor. Ive been able to reduce voltage too by playing with RAM timings so something is going on between the two pieces of hardware.

Ive got ddr3 1600 running at 7-8-7-20-1T at 839Mhz 19*210 and I think there is room to increase the 210 significantly with optimised speed timings.

Ive got more testing to do before getting back with screenies but Im optimistic right now. CPU volts 1.35 in CPU-Z and QPI 1.35 in BIOS still.
Ram volts as standard 1.66 in BIOS.

[SteveRo]

Yep, I know. I'm going to try and see if it's still stable with it lowered next. I needed it to get stable even at lower overclocks with these 2 gig modules. I need some 1 giggers like yours so I can cheat. :P

Edit: But isn't the QPI/VTT voltage max more of a thing referencing the max DRAM voltage which they only had to specify because the old DDR3 memory couldn't run at low enough volts to stay within 0.5 volts of the stock QPI/VTT? So if you are running 1.9v on your memory you need at least 1.4v QPI/VTT (probably more lke 1.5v) and this is 'safe' with sufficient motherboard cooling, I think.

So far I am unable to get stable at all ever with 3 x 2gig modules at tight timings at 4.0ghz and beyond without exceeding 1.35v QPI/VTT. Actually, I needed 1.375v QPI/VTT to pass long prime blend runs even with 2 sticks.

Ya, 3x2gb seems to make things much harder for a lot of folks.
I was worried that 3x1gb would not be enough - that I would be going into virtual memory too much.
As a mater of fact I am sure I am going into vm a lot - I don't even notice it though because I am running raid 0 .1ms SSDs.

[Hooti3]

anyone done any o/cing on the 3839A batch..just got one of these..

wondering what they are like...in terms of oc capability...

haven't been able to find much online..so i have come here

[aussie-revhead]

Couldnt you find a good batch first then buy ?

[TheScavenger]

Couldnt you find a good batch first then buy ?

I had my buddy at Microcenter dig through a few boxes to find a particular "B" batch I was looking for

[Hooti3]

Couldnt you find a good batch first then buy ?

nah they said they couldn't because it was picked randomly from their supplier

we will see how it goes...got a decent board now...so will give it a crack

going to make a custom bracket for my all in one swiftech kit tomorrow

should be able to push it a little more with that.

[Postmodum]

anyone done any o/cing on the 3839A batch..just got one of these..

wondering what they are like...in terms of oc capability...

haven't been able to find much online..so i have come here

Hooti3, i have one of those! Really don't know if good or not, never took that in consideration ... but i'm doing 3.6Ghz with stock settings only BCLK @ 180 a multi @ 20

Stress





Idle after the test finished.



And a run of 3DMark Vantage

[Splave]

anyone done any o/cing on the 3839A batch..just got one of these..

wondering what they are like...in terms of oc capability...

haven't been able to find much online..so i have come here

4ghz ht on 1.4 vcore Naekuh
http://www.xtremesystems.org/forums/...d.php?t=213105

[Postmodum]

Btw, my Batch is #3839A569

Don't know anything if its a good one or bad one, but ok! @ 3.6 never had a single problem with it. @ Full it's @ 67ºC so ... its fits for me

[eva2000]

the test isn't a very good comparison, can you run it again , without HT on the 750GB hdd please, then we can get an idea of the gains from the SSD and also the benefit (or lack of) HT

Thanks
Day

I use HT all the time. It speeds up my rendering times immensly. For instance....

My 4.4GHz E8600 rendered a scene I created in 6m 6s.
My 4.0GHz i7 920 renders the same scene in 3m 9s with HT off.
My 4.0GHz i7 920 renders the same scene in 2m 2s with HT on.

Real world proof in VUE 7 - no synthetic benchmark crap. Now just think of what it does when I have 2-3 day render jobs!

THAT is precisely why I bought the i7 vs. a Q9550 or something...

Guys i'm doing dvd-video to xvid and divx conversions in tmpgenc xpress 4 latest version supports core i7 but seems the gain is with ssd as source drive as HT and no HT result in about same times - with no HT slightly faster

so what other encoding/conversion app would you guys suggest that's easy to setup and run ?

[Chicken Patty]

not a bad CPU score, about 200 points higher than my vista run, card was clocked lower here so overall score is less, but CPU performed better

[TheScavenger]

not a bad CPU score, about 200 points higher than my vista run, card was clocked lower here so overall score is less, but CPU performed better

Looks good. Whats your batch#?

[Pisklink]

Is there anyone who uses a D-tek Fuzion yet on an i7 CPU? I would like to know which nozzle (if any?) to use best with Core i7 and D-Tek Fuzion V1. I guess the quad nozzle isn't of any use anymore with i7 since all cores are intergrated into a single die. Anybody a suggestion?

[T_Flight]

They're saying don't use any nozzles in the DTek. Since the i7 is not longer two separate cores, and is on a single die you want all that water going right at it. Head over to the Water Cooling forum. There's alot of info on that that's already been posted. If you need help, just ask in there. It's one of the best if not THE best Watercooling forum on the planet.

There are a couple people using those DTek blocks, but they had custom made mounts done up for them. DTek still hasn't got the Pro Mounts done for them yet. One guys is making plates for them though.

Warning: Do not buy or sell in the watercooling section. It's a good way to get gone, and they will never let you in the FS section after that again. That includes PM's too. Do it in the For Sale section. I saw a warning about that in there so just a freindly reminder.

[JudgmentJay]

anyone done any o/cing on the 3839A batch..just got one of these..

wondering what they are like...in terms of oc capability...

haven't been able to find much online..so i have come here

Just built my system and am definitely still tweaking, but it seems like a decent batch. With HT off, 200x20 has been priming for ~3.5 hours at 1.328v with a max temp of 66c (TRUE120). I would enable HT, but my board doesn't seem to like it very much... going to have to play with that

Over 4ghz takes a significant voltage increase, but my temps were still decent. At 200x21 (Bloodrage Turbo always-on feature) and 1.424v vcore I was only at 75c max after 2 hours of prime, but it errored out. At 210x20 it wasn't even stable at 1.46v vcore and 1.4v VTT/QPI. Are such absurd voltage increases over 4ghz normal?

[RealTelstar]

not a bad CPU score, about 200 points higher than my vista run, card was clocked lower here so overall score is less, but CPU performed better

Can you do a comparison with everest?
In my case W7 is SLOWER than vista.

[Bill Reid]

OK guys total newbie here... can you help me out? Got my first watercooled rig together and this is my first attempt to overclock. Please take alook at my setup and let me know what you think. Any advice much appreciated.
Thanks,

920/P6T 1102 Bios, CPU Batch 3841A403
G.Skill F3-12800CL8T-6GBHK

I lifted these settings from another member that posted them here. So far no hangups...

CPU 1.35V
QPI/DRAM 1.35V
IOH 1.18V
IOH PCIE 1.50V
ICH 1.2 V
ICH PCIE 1.50V
DRAM BUS 1.64V
CPU Differential Amplitude 700mV
CPU Clock Skew Delay 100ps
CPU Spectrum Disabled
IOH Clock Skew Delay 100ps
PCIE Spread Spectrum Disabled

Here's my rig:


My results: What else do I need to call it good? BTW, regarding temps, pic was taken about 30 seconds after test completed.

[Pisklink]

They're saying don't use any nozzles in the DTek. Since the i7 is not longer two separate cores, and is on a single die you want all that water going right at it. Head over to the Water Cooling forum. There's alot of info on that that's already been posted. If you need help, just ask in there. It's one of the best if not THE best Watercooling forum on the planet.

There are a couple people using those DTek blocks, but they had custom made mounts done up for them. DTek still hasn't got the Pro Mounts done for them yet. One guys is making plates for them though.

Many thanks. I already asked that question in de WC section but only two members replied. I thought that because there aren't many i7 D-tek users yet that there would be a lack of experience. That's why I asked here as well and was hoping someone who already started OC'ing his i7 with a D-tek could provide an answer. In the WC section they tol be the same thing as you did. I guess I will give it a go then without any nozzle and see how it works out.

[SteveRo]

OK guys total newbie here... can you help me out? Got my first watercooled rig together and this is my first attempt to overclock. Please take alook at my setup and let me know what you think. Any advice much appreciated.
Thanks,

920/P6T 1102 Bios, CPU Batch 3841A403
G.Skill F3-12800CL8T-6GBHK

I lifted these settings from another member that posted them here. So far no hangups...

CPU 1.35V
QPI/DRAM 1.35V
IOH 1.18V
IOH PCIE 1.50V
ICH 1.2 V
ICH PCIE 1.50V
DRAM BUS 1.64V
CPU Differential Amplitude 700mV
CPU Clock Skew Delay 100ps
CPU Spectrum Disabled
IOH Clock Skew Delay 100ps
PCIE Spread Spectrum Disabled

Here's my rig:


My results: What else do I need to call it good? BTW, regarding temps, pic was taken about 30 seconds after test completed.

Bill, Beautiful set up! looks like you are on your way!