Core i7 Overclocking by Batch #

[cky2k6]

As I think a lot of us have noticed, there are almost two lines of bloomfield cpus, with the two main batches (a&b) differing greatly in the vcore needed to achieve an oc.

So to all core i7 owners, please respond to this thread with your batch #, a screenshot of cpu-z, realtemp or coretemp, and prime95 on small fft showing your load temps, and what cooling you are using. There is a good deal of variation among core i7 chips with load temps and voltages, and its interesting to see the correlation with the batch #s.

For those that don't know, the batch # can be found on the retail box, on the bottom of the sticker, under the label of FPO/Batch #. If this thread gets a large enough sample of batches, I will post up a chart of the varying results.

To start off, here is mine: 3837B076


cooled by cooler master 212 (under vigor brand)

Now as promised, here is a chart of all of the results posted here and the ones that Avan gathered from the main oc'ing thread. http://www.xtremesystems.org/forums/...7&postcount=35

****New Announcement**** I have published this spreadsheet for all to edit on google docs. If you don't see your info posted, feel free to edit in your information while adhering to the layout http://spreadsheets.google.com/ccc?k..._RqTL5QA&hl=en

[SmokeyTheBandit]

I'll be nr 2 then.
This will be great for anyone buying an i7

Mine is cooled by an EK Supreme (WC)
Batch : 3837A909

[Nilos]

Mine is 3837B194



Aircooled by a Thermalright Ultra Extreme. Date must be 03/01/2009 of course!

[Leeghoofd]

Mine is 3837B194

Is that my new CPU nilos You already on the Giga Extreme version now ?

[ReverendMaynard]

batch in sig.

[Splave]

sub-d will edit with info when i get home

[FZ1]

I'll be nr 2 then.
This will be great for anyone buying an i7

Mine is cooled by an EK Supreme (WC)
Batch : 3837A909

That's a great chip! I haven't been able to boot higher than 3.8GHz on my "A" batch, at least below 1.33v although I haven't had time to mess with it though.

[cky2k6]

come on guys, lets keep this going.

[dengyong]

This is not max, it's my 24/7 settings......still tweaking
Batch #3838B079
1.296v under load, HT on
The rest is in my sig.

[scaramonga]

HT off.



Batch #3838A505

[CodeRush]

Batch # 3835A699
24/7 settings, HT off.
Cooled by handmade copper waterblock.

[beast200]

Hi

i have a 920 batch 3835A766 @ 21 x 200 with tubro on and HT off
cpu volt 1.425
watercooled
temps idle 39 40 42 40
load 69 69 72 70
temp at idle abit high but no problems

cheers

mark

[dengyong]

[mikesrex]

I have an i7-920 batch 3841A507 and will let you guys know what it does on air with a TRUE when I get a mobo and some RAM to run with it.

[JWMc]

i7-920 3841A383 from Newegg:

These are all with LinX x 20 and Prime blend stability:

4.0ghz (20x200 turbo off) @ 1.30000v bios, 1.264v load with HT off, 1.335v QPI, 8-8-8-21-1T, 1.65v DRAM (1.66 BIOS), max LinX temp 67c on TRUE 120 push-pull
4.0ghz (20x200 turbo off) @ 1.40625v bios, 1.360v load with HT on, 1.375v QPI, 8-8-8-21-1T, 1.65v DRAM (1.66 BIOS), max LinX temp 83c on TRUE 120 push-pull.

Edit, added recently:
3.8ghz (19x200 turbo off) @ 1.21875 bios, 1.184v load, 1.20v idle, HT enabled, 1.275v QPI, 8-8-8-21-1T, 1.65v DRAM (1.66 BIOS), max LinX temp 63c on TRUE 120 push-pull. I got HT disabled down to 1.154v load I think for LinX but I had to bump it all the way back up to 1.184v for Prime blend even with HT off.

Higher voltages were required for Prime blend stability vs LinX x 20 even though LinX has highest temps. I think this is due to memory system load because I had to increase vQPI a bit for HT as you can see.

http://www.xtremesystems.org/forums/...postcount=1273

[Bullu$]

batch:3837A909 (Same as SmokeytheBandiet)

HT >enabled
Turbo >disabled
max bclock 222.5 x 19

[jcool]

My 3838A685 920 sucks. It hates the 20x and despises the 21x, only runs well with 19x which means I can't OC this sucker past 4,05Ghz stable really. Currently on 190x21 semi-stable, needs 1,5V Vcore though (+ I am on phase)
Avoid 3838A...

[joger]

3837A715 here, under the stock cooler @1.352v under stress, 1.094 idle.


My mobo is the EVGA X58 TRI-SLI with the original bios. I'm having a very hard time overclocking this thing, uncore and QPI multipliers are on minimum and I still get instant bluescreen at 3.7ghz.
Mem is 3x G.Skill DDR3-1333 CL9, but it won't work if I set it to multi's so that they're running DDR3-1375 at 1.65v (stock is 1.5).

EDIT:

I updated the bios to the newest one, and got the mem running at 1372mhz CL9. Still no go about reaching past 3.7ghz, waiting for a new cooler.

[=[PULSAR]=]

bump

[jcool]

New CPU, new luck I guess: i7 940 3835A730

4,4 Ghz, HT enable, 1,5V:



Needs 1,304V for 4Ghz and around 1,38 for 4,2, haven't tried that one really hard yet. Likes 21x and 23x mult equally well and hates the 22x.

To all the people using IntelBurnTest on i7 with HT on: It's pretty useless as it only loads up half the cores, use eitehr prime blend or LinX for now.

[The-Fox]

So can we confirm that the earlier the batch the better ? the trend is getting clearer.
And even multies are i7's enemies.
Btw, that is on cascade, I wonder if it can hold 4.4Ghz w/ HT=on using water cooling.

And you can use two instances of IntelBurnTest to take 100% CPU for i7

[jcool]

Are you refering to me?

If so No, that is on my maxed-out 1/3hp Single Stage which I run 24/7
Temps are pretty bad under load (CPU gets way into positive range) so yeah, I guess my chip could do 4,4 with HT on on high end water as well.
And yeah, you can run 2 instances IBT but people here are only running 1
And running LinX is way easier and more comfortable than 2x IBT, it's like back in the Dualcore days with prime vs. Orthos

Although I am sure IBT will be adapted soon enough.

[The-Fox]

Yahh jcool I was referring to you.
Are you sure linX is using all cores/100% CPU on i7 ?

BTW, what is the earliest Retail i920 (wee/batch), so far I have seen week 35 ?

[rge]

Batch 3836A405
12 hrs prime stable with 21 multi (i940 turbo off) HT on
4.0ghz with 1.31 vcore cpuz load
4.1ghz with 1.34 (pic)
4.2ghz with 1.376.. 24/7 (post in sig)

[JWMc]

We need apples to apples comparisons for load testing. I am using less voltage for LinX vs Prime blend, every single time.

Edit: I'm making another post on the subject of comparisons and I also think I found out why I saw this strange behavior. I think I needed a slight vCore bump for Prime blend only because I was running specifically @ 20 x 200 with the settings I had. Honestly I can't even say for sure that I needed it at all but it was the last change I made and it passed the test, could have just been luck. I made this post when I had just started learning what the hell I was doing.