Core i7 Overclocking by Batch #

[STi489]

I have 3835A538...It's ok i guess..I would like to try a B chip.

l

[kpablo]

#Batch 3841A431 (From Buy.com)
Core i7 920@4011MHz (191x21) Turbo & HT ON - WC (Apogee GTZ)
Vcore: 1.41875v in BIOS, 1.392v in IDLE, 1.376v in FULL (CPU-Z)
QPI: 1.35v
Mem: 1.64v

QPI x36
Uncore x16
Mem x8

MB: Gigabyte EX58 Extreme (F4m)
MEM: 6GB Corsair Dominator PC12800 CL8

LinX x20:


CPU-Z Validation:
http://valid.canardpc.com/show_oc.php?id=489510

Temp Ambient, 30C

[icecold316]

#Batch 3841A431 (From Buy.com)
Core i7 920@4011MHz (191x21) Turbo & HT ON - WC (Apogee GTZ)
Vcore: 1.41875v in BIOS, 1.392v in IDLE, 1.376v in FULL (CPU-Z)
QPI: 1.35v
Mem: 1.64v

QPI x36
Uncore x16
Mem x8

MB: Gigabyte EX58 Extreme (F4m)
MEM: 6GB Corsair Dominator PC12800 CL8

Temp Ambient, 30C

I got the same batch i7 from buy.com(ebay store), same memory, almost same mobo UD5 running at 3.8Ghz HT ON with CPU : 1.3V and QPI : 1.3V, Mem: 1.64V (7-8-8-20), temperatures are way high since I am on Stock Cooler. Scythe Mugen 2 is on its way.

I thought 3841A431 would have made into 4Ghz under 1.4, but I guess I am wrong.

[kpablo]

I got the same batch i7 from buy.com(ebay store), same memory, almost same mobo UD5 running at 3.8Ghz HT ON with CPU : 1.3V and QPI : 1.3V, Mem: 1.64V (7-8-8-20), temperatures are way high since I am on Stock Cooler. Scythe Mugen 2 is on its way.

I thought 3841A431 would have made into 4Ghz under 1.4, but I guess I am wrong.

Tried to put 20x200 and HT on, Turbo OFF before... but to stabilize i need to put in Bios like 1.4325v (1.408v in CPU-z)... imo not worthed to do it

[headman]

Batch# 3837A715

[roller11]

I got the same batch i7 from buy.com(ebay store), same memory, almost same mobo UD5 running at 3.8Ghz HT ON with CPU : 1.3V and QPI : 1.3V, Mem: 1.64V (7-8-8-20), temperatures are way high since I am on Stock Cooler. Scythe Mugen 2 is on its way.

I thought 3841A431 would have made into 4Ghz under 1.4, but I guess I am wrong.

This is a good result, but it blows my theory that the later the chip, the worse the OC, especially week 41.
I have witnessed bench of 3 different 3841A43x. 3841A434, 3841A436, and another one. All were duds, and one overheated severely under load. Not exactly the same batch, but close. All came from amazon.com and buynow.com (same distribution point), so I wonder if that warehouse just happened to get a bad batch of chips?

[kpablo]

Anyone got the OEMs of 920s from Tigerdirect? If is... what batch # you got?

[icecold316]

This is a good result, but it blows my theory that the later the chip, the worse the OC, especially week 41.
I have witnessed bench of 3 different 3841A43x. 3841A434, 3841A436, and another one. All were duds, and one overheated severely under load. Not exactly the same batch, but close. All came from amazon.com and buynow.com (same distribution point), so I wonder if that warehouse just happened to get a bad batch of chips?

I was under impression Heat was directly proportional to CPU V and QPI V. So lower it can go, less the heat the chip generates.

So shouldn't it be, if the batch is running at less voltage less overheating?

I tried to set the same Bclk, CPU v and QPI as of 3841A383 on mine and it didnt work for me, so the last three digits matter a lot too.

But, if my i7 is doing 3.8 ghz @ CPU 1.3 V and QPI @ 1.3 V and keep temperatures low, I got no reason to complaint.

[GenTarkin]

This is a good result, but it blows my theory that the later the chip, the worse the OC, especially week 41.
I have witnessed bench of 3 different 3841A43x. 3841A434, 3841A436, and another one. All were duds, and one overheated severely under load. Not exactly the same batch, but close. All came from amazon.com and buynow.com (same distribution point), so I wonder if that warehouse just happened to get a bad batch of chips?

Well it may not entirely overrule your theory. In general with intel throughout the last couple years I have noticed that the later a stepping is in week they generally dont OC as well. Probably beceause the process matures and they start binning them more accurately. But then there is the other side also, which is demand and they are forced to bin what would be higher quality chips for say a 940 or 965 part as 920's because the 920's are selling a lot faster then their appropriate bin right now....
The above could easily be the case with these newer 920's.
Just my guestimate anyways =)

[roller11]

My theory as stated is based strictly on observations of ppl in this forum, and my own testing. I think I would rephrase your comment a little like this: Early on, Intel doesn't want a lot of exposure (risk of returns, bad publicity) with their chips so they are very conservative, meaning they want extra saftey margin which means the chips that make it through will have extra good AC characteristics. As time goes by, they get more confidence that their chips will meet specs, so they allow worse chips onto the market which will of course still meet warranty 2.67Ghz. In other words, perhaps early on the Intel's internal threshold might be, say, 3.4Ghz, then later on, they may relax that to 3.0 Ghz. Which I believe is what you are saying. To balance that, as the process matures, one would expect the overall quality (i.e. AC characteristics) to improve so the percentage of good chips (superior AC characteristics) will rise. Of course I'm just speculating, trying to apply common sense.

[Khalmyr]

Doesn't explain why the 3838A's are so much worse than the rest :P

[anzial]

'cause it's just a theory

[The Nemesis]

Every review/overclocking guide I've read all pretty much say that an i7 920 will do 4GHz @ 1.4v-1.425v core and 1.3v-1.325v cpu vtt/qpi when run memory at ddr3 1600 and below. So really a cpu shouldn't be considered a substandard until it can't achieve this. HT or cooling play a big part in this though. the 4GHz is not to be had with stock or non high end cooling.

[liftedcj7on44s]

mine is a 3837A732. anyone got this? how old would this be?

[vikingboy]

mine is a 3837A732. anyone got this? how old would this be?

ive got one of these - mine is stable at 210 * 19 @ 1.36v (and improving still.....)

[liftedcj7on44s]

ive got one of these - mine is stable at 210 * 19 @ 1.36v (and improving still.....)

Is this with HT on? What cooler and what settings are you using? I am running the 211 bclk and the same multi and vcore. I have HT on but cant get it prime or linx stable. I have been running it like this now for a week without an issue but as soon as i start prime or linx i get a BSOD within 1 minute with HT turned on. If i turn off HT it is completely stable.

[MACMAC]

Found a 3831A ES on ebay earlier. This must be a real early run, I wonder what the earliest week they made was.

Mine is a i965 3830A ES and it is actually half-decent for air benching (~4.5Ghz @ 1.50V).

[scaramonga]

Well, I've played with this thing for days now, and finally got it stable with HT & Turbo ON. I was previously sitting at 4GHz stable but only possible with HT & Turbo OFF.

Any setting of BCLK above 190 and/or a Multi of 20 with HT/Turbo ON resulted in Prime & LinX crashing with BSOD within 5 minutes, no matter what voltages were changed, believe me, I tried them all!

The only highest settings that will now work with HT & Turbo enabled are:

BCLK - 190 (x21) 3990.04 @ 1.44vcore - vdimm @ 1.65

Everything else voltage wise on AUTO.

Prime & Linx now running more than an hour, so I'll consider that stable as I was lucky to last 5mins before.

Tested on 2 versions of BIOS - 0805 and 1001, same results for both.

Its a pain that my RAM is running slightly under spec (1526 or summit? cant recall ATM) and that I cannot, whatsoever, run at over the 4GHz mark with HT & Turbo enabled, but I'll settle for that just now as irritation is taking over my mood.

Oh and I'm - Batch #3838A505

[jcniest5]

How important is it to set CPU PLL Voltage when overclocking? Is it okay if it's set at Auto or is it better to assign a number to it (and what is a good number if so)?

[zlojack]

My batch # is 3837A712

I'm running Prime 95 (Small FFT, Large FFT, Blend 8+ hours), LinX (50 Passes) and HCI Memtest (400%) stable with 1.35v bios (under load it goes up to 1.38v according to E-LEET) and the cores get up to around 69-74 (0-74 1-71 2-70 3-69) is the average usually.

Chip is at 206x19 HT on. QPI PLL is set at 1.35v VDIMM 1.65 CPU PLL 1.85

[ginfest]

mine is a 3837A732. anyone got this? how old would this be?

Same chip here I'm not sure how "old" these are, got mine the second week of release.
Currently running at 19x200=3.8 +Turbo, HT on
1.35 Vcore, 1.35 QPI/VDIMM (VTT), the rest on Auto

[Avan]

mine is a 3837A732. anyone got this? how old would this be?

I've posted this in another thread before, but here ya go.

Example: L149A463-0726
1st letter or digit = plant code (Malay)
0 = San Jose, Costa Rica
1 = Cavite, Philippines
3 = .............., Costa Rica
6 = Chandler, Arizona
7 = .........., Philippines
8 = Leixlip, Ireland
9 = Penang, Malaysia
L = ............, Malaysia
Q = ..........., Malaysia
R = Manila, Philippines
Y = Leixlip, Ireland
2nd digit = Year of production (2001)
3rd & 4th digits = week (49th week )
5th - 8th digits= lot number
10th - 13th digits = serialization code

[Splave]

mine is a 3837A732. anyone got this? how old would this be?

I have a week 37A clocks 4.3ghz without HT stable 1.425vcore on water

with HT over 4ghz HT on is a struggle

[liftedcj7on44s]

i have not tried more than 1.425 on mine. with HT on and its not stable at 4ghz. with HT off though its stable at 4ghz using 1.36 vcore.

[scaramonga]

with HT off though its stable at 4ghz using 1.36 vcore.

Run a few 'hours' of Prime or LinX and I bet you will get errors and/or BSOD.

But yup.......HT seems to kill most peeps clocks dead for sure, even with a decent cooling system installed.

On my REII, BCLK and Multi are a big factor as in, 193-200 it definitely don't like, along with 20x Ratio, any of these combination's, with or without HT causes problems.

I've only been up to 1.45 vcore, as I'm wary of giving the i7 any more than that, although MAX temps with that were only load 77 (with HT on) at any time testing, still I heard that the proper MAX should not exceed 1.44 vcore and I dunno if that is just being safe or bull?