Core i7 Overclocking by Batch #

[ChaosAD]

I just found an 3843B276 in a local shop. Do we have any info on this batch? The fastest reply, the best

[Scooby]

I have posted this at the i7 stable OCing database thread and I'd like to contribute to the OCing batch list if it's OK. I haven't seen to many of this batch around but if you guys fine one grab it, it a killer chip.

CPU: i7 920
CPU MHZ: 4515MHz
HT: ON
CPU BATCH: 3849B203
CPU S-SPEC: SLBEJ
CPU VOLTS: 1.35
QPI/DRAM 1.34v
IOH 1.11v
IOH PCIE 1.51v
ICH 1.11v
ICH PCIE 1.51v
DRAM BUS 1.5v
CPU COOLING: AIR
BRAND AND MODEL OF COOLER: TRUE Copper w/ Scythe Gentle Typhoon Fans
MOBO BRAND AND MODEL: Asus Rampage II Extreme
BIOS VERSION: 1502
OS: Windows 7
RAM BRAND: Corsair
RAM MODEL: TR3X3G1600C9, 6 x 1Gb (Dominator GTs 2000Mhz died during stress testing - Awaiting RMA return)
RAM MHZ: 1723MHz
RAM TIMINGS: 10-9-8-20-1N

[SwIeRsZcZ]

Does anyone know anything about this chip: 3835A732 (i920) ? (aim 4.2GHz on Air with considerable temperatures) ???

[unclewebb]

Scooby: Your screen shot clearly shows turbo throttling. Your multipler is at 20X at full load which equals about 4308 MHz.

At idle it jumps back up to 21X or 4523 MHz but it's not running at that speed at full load.
You need to make an adjustment in your bios to get the 21X multi to stick at full load.

Check out RealTemp 3.30 and i7 Turbo V8 so you can watch your multi at full load.

http://www.fileden.com/files/2008/3/...alTempBeta.zip

[Scooby]

Thanks unclewebb, I'll try that. I'll follow-up on this on the other thread.

See you there.

[DashTrash]

I was able to get a 3849b040 from Microcenter in Cambridge today. I'm visiting from Oregon (no Microcenters), went in with a list of batches I wanted. First salesman was not real helpful, but the second one took my list and found a 3849. I went in originally going to get a s775 chip, but thought I couldn't pass up the opportunity. Will be a while before I'm able to test it, though. Anyone had any luck with this batch?

[poke349]

i7 920 D0:
Batch: 3914A114

Asus Rampage II GENE
Cooler Master Hyper N520 with 2 x Delta FFB0912SH-F00 (@ 50% duty cycle for 24/7 use, @ 100% for benchmarking - they are VERY LOUD)
Stock ram speed: 1333 MHz

Stock vcore = 1.13v

3.36 GHz (21 x 160) stable with 1.13v
3.50 GHz (21 x 167) stable with 1.20v (my 24/7 configuration - at least 10 days stable under a near-equivalent prime95 load)
3.80 GHz (21 x 180) stable with 1.225v

There were no problems here. The vcore I used here are simply wild guesses. I haven't pushed it to see what the limits are.


Here's where I decided to see how far 1.225v would go.

4.0 GHz (21 x 191) stable with 1.225v
4.2 GHz (21 x 200) ~about a min. prime stable with 1.225v
4.2 GHz (21 x 200) stable with 1.250v


All other BIOS options set to auto except I never allowed the vram to go above 1.6v.


Good? Bad? Middle?



EDIT:

Here's my 24/7 config:
CPU - 3.34 GHz (3.50 GHz Turbo) - 20/21 x 167
Ram - 1333 MHz
vcore = 1.20v
vram = 1.60v (undervolted from 1.67 to reduce temps)

As mentioned: It is at least 10 days stable under a near-equivalent prime95 load.
I'd clock higher, but I run my rig under a prime95-equivalent load 24/7 and I want it to last a while.

Why 3.34 GHz?
It works well with my 1333 MHz ram: 167 x 8 = 1336 MHz

[SwIeRsZcZ]

i7 920 D0:
Batch: 3914A114

Asus Rampage II GENE
Cooler Master Hyper N520 with 2 x Delta FFB0912SH-F00 (@ 50% duty cycle for 24/7 use, @ 100% for benchmarking - they are VERY LOUD)
Stock ram speed: 1333 MHz

Stock vcore = 1.13v

3.36 GHz (21 x 160) stable with 1.13v
3.50 GHz (21 x 167) stable with 1.20v (my 24/7 configuration - at least 10 days stable under a near-equivalent prime95 load)
3.80 GHz (21 x 180) stable with 1.225v

There were no problems here. The vcore I used here are simply wild guesses. I haven't pushed it to see what the limits are.


Here's where I decided to see how far 1.225v would go.

4.0 GHz (21 x 191) stable with 1.225v
4.2 GHz (21 x 200) ~about a min. prime stable with 1.225v
4.2 GHz (21 x 200) stable with 1.250v

What temps do you have stress/idle on 4.2Ghz (on air) ??

Thanx for info

[poke349]

What temps do you have stress/idle on 4.2Ghz (on air) ??

Thanx for info

Stress: 84/79/81/77
Idle: 40 - 45

EDIT 1: These weren't on prime95, but it runs just as hot as prime95 large FFTs.

EDIT 2:

Here's prime95 @ 4.2GHz... (click to enlarge)
All hardware monitors are open so you can read the volts.
The only thing is that the CPU fans are attached to the "OPT_FAN1" port and a case fan is on the CPU port. So the CPU fans are running at 4000 RPM, and a case fan is running at 3000.


EDIT 3: This is on the borderline for Turbo-Throttling. This explains why RealTemp reads the multiplier as 20.5.


I'm pretty sure this chip has more to it... But since it's my only functional rig right now, I don't have the guts to push any higher.

[oggie]

Got myself batch 3907A360, from what I understand these are a bit of unknowns......?

[oggie]

Ok playing around a bit with my cou, whats highest cpu vtt allowed ?

[poke349]

Ok playing around a bit with my cou, whats highest cpu vtt allowed ?

The intel specs say 1.375. Which everyone says is very conservative.

I wouldn't go higher than that for 24/7, but for benchmarking it's probably fine.

[freshy]

Batch "3849B018"
Cooled with an EK supreme, D4, and 3 x 120m RAD.

This chip definitely has more in it, but my temps start to get to high for me.
So far max temp is 80 in Prime and 75 with WCG.

i got same batch, still priming after 5 hours @ 4.2ghz @ 1.255v and qpi/vtt @ 1.22v

[Osterman]

I habe been offered 2 different kind of batch.
One kind is 3849B wich is known that some can to 5ghz minimu with dice. I think the complete batch is 3849B028.
And the other batch is a 3845A wich was tested and it can do 4.2 with 1.25vcore

Well, does someone has one of those so I can hear at least one more opinion.

Thanks in advance.

[systemviper]

3845B027. This CPU clock, 5383.73 Mhz, is #1 on HWbot for both i7 920 and W3520.

http://valid.canardpc.com/show_oc.php?id=593562

Nice, i put this aside when they first came out from tankguys, i just looked at it for a build i am doing, an saw it's 3845B027, should be some interesting testing....

[road-runner]

3849B018, Heatkiller copper block...

[drewxbk]

Finally got my account validated, I have batch 3904A361. Haven't heard anything at all about this batch yet..so far I've gotten 3.7 at 1.1vcore

[Osterman]

Nice, i put this aside when they first came out from tankguys, i just looked at it for a build i am doing, an saw it's 3845B027, should be some interesting testing....

SHould be nice to know if the same batch but with an A instead of a B can do the same.
As I already said there are some 3845A which can do 4.2ghz with 1.25 vcore.
Who know what they can do with 1.35 or 1.36

[poke349]

SHould be nice to know if the same batch but with an A instead of a B can do the same.
As I already said there are some 3845A which can do 4.2ghz with 1.25 vcore.
Who know what they can do with 1.35 or 1.36

Is it still that easy to find 3845B now? That's last year's batch... lol

I got my i7 some 2 weeks ago and it's 3914A... It to hit 4.2 GHz @ 1.25 vcore without trying.

Maybe we should start checking if any of the newer batches are as good as 3845A.

[tomslick.ca]

Nice, i put this aside when they first came out from tankguys, i just looked at it for a build i am doing, an saw it's 3845B027, should be some interesting testing....

I was able to do 4.6Ghz @ 1.35V on air with that batch
http://valid.canardpc.com/show_oc.php?id=581572

running 4.0Ghz @ 1.168V HT on 24/7
http://valid.canardpc.com/show_oc.php?id=575750

[SwIeRsZcZ]

See some people have batch #3849B , but I found at local store this one : #3849A... Any info about it ? Good one or rather average ?

[systemviper]

I was able to do 4.6Ghz @ 1.35V on air with that batch
http://valid.canardpc.com/show_oc.php?id=581572

running 4.0Ghz @ 1.168V HT on 24/7
http://valid.canardpc.com/show_oc.php?id=575750

your getting me excited, got to get playing

[road-runner]

Is it still that easy to find 3845B now? That's last year's batch... lol

I got my i7 some 2 weeks ago and it's 3914A... It to hit 4.2 GHz @ 1.25 vcore without trying.

Maybe we should start checking if any of the newer batches are as good as 3845A.

I went to microcenter to get one of the $200.00 920s and saw a B batch setting in the case and told them to give that one...

[poke349]

I went to microcenter to get one of the $200.00 920s and saw a B batch setting in the case and told them to give that one...

I had some time to play with my chip a little more last night.

To restate my specs:

920 D0 - Batch #3914A114
Asus Rampage II GENE
6 x 2 GB Patriot Viper @ 1333 MHz

All memory timings set to auto.
All CPU features (including HT) are enabled.
Memory speeds set to the highest speed below stock (1333MHz).
Memory voltages set to auto, but never allowed above 1.6v.

Stress testing was done using:

• prime95 small FFTs
• y-cruncher, Multi-Threaded Pi Benchmark

Stock vcore is either 1.13 or 1.16 (bios and turboV give me different values)

vcore = 1.15v
uncore = 1.200v

3.8 GHz (181 x 21) - stable
3.9 GHz (186 x 21) - appears stable
4.0 GHz (191 x 21) - 10 min. prime stable


vcore = 1.175v
uncore = 1.200v

3.8 GHz (181 x 21) - stable
3.9 GHz (186 x 21) - appears stable
4.0 GHz (191 x 21) - appears stable
4.1 GHz (195 x 21) - 5-10 min. prime stable


vcore = 1.200v
uncore = 1.200v

4.0 GHz (191 x 21) - appears stable
4.1 GHz (195 x 21) - 5-10 min. prime stable


vcore = 1.225v
uncore = 1.250v

4.0 GHz (191 x 21) - stable
4.1 GHz (195 x 21) - appears stable

----------------------------------------------------------------

The rest are all @ 4.2 GHz (200 x 21):

vcore = 1.225v, uncore = auto to 1.39v
unstable: 2 billion digits of pi failed after 2 min.

vcore = 1.250v, uncore = auto to 1.39v
appears stable: 2 billion digits of pi success - twice in the row (21 min. for each run, 42min. total)

vcore = 1.250v, uncore = 1.250v
5-10 min. prime stable

vcore = 1.250v, uncore = 1.275v
appears stable

• > 30 min. prime stable
• successful computation of 2 billion digits of pi - twice in the row. (21 min. for each run, 42min. total)

I stopped here. Any higher and it hits 80C which is the threshold for turbo-throttling.

In this last test @ 1.250v/1.275v, both prime95 and Pi were sustaining 77-78C and turbo V8 was already showing a bit of turbo-throttling.



In light of these results, I've lowered my 24/7 config (167 x 21 = 3.5GHz) vcore from 1.2v to 1.15v and locked the uncore to 1.20v

[freshy]

Nice, i put this aside when they first came out from tankguys, i just looked at it for a build i am doing, an saw it's 3845B027, should be some interesting testing....

nice batch

3849B018, Heatkiller copper block...

nice , what was vcore and qpi/vtt voltage in bios for 4.5ghz ? and what cooling and mobo ?

i havent pushed past 4.2ghz @ 1.255v yet, i had to rma my 2000mhz blade kit, so i was stuck testing the mem for a few days and got sick of it , but might push cpu a bit more today but ive only got a 2x120 rad at the moment and my temps are already hitting 70c when priming @ 1.255v so not sure how far il be able to push till i get another rad