Which i5-3570K Batch To Keep?

[geokilla]

So I was able to pick up two Core i5-3570K from Micro Center... Which batch should I keep? One is 3220B501 and is made in Costa Rica, while the other is L206B682 and is made in Malaysia. I haven't found anything on Google, but I did find http://www.xtremesystems.org/forums...ridge-i5-3570K-i7-3770K-batch-and-o-c-results which is good, but doesn't seem to have my batch numbers.

 

[abekl]

You can't pick the over-performing chips simply by batch number or location.

 

[ShintaiDK]

You can't pick the over-performing chips simply by batch number or location.

^^ :thumbsup:

 

[Anarchist420]

I heard that some fans were better than others. I got a kind of lousy 2500k because it won't idle below .93vcore. I should've gotten a 2600k because they run cooler and because they have hyperthreading but it was out of my budget. I want several upgrades although they'd really be meaningless due to the fact that there are no decent monitors in existence. The upgrades I want are a 7990 GHz edition (against my better judgement) with ball bearing cooling fan, a 2600k, a better CPU cooler, a GTX 670 with 680 PCB and 680 HSF, and another WD Black 500GB HDD.

 

[Meghan54]

You can't pick the over-performing chips simply by batch number or location.

Exactly. And you cannot simply look at OC'ing results by batch no. because there are too many other variables involved, such as motherboard, memory, quality of your ps's output, cooling, etc., that are involved with hitting whatever OC you're trying to obtain.

 

[bl00tdi]

You might just have to install them both and mildly OC them to see which has the lower VID increase with frequency. This will generally point to the better chip. What does MC's return policy say about open box processors?

 

[Ferzerp]

This myth is still around?

The die itself is made in the USA or Israel for 22nm if I recall correctly. I don't know if the wafers are cut there, or shipped whole (I'd think they cut them here to make transport easier), but the are shipped for final assembly to costa rica or malaysia, but the really precise parts are all done in totally different locations.

edit: checked. 2 fabs in AZ, 2 fabs in OR, and 1 in Israel for 22nm now.

 

[Bill Brasky]

What does MC's return policy say about open box processors?

I think you can only exchange open box processors, not return.

 

[bl00tdi]

BTW OP, if you can tell me what the VID's are for each frequency I will buy the worst of the two off of you. I think I might have the worst 3570K made so far. I require 1.35v+ to run stably at 4.5. This chip refuses to be stable at 4.6, regardless of voltage. The VID @ 4.5 is 1.320v! POS!!


:thumbsdown:

 

[BonzaiDuck]

BTW OP, if you can tell me what the VID's are for each frequency I will buy the worst of the two off of you. I think I might have the worst 3570K made so far. I require 1.35v+ to run stably at 4.5. This chip refuses to be stable at 4.6, regardless of voltage. The VID @ 4.5 is 1.320v! POS!!


:thumbsdown:

What temperatures are showing with that bad-boy when you're @ 4.5 and loaded?

 

[cmdrdredd]

BTW OP, if you can tell me what the VID's are for each frequency I will buy the worst of the two off of you. I think I might have the worst 3570K made so far. I require 1.35v+ to run stably at 4.5. This chip refuses to be stable at 4.6, regardless of voltage. The VID @ 4.5 is 1.320v! POS!!


:thumbsdown:

What motherboard? There are boards that are better than others at providing proper and steady voltage.

 

[bl00tdi]

http://i95.photobucket.com/albums/l153/nismology/baa7acee.png


Z77 Sabertooth
Xigmatek Aegir Push/Pull in East-West heat pipe config
Prolimatech PK1
27C ambient


AIDA64 stress test = 80C @ 1.360v

I've tried tweaking the VRM related stuff and messed with other settings/voltages and 4.6 semi-stable still requires 1.42v+ (yes PLL overvoltage is enabled, etc) but still crashes in Cinebench. I've never hit 100C at those voltages (more like mid-90's) but it's still an absurd voltage for that freq. The VID @ 4.5-4.6 alone tells me this chip is on the crappy end of the scale. Have you guys encountered anything worse?


Edit: The 88c in core temp was from a previous run at a higher ambient.

 

[MustangSVT]

BTW OP, if you can tell me what the VID's are for each frequency I will buy the worst of the two off of you. I think I might have the worst 3570K made so far. I require 1.35v+ to run stably at 4.5. This chip refuses to be stable at 4.6, regardless of voltage. The VID @ 4.5 is 1.320v! POS!!


:thumbsdown:

its TIME TO DE LID!!!!!!!!!!!!

 

[bl00tdi]

its TIME TO DE LID!!!!!!!!!!!!

+1 to delidding! I actually have a tube of liquid ultra on my desk that I bought in anticipation of delidding, but this particular CPU is just too much of a stinker for it to be worthwhile. If you look at my temps above, I'm actually running into a voltage wall first.

 

[RU482]

You can't pick the over-performing chips simply by batch number or location.

IIRC, in the past - like years ago - this was a common technique, as some batches or fabs had better OC capabilities than others. Might have been AMD...I dunno, I just remember it being a big deal if you had the right numbers

 

[Rvenger]

OP: The Malaysia batches are known to run cooler than Costa Rica. I read that on OCN somewhere I believe.

I can vouch that my 3770k and 3570k are both Costa Rica batches and they are all hitting 80c on core 2 with a mild OC.

 

[ShintaiDK]

Faith can move mountains.

Its a myth.

 

[Anarchist420]

OP: The Malaysia batches are known to run cooler than Costa Rica. I read that on OCN somewhere I believe.

I can vouch that my 3770k and 3570k are both Costa Rica batches and they are all hitting 80c on core 2 with a mild OC.

That could be true because my 2500k was from cr and it runs a few degrees celsius higher than the average temp I've heard for 2500Ks. Then again, i may be using a cooler even worse than the stock one and I didn't apply the ac mx2 like a boss.

 

[Rvenger]

That could be true because my 2500k was from cr and it runs a few degrees celsius higher than the average temp I've heard for 2500Ks. Then again, i may be using a cooler even worse than the stock one and I didn't apply the ac mx2 like a boss.

Not sure it correlates with SB and IVB. I had a costa rica SB and it ran cool and at low volts.

Im thinking the malaysia chips are maybe getting a better application of TIM resulting in better contact between the IHS and die.

 

[geokilla]

Is the batch thing really a myth? I remember years ago, overclockers would go nuts trying to pick out the best batch and from what die and fab, because one was supposedly better than the other.

What about this? One of my 3750K are rated at 95W, while the other is 77W. Given this info, the Costa Rica (77W) should be cooler and overclock higher. Keep in mind this is the 3220B501 batch, which seems to be the better overclocking batch. Thus maybe debunking the saying that choosing from a good batch is a myth?

 

[Phynaz]

Is the batch thing really a myth?

Yes.

Intel's manufacturing concept is "copy exact". Every line in every fab is an exact clone of every other.

The only time there could be some substance is if a production tweak is put into place that happens to change a manufacturing parameter that results into a fundamental change of overclockability, etc. Then of course any chips made prior to this change would not benefit from it.

I really don't think the above happens very often without a stepping change.

What about this? One of my 3750K are rated at 95W, while the other is 77W. Given this info, the Costa Rica (77W) should be cooler and overclock higher.

False. The power rating is for cooling design only. It says nothing about how much power a chip will actually dissipate, or what its overclockability is. For example, the entire C2D line had the exact same power rating. Now we all know that every model of the C2D didn't use the same amount of power.

Intel currently produces nearly a million chips a day. You have a sample size of 2. You can't draw any conclusions from your experience.

 

[vista1984]

Yes.

Intel's manufacturing concept is "copy exact". Every line in every fab is an exact clone of every other.

The only time there could be some substance is if a production tweak is put into place that happens to change a manufacturing parameter that results into a fundamental change of overclockability, etc. Then of course any chips made prior to this change would not benefit from it.

I really don't think the above happens very often without a stepping change.

False. The power rating is for cooling design only. It says nothing about how much power a chip will actually dissipate, or what its overclockability is. For example, the entire C2D line had the exact same power rating. Now we all know that every model of the C2D didn't use the same amount of power.

Intel currently produces nearly a million chips a day. You have a sample size of 2. You can't draw any conclusions at from your experience.

+1,
both tdp are 77w.
test them before you decide which to keep. my opinion is go undervolted, who survives, who stay. don't get too serious for this.

 

[Ferzerp]

The first run of boxes shipped out with 95W printed on them instead of 77W is all.

It's still silicon lottery.

 

[ShintaiDK]

The first run of boxes shipped out with 95W printed on them instead of 77W is all.

It's still silicon lottery.

+1

 

[Madpacket]

I guess it's different today. I used to bin chips, hundreds of them back in the Athlon XP / Pentium 4 days and the stepping and country of manufacturer made a large difference in overclocking / undervolt potential. This was for a mom and pop shop where I built and sold a ton of overclocked computers for at least a few years.

If I recall correctly it was generally Malay chips for Intel and German chips for AMD that binned the best.