Whats your procedure for binning IB?

[ZMBKLN]

Hey guys...

 

Was curious to see what you guys are doing to bin your IB chips...

 

It's obviously not as simple as SB was...

 

I'm sure it's the old "more than one way to skin a cat" thing as well...

 

But I'd like to see how you guys are doing it.

 

-David

[TaPaKaH]

Lowest voltage for 5GHz suicide/32M on air. Below 1.25/1.30V = good.

[Mafio]

nah, it's the same shit: buy as much chips as you can

[Splave]

pay someone alot for a binned one....cheaper in the end

[K404]

pay someone alot for a binned one....cheaper in the end

 

HA! Yea, I was gonna post something similar

 

Binning procedure: 1/ Buy binned chip, 2/ confirm the advertised MHz. 3/ beer

[GENiEBEN]

4/ kill it in the attempt to see if it goes any higher than advertised speed.

[TaPaKaH]

5/ write to the seller that it was degraded and could never hit advertised speed and ask for a refund

[M.Beier]

Well it all depends on what you are binning for, for max clock in 2D tests, teh way Sam suggested is pretty spot on, though I still think my chip with 1.308v 5ghz 32M is pretty decent

However, for IMC binning its different, and again if you are going for 3D, start binning by benching 3D

 

The buy prebinned, it all depends on your network if its cheaper or not, and if you are from USA or not.. In USA you can buy and sell with a 20 USD/cpu loss, other places expect 120 USD/cpu loss instead.

[DrWeez]

pay someone alot for a binned one....cheaper in the end

 

who is this "someone" you are talking about?

 

[Blue Storm]

Someone who sells it on XS...sometimes overclockers sell there binned chips but for very high price

[Hiwa]

600 to 700 for 6800

Over 1k for 6900 cpus are all normal prices.

I dont bin chips for about 2 years, my last binning were with 980x time and i sold some chis for 2k or so

3770k costs 330 here, normaly for 6900 u need over 50 chips (unlucky guy ) or maybe 100

U have to sell crap chip for 270 to max 280 so u loose 50 on each cpu that means 2xxx on 40/50 chips.

 

If it goes nice u buy 20 and u find 15 6900 cpu

My e8600 time, i bought 20 cpus and 98% of them could do 6500, so i changed my car that year lol ( kidding caz if u bench seriusly u spend 20 per day just for beers )

[I.nfraR.ed]

I hit 6.9+ GHz valid with my second chip and sold for 400euro. It was also a very good 3D chip. Now I understand it was a bit silly low price.

It sold ~10 min after I posted the thread and took me more to write the text in the thread, lol.

Edited November 5, 2012 by I.nfraR.ed

[Blue Storm]

Over 1k for 6900 cpus are all normal prices.

I know that this prices are suitable for that pieces of binned chips. I only meant that they are much more expensive compared to new ones.;)Price of binned chips are good when you realize how much work it took to bin and how much money that man invested into new chips (sure he can get it directly from Intel:p), etc...

[pro]

LN2 ALL OF THEM

 

dont waste time with air

[Vivi]

1000 dollar 3770k is much better value for money than 1100 dollar 3970x

[ZMBKLN]

Great Input here guys...

 

Thanks a bunch...

 

I am curious though what you guys are "starting off with" for settings such as VCore and Etc while testing under LN2...

 

I'm gonna give this 5.0 @ 1.25V 32M test a try also to see if I can find any pattern in it.

[I.M.O.G.]

I had read every post in the IB results/binning thread, up until 2 months ago anyway. Air testing was not reliably accurate - plenty of good air chips come out mediocre 6.2-6.5 cold chips. I've had chips that ran the 5GHz sp32m on ambient temps (with a CPU pot on and just a dash of LN2 here and there to maintain temps) at 1.25V-1.35V, and there was no pattern to what did better under cold.

 

I agree with the take them all cold philosophy and see what they will do. I started all on air at 1.4V@5GHz though, every chip I've bought can do that and seem reasonably stable (or something like stable).

 

My binning settings for 3D Frequency:

HT off, 4 cores enabled

107MHz FSB

1.7V PLL

1.1 IO/SA

1.65V DRAM

1.55 CPU, pull down temps, go to 1.65 around -100Cish and apply load (helps prevent coldbug in my experience), continue pouring to take it to -190C, once full go to 1.7V and find max frequency using CPU multiplier, crash, 1.75V find max frequency using multiplier, 1.8V find max frequency using multiplier.

 

That is enough to see what the chip wants to do, and how it scales with additional voltage. For the chips that showed promise, I'd screw around more with FSB and fine tune the voltages.

 

I don't remember how many chips I've binned, only 7 or 8 maybe. I've killed 3 I think. 2 of the dead were the best I've had, not stellar. Some chips I bought discounted at microcenter and sold to break even including price/tax/$30 gas round trip to microcenter. Some chips I took about a $50 loss on total, to move them quickly so I could buy more.

 

I've seen other guys recommend 1.25V on IO/SA, or keep one a little higher than the other. Also seen other guys saying tweak PLL a lot. For me, none of that has changed a good chip into a bad chip, or a bad chip into a good chip... Just the good chips ran faster with pretty much the same settings, while the bad chips I could tune them in for less than 100MHz extra, but they didn't turn from sucking into being good.

Edited November 6, 2012 by I.M.O.G.

[Viss]

I start 50x100 4/8 with 1.30v aircooled. If this boots into OS ill start to lower vcore to see whats lowest for this settings. If it doesnt boot i disable HT, disable 2 cores, diasble 3 cores. Stil no boot then increase vcore to find out what it needs to boot 50x100 1/1, if thats over 1.375v i dont bother to test it cold.