cherry picking

[Sil3ntSnip3a]

hey guys

i just got a 965 from hokiealumnus. codes are as follows:

HDZ965FBK4DGM - so i know its a c3 125w
CACAC AC 1009GPMW - here is where i get lost. this is the big picture for cpu cherry picking right?

need help understanding the numbers so i can cherry pick my cpus in the future!

thanks ya'll

[chew*]

hey guys

i just got a 965 from hokiealumnus. codes are as follows:

HDZ965FBK4DGM - so i know its a c3 125w
CACAC AC 1009GPMW - here is where i get lost. this is the big picture for cpu cherry picking right?

need help understanding the numbers so i can cherry pick my cpus in the future!

thanks ya'll

not really, week matters most, and only with having seen results from those weeks.

[crazydiamond]

like chew said, the week is usually the most important thing when picking cpu's but other thing people will look at is the below the week stepping, like if you look at this pic, below the 1015 dpaw you will see a bunch of numbers that end with "313" that is what people are talking about, usually they want lowest number you can find for that. but since week is most important that is what you want to look at first. like a good week with high number would be better than a bad week with a low number

[chew*]

So taking what crazy said into consideration,

week 10 for x4- x2 cpu's seemed cherry.

week 15 for x6 was cherry.

Also if you work at at pc store like some people you can take this to the next level and bin each cpu by VID, ref vid is diff chip to chip.

So although they are not quite "binning" on ln2......they don't have to. They can just look for chips with the lowest ref vid and know they are getting a high leakage chip.

lower default vid = higher leakage, worse for air sometimes great for ln2, higher default vid = lower leakage, better for air sometimes.

Had someone acess to do that with chips, they could preety much dominate without much effort.

To understand the leakage thing, you need to look at tdp becasue at first you would think the analogy is backwards.

AMD specs XXXX part to operate under say 140w tdp, so chips that have lower vid would exceed tdp likely with a higher vid aka they have higher leakage. Leakage is more so defined by tdp and not the chips ability to withstand higher voltage.

Me personally I tend to like the chips one off from the lowest vid, they tend to be more voltage tolerant when running in high ranges on ln2.

[Sil3ntSnip3a]

So taking what crazy said into consideration,

week 10 for x4- x2 cpu's seemed cherry.

week 15 for x6 was cherry.

Also if you work at at pc store like some people you can take this to the next level and bin each cpu by VID, ref vid is diff chip to chip.

So although they are not quite "binning" on ln2......they don't have to. They can just look for chips with the lowest ref vid and know they are getting a high leakage chip.

lower default vid = higher leakage, worse for air sometimes great for ln2, higher default vid = lower leakage, better for air sometimes.

Had someone acess to do that with chips, they could preety much dominate without much effort.

To understand the leakage thing, you need to look at tdp becasue at first you would think the analogy is backwards.

AMD specs XXXX part to operate under say 140w tdp, so chips that have lower vid would exceed tdp likely with a higher vid aka they have higher leakage. Leakage is more so defined by tdp and not the chips ability to withstand higher voltage.

Me personally I tend to like the chips one off from the lowest vid, they tend to be more voltage tolerant when running in high ranges on ln2.

chew* can you put this in more understandable english . tdp? vid?

my chip is as follows.

HDZ965FBK4DGM
CACAC AC 1009GPMW
9C82851C00335

so decoding this chip would be as follows


HDZ965FBK4DGM Z=Black Edition 965=duh DGM=140w aka C3 stepping.

CACAC AC 1009GPMW - pretty much worthless ?

9C82851C00335 - 0335 means that this chip was made in week 35 of 2003.

am I getting this so far?

[skier]

chew* can you put this in more understandable english . tdp? vid?

my chip is as follows.

HDZ965FBK4DGM
CACAC AC 1009GPMW
9C82851C00335

so decoding this chip would be as follows


HDZ965FBK4DGM Z=Black Edition 965=duh DGM=140w aka C3 stepping.

CACAC AC 1009GPMW - pretty much worthless ?

9C82851C00335 - 0335 means that this chip was made in week 35 of 2003.

am I getting this so far?

if i understand right thats week 9 chip off batch 335? no one actually SAID what the 00335 means, just thats it's important to be low. and LOL not 2003, Phenom II's were certainly not out then, that was the pentium4/athlon xp/64 days

my 1055 is 1022BPMW/00305

[Sil3ntSnip3a]

and while we are here... I also have an intel pentium 4 northwood. SL6WK stepping. what are the significant figures to look for when picking an intel chip. i'm not asking if my chips are good, i am simply trying to understand the process behind finding a good chip, and knowing a good chip for overclocking. i know its all a shot of hit and miss, but for the most part it is pretty damn accurate!

let me post you some pictures of the chips.

[BeepBeep2]

chew* can you put this in more understandable english . tdp? vid?

my chip is as follows.

HDZ965FBK4DGM
CACAC AC 1009GPMW
9C82851C00335

so decoding this chip would be as follows


HDZ965FBK4DGM Z=Black Edition 965=duh DGM=140w aka C3 stepping.

CACAC AC 1009GPMW - pretty much worthless ?

9C82851C00335 - 0335 means that this chip was made in week 35 of 2003.

am I getting this so far?

No, I believe you got (almost all of) that wrong.

HDZ965FBK4DGMZ or (HDZ965FBGMBOX)= It's just a product code. Z = BE, M = C3 stepping (125w!), I = C2 stepping (140w).
CACAC AC 1009GPMW - 1009 means 2010, Week 9. GPMW talks about the day of week, what batch, where, and what kind of batch it was. (large or small)
9C82851C00335 - 0335 means that that CPU was 335th from the middle of the wafer. (I believe, or at least from somewhere on the wafer)

[Sil3ntSnip3a]

if i understand right thats week 9 chip off batch 335? no one actually SAID what the 00335 means, just thats it's important to be low. and LOL not 2003, Phenom II's were certainly not out then, that was the pentium4/athlon xp/64 days

my 1055 is 1022BPMW/00305

didn't make sense to me either. hence the post

[Sn0wm@n]

it wasnt built in 2003 for sure ... so youve got something wrong for sure .... its probably more like 1009 ... 10th month of 2009 ... probably and those letters probably means wich plant did it or something like that .. maybe someone might clarify wich part of the code means what

[Sil3ntSnip3a]

No, I believe you got (almost all of) that wrong.

HDZ965FBK4DGMZ or (HDZ965FBGMBOX)= It's just a product code. Z = BE, M = C3 stepping (125w!), I = C2 stepping (140w).
CACAC AC 1009GPMW - 1009 means 2010, Week 9. GPMW talks about the day of week, what batch, where, and what kind of batch it was. (large or small)
9C82851C00335 - 0335 means that that CPU was 335th from the middle of the wafer. (I believe, or at least from somewhere on the wafer)

so the GPMW doesn't have any significant value?

[Sn0wm@n]

so the GPMW doesn't have any significant value?

of course it has .. but we dont know what it actually means so we cant know exactly if it has any significant value ...

[Sil3ntSnip3a]

blah I'm confused... so when cherry picking a cpu you base it off of results posted from other users. and then you usually look to match there stepping and week released number?

[Sn0wm@n]

yep .. the best method is to usually do like chew* told ...

[Sil3ntSnip3a]

so based off of that.. my chip should be pretty good! its 1009gpmw which is week 09 @ 2010.

[crazydiamond]

1009 = 2010, 9th week.... 1015 like my chip = 2010, 15th week

gpmw, dpaw, etc. are telling us what day of week, what plant, etc.. hard to use that for binning but if someone had access to enough chips they might see some trends. nothing worth changing chips over afaik

the 335 like your chip shows or 313 like my chip showed is telling us how far from the middle of the wafer the chip was. from what i have heard the chip in the middle of the wafer starts at 1, and as they move closer and close to the edge, the numbers get larger. from my understanding apparently chips from the middle of the wafer are better which is why people look for the lower numbers. but like i said in my first post the week is more important, so hypothetically speaking, a 1015 chip (15th week 2010) with a high number like mine (313) would be better than say a 1020 chip (20th week 2010) with a low number like 5 or something

[Sil3ntSnip3a]

mmm makes sense now! so week is most important and within that week if you can get a peach wafer placement then you are golden!

and finding how to cherry pick you cpu for the week you need to find is all based off of user combined data!

[BeepBeep2]

No, I believe you got (almost all of) that wrong.

HDZ965FBK4DGMZ or (HDZ965FBGMBOX)= It's just a product code. Z = BE, M = C3 stepping (125w!), I = C2 stepping (140w).
CACAC AC 1009GPMW - 1009 means 2010, Week 9. GPMW talks about the day of week, what batch, where, and what kind of batch it was. (large or small)9C82851C00335 - 0335 means that that CPU was 335th from the middle of the wafer. (I believe, or at least from somewhere on the wafer)

so the GPMW doesn't have any significant value?

of course it has .. but we dont know what it actually means so we cant know exactly if it has any significant value ...

GPMW talks about the day of week, what batch, where, and what kind of batch it was. (large or small)

0335 means that that CPU was 335th from the middle of the wafer

Then everything I said restated again:

1009 = 2010, 9th week.... 1015 like my chip = 2010, 15th week

gpmw, dpaw, etc. are telling us what day of week, what plant, etc.. hard to use that for binning but if someone had access to enough chips they might see some trends. nothing worth changing chips over afaik

the 335 like your chip shows or 313 like my chip showed is telling us how far from the middle of the wafer

mmm makes sense now!

[Sil3ntSnip3a]

thanks guys!

[BeepBeep2]

thanks guys!

BTW:

If you get two CPU's that both say for example:

1015CPMW

chances are they are from the exact same batch.

[stangracin3]

so what are the weeks we want to stay away from?

[chew*]

mmm makes sense now! so week is most important and within that week if you can get a peach wafer placement then you are golden!

and finding how to cherry pick you cpu for the week you need to find is all based off of user combined data!

well and VID, vid is the ref voltage the board recognizes chips at.

so a chip that shows up at auto at 1.20v is super duper high leakage, and a chip showing up at 1.40 ref is super low leakage.

[Alex-Ro]

Also take in mind that letter written on the PCB.For example common letters for x4 and x6 are T and N.T is something common,N acts better in overclocking.For example,all 970BE i've seen in reviews had N written on the PCB

[chew*]

Also take in mind that letter written on the PCB.For example common letters for x4 and x6 are T and N.T is something common,N acts better in overclocking.For example,all 970BE i've seen in reviews had N written on the PCB

N means less than you think, N wafers die easier at high volts IME.

I can show you 2 N semprons, one is a total POS, other is preety damn good.

[Mad222]

Chew was right on the VID
Alex was right on the PCB letter which was mentioned months ago
However, the new batches with “N” PCB are something diff. from early batches like 1007/1008/1009; strong but hardly be OCed to a very high CPU Freq.

Thuban & Deneb are totally diff. animals & have diff. criteria for binning

Production week with letters & PCB could only give you hints to find a good chip on air. You need something more to get a monster on LN2. I’m still trying to ID the differences !

This is a new one I just bought. I need another LN2 session to understand this chip more….