Figuring out G.Skill's SNs

[websmile]

Last tests as I am making break from memory oc - Gskill were random buy at blowout sale on a shop(10321040), Team is weaker one of two kits I bought unopened and sealed lately at german forum



Still have a couple of kits equal or better, but in spite of legends, Gskill is not only brand to make happy

[Splave]

hopefully not from one of you guys no offense

[MaJ0r]

Last tests as I am making break from memory oc - Gskill were random buy at blowout sale on a shop(10321040), Team is weaker one of two kits I bought unopened and sealed lately at german forum



Still have a couple of kits equal or better, but in spite of legends, Gskill is not only brand to make happy

I got two kits Team Group LV DDR3 2000 one of them has PSC chips, but nothing special, SuperPi32M 2400MHz with 7-11-7 needs 1.75V, 2600 postcode 55. Second kit - has BBSE chips which are totally crap ) But in Ukraine Team Group with BBSE or PSC much easier to buy than G.Skill rate 2000MHz or higher

[websmile]

Quality on TeamGroup kits varies a lot, I other psc 2000lv kits that weren?t bad but far from the ones I have now, at auto timings these do 1300 8-12-8 at 1,70v. Voltage and settings are not maxed on the run at the screen, this were only some last tests on function as cpu is sold and board will follow.

[Dumo]

Got this pi kit. Probly from someone here

[XBrytanX]

I think you got it from reg. Joe ,i was biding on it ,enjoy

[Dumo]

^Just hope it will clock decent

New batch of TX 2600 C10. Seems to like v more than 2666 C10 hyk0

[Splave]

^ohh very nice! show us 4 sticks now

[websmile]

Did you remove Heatspreaders? If scheme on Gskill is still what several people observed, 2500 should be HYKO - at least you can find pics here showing this^^
http://www.xtremesystems.org/forums/...=1#post5153641

[Dumo]

^I don't wanna ruin these beautiful sticks for now. Lol

Still in the hunt for early batches 1240~1248 TridentX 2600C10 though

Got a few TeamX 2600s and so far not too good

[websmile]

Yes, they do like they do - and result looks really good - my best 2666c10 were 1244, and did quite well, easy 2600 9-11-11 32m and 1333 9-11-11 1m at 1,73v - keep on hunting

[Splave]

binning bbse is turning out to be a nightmare,
3x good sticks = 8-10-7-22-tWCL6 2500mhz 1.85v or less
3x okay sticks = 8-10-7-22-tWCL6 24500mhz ish 1.85-1.92v
5x garbage sticks = 8-10-7-22-tWCL6 2400 no boot 1.8-1.95v

Some sticks wont boot over 1.88v, others like 1.95....seems like a very dynamic IC lol

[Alex-Ro]

i know the feeling,elpida bbse is a interesting IC for sure,for some reason i thinking finding very good bbse is easier than psc....

[l0ud_sil3nc3]

ic pic from the 2650 c6 dimms

[l0ud_sil3nc3]

Clean bench are with everything ready to go



And another obscure G.Skill serial, first time I have seen the "G"angsta edition (thanks Sam)



*follow up

[bullant]

Some G Skill Pis 2300Mhz 8-11-8-28,on air no stability testing yet,will do this later.Sticks are from 2 different batches but they seem to clock the same in the little testing Ive done



Some Patriots I was working on a few nights ago,not bad from low bin kit 2000 9-9-9-27 1.65v PSC I believe

[Reefa_Madness]

During a discussion over at OCN about the GSkill numbering scheme with respect to the Hynix "1400" vs "2400", TapaKaH (Sam) made the following post and it seems to be valid when I've looked back at various GSkill parts since the April 2012 change in the coding.

My current guess is that 1400/1500 means that ICs have 1333 9-9-9 JEDEC bin and 2400/2500 is they have 1600 11-11-11 JEDEC bin.

With the above in mind, I was wondering if you guys that have modules with the spreaders removed could look at what I've recapped below and see if it holds up...it is a copy of a post that I made over at OCN after Sam made his.

That's an interesting take and you may be right. That would fit the pattern with Samsung, as well. The HCH9, a 1333 part was being shipped with the S/N "1500" in those 2400 C9 kits and the HYK0, a 1600 bin was being shipped with the "2500" S/N.

Following that logic then, a 2x8GB kit of Hynix based MFR PBC would have a S/N with a "2400" in it, but one made using the older MFR H9C (if used) would sport a "1400". Would you agree?

http://www.hynix.com/products/comput...=computingDDR3

Over at XS there was a pic posted of the ICs of one of these 2666 C11 kits with the "1400" and it was in fact CFR H9C (a 1333 part if I'm reading the info from the Hynix site correctly).

http://www.xtremesystems.org/forums/...=1#post5153644

Therefore a 2x4 double sided kit with Hynix timings and sporting the "2400" number should be expected to be using CFR PBC...a 1600 bin part?

http://www.hynix.com/products/comput...=computingDDR3

So if you know whether it is a single-sided, or a double-sided module, then based on whether you are looking at a 2x4GB kit, or a 2x8GB kit and whether it has the 1400/2400 for Hynix or the 1500/2500 for Samsung, you have a pretty good chance of guessing what is under the hood.

Does this look right?

For 4GB modules:
single-sided 1400 = MFR H9C (1333 part)
double-sided 1400 = CFR H9C (1333 part)
double-sided 2400 = CFR PBC (1600 part)

single-sided 1500 = the Samsung 4Gbit HCH9 (1333 part)
double-sided 1500 = the Samsung 2Gbit HCH9 (1333 part)
double-sided 2500 = HYK0 (1600 part)

for 8GB modules:
double-sided 1400 = MFR H9C (1333 part)
double-sided 2400 = MFR PBC (1600 part)

double-sided 1500 = the Samsung 4Gbit HCH9 (1333 part)
double-sided 2500 = ??? (I don't know it)

There is also a Hynix AFR that I've gotten on some generic 8GB modules and it is like MFR so it is possible to get these instead of MFR.

http://www.hynix.com/products/comput...=computingDDR3

There are also some Samsung variants to HCH9 that have popped up from time to time, so there are no guarantees, but generally speaking, most Hynix based stufft these days seems to be either CFR or MFR, and most Samsung would be either HCH9 (2Gbit or 4Gbit) or HYK0.

I also recently received generic 4GB DDR3-1333 CL9 sticks with Hynix "H5TQ2G83EFR PBC" instead of the typical BFR or CFR. Haven't tried them out yet. Has anyone else had a chance to test any EFR based modules?


http://www.hynix.com/products/comput...=computingDDR3

[Alex-Ro]

Yea,massman did,he said they were absolute crap...

[Don_Dan]

I think Tapakah is right, I've checked the pic of a "0800" kit I saved, and the IC is indeed a 1066 7-7-7 part.

Concerning EFR.

[Reefa_Madness]

Thanks for the confirmation...and "Concerning EFR", I threw in my pair right after my post and I couldn't do 2133 either, so my experience is consistent with his.

Off they go to the mother-in-law's build, or similar.

[TaPaKaH]

Already seen ver 5.13 on Corsair memory ... careful with EFR

[Reefa_Madness]

@Sam...thanks for that heads up.

Note to self...put "ver5.13" on the IGNORE list.

[stunned_guy]

Have here Hynix BZR on 1600C9 RipJaws.

[Don_Dan]

02x0, x400 or something else?

[stunned_guy]

It is 0240