Alex@ro

for example in romania and germany after asking 20+ shops,they all have samsung based 2666C11,so the rules are fair enough

strong bdbg!

thanks for clarification

oh,for mfr they are very real for samsung or hynix cfr not quite... so dfordog is clearly mfr due to validation and picture: This is regular samsung kit we find in retail,again look at smd-s upside contacts:

hey george,yes it should be samsung as all 2666 predator we know of are samsung,however looking at smd arrangement it is clearly single-sided hynix mfr.also,it is very hard if not impossible for samsungs to reach 3300+ for valid.looking for a clear answer on this,i mean if some region can buy retail kingston mfr and other ones can't how can we all have equal chanches?

only submit 1 picture of your rig,then add the others by editing,i got the same error too,when using multiple larger images. btw,where is this mfr coming from,cause i\m sure as hell that double sided cfr won\t do this? http://hwbot.org/submission/2437885 http://hwbot.org/submission/2437846 since all kingston's lineup in europe(i called over 20 shops) and usa are using cfr,how are we supposed to compete with mfr guys?

binning on air .. http://valid.canardpc.com/gpw3m7

this gives me hope,my best chip can't prime 1.15 on air with top high end cooler,reaches 95+ lol

i do believe so,but you need to test it on cold.my best 4670K\s do wprime at 1.12 and 1.14 4.5 but they do 5 ghz @1.26 and 1.31,so this means a much nicer volt/mhz curve.Ofc cold could change some things....

yeah,the haswell wall is something interesting,i had some chips that could do 4500 with low-voltage but would not hit 5 ghz for enterin os even with 1.55....perfect chip would be a chip that has low cb,high-leakage and no freq wall....

will someone complain if my 2666 predator kit run 2933 9-12-12 for 32M ?Seriously hope not....//

Damn man,that is one crappy batch,hopefully it scales somewhat on cold...

Hello,i have another 4670K for sale,basically as new,only tested for 5 ghz,comes with full box and warranty papers Strong chip,can do 5 ghz@1.31Vcore with 5000 cache(ring)@1.42 for 32M Batch: L311B436,no idea about CB as it was never tested under cold,only 30 minutes air. Price: 300E +shipp Regards.

funds received,SOLD,can be closed,thanks hwbot

pending

For light benches like pi,pifast,2001 aquamark up to 1.5 is fine on air,i have not seen any degradation whatsoever

With sadness i have to sell my best kit of psc due to some problems with car and need of money.This is my result of binning dozens of kits with this coming as the best from all of them. They do: 2600 7-12-7-27 twcl 6 ultra tight 32M at 1.96V all air 25 ambient 2666 8-12-7-27 twcl 6everything ultra-tight can run 32M all day at 1.87V(25 celsius ambient) 2722 8-12-7-27 twcl 6 ultra-tight 32M at 1.95(~20 ambient) 2600 7-12 2666 6-11-6-28 ultra-tight twcl 6(could be tighter but was first play) on dice -60 degrees and 2.16(can run also @2.11 but had that just to be sure) 2712 6-11-6-28 twcl 6made half a run of 32M at 2.20 still on dice. When i say ultra-tight i mean tweaked for 32M and stretched almost to limit,i do not test with twcl 8 or something. 8155 pcb guarantees no problems with cold so these should be prime candidates fo 2800 cas 6 run. They come with full package,excellent condition vizual and functional and all results are guaranteed given you provide them a decent IMC and similar parameters as mine. Price: 500$ or 400E,+ shipping(about 20E to europe and us,other countries please pm me first).

At 5 ghz only 32M since i have yet to found a chip that does prime For lower frequency stability,say 4.5 i use prime with avx.

Haswell is such an interesting animal...so,after 15 chips.... -general rule for me,the better the core(lower voltage for 5 ghz) the better the cache/ring frequency -some chips can do 4.5 stable with 1.15 let's say but they end up not doing 5 ghz at all,while some require 1.2 and they do 5 ghz under 1.4 -VID could be an indicator of chip quality but not main factor,got 2 0.97 that would require 1.45+ for 5 ghz while my best chips*under 1.3 * had 1.02 -imc quality varies,but not related to core/cache,the best imc i encountered*3400 stable 32M with 1.15 vccio a/d had almost the worst cores i've seen requiring 1.3V just to enter os at 4.4 ghz -chips behave differenty when searching for lowest voltage for a given frequency,example while testing for 5 ghz 32M some chips would give not exact in round or sqrt,some would simply freeze the screen while others will do classic 101 or 124. -ram frequency play a role in vcore required for stability but not to all pieces,testing stability at DDR3-3000 some would require +0.02-0.3 over DDR3-2133,some would need the same vcore. Still have some things to figure out,maybe other members can chime in with experiences and binning multiple haswell so we can create a has$$well wikipedia here

yeah.while sorting out haswell chips,there was a direct correlation between cache(ring) clock and core,the better the core the better the cache frequcny,shitty core meant shitty cache freq.

best piece does 5 ghz 32M with 1.275,also 5300 validation on normal air cooling,will be testing it under cold soon http://valid.canardpc.com/c0r867 second one is 5 ghz with 1.315,both are very strong in ring frequency,5000 32M stable with 1.45 for first,second only couple of loops

4670K run in the same park as far as i am concerned,had no luck with 7 4770K,only junk mostly,switched to 4670K,found 3/7 that can do 5 ghz under 1.4V,L309B384

was sold on local forums,can be closed,thanks hwbot

big difference between 1m and 32M,biggest i have seen so far,all my chips needed 0.02-0.03 between 1m and 32M

great chip nethertheless,u aiming for higher vcore?