Everything about INTEL IVYBRIDGE...

[sin0822]

nice action

[Eldonko]

5650 was max I could do with SS. Heat from these things is stupid, you need LN2 to get anywhere.

[@rne]

Not bad. Max 5,3GHz with Single Stage here and 3570K.

[SoF]

5650 was max I could do with SS. Heat from these things is stupid, you need LN2 to get anywhere.

just had 2 crappy chips that would hardly do 5.7 on cascade


first post updated!

[sin0822]

max 5.3ghz here on air. Thats validation tho.

[Calathea]

so Ivy sucks for everything but ln2 right? (Ok, it's a great chip for everyday comp, whatever). I, Mean Machine and ME4ME tested two retail chips this weekend, pi stable at around 6.3-6.4 GHz @ 1.75 vcore. Any estimations on SS -50?

[SoF]

I think they need to push much higher vcore for 6.5+ and you need to run fullpot.

heard about petri's pot doing well (I think he has most exp with LN2 and fullpot on AMD and optimized his design therefor so that sounds very plausible).

stummerwinter needs 1.9+ for running past 6.5 on his retail.

Best and wors chips in a batch are within 200 MHZ under cascade - if you have a good one it is real fun! saw 6 gig 1m and pifast on a good -100 cascade already in Germany.

[Calathea]

We did test with 1.80-1.85 vcore and of course full pot all the time No scaling past 1.75 really... does that sound odd?

[Eldonko]

Unless you are using LN2 your chip will stop scaling because there is too much heat. For me on SS above 1.7 did more harm than good other than for a suicide SS. With LN2 chips should scale up past 1.9v from what Ive seen.

[SoF]

First stepping list - cascade is only -80°C so you might want to add 100 MHZ if you are in the -100°C club

[MaJ0r]

SoF which Vcore did you use?
I tested 3 chips, with SS(about -40C):
two could run SuperPi 32m at 5600MHz with 1.7V and best one could 5700MHz.
With LN2 I killed best chip and tryed another(random from two). Max valid http://valid.canardpc.com/show_oc.php?id=2341144
full pot whit high memory divider. I think these cpu are bad

[SoF]

which stepping you have?
also for quick testing P67 seems more easy for some reason. need to spend some time to max out a chip on mvg...
last chips btw can do 5700 with some work...just not so "easy" and so no need to waste time on them.
plug in my good chip 5.8 to 5.9 easy...1.72-1.76V seems to be best sometimes a bit lower depending on bench.

want to test one of the last chips again on MIVE just to be sure. don't have much time these days unfortunatly

[MaJ0r]

I tested L153A673 Malay retail
Stepping 9(E1 or so)?

[SoF]

yeah last chips from 2011 - saw one of them doing 6 gig on cascade here.
tend to be really OK for first retails.

€: link fixed

[SoF]

My girl is inspired by the modjo family and ivy

[Stelios]

SoF which Vcore did you use?
With LN2 I killed best chip and tryed another(random from two). Max valid http://valid.canardpc.com/show_oc.php?id=2341144
full pot whit high memory divider. I think these cpu are bad

Just saw that ... That sucks man ...

How did you kill the CPU under LN2 ? What voltages ? High PLL ? High VCCIO ? VCCSA ?

[MaJ0r]

I killed one 3770K very quickly and I didn't understand how ) Just tried boot with 1.92V with temp about -100C )
PLL=2.0V, VCCIO=1.25, VCCSA=1.25
With second cpu I used "safe boot" at 6018MHz 59x102 with 1.75V core and -140-160C than I changed multiplier in OS, but I had to use ASUS AI suite II and Turbo evo(too many process in OS )

[M.Beier]

I killed one 3770K very quickly and I didn't understand how ) Just tried boot with 1.92V with temp about -100C )
PLL=2.0V, VCCIO=1.25, VCCSA=1.25
With second cpu I used "safe boot" at 6018MHz 59x102 with 1.75V core and -140-160C than I changed multiplier in OS, but I had to use ASUS AI suite II and Turbo evo(too many process in OS )

Question... Did you ever get a higher overclock by increasing VCCSA?
1.92vcore shouldnt be the killer, nor the VCCSA, but... Hmm... PLL 2V? Hmm... 1.7-1.8v would make much more sense.

[Stelios]

I killed one 3770K very quickly and I didn't understand how ) Just tried boot with 1.92V with temp about -100C )
PLL=2.0V, VCCIO=1.25, VCCSA=1.25
With second cpu I used "safe boot" at 6018MHz 59x102 with 1.75V core and -140-160C than I changed multiplier in OS, but I had to use ASUS AI suite II and Turbo evo(too many process in OS )

I don't like the combination of high vccio & vccsa . We killed two 2600K by doing this .

As Marc wrote , 2.0V PLL is also not recommended .

Only time will tell what is killing them . Hope we find out soon and not kill any gold CPU's !

[SoF]

I killed one 3770K very quickly and I didn't understand how ) Just tried boot with 1.92V with temp about -100C )
PLL=2.0V, VCCIO=1.25, VCCSA=1.25
With second cpu I used "safe boot" at 6018MHz 59x102 with 1.75V core and -140-160C than I changed multiplier in OS, but I had to use ASUS AI suite II and Turbo evo(too many process in OS )

Plll 2V is ok -80°C - many chips seen that in my hands, all took it with a smile
As for IO/SA I'AM A :banana::banana::banana::banana::banana: - I admit, but I usually don't exceed 1.15V at all...not needed!
1300+ on air with 1.075/1.1 and on cold that only gets you nuts on bad chips - had 3x 5.7 in a row and tended to f*** them...plugged in my "be-happy"-chip and accepted they are no good.
once you have a good chip --> work your way from there and don't waste time...binning is still painfull at a certain point...

I don't like the combination of high vccio & vccsa . We killed two 2600K by doing this .
As Marc wrote , 2.0V PLL is also not recommended .
Only time will tell what is killing them . Hope we find out soon and not kill any gold CPU's !

stay within 1.2 io/sa / 2V pll first, run 1300+ cl9 and max the chip mhz-wise on ALL 2D and classic 3D benches - then do what you think it leads you to heaven

[tiborrr]

L209C207 #1579... max valid @ -95°C 5844MHz, 1.832V:
LN2 next week. Sry, I'm too lazy to translate now.

http://oc-lab.si/forum/viewtopic.php?f=5&t=6056

Ivo prvič na poštenem mrazu
Gornja Radgona, 27.04.2012

V četrtek popoldne sem po pošti od Simona fasal njegov prvi Ivy Bridge (IVB na kratko), ki ga je uspel kupiti potem kitajskega eBaya pred kakšnim tednom dni. Ker je bil včerajšnji dan, t.j. petek, rezerviran za hribe (domov prišel šele v soboto ob pol enih zjutraj), sem se navijanja lotil šele danes. Seveda sem se skuhal kot rak ("kreme za sončenje so za šleve" mentaliteta...), zato danes nisem v vizualnem stanju niti za v gostilno . Na žalost se nama s pajdašem Moonmanom ni uspelo dobiti, da bi zadevo 'razdevičila' v dvoje (o kriza, kakšen stavek ), zato se bo tortura nadaljevala po praznikih, najverjetneje v družbi parih solidnih grafikih in tekočega dušika. Takrat 6GHz+ ne bi smelo uiti.

Pripekalo je in se vdiralo do riti...




Tukaj so moje hitre ugotovitve glede tega čipa:
- čip prav ljubi mraz, vsakih 10°C nižja temperatura da v grobem ~80MHz višjo frekvenco
- čip požira volte in z njimi skalira kot za stavo, obstaja velika nevarnost kurjenja čipa na krilih samozavesti (2Vcore je po izkušnjah Ukrajinca Maj0r-ja tudi na -190°C smrtno)
- Pi32M ni ravno hiter (?)
- Ivy je E8600 za leto 2012! Sedaj se tudi s povprečnim čipom, kompenzirano z mrazom in volti, da dosegati luštne rezultate. Koga briga, če je čip IVB za 24/7 slabša izbira za nedeljskega navijalca kot SNB.

To ni prvič, kar sem se igral z IVB-jem. Že v torek, 24.04.2012, sem na Simonovo povabilo odšel v Portorož, kjer sva na hitro ob skodelicah vroče kave hitro preizkusila njegovo najnovejšo igračko, Core i5 3570K 'Ivy Bridge' procesor, prvi retail primerek na slovenskih tleh, na njegovem single-stage kompresorskem hlajenju (cca. -44°C v loadu). Po urici ali dveh igranja sva hitro prišla do omejitve čipa, ki je znašal okoli 5.48GHz (SuperPI 1M - 1.725V) oz. 5.32GHz (Wprime 1024M - 1.6V). Omenjene podatke delim zaradi lažje predstave o samem skaliranju (najin sistem je z izjemo rama identičen):




Sistem:
- Intel Core i5 3570K retail (L209C207)
- ASUS Maximus V Gene (0701 bios)
- 2x2GB Kingston HyperX DDR3-2133 (Elpida)
- PoV GeForce 8400GS 256MB PCIe
- Teamgroup 8GB SSD
- 750W Seventeam V-Force PSU

Hlajenje:
- CPU: dvo-stopenjska kaskada (R-507/R-1150) @ ~ -95°C

Izolacija oz. bolje rečeno zaščita pred vlago:
- stiropor (jeba, ni bilo armaflexa nikjer pri roki) za pod ploščo
- gumice okoli CPU-ja
- živilska folija (za zaščito gumic pred svinjarija + dodatna zatesnitev)
- papirnata brisača (da vpije morebitno vlago okoli uparjalnika)



CPU-Z suicide shot:


2D rezultati:
- Pi1M @ 5821MHz, 1.83V
- PiFast @ 5757MHz, 1.75V
- Pi32M @ 5768MHz, 1.73V
- Wprime 32M @ 5655MHz, 1.77V
- Wprime 1024M @ 5617MHz, 1.75V









Glede na moč procesorja in ustrezne pogoje, ki jih je imel, ni treba izgubljati besed, da so vsi dobljeni rezultati novi slovenski rekordi. Ker bodo čipi že ta teden postali široko dobavljivi tudi pri slovenskih trgovcih, se je kaj kmalu za nadejati konkurence!

Lep OC pozdrav,
Niko :drunk:

[SoF]

Good to see moar chips! 3570K doesn't seem to do that well
You deserve a 3770K that "bluescreens" at 5.9 boot @-80°C

Just tested some bclk 3D and MVG tweaking today...ppc is very awesome with the right setting.





on 1000mhz air-gpu 100 mhz cpu more will give you ~2000-2200 points in final score

[MaJ0r]

I don't like the combination of high vccio & vccsa . We killed two 2600K by doing this .

As Marc wrote , 2.0V PLL is also not recommended .

Only time will tell what is killing them . Hope we find out soon and not kill any gold CPU's !

Guys those voltages which I used were at LN2 mode ))
I saw Mad222 and other guys did cpu pll mode to increase it.
Vccio/Vccsa - I dont think these voltages could kill my cpu. I think Vcore was too high for low temp.
Anyway thanks for yours attention and comments, next time I'll be more careful.

[Cecil]

At ~0C or slightly below (maybe -10C) Sandy would probably clock higher, but which would be fastest at highest speeds obtainable in that temp range? Thinking of doing some more vid card benching.

What about at -40C SS range?

[sin0822]

i have used upto 1.3-1.4v on vccsa and vccio with LN2 full pot and no death, on more than 1 CPU.