new toys: 1090t (4.25ghz blend stable), chIVformula, supertalent hypers, g.skill SSD

[BeepBeep2]

With your ram I suggest running a 3000 CPU_NB speed...does the NB scale any over 1.3v?

[FlanK3r]

nice Diamont, can u try absolut maximum with 1.55V (1h blend?) -for interesting. Thx.

[crazydiamond]

my nb wont do it for 24/7 beep. it can barely bench over 3000. probably somewhere around 2900mhz w/ reasonable volts. it scales over 1.3v but very poorly, ill be testing it more soon.

ill check it out flank3r. i have 30min. run at 4300mhz 1.57vcore load, and 30 min. run at 4280mhz 1.55vcore load but i think this one vcore could come down a little bit. ill try to find max mhz for 1 hour prime with 1.55vcore load today

[crazydiamond]

here we go flank3r but i admit i cheated and used 1.56vcore dont taze me bro! also, cpu-nb is set to 1.25v in bios, cpu-nb llc off

1hr blend
4320mhz - 1.56vcore load - win7 64bit - 71f ambients

[Daveburt714]

Very Nice CrazyD!
54c max Tcase isn't shabby either.

I usually try to run my IMC up at the same time (~3Ghz).
I think my chip might do that too if it weren't for the locked multis/HtRef limit..

[crazydiamond]

here we go dave, went at it again with higher nb. i tried the same settings as before, 4320mhz 1.56v & this time nb up 1 multi at 2960mhz. started with 1.33v on cpu-nb and windows froze as soon as prime started. tried 1.37v on cpu-nb and windows froze about 2 min. in. tried 1.39v and about 6 min. in, windows said prime encountered error, needs to close. 1.425v on cpu-nb and got same error 8 min. in. i think i needed 1 more bump on vcore to get that high cpu & nb together so to keep w/ flank3r's 1.55v max, instead of bumping voltage i dropped ht by 1, and lowered vcore by 1 (in aod, .0125) this gave me 4300mhz, 1.55v load - 2950mhz nb, never restarted so still at 1.425v on cpu-nb lol. let it prime & sure enough no problem so like i said im gonna leave nb testing for another day to find out minimum it needs. ill also try 4320mhz 2960nb again with 1 bump on vcore to see if i was right in thats why it was failing

4300mhz - 1.55v load - 2950mhz nb, 1.425v - 71f ambients - bla bla =)

[FlanK3r]

This like me ,-) !

[BeepBeep2]

That's better

Nice NB clock, CPU clock, everything clock.

[crazydiamond]

spent a few hours playing w/ cinebench tonight got a pretty good improvement over my last run. tried to get a vantage run in for you titan w/ same settings as cinebench below, but cpu test crashed 2x in a row & i didnt feel like tweaking anymore after 2hrs of cinebench xD think i could pass if i back down on nb/imc just a tiny bit, i can get cpu way more stable with around 2980mhz nb rather than 3000mhz+ 5850 passed 3d tests @ 1030mhz core 1275mem no problem, definately more left in it. ill tackle it someday

heres my cinebench run, tried to run it again cuz sometimes i get better score on 2nd run but errored out

4402mhz - 3006nb = 7.89 --- 71f ambient - win7 64bit

[FlanK3r]

congratulations, looks good your chip

[crazydiamond]

thanks flank3r i cant stop tweaking/playing with this thing! pretty soon im gonna do some more prime on the cores, find out vcore for clocks i havent check yet, 3.8, 3.9, 4.1, etccc. then after that will do same tests for nb, just test minimum volts for stability at 2600mhz, 2700, 2800, 2900, max... THEN after that i was wanting to try to test some of these cpu/nb clocks but with TURBO CORE enabled, but i think we need better bios for that. i think when turbo activates, the other cores which are not using turbo are supposed to downclock to low p-state but i think the way it is now they stay at normal i want to test 4 core, 3 core, 2 core turbos by using same amount of threads in prime95

[FlanK3r]

i found my "summer profile" 4 GHz with 1.328V :-)

[crazydiamond]

thats a very nice combo of clocks/volts if turbo was working properly you could probably have that + 4.3 or 4.4ghz with turbo

[Barr3l Rid3r]

here we go dave, went at it again with higher nb. i tried the same settings as before, 4320mhz 1.56v & this time nb up 1 multi at 2960mhz. started with 1.33v on cpu-nb and windows froze as soon as prime started. tried 1.37v on cpu-nb and windows froze about 2 min. in. tried 1.39v and about 6 min. in, windows said prime encountered error, needs to close. 1.425v on cpu-nb and got same error 8 min. in. i think i needed 1 more bump on vcore to get that high cpu & nb together so to keep w/ flank3r's 1.55v max, instead of bumping voltage i dropped ht by 1, and lowered vcore by 1 (in aod, .0125) this gave me 4300mhz, 1.55v load - 2950mhz nb, never restarted so still at 1.425v on cpu-nb lol. let it prime & sure enough no problem so like i said im gonna leave nb testing for another day to find out minimum it needs. ill also try 4320mhz 2960nb again with 1 bump on vcore to see if i was right in thats why it was failing

4300mhz - 1.55v load - 2950mhz nb, 1.425v - 71f ambients - bla bla =)

What cooler? Looks nice, I can prime at 4.1Ghz / 1.48v on air, but my NB is sweet 2940Mhz stable at 1.32v (not as sweet as my 1055T but anyway is good.)

What a hell is that 120C temperature?

thats a very nice combo of clocks/volts if turbo was working properly you could probably have that + 4.3 or 4.4ghz with turbo

Why use turbo if we have CnQ? he doesn't even know wich ones are the good cores.

[crazydiamond]

What cooler? Looks nice, I can prime at 4.1Ghz / 1.48v on air, but my NB is sweet 2940Mhz stable at 1.32v (not as sweet as my 1055T but anyway is good.)

What a hell is that 120C temperature?



Why use turbo if we have CnQ? he doesn't even know wich ones are the good cores.

thx barr3l.. its water cooled, you can see the loop in my 1st post, sig tells hardware.

turbo can be used w/ cool & quiet. ideally, if they both worked as they were supposed to, flank3r could have clocks like this:

idle/no load = cool & quiet p-state (1ghz/1v range)..
6 cores full load = normal p-state (4ghz/1.33v)..
2,3,4 cores loaded = turbo p-state (4.3-4.4ghz/1.45v range), then when in turbo, whatever cores arent being used will be put in cool & quiet p-state

it shouldnt make a difference what cores are the good ones etc. since he can get 4.3ghz stable on 6 cores it should be fine w/ 3 or 4 cores, maybe even able to get 4.4ghz for turbo since its not all cores at once

[crazydiamond]

some more blend today waiting on another half hour of 3.8ghz 1.284v i have min. volts for 3.8-4.3ghz now, might just skip to like 3.5, 3.0, 2.5ghz then start testing nb

4100mhz - 1.392v load - 75f ambient (this took 1.53v on deneb )


3900mhz - 1.308v load - 75f ambient

[BeepBeep2]

*claps* 4100 takes 1.65v on my CPU

[crazydiamond]

thx beep, i agree w/ the claps, amd deserves a standing ovation

3800mhz - 1.284v load


3500mhz - 1.2v load

[FlanK3r]

think 3500 MHz u can set lower a bit . Try it. And dont forgot at LINX

[crazydiamond]

1.18v would give me math error on worker 5 twice in a row, first was 9 min. & thought maybe memory problem cuz i had a ton of progs running.. tried it again w/ nothing else running at gave me same error on same worker, just this time 19min. in. that was my weakest core when testing valids too. where is ACC when i need it right now im testing 3200mhz 1.104v - 34c load xD EDIT: failed bout 20 min., bumped to 1.116,,, this should pass for sure

any specific clocks you want me to test on linx? how long should i run it?



edit: 3200mhz - 1.116v load

my 1090 said he wants a break for awhile

[Barr3l Rid3r]

here it goes very good and very happy also

[crazydiamond]

wow very nice barr3l, great clocks all around!

[Barr3l Rid3r]

wow very nice barr3l, great clocks all around!

Im going to C7 now, it seems better. 1800+ easy w/ less voltage and better everest performace, what was I waiting for?

[BeepBeep2]

Im going to C7 now, it seems better. 1800+ easy w/ less voltage and better everest performace, what was I waiting for?

C6 is optimal. C7 unless you are running 2000+ don't bother.

[Barr3l Rid3r]

C6 is optimal. C7 unless you are running 2000+ don't bother.

optimal for what? I can choose between 1680 6-6-6-21 or 1840 7-7-6, bandwith results were better in the 1800 range.