Printable View
LO39B214
Almost done with my testing here...
My 24/7 settings will be 4.6ghz 1.35V
Everyting in the bios was left as DEFAULT, AUTO, OR STOCK, except for:
By all Cores
46
CPU Voltage 1.35V
RAM Voltage 1.5V
I was able to test 4.8ghz 1.4V with Prime95 for 1H using the IN PLACE LARGE FFTs test but I'm not confortable with 1.4V. However, I experienced some voltage drops as low as 1.376V
http://img.photobucket.com/albums/v4...n23/48ghz3.png
great for me too about the same , 4.6 ghz will require 1.35-1.375 to be full stable , 4.7 Ghz 1.4.-1.425 and 4.8 Ghz 1.45-1.47 , interesting to see that the core still responding well to voltage and giving more stable clock , strange but look those chip like voltage hope they can handle 1.4-1.45v for long (lol 1 year or so)
They will probably not keep the speed if you guys keep abusing them with LinX(SP1)+AVX/prime at these voltages. Several people (including C-N and the author of the ud7 overclock guide) have gotten degrading overclocks after abusing the chips with repeated stress testing. After seeing this, I'm not going to run prime at high voltages anymore; it's just not worth risking the chip. I'd rather game for a day rather than burn my chip out running prime for 24 hours...
I've never been a fan of stressing a CPU with these programs...in fact none of my previous rigs have ever been Prime stable for long, not even 5 minutes for my last Q6600 @ 3.6ghz, yet it almost never crashed on me.Quote:
Originally Posted by Falkentyne
This last build has been very unstable, it even BSOD on me while watcing a movie or trying to reply to a thread (did happen) so for the 1st time I felt the need to test it some more with Prime95 and other stressing software.
Im glad people are testing the chips hard. Its the only way to see what they can take, or what kills them. Im ok with learning from someone else's mistake. :rofl:Quote:
Originally Posted by Falkentyne
Yeah I'm ok with this too. I degraded too many chips from stress testing in the past.Quote:
Originally Posted by OC Nub
I'm at 24/7 at 4 ghz for now. 4.5 ghz works with all auto even with LLC off.
At 5 ghz, I just run my 8 thread chess engine for a stress test. If that doesn't crash or error out, I'm usually good to go in any other game. I wouldn't mind using Prime, but NOT at 90C. And that is with a Venemous-X, too. I actually tried my TRUE Black (the mount was EXTREMELY secure and it refused to swivel at all), and some MX4, and the temps were slightly worse. Went back to my Ven-X (which does swivel if I apply pressure, but is otherwise secure), and tried Chill factor 3 (apparently this is the same as MX3), and had slightly better temps than even MX4.
I guess the high temps is due to the overvolt to 1.456 during load. Fan is a 103cfm panaflo. Maybe I'll try my 190cfm Delta later, but I doubt that will do much due to how FAST the temps reach 80C.
I had a nice full map rotation marathon on Black Ops last night at 5 ghz. Felt great not having the FPS EVER Drop below 100 when the spam started going off...
So total vcore needed for load stability seems to be around 1.45v for 5 ghz with LLC overvolt from 1.4 (manual)/1.39v (auto) bios setting).
Stuck @ 51/52 multi max... air or SS:(
All the same batch @ Micro Center Yonkers NY last week
http://img840.imageshack.us/img840/3...nshot001mw.jpg
Got the same mobo and cpu with exact batch. Looking like it's not a gem. Just waiting on Amazon for my heatsink to finish build.
Quote:
Originally Posted by Dumo
It is what it is & what you call it depends on how you look at it. I don’t see it as degrading as such its more like maturing to true stability. I would rather beat the golden edge off a chip in a few weeks than have to add a vcore bump every month for 6 months. I don’t like the D word & think many people have called a chip degraded when they were close to the top of what a chip could do temp wise only to find it wont do it anymore as it needs more voltage they cant cool.... SB is different my chip still does 5GHz @ 1.45v load CPU-Z & still runs in the mid 70's on air in my front room with a 15~20 ambient.:)Quote:
Originally Posted by Falkentyne
2600k In Da House. Batch# L040B165 :).
http://img.techpowerup.org/110125/Capture086.jpg
Finally found my babe, but still no MB.
Still trying to find a good MB for a compact build, but I'm torn apart between ASRock P67 Extreme6 , ASUS P8P67-M PRO or the little cute ZOTAC H67ITX-A-E.
I know, it's all over the place, LOL, I'm going to use a few days to decide between size and OC-headroom .
Can you give me a rundown (or template) of your BIOS setting? I'm also running Asus P8P67 Pro. I want to see if I can get mine to run stable at that speed. Right now, I can't even run 4.8Ghz stable. My batch is: L040B305. Thanks...Quote:
Originally Posted by Peter949
I've got the same batch, and I'm running @4.5Ghz with 1.275v also on a p8p67 pro. Will try 4.7 later on, but seems a big bump in vcore for 200mhz more ? If it's the mobo that makes the difference I would say +1 for asus !Quote:
Originally Posted by aytacaksel
batch L036B120T ,Asus p8p67pro, my 24/7 clocks
I'm sorry I am too lazy for screenshots but for the sake of adding more info to this thread here are my chip stats:
L040B165
4.7 @ 1.32volts PLL Overvolt OFF
4.8 @ 1.38volts PLL Overvolt ON
4.9 @ 1.45volts PLL Overvolt ON
Those settings are Prime and Linx stable. I run 4.7 because I do not believe the PLL Overvolt setting to be safe for 24/7.
batch#L040B671
4.9-1.4v
5.0-1.45v
http://img13.imageshack.us/img13/9743/51135390.jpg
Uploaded with ImageShack.us
I ran Hyperpi32m and it completed without any problems. I also was able to game last night for a few hours at my prime unstable 4.9ghz with 1.395 vcore. The night before I was bluescreening with games within a half hour at those exact same settings. Still prime unstable though. There's nothing more vexing than inconsistent instability!Quote:
Originally Posted by Sam_oslo
Hey One-Hertz, did prime stability require significantly more vcore than linx stability? Did you run prime blend or small fft's, and for how long?Quote:
Originally Posted by One_Hertz
Yes, I found prime much more difficult to pass than linx. I ran blend. I tested linx and prime for about 40 minutes each to call the settings stable.Quote:
Originally Posted by drnick
No boot with 1.60v @Bios.. also no boot with 55x102 and 1.60v. i don't wanna give more than 1.60v with air. i am afraid to damage the cpu. :shrug:Quote:
Originally Posted by Alex-Ro
:wave:
Heh, lucky man, I have opposite problem, board in the house, CPU not :)Quote:
Originally Posted by Sam_oslo
http://www.ulozisko.sk/obrazky/363843/deska22.jpg
LOL, I would send the CPU to you, but not that batch#. I have to try that babe myself. :rofl:Quote:
Originally Posted by FlanK3r
You went the wrong way. It's a good idea to hunt-down a good early-batch and wait for MB-laboratory-rabbits to iron out MB-bugs.
Usually I buy Intel CPU at the first day, because early-batch#s tend to have a better OC-headroom.
But in this round, I've studied CPU-laboratory-rabbits in several days, and then hunted down this baby.
Pretty cool, I got my components yesterday. I'm from Norway (currently live in sweden), and my batch on the CPU is #L040B167 ;)Quote:
Originally Posted by Sam_oslo
The little sh** clocks like there was no tomorrow - my first inital test it did 4.9Ghz / vcore 1.385 / Internal PLL enabled / Extreme profiles.
I love this new generation, I was actually a bit disappointed they changed overclocking methods this drastically - but it has certainly made me change my mind!
I just took delivery of my CPU, which may be the newest in this thread. To be built with a GD65 and 4Gb XMS3 when I get home from work:
i5 -2500K
L049A796
Anyone know what the significance of the A is? Most other CPUs in this thread have a B in that place.
/Edit/ 45x100 set in BIOS seems stable at 1.25V. CPU-Z shows 1.225V idle and 1.264V under Prime95 load. MSI Control Center has 1.2150V displayed. All other voltages on Auto with Vdroop control on low.
Different cpu wafer, production sample. CPU production is divided in unequal wafers as their lifecycle advance.Quote:
Originally Posted by roadie
http://www.treehugger.com/intel-cpu-wafer-d001.jpg
In a perfect world, all end-users would get a production sample from the inner parts of the wafer, as they have the least flaws.
As you get to the outer area of the wafer the % increases, although Intel has a limit of "fault tolerance" for each batch.
2500k L042B291 4.8ghz@1.35v passed linx. Gonna try a 5ghz run.
edit: just a average batch. looking like i'll need 1.37v just to get 4.8ghz 24/7 stable / 8hrs prime stable.
So, try to get batch with the smallest last three number as possible, correct?Quote:
Originally Posted by tentp0le