Splave Posted May 23, 2012 Posted May 23, 2012 ^ very true and I think alot of people rely on cpu-z as actual volts not dmm values. Quote
Viss Posted May 23, 2012 Posted May 23, 2012 (edited) I suspect the rule is true, but has to be approached from 2 directions. I *expect* that all 6.5GHz+ chips do 5GHz at 1.275-1.30v? But not all chips that do 5GHz at 1.275-1.30v? will do 6.5GHz+ Depends on what your booting at (number of cores and HT). I used 50x100 4c/8t for testing on air, the chips that booted with these settings @ 1.375-1.425v also did (or really close to) 6500mhz spi 1M on LN2. For air tests i used P8Z77-V Deluxe, Prolimatech Mega and LCC set to High, ambient 20 deg C. My best so far boots 50x100 4c/8t @ 1.3625v and does 1M and pifast at 6700mhz affinity set to strongest core. I also found several chips that could not boot 50x100 at 1.55v in exact same batch. Edited May 23, 2012 by Viss Quote
Massman Posted May 23, 2012 Author Posted May 23, 2012 I use three pointers: 1) 5G+ on air 32M (really bad air cooler) 2) 5.5G wprime at -30°C 3) 6G wprime at -90°C If the chip meets these criteria, it's a good one (imho). So far, I only found 2 chips that could do that. Quote
K404 Posted May 23, 2012 Posted May 23, 2012 Depends on what your booting at (number of cores and HT). Exactly Most of Ivys uses are single threaded. Even the 06 scores that are going up are without HT. needed ~1.4V also to boot 5 gig on air. Settings I usually used: 1.7 - 1.75V -80°C 57x100 or higher bclk usually doesn't made a big difference at that point I think 6 gig 2D on cascade (-80 till -100°C) is a good indication of an ok chip How many cores? Quote
Hiwa Posted May 23, 2012 Posted May 23, 2012 (edited) L214C112 5GHz Air 1.35 4/8 Ln2 GD65 6580 1.77 Wprime 6340 32m test : need to set all timings manually or very bad performance if set auto, bad subs setter, nice memory clocker and i hate Control center, you need windows managment service enable to open CCII and need framework 4 on 2D OS ..... oh no i like tiny OS * Ln2 M5G 6610 1.77 Wprime 6390 32m test : Simply Rock but loves nice memories, with normal memory is a bit lazy board, Best memory Clocker and very stable board. Ln2 UD5X 6470 1.77 Cold problem Wprime 6390 32m test: Simply Rock like M5G, nice memory clocker and nice performance, Great tool for OS OC much better than asus and msi tool, on LN2 killed me with CBB. Cpu don't scale after 1.80 +2900 IMC 1.25 VTT on all Boards with samsung Edited May 23, 2012 by Hiwa Quote
Hiwa Posted May 23, 2012 Posted May 23, 2012 Clearly a fail chip Better sell it cheap :p i really don't care about +6600 i care about that +2900 IMC this chip can do 2800 with 1.00 VTT and 0.99 VCCSA lol Quote
I.nfraR.ed Posted May 23, 2012 Posted May 23, 2012 L212B228 #1700 3770K on M5G 5000MHz 2c/2t takes 1.4V to load XP and run 32M on water. 4700MHz 4c/8t 1.28 LinX + AVX stable Quote
I.M.O.G. Posted May 24, 2012 Posted May 24, 2012 Not sure if I posted earlier in this thread... Thought I did. Anyways, just figured out my MVG is coldbugged. Tested 3 chips in it so far. One chip was pretested to have no coldbug in another MVG, but coldbugged in mine around -110. The second chip coldbugged in this board at the same points, but ran full pot in another MVG immediately after that. Third chip also coldbugged in my board, and I resold it without testing in any other board. Replaced the board, used the same insulation method, and did the same thing to pull temps down... New MVG and same chip as before runs full pot without a problem. So confirmed I just had a bum MVG the first go around. Will be back with chip results. Quote
K404 Posted May 24, 2012 Posted May 24, 2012 Here's a question.... Who has taken a poor-average CPU and replaced the IHS TIM? What if *some* of the variation is down to IHS attachment difference? ....who has done that with a CPU that was already <1.40v for 5GHz? Quote
Splave Posted May 24, 2012 Posted May 24, 2012 ^ I will today max chip runs right now is gene volts maxed out (still scaling) 1.92v set = 1.94 and runs 6527 max superpi1m and pifast, max cpu-z is 6568mhz and wprimes nothing special. Max 06 cpu test 6275mhz. Quote
Splave Posted May 25, 2012 Posted May 25, 2012 woooooow PB4 and PD2 PCB codes from superbiiz top one needs 1.48v just to boot os x50 2c2t bottom one needs 1.51v just to boot os x50 2c25 FAIL! Quote
Pt1t Posted May 26, 2012 Posted May 26, 2012 L212B308 5Ghz 4/8 1.32V wprime 5Ghz 2 1.25V 32M boot 2933 with samsung ic's Quote
MaJ0r Posted May 26, 2012 Posted May 26, 2012 L212B308 5Ghz 4/8 1.32V wprime 5Ghz 2 1.25V 32M boot 2933 with samsung ic's promising chip ) Quote
MaJ0r Posted May 28, 2012 Posted May 28, 2012 6.6 max Hmm, my theory that good cpu at air=good LN2 clock is destroyed Do you have any coldboot/coldbug issues? Quote
uncle fester Posted May 28, 2012 Posted May 28, 2012 L215C040 weird Sheepeeuuuuu, bench wprime 32m at 6.390, 1024 @ 6.3 but struggle to clock to 6.6 nearly got 1m there, only clocks ok with 4C/8T. Scales with VTT voltage. looked very promising but seem very stupid lol Quote
K404 Posted May 28, 2012 Posted May 28, 2012 (edited) Well..... what if the silicon is able to do the MHz... but the BCLK or multi holds it back? If A CPU needs say....1.50v for 5GHz, it's unlikely to be a 6.5GHz+ CPU unless the volts are pushed mega hard. Maybe to some extent we should draw parallels between Ivy and Cedar Mill binning...... I remember now.... we were all hoping for a 125 BCLK strap because SB-E has it.... Edited May 28, 2012 by K404 Quote
Splave Posted May 28, 2012 Posted May 28, 2012 for me so far booting 5ghz 1.48v = 6.449mhz 1m/pf 1.51v = 6.287mhz cpu-z 1.48v = 6.312mhz cpu-z 1.51v = 6.285mhz cpu-z 1.33v = 6.527mhz 1m/pf 2 more on tuesday and still have to test 6.527mhz chip with ihs off on ln2 Quote
Hondacity Posted May 29, 2012 Posted May 29, 2012 (edited) bad batch: 3570k 5.5ghz max tested on two motherboards tim was tested ok coldbugys @ -110c WTF notes: "55" was cpu related. Its coldbugging outer memory channel. weird problem. both problems existed on two boards. Edited May 29, 2012 by Hondacity Quote
NoMS Posted May 29, 2012 Posted May 29, 2012 (edited) 3570K Batch: 3209B767 #1699 Runs Pi 32M @ 5GHz with 1.47V on Corsair H70. Will give a try with DICE tomorrow. Edited May 29, 2012 by NoMS Quote
Splave Posted May 30, 2012 Posted May 30, 2012 2 more for me, L206B383....but guess I have to test how low voltage will go at x16 lol damn! Quote
NoMS Posted May 31, 2012 Posted May 31, 2012 (edited) Tested my 3570K (3209B767 #1699) on DICE. 5700 1.7V valid 5500 1.68V 3dm06 5600 3dm03/05 The memory controller seems to be a POS on this chip... On H70 max I can get is 2200 no matter what I try, on DICE it easily go 2200+ with exactly the same settings lol Edited May 31, 2012 by NoMS Quote
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.