Thats a pitty,chances are it will need slightly more volts for 4ghz.:(Quote:
Originally Posted by felix_w
Printable View
Thats a pitty,chances are it will need slightly more volts for 4ghz.:(Quote:
Originally Posted by felix_w
its not to hard, because one my chip can 4 GHz stable with 1.325V only (air) and second with 1.365V
So far i've tried to see how high i can take my 8GB-8800 Pi Black kit...seems that all that time the problem was the IMC of my previous 940BE....
Running LinX Full Ram for 16.5x210 @ 1.4V...All other Voltages Auto, except RAM=1.9V and Turbo,CPU & PCIE Spread Spectrum Off...4 runs and moving strong...
Everest CPU temp is 41-42C at 100% load (Cores @ 31-32C), Coretemp is 41C with 9deg offset (trying to see if i can calibrate)...
Water Temp Before CPU is 25.3C (acc. to Phobya Sensor) and Ambient Temp is about 22C...
When i finish with LinX, i'll try CPU/CPU-NB clocks with 1.4V and then overnight Prime95 x64 v26.6.1 with what i think to be stable...
I hope to find 24/7 Prime Stable Max clocks till Friday-Saturday...
EDIT : I am having a difficulty getting 4G stable...at 1.45Vset in bios,i'm getting a big Vdrop under stress...1.413V is what the cpu gets while running prime...
System is inside my case and running 4 sticks of ram....these could be reasons for holding it back a little,aren΄t they?
You guys are pretty lucky. None of my Thubans (three to date) have been happy doing 4.0 at anything less than 1.475V. The 1100T that was given to me as a gift is the one that does 1.475. It acts mostly stable even at 4.1-4.2 GHz, but it's clearly not since random crashes are eliminated by decreasing frequency. The 1090T I used before it wasn't ever fully happy at 4.0. It would do 3.8-3.9 at 1.55V with total stability, but it couldn't do 4.0 regardless of voltage without periodically crashing. It "acted" stable-ish up to 4.1 GHz. The 1055T I played with originally seems to do 4004 at 1.55 most of the time, but it also has some issues. Anything less voltage wise and it bombs out fast.
wow that sounds pretty bad,
i got my 1055T stable at like 1.45v for 4ghz (i have the screenshot, never uploaded it to the overclocking sheet since i dont really use it at those speeds)
i wonder if you could be heat limited then?
I don't think that is the case. It's a relatively cool environment with a top performing heatsink and the best thermal paste around. (Tuniq Tower 120 Extreme + Shin Etsu X23-7783D) I'm more inclined to believe my chips just all fall on the poor side of normal, considering the graph. I also think a lot of people consider their system stable when it isn't. Ideally we'd have 48 hours of OCCT. I've seen a fair share of chips fail out after only 25-30 hours. It may not appear unstable at first, but it does cause crashes eventually during normal use which a user might attribute to something else due to the infrequency.
i had a setting thats stable forever on benchmarks
crashed when my friend was playing facebook flash games due to idle voltage being too low, but even that was passing every test i threw at it
some things in life just act weird
Sorry for the delay,i was all day out and stuck on a crappy traffic jam in the evening...right now i'm running 1hr so far at 20x200 @ 1.55V set at bios...all other settings at auto and ram at 800mhz...temp is 48C max....voltage at load is 1.507-1.510 as cpu-z shows...
tomorrow i'll leave it priming with 2x2GB Blades to see if 4 sticks is the issue or i got a banana cpu....
hi!Quote:
Originally Posted by I.nfraR.ed
how did you set the load line calibration on the ch iv formula?
what is the max safe voltage of the NB voltage (not the cpu-nb voltage).
thank you
ps
awesome ram kit :)
Must get Thuban!
Your GTX2 is just fabolous, prime stable @ 1850 6-6-5 1,70V How many sticks do you have?Quote:
Originally Posted by I.nfraR.ed
can I have a picture of your bios subtimmings?
Phenom II x6 1090T @4.05
LinX without error
http://image.ohozaa.com/i/7ea/38bus1800665picture2.png
- CPU Model: AMD Phenom II X6 1100T BE
- CPU Stepping: CCBBE CB 1105MPM
- CPU Frequency: 4050 MHz
- CPU vCore: 1.575V
- CPU Multiplier: 18.0x
- CPU Turbo: Disabled
- CPU NB Speed: 2925 MHz
- HT Ref Speed: 225 MHz
- RAM Speed: DDR2-1200
- RAM Timings: 5-5-5-18-2T
- RAM Configuration: 2 x 2GB
- RAM vDIMM: 1.90V
- Motherboard: ASUS M4A79 Deluxe
- Chipset/Socket: 790FX + SB750, AM2+
- Cooling: Water (D-Tek Fuzion V2)
- Temps: 31C Idle / 48C Load
- Operating System: Windows 7
- 32/64-Bit: 64
- Stable/Suicide/Untested: Stable
http://i256.photobucket.com/albums/h...able8Hr-th.png
Though Vcore is set to 1.575V i measured with DMM to be 1.545V during load...
I have one question though for fellow members....how is it possible to have passed 8+hr of prime blend and when running PCmark 7 system crashes with BSOD error #124 with the same settings ?
Updated
I'd lean on the RAM. Increase the Mem usage in Prime to max out the RAM and see how it does then.Quote:
Originally Posted by felix_w
Thank you charged...:up:Quote:
Originally Posted by charged3800z24
so prime 8hr stable settings is just a first step to stability, right ?
its near impossible for any one test to check full stability, just live life and if it never crashes, its stable enough for you. for me i just cant have it crash in WoW, which it tough since i played at someone else house, where i think the power lines fluctuated the voltage too much, since its perfectly stable at my home. so while it passes how ever many hours of however many tests i threw at it, i got different results at different places.
thanx dan...i just tried about 2-3 weeks with my spare time to geti this thing to stable (?) and you can imagine my face when system crashed while running transcoding test in PCMark 7....
with cpu frequency a nitch down, i can pass, but now i'm really trying to run every benchmark possible for many hours to see if it's an IMC thing or my ram...
i have a second pair of blades , so i'm now testing those...
i had my super stable pc crash due to facebook flash games that never pushed the cpu past 5% usage.
some things will never make sense, uping the Vid by another notch did fix it though
Motherboards tend to overcompensate for high loads. The flip side of course is that your idle state is receiving less voltage than your load state, so eventually your CPU causes a problem that results in a fatal exception. The scenario where you're fine under full load but crash at idle until you up vCore a smidge is a rather common story because of that.
with an asus motherboard i know that LLC has a major part in what actual voltage is that we almost have no clue of.
i actually have it off so that my idle voltage is high, and when load kicks in the voltage droops rather than overcompensates, so i know that in between idle/load it should be more stable as it changes pstates
in the end it was stable for everything at every pstate, except that damn flash game. it was my roommate who noticed it, since hes the one playing those lame things, and i upped the vid and it never happened again.
Even without LLC, that circuit isn't without any form of compensation. All regulated supplies attempt to compensate, just not all do it so aggressively.
Performed extensive sessions with memtest 4.20...
any combo of 4.0GHz and NB over 2600MHz with ram at 1150 or 1200 would cause errors ( test #5 move block) after 20-30 loops...
4.0Ghz & NB 2400MHz with mem 1200 5-5-5-18-30-2T (1.88V) would pass even 50-60 loops...
So i think that the IMC is the thing, not my ram...have 2 sets 1200 ram with Active Cooling, that do fine on other systems or this system @ lower cpu/nb clocks...
Quote:
Originally Posted by konlaos
wich memory is this? what a weird XMP, 3500Mhz?
Elpida Hypers in a 1333Mhz kingston? WTF?
@Barr3l Rid3r
KHX2000C8 (Elpida Hyper MGH-E)