How high can you go on NB clocks with stock CPU-NB ? Do you have time to give it a try ?
Thanx
Printable View
If you mean stock volts for NB which are 1.15V then I've been able to run it stable at 2.6GHz
2.8GHz required 1.2V for full stability
3.0 GHz required 1.25V for full stability
Here is a post of mine with quite demanding chess app. running at these settings:
http://www.xtremesystems.org/forums/...&postcount=203
It probably can go to 3.1GHz with moderate voltages but I prefer to gain few more MHz on CPU cores instead.
All this using only AIR cooling (TT120Ultra + 1 FAN) enclosed in case and normal room temperatures (Ambient 18C).
Thanx, i thought that for high NB clocks stability, the voltage that made the difference was CPU-NB, not NB Voltage.
Compared to your previous cpu ( 940BE i think) did you see improvements ( NB stable clocks needing lower voltage than relevant 940BE's clocks ) ?
Charts updated.
I love the copy and paste action, folks. My chart maintainer program can auto-parse most of your guys' entries with very little manual conformity cleanup. :) Easy updates = responsive updates
I'm going to start requiring proof of stability in a standardized form via screen shot for future entries. While I don't think we have anyone trying to cheat the charts for the sake of notoriety, I do think charged has a point. It will help build credibility and also serve to regulate a standardized definition of "stable".
I updated my post, for some reason I put N/A for cpu multiplier, lol
And so it shall be reflected. :) Updated.
To those users who have posted results classified as stable, I request that you please run y-cruncher in stress test mode for eight hours and take a screen shot. I know that's kind of a pain, but in the world of stability testing, short-term tests just don't tell us much. This duration I believe to be a reasonable compromise between accuracy (a couple of days) and what we've got now (as little as 20 minutes). It's the kind of thing you can start and then go to bed. I'm requiring a standardized proof of validation going forward and would like to update the chart to reflect this new standard.
Edit: IBT, OCCT, Prime95, and LinX are also acceptable if 8+ hours but y-cruncher is still strongly preferred.
Hows failing Prime at stock with turbo disabled for stable:shrug:
1090T on Biostar TA785G3:down:
Crunches fine though:up:
Still waiting on DFI..........:slap:
CPU Model: AMD Phenom II X6 1055T
CPU Stepping:
CPU Frequency: 4018 MHz
CPU vCore: 1.425V
CPU Multiplier: 14x
CPU Turbo: Disabled
CPU NB Speed: 2870 MHz
HT Ref Speed: 287 MHz
RAM Speed: DDR3-1529
RAM Timings: 7-7-7-18-1T
RAM Configuration: 2 x 2GB
RAM vDIMM: 1.76V
Motherboard: ASUS M4A79T Deluxe
Chipset/Socket: 790FX + SB750, AM3
Cooling: Air (Noctua U12P)
Temps: 30C Idle / 51C Load
Operating System: Windows 7
32/64-Bit: 64
Stable/Suicide/Untested: Stable
http://img511.imageshack.us/img511/5241/stab.png
I have to be doing something wrong. You guys with the 1055t are all around 4ghz-3.8 with like 1.4-1.45 volts. even at 1.525 mine gives me errors in prime. I dont get it.
hmm. Errors on just one core?
This thread was looking promising.
8 hrs on a second tier stability test.
Ill pass.
You aren't alone, zaraza. Mine is the same way. It might go for 5 minutes or 12 hours but even at 1.525V mine can't muster 4004 MHz either despite being pretty close. I play BC2 on it all the time.
@Forsaken1: y-cruncher's stability test is more vicious than Prime95, but then again what isn't these days? If you don't want to run it, that's fine with me but don't knock the quality of the result. The results are of higher quality than requiring Prime95 or many of the other more well-known tests. That being said, I will accept others so long as they're eight hours. Even then that doesn't really prove stability but it's good enough to be a useful mark of reasonable likelihood.
Which test do you regard as superior/preferable?
Favor to ask:
Can one of you guys with a 1090T BE install BOINC,pick WCG and run the benchmark?
Would be helpfull if the chip was at 4000 for the test.
Thanks..
BOINC is here:6.10.18 for 64 bit OS
http://boinc.berkeley.edu/download.php
Yes, you're correct! CPU-NB voltage needs to be adjusted to gain stability above certain clock.
Compared to my previous CPU I see quite big improvement. My old CPU-NB could do 2.38GHz stable at stock volts (1.15V) where this one is super stable @2.6GHz (1.15V) and bench stable at 2.8GHz which suggest that the real max. stability lies somewhere around 2.7-2.75GHz stock volts.
That's 400MHz more. On top of that on 940BE I needed 1.35V to reach 2.9GHz NB compared to only 1.21V needed on new 1090T.
Of course!
http://img689.imageshack.us/img689/5840/boinc4ghzx6.png
@Particle
Do you think BOINC 8h is good enough? Because this week we all should be crunching and not wasting our CPU time for pure stress testing.
If you all think BOINC loads are too easy for CPU then I will run y-crunchera after May 8th.
Particle, did you buy yours from frys? I remember buying a 720 BE that was defective. It would crash in prime at stock speeds. Im gonna put my cpu to stock and see if it doesnt pass prime.
I think if it runs BFBC2 with no problems then its stable, maybe prime has problems with it? im just having wishful thinking. Im most likely not overclocking correctly.
Nice to know that im not alone. Im afraid to pump too much CPUNB volts. My errors in prime are only on 1 core. It can run fine for 2 hours, then it gives me 1 error on core 6 or 4 or 5. Its wierd.
anyway to enable the 14.5-16.5 multipliers on the 1055t... lol.. I see them in the bios and it lets me select them but ofcoarse it doesnt boot with it. Its like a super big tease!.
just a quick test
* CPU Model: AMD Phenom II X6 1055T
* CPU Stepping: CCBBE CB 1014GPMW
* CPU Frequency: 3976 MHz
* CPU vCore: 1.475V
* CPU Multiplier: 14x
* CPU Turbo: Disabled
* CPU NB Speed: 2556 MHz
* HT Ref Speed: 284 MHz
* RAM Speed: DDR3-1514
* RAM Timings: 8-8-8-22-1T
* RAM Configuration: 2 x 2GB
* RAM vDIMM: 1.65V
* Motherboard: ASUS M4A785TD-V EVO
* Chipset/Socket: 785GX + SB710, AM3
* Cooling: Air (Noctua NH-U12P)
* Temps: 21C Idle / 52C Load
* Operating System: Windows 7
* 32/64-Bit: 64
* Stable/Suicide/Untested: Stable (30 min linx)
power draw idle (c&q + c1e) 68W
load (linx) 340W
4GHz isn't stable atm :(
So what should if the cpu does not want to post above 3710MHz. What voltage should i raise. Only the CPU voltage is raised to 1.475v. does something else need to be raised or set somewhere so I can attempt to get this at 4.0 stable which is my only goal.
Since you have to use the HT Ref clock to increase core frequency which also raised the RAMs clock as well, where is your ram clocked? If it is above what it can do speed wise it may need to have the divider dropped or the voltage raised. This may be the reason it won't post. Of course NB and Ht Ref clock in general could be a factor as well. Can ypu post your overal settings?:)
I don't think that's a good metric to go by. In the case of WCG, you may not even know there's a problem for days until your returned WUs get marked as invalid. I know I had that happen at 4GHz. It might go all night doing WCG or BSOD after two hours. All the while it was producing invalid WUs.
This is really stupid!
I got fed up with my cpu not passing prime, so i upped the voltage to 1.575, while running prime cpuz showed up to 1.6v.
I still got the same damn error on the same damn core. I get it at 3.8 - 3.9 - 4.0 - 4.1 I have not tried 3.7, no matter what volts i give. Something has got to be wrong. This thing probably isnt even stable at stock speeds. :banana::banana::banana::banana: like this always happens to me lol.