these numbers have been marked as stable now
http://img832.imageshack.us/img832/3453/stableb.th.png
Printable View
these numbers have been marked as stable now
http://img832.imageshack.us/img832/3453/stableb.th.png
dont waste your time updating with my last results, due to the fact it was stable for 10 hours kinda made me pissed cause i knew i could get more out of it
so i found out 3780 isnt stable, and currently trying 3752, will have an update when i get off work tonight.
i honestly didnt expect this to be stable, no idea whats working considering a few weeks ago i couldnt get 3.4ghz stable at this voltage, lol.
Might be a bit higher then what it's showing maybe 10C higher then.
maybe the a lower ambient is helping ?
My cores max speed is 4.3ghz at 1.55 volts
and NB can hit 3.2ghzat 1.5 volts
too much of a heat limit for me to run both at speed.
well i think my temp sensor is way off, the bios showed like 18C idle, when the room is 22, and when in windows its showing 23C which might be right, or the same number as the bios just 5C hotter due to being in windows instead of a bios. so when its at 45C i think its possible the cpu is closer to 50C, which is a the touchiest point for these chips. and i think i remounted and got much better contact so when i use to dump on that extra heat, it wouldnt go anywhere, and now keeps it cooler and at lower volts (thuban prefers temp over volts if i had to summarize it for someone else)
i also still wonder what will happen if i raised it up a notch more on the volts, and dropped my HT (NB is stuck and thats whats screwing me over, i havnt been able to load windows at 270ish bus ever, nomatter how low everything else is)
I was looking awhile back for a thorough article on NB frequency effects, and I just stumbled upon one (that's quite old, actually, don't know how I missed it) that does a good job of it:
http://hwbot.org/article/news/hwbot_...rmance_scaling
They even reference back to Xtreme Systems here
And here was my own personal struggle for NB frequency.
Bottomline, there is no universal ratio where anything less bottlenecks ALL apps. But, generally, you want to strive for at least 2:1 NB-to-RAM ratio (or at the very very least, 1.7:1, eg. 2300NB:1333RAM). 1090T owners have a supreme advantage here, since the NB multiplier on 1055T's won't go past 10x, so the frugal ones like myself are held back by FSB limits. But it breaks my heart (well, not really, but for dramatic effect...) to see a 1090T at over 4GHz, RAM at over 1800... and the NB still sitting at 2000.
are you sure about that limit of 10x? then i wouldnt be able to adjust my NB at all, since my bios is very funny. i cant lower it from stock, and i can only raise it at the same rate as my HT link. basically im just short on adjustment options, lol.
also, what do people recommend for testing turbo stability? something that only uses 2-3 cores at a time, without me forcing core affinity.
On the 1055T? Yes, I'm sure. The only way to raise it is with FSB (which starts at 200, then 10 mult =2000 default)
Around the 2400 area (ie ~240fsb) you'll need to start adding "CPU-NB" voltage (not the one labelled "NB Voltage")
Regarding the HT Link: you won't see a benefit in performance beyond stock, but supposedly, for ultimate stability, it's best left in-sync with the NB freq (then again, for ultimate stability, AMD's current implementation of Turbo is best turned off... the proof is in the pudding: my out-of-sync settings are plenty stable), so if you can't modify the multipliers at all, don't sweat it.
So, at your current 257FSB, that means your NB is running at 2570... if you haven't already increased CPU-NB voltage, you need to do so. If you can't, then it's unfortunate, but you'll have to back way off on your OC. Feel free to flip your MB the bird.
As for the CPU frequency: you can generally get 3.6 to 3.7 on stock voltage, so you're fine there, but your turbo is probably giving you problems. Try setting the turbo multiplier to 15.5 for just-under 4GHz. Can't change that multiplier either? Well, then... feel free to flip your MB the bird again. (or just disable turbo... for the love of crap, please tell me your MB at least does THAT. If not, it's back to scaling your OC waaay back so that the Turbo freq is stable)
i hope your not talking about in AOD, that program is not interactive with my bios. i only have one voltage for NB in my bios, and i honestly do not know if it helps any, so just put it up to a decent value so its not stupidly dangerous
those are my old clocks too, my current is 266x14, and at 270x14 is unstable to where it locks up, rather than spiting out an error.
i can adjust my turbo multi, and i have that at 16 now instead of 16.5x so its right around 4.3 still. which has never really given me an issue, but i still havnt given it a good few hour test.
Here is mine for now.
*CPU Model: AMD Phenom II X6 1090T BE
*CPU Stepping: -
*CPU Frequency: 4219 MHz
*CPU vCore: 1.392V
*CPU Multiplier: 21x
*CPU Turbo: Disabled
*CPU NB Speed: 3000 MHz
*HT Ref Speed: 2000 MHz
*RAM Speed: DDR3-1600
*RAM Timings: 8-8-8-21-2T
*RAM Configuration: 4 GB (2 x 2 GB)
*RAM vDIMM: 1.65V
*Motherboard: Gigabyte 890FXA-UD5
*Chipset/Socket: AM3, 890FX + SB850
*Cooling: Air (Hyper 212+)
*Temps: 17C Idle / 65C Load
*Operating System: Windows 7
*32/64-Bit: 64
*Stable/Suicide/Untested: Untested
A final question. I must run LinX for 8 hours at least for be added to the stable table rig? I ask this because maybe in the other 24 pages the requirements has change. Thanks for the reply.
saint 19, looks good, only with 1.4V? Show more for us, maybe will great chip (1.48V and 4300 MHZ +???)
Here are some little test that I did.
http://img243.imageshack.us/img243/1128/pifast.th.jpg
http://img209.imageshack.us/img209/2...perpi1m.th.jpg
http://img837.imageshack.us/img837/1...rime32m.th.jpg
http://img221.imageshack.us/img221/2...me1024m.th.jpg
I was thinking the same about 4300MHz I will try it this weekend I hope don't kill the chip :rofl:
think not, 1.48V is safe for 24/7
So FlanK3r and Manicdan what voltage and temp do u thing that is safe? With my 955 C2 hit the 70ºC without problems.
turbo adds like .2v to whatever your settings are. so stock it gives it about 1.5v
check the very first page of this thread for what kind of clocks and cooling people have used for 8 hours of stress testing.
i would say 1.55v is probably the highest you can go for 24/7 and not get any degradation. 1.6 should be fine if you have water
and temps on thuban seem to be better lower in any case. as ive found out, any voltage you want, probably wont help if your over 60C
just read like everything from the very first post and it will give you lots of ideas for what is possible. (temps, clocks, volts, cooling methods, motherboards, max NB/HT, etc)
for Deneb up 1.5-1.55V (if your temps OK, means about 55-58C only) and about 1.45-1.488 for Thuban
i had come across a weird conclusion this past weekend.
my clocks were not being stable at 3.71ghz @1.368v. the temps would hit 42C or so which caused my fans to speed up, after a while it would stable out around 44-45C, but never more. but i got constant errors in y-curncher
so i dropped the voltage down to 1.346, and guess what, the temps never broke 40C (which is when the fans start to speed up), and after an hour or so i was still rock solid and temps still under 40C. its crazy to think that going down in voltage helped. but now my rig is completely silent, and seems to be very stable. i wont have time for the full 8 hours until wed though.
so my tip to people out there, max out the clocks on the lowest voltages first, and increase them later since its a very slipper slope once temps get just a hair too high.
Ok guys 4.3GHz @ 1.48V.
http://valid.canardpc.com/cache/banner/1468598.png
You can see more pics here
but what about some LINX or PRIME???
I will do this weekend, I'm waiting my C300 SSD to get better results. ;)
Quite tempted to grab a newer batch 1090t, now a days the majority seems to do really well with low voltage :)
Hi guys.
If I run LinX 8+ I must run y-cruncher too, or with LinX is enough. What are the advantages of y-cruncher?
1023GPMW | 24 degrees ambient
http://b.imagehost.org/t/0145/4_1GHz.jpg
I'm not finished with it, but it seems that it's weaker than my 1010MPMW. The real voltage was steady all the time (1389 - 1391mV).
At least it runs cooler. The fifth core is the weakest one, others don't need this vCore.