yeah, for my board 284 was the highest i could go with all the other components at their lowest multiplier. Vcore was +200mv, so the final Vcore was 1.264 according to cpuz.

the main limiter with the 9150e is the locked multiplier. without headroom there you're limited by your max FSB, and the highest most can go doesnt need more than 1.3V

nice temp if its true...
25 is almost imposible for me...
now im idling at 28c..also quite weird...
u using what cooler what mobo sir?

u increase..i decrease..now run at 3.55ghz 1.26vcore (not yet stable) but the temp realy weird.. i wonder..
calling for another ta790gx user..pls...

Miahaha tabik spring if it's really only 25c

what board u using?nb volt been pump? the mobo volt... i wud love to see that chip run at 3.0ghz at least..huhuhuh

cloudwan..ur fotress easy to reach the proc? can u pls to some hand job at the proc ( the point part,that not covered by cooler)..or mebbe the cooler base..i cant sense any heat from there...

The casing is no prob but its the CCF and fan that's blocking me to do a hand job on my cpu to test the temp

i did touch all the heatpipes that very near to the base..and did touch the other side of my proc... also the cooler base -while IBTing
only the slight heat can be detected..deng... im worry bout it..

any 550be + ta790gx user here? pls help

x yh riso...just low vcore will do..

yeah... but if the cooler not well?mean if it sick..hehe
crap cooler..cant expect much...
and at aod..i can set the cpu and nb-cpu volt..but both will be exactly the same.. the slider will show same value after i apply..
haish..too much prob ar this board..sell off and get a gxe instead.

blend = 1sec then 1 worker fail
small = can run until i stop at 10mins
large= can run until i stopa at 7mins
IBT maximum 7loop = pass

what is the prob?cpu and ram can run atleast 5mins stable..but if blend then cannot..

here some ss..

it said cant utilize FFT code..what is that?


This post has been edited by kuntawakaw: Jul 16 2009, 12:12 AM

If I were to get a heatsink...

Xigmatek Dark Knight

vs

Core Contact Freezer

CCF has a little bit better performance but Dark Knight looks too sexay..

blend test is good for testing ur system stable..

everyone knows that...the thing that amuse me is torture run with 0 errors but stopped..and said FFT code cant utilize..
means the torture is halted before working... im

---

Added on July 16, 2009, 12:21 ameveryone knows that...the thing that amuse me is torture run with 0 errors but stopped..and said FFT code cant utilize..
means the torture is halted before working... im

This post has been edited by kuntawakaw: Jul 16 2009, 12:21 AM

Since blend cannot..There must be something wrong with your RAM and NB Frequency tuning....
For now..try tweaking the RAM and NB Freq..You gotta try..no matter what
If we dont try..we wont know..

the board is a DFI JR 790gx mr2s, as stated in my siggie. NB was at 1.3 at the time. to reach 3GHz with that proc you need 333 fsb, nothing less =x the multiplier maxes out at 9x.

Btw fellas, i've been priming using 4 instances running custom with 400mb of ram to each. does this sound sufficient to test for stability? i couldnt run all the instances with blend because it tends to use almost all the available ram, so unless you run 1 blend and the other 3 cpu only i'm guessing what i'm doing makes sense?

This post has been edited by strife_personified: Jul 16 2009, 03:59 AM

nope nope nope...i hope u read my post properly...



small >>>pass..proc stable
large>>> pass..ram stable

@bolded it stopped before it start...
and i just restart, and then primable... i donno for how long...wake up it failed..

and ts..i did mention few times edi.. i know my ram limit edi..and nb freq, i roughly know.. 3.6ghz can up to 27XXmhz..3.8ghz have to lower down to 25XX


im tired of burning in...
can i just make it 2hrs only? here some ss... my undervolt program..


This post has been edited by kuntawakaw: Jul 16 2009, 08:16 AM

thats makes me blur..the theory is Core speed increase..NB speed must increase too..but in ur situation is different..sometimes need to lower a bit..dunno which one is true..hard to OC in AMD..

i think... nb speed need to increase to able to coupe with the fast proc... so no bottle necking of nb occur..
but increase nb will sacrifice stability..so in order to stable at that vcore and nb volt, i lower down the nb freq... i did tried with 28XX at 3.6ghz..its cant even prime at 10mins..lower down abit, i can...

correct me if im wrong then

could you post your settings? i'm not sure how the options are in prime 95, but when you use stress prime (which is a prime 95 front end) and you run more than one instance of blend, it does indeed cause one of them to fail due to lack of ram (they both fight for the whole thing, i think).

Then..i said again..lower more ur NB Freq..maybe around 1800?Since it help reduces stability..We are finding max clock for CPU...later time we will manage the NB..

at default... just press blend..i guess the aod thing la...just installed aod last night

the figure i gave u earlier is the max i can obtain..the 3.8ghz @1.392 is stable with the nb freq at 19XX..so i guess the figure is chun edi..or i need to reduce a bit more?

it's ok to sacrifice some clock in order to maximize d bandwidth mate, give & take we must.

nice freq on d ram there, vdimm needed is?

biostar boards are known to be AOD unfriendly, i only use it for ACC adjust most of d time coz it's too unstable for even voltage adjustment. it's a fact that u gonna live w/ it as long as ur on biostar 790gx board. alternatively u can use biostar oc tool to tweak d volts...

blend test fail mostly related to mem, be it timing, vdimm or imc.

a fast google session tells me that it (d error ur gettin) could be page file related.

Source

xigmatek blowin up

ccf blowin towards exhaust fan

i use p95 & my habit is run custom, leave everything untouched (min @ 8k, max @ 4096) except amount of ram & time. amount of ram @ 1024MB (i'm on 4GB), time @ 1min (fastest way to complete 1 full cycle > near to 3 hours compared to 15Min which require almost 8 hours to complete 1 cycle). in case u dunno, p95 stresses systems more than orthos does.

u ain't need 28xx nb freq for 24/7 especially when no benchies are ran, a 2200MHz is more than enuf ledi...

This post has been edited by ah_khoo: Jul 16 2009, 09:44 AM