"ASUS P5Q Deluxe discussion".

[sxs112]

ASUS P5Q Deluxe 1611 Beta Bios
[ 1611 ]
----------------------
1. Fix chassis fan mode can't be saved after resume from S3.
2. Fix display priority is not correct when plugs two PCIE VGA cards.

[Leeghoofd]

Same changelog as P5Q3 Dlx SXS112 ?

[sxs112]

Same changelog as P5Q3 Dlx SXS112 ?

ASUS Now no P5Q Deluxe 1611 Changelog

[x3mgamer]

1) use realtemp or coretemp to properly readout the values ( verify the corrrect TJcasemax : Everest is probably reading 5°C too much )
2) you claim stability based on what ? if the CPU is stable at small FFT ( at least 8 hours) I can hardly imagine you need to up Vcore to 1.225 to get blend stablity
3) You need to take these templates as a guideline, final tweaking is for you, if you are 100% sure 1.1375 made the CPU stable before, why not try 1.21Vcore and properly adjust PLL and FSB Term voltages...and try go down...
Each CPU, mobo,ram, PSU combination is different and even though a template can work fine, it doesn't mean you cannot go lower on certain voltages...
Blend stability is also over a long period of time, you claim stability again based on what ? a 60 min run ?
4) Which cooler do you use on ya setup ?

1. ya everest is always reading too much, realtemp and coretemp r more like it

2. another test failed after a few minutes so no stability yet

3. if i lower vcore i shouldnt worry about other values right? cpu voltage isnt influencing other values, or am i mistaken, other values being fsb term, pll, gtl references, nb, sb, pcie sata voltages?

4. OCZ Vendetta 2

[Martin.v.r]

edit
bios 1611
no vdrop,but look here
vcore go up,by load

[CryptiK]

I'm not sure what you are trying to show there?

[Martin.v.r]

vcorego up,by load

[CryptiK]

Ah ok, my P5Q-D did that with all BIOS versions -> 0.008v increase under load.

[x3mgamer]

about cpu gtl reference voltages:


think of it more as a signaling threshold rather then actual voltage applied to the CPU. Although the data that is transmitted across the FSB is digital, the signal response isn't completely on and off, there is some transition time between there.

http://www.thetechrepository.com/att...1&d=1169236522

In the figures from the post here, think of Vtt and Vss as on and off. The GTLREF voltage is the line that the signal has to cross before the signal is seen to have changed from on to off or vice versa.

http://www.thetechrepository.com/att...1&d=1169078346

As the speed of the FSB scales, the response to changes from on to off can change, resulting in a need to adjust the point that the signal is seen to change up or down. By doing this, you can place the reference point somewhere that it wont be freaked out by the noise in the signal as it changes.

In summary, you aren't actually adjusting the voltage that is provided to your CPU, you are just tweaking the point that it will see as a change in the signal the FSB is sending it to try and squeeze out a bit more stability at higher FSB speeds.

[Martin.v.r]

not here,best bios ever 1306,but asus has It no more

Now, I suffered continuing vocational test with bios 1611

[CryptiK]

Don't you have it saved somewhere? I still have a copy of 1306 if you want I can email it to you. PM me your email address if you want me to send you it.

[Leeghoofd]

about cpu gtl reference voltages:


think of it more as a signaling threshold rather then actual voltage applied to the CPU. ....it will see as a change in the signal the FSB is sending it to try and squeeze out a bit more stability at higher FSB speeds.

I think it's best to start from scratch here and try to determine if we can get it stable even at just stock settings... if you want to do it ofcourse. But trying templates, changing voltages on the fly will probably not give you the result you want...

Best bios to use is 0803 or 1406 (haven't tried 1611 here)

First pull one set of dimms... and use 4GB only... reset bios to defaults, reboot and set just the Vdimm to a correct value (I would set 0.08 lower than required eg 2.02 for 2.1Vdimm ops), ram timings manually 5-5-5-15 and TRFC 55 and NB to 1.2

Then boot into windows and launch Prime and run Small FFT test : (make sure you got CPU-Z open to verify Vcore) and Real temp eg to monitor coretemps... if the CPU passes eg 1 hour without an issue (no cores capping out), reboot the rig and set Vcore yourself to a value lower than monitored before in CPU-Z...test again till one of the cores fails, then give her some more juice... then test the rams with Hyperpi, HCI Memtest,...

if all is stable swap the current 4GB kit with the other 2 dimms... retest ram stability again with that kit (Hyperpi 4 instances 32MB, HCI memtest etc...)

If that is stable report back with ya working Vcore setting etc... then we can see if we can get the 4 sticks stable...

I think it's better to do it like this because what works for us, might not work for you... there are enough kind peeps here that will guide you through it mate

[x3mgamer]

will do sir im on it (bios 1406)

[x3mgamer]

I think it's best to start from scratch here and try to determine if we can get it stable even at just stock settings... if you want to do it ofcourse. But trying templates, changing voltages on the fly will probably not give you the result you want...

Best bios to use is 0803 or 1406 (haven't tried 1611 here)

First pull one set of dimms... and use 4GB only... reset bios to defaults, reboot and set just the Vdimm to a correct value (I would set 0.08 lower than required eg 2.02 for 2.1Vdimm ops), ram timings manually 5-5-5-15 and TRFC 55 and NB to 1.2

Then boot into windows and launch Prime and run Small FFT test : (make sure you got CPU-Z open to verify Vcore) and Real temp eg to monitor coretemps... if the CPU passes eg 1 hour without an issue (no cores capping out), reboot the rig and set Vcore yourself to a value lower than monitored before in CPU-Z...test again till one of the cores fails, then give her some more juice... then test the rams with Hyperpi, HCI Memtest,...

if all is stable swap the current 4GB kit with the other 2 dimms... retest ram stability again with that kit (Hyperpi 4 instances 32MB, HCI memtest etc...)

If that is stable report back with ya working Vcore setting etc... then we can see if we can get the 4 sticks stable...

I think it's better to do it like this because what works for us, might not work for you... there are enough kind peeps here that will guide you through it mate

ok i pulled one set of dimms so now only slots 1 & 3 r populated, i disabled marvel ide, express gate , starting logo, firewire and all that bs i dont need

ive set everything exactly as u said (TRFC being row refresh cycle time? cant see what else could it be)

all is idle and cpuz is reporting vcore oscilating between 1.104,1.112,1.120,1.136

core speed is also reported being 2003 MHz coz of C1 and speedstep being enabled

realtemp 2.70 is showing minimums of 39 36 31 31

starting prime 95 small fft............................................... ..........................

.......

4 cores 100 % load each, cpu-z is reporting constant vcore of 1.136

realtemp is showing maximums of 54 52 53 53 after 5 minutes of small fft

to be continued...


small update after 40 mins prime cpu-z is showing vcore 1.144 and realtemp maximums r 55 54 53 53

[x3mgamer]

1 hour prime finished without errors.......reboot.......vcore set to 1.12500 being the lower value than 1.136 first reported by cpu-z under load

idle:

cpu-z vcore same oscilations, realtemp minimums 38 34 30 30

load fft:

cpu-z vcore 1.120, realtemp maximums (after 5 mins) 53 49 51 51, this test will keep on going for approx. 9 hours more till i come home from work, if it doesnt fail that is

[Leeghoofd]

Is there any real benefit to enabling the Static Read Control?

Good increase in ram bandwith and drop on latency, though most users disable it as it's one of the main causes of instabilitty

What's the final verdict on the operation/options for CPU Margin Enhancement??

For Dual cores leave on optimised, for 45nm quad cores try performance or comaptible

The lower FSB Strap to North Bridge performs better on other boards, is there a specific reason most of you are leaving it AUTO on the P5Q ?

If you are not seeking for max performance leave on auto if you want max performance manual selection is required and good NB voltage/GTL tuning, hence why most leave it on auto...

I keep seeing a recommendation of 55 for the Row Refresh Cycle Time, again is there a specific reason to leave it loose on the P5Q?

55 is recommended for 2GB dimms, so in your case you can tighten it... most 1gb dimms run happily at 42 and lower

1 hour prime finished without errors.......reboot.......vcore set to 1.12500
idle: cpu-z vcore same oscilations, realtemp minimums 38 34 30 30

load fft: cpu-z vcore 1.120, realtemp maximums (after 5 mins) 53 49 51 51, this test will keep on going for approx. 9 hours more till i come home from work, if it doesnt fail that is

K Good start then, temps are already pretty high for just 2.8Ghz ops (is that Tjcasemax 100 set ?) I already would start to think to buy a cooler with a decent mounting (meaning backplate with screws) Those pushpins are really the most crappy invention ever, as they severely bend the mobo, making the heatsinks loose contact with the mosfets etc...

If it surivives the small FFT you can try a blend run then... if that passes, swap the dimms for the other pair and don't change a thing...

[sxs112]

ASUS P5Q 1611 Beta Bios
[ 1611 ]
----------------------
1. Fix chassis fan mode can't be saved after resume from S3.
2. Fix display priority is not correct when plugs two PCIE VGA cards.

ASUS P5Q Pro 1611 Beta Bios
[ 1611 ]
----------------------
1. Fix chassis fan mode can't be saved after resume from S3.
2. Fix display priority is not correct when plugs two PCIE VGA cards.

[pangingIII]

why asus removes all the rest of final/release version of BIOS's (for P5Q-D) from their ftp servers but the 1406 version, do you happen to know, sxs112?


P5Q-ASUS-DELUXE-1611.zip
ftp://ftp.asus.com.tw/pub/ASUS/mb/socket775/P5Q_Deluxe/

[Leeghoofd]

Probably that for most users 1406 is the better bios ? probably another imaging error like we had already a zillion times with Asus

[x3mgamer]

K Good start then, temps are already pretty high for just 2.8Ghz ops (is that Tjcasemax 100 set ?) I already would start to think to buy a cooler with a decent mounting (meaning backplate with screws) Those pushpins are really the most crappy invention ever, as they severely bend the mobo, making the heatsinks loose contact with the mosfets etc...

If it surivives the small FFT you can try a blend run then... if that passes, swap the dimms for the other pair and don't change a thing...

wow 8 hours 48 mins prime and no errors

minimums 38 34 30 30

maximums 53 53 51 51

starting blend................... ( i could probably lower vcore further and further and test with fft until cores start failling but whats the point right?)

[x3mgamer]

ok first core failed after 8 mins, second 15 etc.


Leeghoofd heeeeeeeelp

[Leeghoofd]

K did it fail in test for the ram, so big numbers (1024K) or was it already busy with eg 8K ?? I have a distinct feeling the mobo doesn't like your rams at all... swap them for the other kit ( so still max 4Gb ) don't change a thing in the bios...and retest blend plz or run an OCCT Mix or even LinX or Intel Burn test program (eg IntelBurnTest.zip )

So you got 333FSB and 1000mhz rams ? or are they at 1066Mhz ( read out via CPU-Z plz ) Plz fill in the ram timings detected by the bios ( via AUTO setting) in the template grab here ) P5Q Dlx template

You are on which bios now ?

[x3mgamer]

first core failed after 2 mins (1024k) fatal error: rounding was 05, expected less than 0.4, same thing with all other cores, im on 1406 bios right now,

cpu-z timing readout:

dram freq: 534.3
fsb:dram : 5:8
cas 5
ras to cas 5
ras precharge 5
cycle time 15
command rate 2T

swapping out with other kit..............


system wouldnt boot if slot 1 wasnt populated, i tried booting with slots 2,4 populated, no go


............i have a bad feeling about this, blend is priming for 15 mins now without errors, what happens if its those first 2 sticks that r making rig unstable? should i rma a kit coz of rounding errors in blend?

[Big Lar]

first core failed after 2 mins (1024k) fatal error: rounding was 05, expected less than 0.4, same thing with all other cores, im on 1406 bios right now,

cpu-z timing readout:

dram freq: 534.3
fsb:dram : 5:8
cas 5
ras to cas 5
ras precharge 5
cycle time 15
command rate 2T

swapping out with other kit..............


system wouldnt boot if slot 1 wasnt populated, i tried booting with slots 2,4 populated, no go


............i have a bad feeling about this, blend is priming for 15 mins now without errors, what happens if its those first 2 sticks that r making rig unstable? should i rma a kit coz of rounding errors in blend?

I have not been following the scenario with your setup, however, have you tried memtest 3.8 in windows with the other ram running on that board yet? Sorry if you allready did this, but if you have not, it will tell the tale rather quickly IMHO.

Larry

[x3mgamer]

Hi Larry, i havent done that yet, what do u have in mind exactly? btw with these 2 sticks blend has been going on for 36 mins now without errors