Foxconn BloodRage Issues / BIOS Bugs Thread

[OverClocker_gr]

Found a bug at G15.

Disabling jmicron 363 gives me instant reboots after raid initialization,JUST before Windows boot.

[saaya]

Found a bug at G15.

Disabling jmicron 363 gives me instant reboots after raid initialization,JUST before Windows boot.

hmmm thanks for the feedback!
what hardware?
with G13 the same config works fine with the same settings?

[BenchZowner]

I'll try to replicate Overclocker's issue later on and see if I get it too

[OverClocker_gr]

hmmm thanks for the feedback!
what hardware?
with G13 the same config works fine with the same settings?

sorry i forgot it

CPU: i7 920
Ram: 3*1GB Patriot PVS32G1866LLK
VGA: 2*4870 @ Crossfire
HDD: 2* 150GB RaptorX @ Raid 0
PSU: Chieftec 850w(CWT)

Yes,same config at G13/P03 works ok.(except the tcl issue that on G15 solved )

[Raja@ASUS]

Is Turbo mode working for you guys without PPM enabled on G15?

I have it enabled and just booted into Windows, but did not notice the multi ramping to 21x under full load.

later
Raja

[BenchZowner]

Is Turbo mode working for you guys without PPM enabled on G15?

I have it enabled and just booted into Windows, but did not notice the multi ramping to 21x under full load.

later
Raja

For the time being at least ( I'm not sure if it's possible ) you must have EIST enabled, else Turbo Mode won't function.
To enable EIST you need to enable the PPM, so...

[Raja@ASUS]

Thanks Bill, sounds about right. I noticed it being avaialble on G15 without the PPM function and thought they'd managed to isolate it.

regards
Raja

[BenchZowner]

Found a bug at G15.

Disabling jmicron 363 gives me instant reboots after raid initialization,JUST before Windows boot.

Just flashed my board with the G15 BIOS man and it looks to be working and booting fine with the JMicron controller disabled.

Give your board a re-flash with the latest awdflash 8.83 with the following command:

awdflash 8A1F1G15.BIN /py /sn /Wb /cd /cc /cp /ch

And try again

Thanks Bill, sounds about right. I noticed it being avaialble on G15 without the PPM function and thought they'd managed to isolate it.

regards
Raja

I'm gonna try it with G15 now since the "Turbo Mode" option doesn't appear to be disabled ( as an option at least ) when I disable PPM.
Gonna report my findings in 5 minutes or so

[BenchZowner]

Confirmed.

Disabling the PPM disables the Turbo function no matter what you do

[BenchZowner]

Damn, it looks like G15 isn't the overclocker's BIOS.
It looks like there is at least 1 bug that f***s up my OCs.
It seems to be ignoring or not mapping the right UnCore and/or memory multipliers.
Anybody else having that ?

For example I tried 200 BCLK, 20x CPU Ratio, 16x UnCore, 8x RAM, 18x QPI, saved and exit, and then no POST... beep beep beep ( no it's not the roadrunner ).
After the overclock recovery power on & off, I got into the BIOS and found that the board decided to use or just show me 20x UnCore 12x RAM 18x QPI.
It also looks like it doesn't apply the QPI Voltage, as I found it on the "Default" setting twice already.

[Raja@ASUS]

(for me) On the G15, only way to get consistent post around 195 BCLK is to use an Uncore no higher than +140mv ~ +180 mv. Not sure if that's PSU related again, but anything higher in Uncore voltage will throw a C1 or non post from reset/cold boot.

Once it boots you can increase Uncore up a bit, but not by much all the time. I've made it up a higher a few times, but find the board will go into overclock recovery on some reboots, so I peg back and start again.


This is all on the 2:8 divider btw, overall this BIOS is a little less stable than G13 for memory clocking.


regards
Raja

[BenchZowner]

Just noticed something on BIOS G13.
In the CPU Voltage (Vcore) setting, in the description box on the right it says "Default value is 1.260V", on the other side the "Target CPU Voltage" field shows 1.160V with the Vcore offset ( setting ) at "Default"
And the trend goes on with any increase.
I checked the Hardware Monitoring page and saw 1.6V reported while the target CPU Voltage in the settings page was 1.5V
I'm gonna rotate the board to have easy access to the measurement points to check it out.

If the HW Monitoring page is correct, then I was running CineBench 10 x64 at 4.3GHz @ 1.6Vcore on a Ultra-120
Poor i920

[Raja@ASUS]

On my board, full processor load vcore usually is within 0.02V of the predicted value in the voltages page. (without loadline calibration enabled)

[Raja@ASUS]

Uncore/VTT appears to be 0.1V under applied voltage, could explain why I have not clocked my memory so high yet ..lol

Running 200 BCLK X20 Prime 95 - 8 thread now. I take back what I thought about this BIOS..lol

later
Raja

Edit: I should add this needs to be fixed pretty fast, the disparity between Vdimm and VTT can be way out if you select XMP. Just bear in mind that you need to measure VTT on your board, rather than take Hardware monitor as gospel.

[Raja@ASUS]

quick update - did not think this was going to happen. needs 1.52VTT (real), but so do many of the other boards up here on 2:10 @ 1T..

Anyway, some polish on the things pointed out in this thread and we're starting to look like we have a more universal contender on our hands...




Need to nail Vcore now for the small fft's but progress nevertheless.

I think this BIOS should be fixed for real VTT voltages before going out wild though and anything Bill uncovers his end as a discrepancy.

later raja

[Raja@ASUS]

Can't get consistent reboot at 200 BCLK. 195MHz is still the limit with the reduced VTT condition. There's definitely something wrong about the way G15 applies voltages after a reset or save and exit from BIOS.

[OverClocker_gr]

Just flashed my board with the G15 BIOS man and it looks to be working and booting fine with the JMicron controller disabled.

Give your board a re-flash with the latest awdflash 8.83 with the following command:

awdflash 8A1F1G15.BIN /py /sn /Wb /cd /cc /cp /ch

And try again

yes i did all that from the first time.
Maybe its the raptors that i have on raid and you dont.

[OverClocker_gr]

and i found and another issue.
I cant set manually the tcl.If i set it manual 7,8,9,10 doesn't matter it ll not post.

so,i continue to have this issue,but now only at cas 8/9.
yesterday when i flashed G15 i only used cas 7.Today i wanted to do some tests and there is no go.

Cas 8/9(cas 7 is ok) gives me no post

[saaya]

sorry i forgot it

CPU: i7 920
Ram: 3*1GB Patriot PVS32G1866LLK
VGA: 2*4870 @ Crossfire
HDD: 2* 150GB RaptorX @ Raid 0
PSU: Chieftec 850w(CWT)

Yes,same config at G13/P03 works ok.(except the tcl issue that on G15 solved )

Did you use the XMP? if yes, then that could be why you couldnt change tcl on G13 and P13 (they are not the same BIOS btw, but very similar)

so you run SATA raid0, and when you disable the PATA controller you cant boot into windows anymore? weird... are you using the ICH10 or the Marvell SAS controller to run raid?

Galerius, i posted a reply in the other thread
this is more of a shipping/newegg rma issue, not really a board issue, and def not a bios issue

Thanks Bill, sounds about right. I noticed it being avaialble on G15 without the PPM function and thought they'd managed to isolate it.

regards
Raja

i dont think thats possible, but can check.
But why would you want PPM disabled? you suspect it has to do with the coldbug right?

Just flashed my board with the G15 BIOS man and it looks to be working and booting fine with the JMicron controller disabled.

your running raid0 too? on the ich10 or on the marvell sas controller? maybe that has to do with it?

I'm gonna try it with G15 now since the "Turbo Mode" option doesn't appear to be disabled ( as an option at least ) when I disable PPM.
Gonna report my findings in 5 minutes or so

hm interesting, ill check with the bios engineers.

I tried 200 BCLK, 20x CPU Ratio, 16x UnCore, 8x RAM, 18x QPI, saved and exit, and then no POST... beep beep beep ( no it's not the roadrunner ).
After the overclock recovery power on & off, I got into the BIOS and found that the board decided to use or just show me 20x UnCore 12x RAM 18x QPI.
It also looks like it doesn't apply the QPI Voltage, as I found it on the "Default" setting twice already.

so optimized defaults, then set 200bclock, 20x multi, 16 uncore, 8 mem and 18 qpi results in no boot? well for that im not surprised, since you need higher volts as well im sure you increased the volts as well, could you please tell me the exact settings you apply to get the no boot problem?
what does the debug LED show? dd?

Just noticed something on BIOS G13.
In the CPU Voltage (Vcore) setting, in the description box on the right it says "Default value is 1.260V", on the other side the "Target CPU Voltage" field shows 1.160V with the Vcore offset ( setting ) at "Default"
And the trend goes on with any increase.
I checked the Hardware Monitoring page and saw 1.6V reported while the target CPU Voltage in the settings page was 1.5V
I'm gonna rotate the board to have easy access to the measurement points to check it out.

If the HW Monitoring page is correct, then I was running CineBench 10 x64 at 4.3GHz @ 1.6Vcore on a Ultra-120
Poor i920

core i7 def likes watercooling
is it fixed in G15 or still the same?
can you meassure vcore on the voltage pads with a DMM and tell me what readings you get?
does this maybe have to do with ppm beeing disabled? do you get the same offset of vcore setting and reading in bios with g15 and when loading optimized defaults and only changing vcore settings and nothing else?

so,i continue to have this issue,but now only at cas 8/9.
yesterday when i flashed G15 i only used cas 7.Today i wanted to do some tests and there is no go.

Cas 8/9(cas 7 is ok) gives me no post

are you using xmp? what speed are you running the memory at?
what vdimm and what vtt? and the only thing you change is cas 7 to cas8 or cas9 and the result is no post? what does the debug led show?
could you create an spd dump with thaiphoon and post it here please?
thanks!

so there seems to be some problem with vtt beeing too low and not applying when set too high in one step
a problem with manual memory timings on the Patriot PVS32G1866LLK kit
a problem with the jmicron disable setting
a problem with vcore?

thanks for all the feedback guys!
ill try to reproduce it all and then show it to the engineers here and get them to fix it asap

for some things like the raid issue and tcl problem and vcore problem i need more details before i can even try to reproduce it, as soon as i get the details ill reproduce it to get it fixed.

PS: can anybody confirm that VTT is off and by how much by reading it in bios and with a DMM on the voltage pads?
thanks!

[OverClocker_gr]

Did you use the XMP? if yes, then that could be why you couldnt change tcl on G13 and P13 (they are not the same BIOS btw, but very similar)

so you run SATA raid0, and when you disable the PATA controller you cant boot into windows anymore? weird... are you using the ICH10 or the Marvell SAS controller to run raid?

No,didn't use.As i said few posts before i rip them off(xmp profiles fro spd) but i was getting the same issue.
i run sata raid 0 on ich10.Yes,when i disable jmicron just before windows boot screen appears, i am getting instant reboots.

are you using xmp? what speed are you running the memory at?
what vdimm and what vtt? and the only thing you change is cas 7 to cas8 or cas9 and the result is no post? what does the debug led show?
could you create an spd dump with thaiphoon and post it here please?
thanks!

no.1600(200*20,8x mem,16x uncore)
it doesnt matter how much vdimm i ll set.1.8-2.0v.
vtt 1.3-1.55
Exactly.See this video about the debug leds caus its kinda weird.
http://rs581.rapidshare.com/files/18...6012009005.rar


In a few minutes spd is coming

[Raja@ASUS]

Hi Sascha,

VTT measurement disparity between the sensor and the measurement pad is around 0.04V at stock voltages (around 1.14V real, 1.18V BIOS reported).

FWIW as VTT draw high current (drawing upto 50 amps and beyond) VTT measurements taken from various points on the power plane (various capacitor legs or the inductor) are not indicative of what the CPU really sees.

If the measurement pad is fed from a trace under the CPU socket, that would make it the most accurate measurement point on the board. It's what the processor really sees that is the important value here.

The Hardware Monitor BIOS value is taken close to the inductor it seems - so is not really what the CPU is seeing.


With regards to the gap between Hardware Monitor and the measurement pad, well, the gap widens the more you increase VTT (makes sense as the current draw across the power plane causes more sag the higher you go).

Below 1.2VTT, the gap is 0.04V (pad is 0.4V lower than reported)
Around 1.3-1.4V VTT the gap is 0.06V (pad is 0.6V lower than reported)
Between 1.45v and 1.6V the gap is 0.1V (pad is 0.1V lower than reported)

Based upon the performance figures and overclocking results I'm getting with this CPU - they correlate exactly with the pad measurement for how it scales per Uncore/VTT volt on other boards.

With regards to the all in one step VTT/BCLK boootup issue, yes anything over 1.34VTT (measured) and my board goes straight to C1 at 195 BCLK.
This is using the OCZ Pro X stream PSU. Can anyone else confirm this behaviour please with their working PSU's?


regards
Raja

[saaya]

No,didn't use.As i said few posts before i rip them off(xmp profiles fro spd) but i was getting the same issue.
i run sata raid 0 on ich10.Yes,when i disable jmicron just before windows boot screen appears, i am getting instant reboots.

ok, ill try to reproduce this... weird..

no.1600(200*20,8x mem,16x uncore)
it doesnt matter how much vdimm i ll set.1.8-2.0v.
vtt 1.3-1.55
Exactly.See this video about the debug leds caus its kinda weird.
http://rs581.rapidshare.com/files/18...6012009005.rar

In a few minutes spd is coming

the video doesnt work... the dl always stops immediatly and the file is only 1.1kb in size... maybe its blocked somehow here at work, could you check and make sure the dl works for you?
i guess the debug leds are just flashing quickly and you cant read anything?
or is it cycling from one code to another in an endless loop?
what bios is this with? g13?
With G15 i cant run Bclock above 166 with 16x uncore on my 920... seems like a bug, as soon as i set uncore to 18x it works fine.

Hi Sascha,

VTT measurement disparity between the sensor and the measurement pad is around 0.04V at stock voltages (around 1.14V real, 1.18V BIOS reported).

yes, i can reproduce this, for me its 1.14v at the pad, 1.16v at the pwm capacitor legs and 1.16 in BIOS. The reason why vtt is not 1.20v is cause the cpu is loaded (bios load) which triggers vtt vdroop. I already asked for a vtt vdroop mod and will post it here as soon as i get it

FWIW as VTT draw high current (drawing upto 50 amps and beyond) VTT measurements taken from various points on the power plane (various capacitor legs or the inductor) are not indicative of what the CPU really sees.

hmmm at stock im getting exactly the same reading on both vtt capacitor legs i meassured. are you sure its not possible to read real vtt properly on the capacitor legs? the cpu cores draw a lot more current and vcore culd always be meassured reliably on the capacitor legs so far...

If the measurement pad is fed from a trace under the CPU socket, that would make it the most accurate measurement point on the board. It's what the processor really sees that is the important value here.The Hardware Monitor BIOS value is taken close to the inductor it seems - so is not really what the CPU is seeing.

i dont know where it reads vtt, ill check

With regards to the gap between Hardware Monitor and the measurement pad, well, the gap widens the more you increase VTT (makes sense as the current draw across the power plane causes more sag the higher you go).

nono, the difference is caused by vdroop
the higher the voltage the higher vdroop will be... thats the intel spec... but im asking the engineers here to please add a bios options to disable vdroop for vtt or at least get me a hardware mod how to disable vtt vdroop

Below 1.2VTT, the gap is 0.04V (pad is 0.4V lower than reported)
Around 1.3-1.4V VTT the gap is 0.06V (pad is 0.6V lower than reported)
Between 1.45v and 1.6V the gap is 0.1V (pad is 0.1V lower than reported)

you mean the difference between the bios reading and the vtt pad reading is increasing? hmmmm ill check...

Based upon the performance figures and overclocking results I'm getting with this CPU - they correlate exactly with the pad measurement for how it scales per Uncore/VTT volt on other boards.

yeah, the bios reading must be off... probabaly reading it at the pwm before vdroop is applied or something? no idea... ill check with the engineers

With regards to the all in one step VTT/BCLK boootup issue, yes anything over 1.34VTT (measured) and my board goes straight to C1 at 195 BCLK.
This is using the OCZ Pro X stream PSU. Can anyone else confirm this behaviour please with their working PSU's?

so it doesnt matter how big the step is, it has to do with setting vtt above 1.34v vtt meassured at the vtt pad?

nooow here are the results of my testing:
EDIT: hmmm i checked again and i was wrong, the problem has to do with the vtt reading...
it seems that the voltage reading pad for vtt is not very accurate... ill post more details later...

And bill, i couldnt get a bclock above 166 with a qpi multiplier of 18 and uncore multiplier of 16.
if i change uncore to 18 it works, if i change qpi to 22 it works as well (although 180 is the max cause of qpi clocks)
this is a bios bug i guess and i reported it to the bios engineers
does setting uncore to 18x solve the problem for you?

about raid not working when the jmicron controller gets disabled, thats a minor bug so i wont test that right now, ill focus on the vtt stuff and the memory problem ocer_gr has once i get the spd. Thanks for the feedback guys!

[Raja@ASUS]

Just been talking to Sascha, the voltage reading at the pad for VTT is right. Ignore hardware monitor as it takes it's reading closer to the inductor. The offset between set and real (with droop) is all that matters and I've found what I wrote a few posts back in terms of the offset seems to be the norm. The droop is normal and actually desirable to ensure that overshoot and ringback effects are minimized.

regards
Raja

[Touge180SX]

Think I am having the same RAID problem. No matter what I do I can't run RAID on this board either on the P03 or G15 BIOS. When I enable RAID it won't boot to Vista CD and gives me a no system disk error but all 3 drives work fine when in IDE mode.

[OverClocker_gr]

ok, ill try to reproduce this... weird..


the video doesnt work... the dl always stops immediatly and the file is only 1.1kb in size... maybe its blocked somehow here at work, could you check and make sure the dl works for you?
i guess the debug leds are just flashing quickly and you cant read anything?
or is it cycling from one code to another in an endless loop?
what bios is this with? g13?
With G15 i cant run Bclock above 166 with 16x uncore on my 920... seems like a bug, as soon as i set uncore to 18x it works fine.


about raid not working when the jmicron controller gets disabled, thats a minor bug so i wont test that right now, ill focus on the vtt stuff and the memory problem ocer_gr has once i get the spd. Thanks for the feedback guys!

the video is ok.Probably its its somehow blocked from the network at your work.

They are cycling.
Boot-b8-b9-00-E8-02-b9-00-E8-02 and the overclock recovery.

I am with G15 bios running 200*20; 8x rams, 16x uncore and 18x qpi.


About voltages..

Setting in bios

Vcore +170mV , target 1.33875
vdroop enabled
Vtt +240
pll 1.913
vdimm 1.89
amplitude 1v
ioh 1.18


Bios Reading

Vcore 1.39
vtt 1.4
vdimm 1.92
ioh 1.2


DMM readings
vcore 1.410
vtt 1.379
pll 1.9
vdimm 1.924
ioh 1.205


About the raid etc issue.
When i have raid enabled on ich10 i cant boot from sata dvd rom.


Attached file is the .thp