Asus P6T Deluxe Discussion Thread

[003]

I can do 222 with my native V1 and a C0/C1 i7

[roller11]

You keep talking about design flaws,

Right, design flaws. A design flaw is when a mobo fails to perform a standard feature at a performance level that is well within the capability of the individual chip that is being used. My chip can comfortably do 21x191 Bclk at 1.184 Vcore.
I know this because I've tested it for EIST in many different mobos. So if there is a failure of my 'system' and the CPU, memory, vid card are not to blame, then by definition, it is a failure of the mobo cause that's the only thing left.

You have this board yourself you should be able to make all these changes yourself to experiment with cause and effect, it is much more useful than looking at my screen shots.

Actually, the opposite is true. Your screens are essential because they help answer the question "is it a design flaw, or just my particular board"? I am unable to replicate your results, so perhaps my board is even more limited than yours.

If I hadIf you set a value that means you're making that value static/set, auto mean's it will be controlled by the predetermined auto parameters/tables.

Right, and I am using the "auto" values. I started with 'load default values', put it in OC mode, and changed nothing but voltage, I even left Bclk at 133. Booted into windows and Vcore failed to vary dynamically even though that feature is built into my chip. You did the same thing, you are seeing EIST work properly at 150 Bclk. Then again, you haven't tested EIST anywhere near the AC limits of your CPU, so your board may or may not support EIST. So the question remains, does EIST work on your board only if you downclock? Or does it still work at a typical CPU clock of, say, 4.0 Ghz?

At any rate here it is "working" on "auto":

I appreciate you taking the time to run this test. It tells me that there is at least some chance that EIST may not be broken after all, that it does work at full performance levels. But since you didn't report what happened when you crank up CPU clock to typical levels, I would suspect there is still a problem,
especially considering that DFI, MSI, Biostar etc. all do EIST at full performance.

[highoctane]

[QUOTE=roller11;3748906]

EIST always works to control the multiplier up to 4ghz this I do know! Its the vcore that will no longer scale lower and higher with clock speed once you get to around 160ish bclock.

You need to enable EIST and have the multiplier and vcore set to auto, settings like the bclock have no auto since outside of overclocking this value doesn't change.

The mode of oc'ing should be set to manual, no ai or anything else, manual should give you all the settings that can be changed off the top of my head.

[raptord]

I think the EIST is little bit complicated on P6T-D (v1 or v2). On my v2, with the setting everything enabled except C1E and multiplier manually set at 21x, Bclck=200Mhz , if I turn on HT, the multipler changes with load (12,20x,21x). But if I go from this setting with HT off (this is only change), the multiplier stuck at 21x, no matter of the load.

Somebody should confirm this.

[chuckbam]

Originally Posted by davewolfs
Anyone care to share what C-State does? I have found that it will reduce my idle temps.

I can confirm what dave says. The C1E doesn't wotk when cpu is o/ced.

Both C1E and SpeedStep work with my Overclock, either separate or together. I am still using the 20x multiplier at 180 BCLK.

I looked all over the Intel website on C-States. I only found a small amount of information related to Mobile Processors.

On the i7, they only list SpeedStep and C1E as supported features. This makes me think the C-States are not supported. And, is why it is disabled in the bios by default.

[roller11]

Both C1E and SpeedStep work with my Overclock, either separate or together. I am still using the 20x multiplier at 180 BCLK.

.

When you say "work" do you mean only multi works, or both multi and Vcore change dynamically with load?
With DFI, MSI, and Biostar, both vcore and multi change at all settings. With giga, multi changes at all settings, but vcore changing only happens up to 139 Bclk. With my Pt6 non-deluxe, multi changes at all bclk, vcore never changes regardless of settings. Since others are getting different results, I'm thinking I should RMA my board.

[highoctane]

Both C1E and SpeedStep work with my Overclock, either separate or together. I am still using the 20x multiplier at 180 BCLK.

I looked all over the Intel website on C-States. I only found a small amount of information related to Mobile Processors.

On the i7, they only list SpeedStep and C1E as supported features. This makes me think the C-States are not supported. And, is why it is disabled in the bios by default.

Cstates are basically the level of sleep state that can be applied, the higher the Cstate the more of the core can be shut down during low to idle useage. In order to get the 22 multiplier with turbo boost you have to enable a cstate high enough to allow the cpu to sleep three cores.

Look up the turboboost white paper on intels site.

[chuckbam]

Cstates are basically the level of sleep state that can be applied, the higher the Cstate the more of the core can be shut down during low to idle useage. In order to get the 22 multiplier with turbo boost you have to enable a cstate high enough to allow the cpu to sleep three cores.

Look up the turboboost white paper on intels site.

I get a small Vcore drop with C1E and not SpeedStep. Both have the multiplier drop two 12x.

I would not want to use anything that would either disable cores or HT. I see no evidence of that with either of the C-State bios settings (@ 20x/ 12x).

http://chuckbam.com/i7_P6T-D/x58_320838.zip

[highoctane]

I get a small Vcore drop with C1E and not SpeedStep. Both have the multiplier drop two 12x.

I would not want to use anything that would either disable cores or HT. I see no evidence of that with either of the C-State bios settings (@ 20x/ 12x).

Its dynamic just like EIST changes the multiplier, you will have no idea cstates are even doing anything unless you have a killawatt meter or other means to measure idle power consumption. As soon as there is a load the cores wake back up when needed, it doesn't cripple the functionality of the chip. If you feel the need to use EIST to save power then you should highly considering using full cstate to reduce idle power consumption further since all four cores are doing little at idle to light load.

The intel white paper on turboboost will give you a good idea of whats going on.

The voltage change you are seeing is from load line calibration being enabled, the bios is actually increasing vcore to compensate for the load. Try disabling load line calibration and see if your vcore continues to go up during load.

[chuckbam]

The intel white paper on turboboost will give you a good idea of whats going on.

Do you have a link? Does it say that the C-States are supported on either the i7 or the x58?

[highoctane]

Do you have a link? Does it say that the C-States are supported on either the i7 or the x58?

Just goto intel and search for turbo or turbo boost and yes cstates are fully supported by i7 as much as EIST or HT are.

[Buggalugs]

Ya i have the P6T vanilla and when overclocked my voltage doesnt change but my multiplier does. On default settings both multi and voltage changes.

My last few Asus and Gigabyte boards did the same thing, i dont really care, its only 200mv i dont think it will affect heat or my electricity bill too much.

[chuckbam]

Just goto intel and search for turbo or turbo boost and yes cstates are fully supported by i7 as much as EIST or HT are.

3.0 C-State
Definitions
Intel® processors based on Nehalem
microarchitecture support core C0, C1, C3, and C6.
C0 and C1 are always supported; the availability
of the remaining C-states may vary by processor
number. Any core within the processor can go into
any C-state independent of the state of the other
cores.
Note: The behavior in a particular C-state on Intel®
processors based on Nehalem microarchitecture
may be different from C-states referred to with
the same number on previous products. One
reason for this is the platform changes, including
the move to the Intel® Quick Path Interconnect,
which results in the elimination of the STPCLK#,
SLP#, and DPSLP# signals.

3.1 C0
C0 is defined as the active state. While in C0,
instructions are being executed by the core. For
Intel® Turbo Boost technology, a core in C0 is
considered an active core.

3.2 C1
C1 is defined as the halt state. While in C1, no
instructions are being executed. For Intel® Turbo
Boost technology, a core in C1 is considered an
active core. This idle state is generally classified
as “ACPI C1.”

3.3 C3
While in C3 the core PLLs are turned off, and all
the core caches are flushed. For Intel® Turbo
Boost technology, a core in C3 is considered an
inactive core. This idle state may be classified as
“ACPI C2” or “ACPI C3” depending on processor
number, BIOS, and operating system.

3.4 C6
While in C6, the core PLLs are turned off, the core
caches are flushed and the core state is saved to
the Last Level Cache. Power Gates are used to
reduce power consumption to close to zero. For
Intel® Turbo Boost technology, a core in C6 is
considered an inactive core. This idle state may be
classified as “ACPI C2” or “ACPI C3” depending on
processor number, BIOS, and operating system.

I am getting lower CPU temps with the C-State on auto. Thanks for the info. I keep-ed looking at the x58 tech pdfs for C-State info.

[roller11]

My last few Asus and Gigabyte boards did the same thing, i dont really care, its only 200mv i dont think it will affect heat or my electricity bill too much.

200mv will most certainly affect heat, in fact, going from 1.100mv to 1.300
vcore means a power increase of 39.6%. That's why EIST is so important, it has a tangible real world benefit. The extra heat translates directly into extra fan noise. So while you're browsing the net, doing Email, doing a slide show for friends, etc. you are unnecessarily subjected to irritaing fan noise because ASUS messed up the design.

I was going to RMA my board, but seems everyone is saying that EIST doesn't work at full performance levels. So there would be no point, I'd just get another board with non-working EIST.

[Buggalugs]

200mv will most certainly affect heat, in fact, going from 1.100mv to 1.300
vcore means a power increase of 39.6%. That's why EIST is so important, it has a tangible real world benefit. The extra heat translates directly into extra fan noise. So while you're browsing the net, doing Email, doing a slide show for friends, etc. you are unnecessarily subjected to irritaing fan noise because ASUS messed up the design.

I was going to RMA my board, but seems everyone is saying that EIST doesn't work at full performance levels. So there would be no point, I'd just get another board with non-working EIST.

OK ya maybe a little. But hey i did some testing on this yesterday and today i fired up the computer and my voltage IS fluctuating a little.

Its going from 1.288 down to 1.280 . All i did was enable Cstate which was disabled by default, and put all voltages on auto. Thats at 3.8Ghz OC. So its working a little now. I guess it would be better if it worked properly but i can live with it.

But ye my computer is silent with the TRUE and a 1200rpm Scythe fan, never hear it, but it would be good if it works properly. Especially if other cheaper boards can do it, but then again those boards are not known for their stability or overclockability so maybe theres a tradeoff.

[ben805]

Anybody has any success running 12GB of OCZ Platinum 1600 on the P6T? I just bought another 6GB kit (2GB x3) but no luck getting them to POST even at the loosest timing at 11-10-10-30

[ben805]

I am using XP x64. Also, I am not the only one who has this issue. Look a bit back into the thread, there is at least one other person to notice it as well.

Same here, running at 21x200 with HT on with only 1.31v, under heavy load in LinX the multiplier would drop down to 20x, all the power saving features(TM, C1E, c-state, etc) are disable in the BIOS and in Vista 64bit OS.

[003]

Same here, running at 21x200 with HT on with only 1.31v, under heavy load in LinX the multiplier would drop down to 20x, all the power saving features(TM, C1E, c-state, etc) are disable in the BIOS and in Vista 64bit OS.

Yes, this is extremely annoying

[ben805]

Yes, this is extremely annoying

this may be one of the feature built-in the Core i7 to prevent damages from overheat/overvolt, not sure what the deal is...probably good for noobie but an annoyance to those who can keep the heat down. To work around this i guess if the motherboard can handle it then use the native 19x or 20x and crank up them bclk. Other than LinX I haven't come across any real world apps or games that could trigger this phenomenon.

[davewolfs]

So I guess if C1E and C State are not have a negative affect on your OC's its an ideal state to leave the machine on? I'm also guessing that these options have little affect if any on real world performance. Can someone confirm this.

[kpablo]

I dont have that problem... and yes, i have a V2 but flashed to V1 (1403) all i did was fixing multiplier to 21, then disable everything in CPU features. Except HT, and cores, ofc.

[003]

What voltage do you used? And you have tried LinX?

[highoctane]

I dont have that problem... and yes, i have a V2 but flashed to V1 (1403) all i did was fixing multiplier to 21, then disable everything in CPU features. Except HT, and cores, ofc.

I've even tested this myself and turbo boost will dial back the multiplier under enough voltage and linx stress testing. The load conditions are definitely not real world and will probably only happen under heavy stress testing at vcore over 1.4ish.

I run my vcore at 1.3875 at 21x191 and it never drops even under linx, at 1.425 mine fluctuates between 21 and 20.5 at around 78C running linx.

EDIT: Using prime, even at 1.425 21x191 the multiplier didn't scale back, only happened to me with linx when temps got around 78C.

The intel white paper gives some insight to this but they don't specify the particular thresholds that actually trigger the safety.

[Efekkt]

Okay, I just set-up my first ever system with an i7 920 and a P6T Deluxe. Currently I can't get it to install windows because it keeps saying a bios update is needed. I tried hooking up a hard disk with an OS and it got in, but the keyboard and mouse wouldn't work, so it was pretty much pointless.

I just updated the bios to the latest (the 1403, I think) but the same damn thing still appears, telling me I should update my bios to see my cpu's full potential. I've tried setting the CDROM as first priority for booting with the XP CD, but it doesn't seem to work. And I seem to get an error that says NTLR (or something like that) not detected. Any suggestions, guys?

Edit: I don't know if you guys need to know this, but I'm on 6GB Corsair Dominators. Thanks for any help!

[Amurtigress]

Okay, I just set-up my first ever system with an i7 920 and a P6T Deluxe. Currently I can't get it to install windows because it keeps saying a bios update is needed. I tried hooking up a hard disk with an OS and it got in, but the keyboard and mouse wouldn't work, so it was pretty much pointless.

Stupid question WHAT tells you that you need a BIOS update? When does that message appear?

Usually when a CPU isn't supported by the BIOS, it wouldn't come up at all.

As for the keyboard and mouse, try a PS/2 one if the not working ones are USB, and vice versa. It is possible that your windows is trying to install USB drivers for the ICH10R southbridge but can't find them until you're installing the latest Intel INF update from your board's support CD.

Something similar happened to me.