C2D only goes to 1.120 idle...anyone get .85ish???

[Black_Knight_MC]

They may be "overvolted" when in speedstep but it does not hurt it. All it mean is the CPU is getting more power than it needs to run at 1.6. Such a wide range is possible, look at the 600 series P4's like a 650. They would drop down and their voltage also, to what the specs said. It doesn't matter the range of the models, if you took a P-D 820 and all the ones faster than it, you could underclock them all because going lower is easy, it is going above the ghz the chip is rated for and having to raise voltage above spec that causes instability (well we rasise the voltage above spec to have the chip ghz run above spec). Point it, the spec says .85 so we should get close at least.

What I find more strange is that from idle to load I go up only 10C. Thats because my idle still runs hot because it can't go that low. I go from 1.120 to 1.280, according to CPUZ, which means from 1.6 to 2.4 I only need .16 more volts? It is so small because the idle voltage is already hich enough to run 1.6, plenty high. I think RMclock only shows what your computer is aiming for or the presets for those multipliers. Beacuse according to RMclock our voltages are dead on, at least accurate by .001, which is really not possible to be that stable, RM says when at 6x I am on the nose of 1.162, never flucuates, underload, jumps to 1.35, nothing in between. If you goto power management you can hand it over to RMclock. Then goto RMclock and choose the VID you want, you will see the lowest option is 1.162 which means when it communicates with the mobo the mobo is not even providing the option for the lower voltages, hence why I think it is a mobo hardware limitation or BIOS limitation.

Please, read the intel PDF i gave, not the whole thing, just the electrical specifications section, it is CLEAR that it is intended for current C2D's not some new process or stepping later. I ran Intel's TAT and had it throttle all the way to 700 mhz for fun and noticed I could idle, at the time and ambient air, at 33C instead of 40C. Now thats just a throttle, not a lower voltage, imagine the heat level if the voltage drops too. I REALLY wish we could talk to TH and see how they got their to run according to spec, they even mention how running at 1.9V at 1.6ghz below their screenshots of CPUZ. On other review sites, the C2D's lose to the same AMD solutions in comparision of idle consumption while on TH they won that comparison seen here:

http://www.tomshardware.com/2006/07/14/core2_duo_knocks_out_athlon_64/page3.html

That is why I think it wouold be a big deal, because now that TH article doesn't mean squat to us real customers. You will also see here that this no longer matters because the AMD runs colder than my C2D, but not the TH one running at spec (I idle at 38-41C, TH at 25C! and AMD at 31C):

http://www.tomshardware.com/2006/07/14/core2_duo_knocks_out_athlon_64/page8.html

You will see this picture of the X6800 idle:

You can see that even the ASUS probe shows .9V, a whole .22 lower than my idle. That is huge when comparing I only go up .16 going from idle to full load. And, my chipset idels at 35C with passive cooling. Now this could be that my 965P runs cooler than the 975X chipset but that would atleast mean my ambient temps are comparable to TH's ambiguous 'office environment' temp.

EDIT- Oh and thanks for contacting them about this, I really would like to hear a word from them. If they say we are all running like they expected and that the spec is just very inaacurate when I will be content that there is nothing I can do but also dissapointed their specs are not what they really run at. I have contacted ASUS already but sometimes those guys are just people reading out of a troubleshooting book and what we need is some engineer to spend 5 seconds saying yes or no and why or why not.

 

[Black_Knight_MC]

beercandy, how is someone to order, say a certain sku from a site when they do not even have a method to select it. On sites there is only ONE model, you dont see the model 3 times with 3 dif sku's.

Also, where is the sku you are talking about in the article? All there is is a revision, I know of those, but those are to fix bugs, not add features they forgot or left out.

Are you asking if my voltage changes if I disable all my USB? I can only think you are asking as it maybe a power problem but wouldn't this instead solve an undervolt problem, not an overvolt?

 

[Black_Knight_MC]

LOL, I know about that, and the way ASUS and all other manufacturer's show different chipset types are by the different models, such as my ASUS P5B being a 965P and the ASUS P5W DH Deluxe being a 975X. My point is that Intel and every site has already said that 965 was built specifically for Core 2 Duo so that area is fine, it is about how the voltages work.

 

[Black_Knight_MC]

Well I don't know what you are trying to prove but my chipset is the 965P, second to only the 975X. Anything other than the 965 series or 975 can't even support core 2 duo's and mine is a performance version of the core 2 duo chipsets. And as far as I can find there are is no documentation for what voltages a mobo is capable of, only what the BIOS allows it to run, hence how BIOS's upgrades can cause more selections.

Oh, and I don't own a non usb keyboard and cant find a ps/2 converter for it, so I can't try turning off all the usb ports.

 

[Black_Knight_MC]

ASUS P5B

http://usa.asus.com/products4.aspx?l1=3&l2=11&l3=307&model=1178&modelmenu=1

I wasn't talking about you proving anything with the USB, that was something i sperately commented on. I asked what you are trying to prove with the "sku's". Like, are you trying so say some of the chipsets that support the c2d's might not completely support them? As in not support all of the voltages supported by the c2d?

 

[crow_smiling]

that sku of that board is not a full featured sku you will want to get a 965G sku which should handle all of the features

You clearly don’t know what you are talking about. The P965 and G965 are very similar chipsets but the G965 supports integrated graphics and the P965 doesn’t.
The G is for Graphics and the same goes for the 945 range where the 945G supports integrated graphics and the 945P doesn’t.
To imply that a board with a G965 chipset is somehow automatically going to be better than one with a P965 is ignorance.

what do you measure for you proc if you disable all of USB in the bios? I know it sounds like a strange question but i was just wondering if it changes

The CPU receives its power from a dedicated source, the VRM. The Voltage Regulator Module that supports C2D is VRD 11. If you look at the Intel Voltage Regulator-Down (VRD) 11.0 Processor Power Delivery Design Guidelines for Desktop LGA775 Socket, you will see that it shows a range of 0.50 to 1.60V is supported by this VRM revision.
The USB buss and everything else in a PC is running off the main power lines, which I think are 5V and 12V.
So how is the USB buss going to affect the ability of the VRM to deliver a proper voltage to the CPU?
Oh don’t tell me, there’s a ‘special’ SKU that takes care of this issue.

Black_Knight_MC: I’d ignore beerandcandy, as he is either ignorant and doesn’t know it or he’s just taking the piss. I think he’s under the illusion that if he mentions the acronym SKU enough people will think he actually knows something.

 

[Black_Knight_MC]

Here, try this link:

http://www.newegg.com/Product/Product.asp?Item=N82E16813131030

Well my mobo supports all 3 FSB and all sorts of features, including its own added features, ASUS is no slouch and I have yet to hear anyone with an intel board achieve the proper voltage. I think it is just mobo makers have not yet made bios's compatible (need to put in more VID's) with the low voltages that the c2d's in speedstep can perform. This doesn't seem that odd since there are bios's being released to make a fix to things like the mobo simply not having core 2 duo ucodes on release bios's.

 

[Black_Knight_MC]

Whoa, well that right there shows that out mobos should be able to run .85 if they are abiding by that voltage standard. Heck, had I found that I would have never started this thread the way I did. I would just assume it is the bios and instead make a thread trying to find who gets the proper voltages. Thats good news to me.

 

[crow_smiling]

Whoa, well that right there shows that out mobos should be able to run .85 if they are abiding by that voltage standard. Heck, had I found that I would have never started this thread the way I did. I would just assume it is the bios and instead make a thread trying to find who gets the proper voltages. Thats good news to me.

It doesn’t necessarily work like that; they very likely build loads of head-room into the spec, as at the time they define it they don’t know the exact voltages that future CPUs will require.
An early revision, VRM 10, allows for a voltage as low as 0.8375V. Do you know any Pentium 4 CPUs that automatically run at that voltage?
It may be that the VRM specs are also meant to encompass the mobile CPUs, which include the LV and ULV parts. Some of the mobile parts are rated as low as 0.762V.

Another possibility is that this 0.85V value is relating to the HALT or Extended HALT Powerdown States. I’m not clear how significant these states are though from a VCore perspective.
Or it could even be related to TM2, as that allows the CPU Clock Frequency and the VCore to be lowered under adverse thermal conditions.
The point I’m making is that there are a lot of different states that a CPU can run in that might potentially require this lower VCore setting. But it doesn’t mean that you should necessarily expect your CPU to idle at 0.85V under normal everyday conditions.

To quote from Intel:

Extended HALT Powerdown State
Extended HALT is a low power state entered when all processor cores have executed
the HALT or MWAIT instructions and Extended HALT has been enabled via the BIOS.
When one of the processor cores executes the HALT instruction, that logical processor is halted; however, the other processor continues normal operation. The Extended HALT Powerdown must be enabled via the BIOS for the processor to remain within its specification.
The processor automatically transitions to a lower frequency and voltage operating
point before entering the Extended HALT state. Note that the processor FSB frequency is not altered; only the internal core frequency is changed. When entering the low power state, the processor first switches to the lower bus ratio and then transitions to the lower VID.
While in Extended HALT state, the processor processes bus snoops.
The processor exits the Extended HALT state when a break event occurs. When the
processor exits the Extended HALT state, it first transitions the VID to the original value and then changes the bus ratio back to the original value.

 

[Black_Knight_MC]

The EHALT state is C1, which if you run Intel TAT you see that they either run at C0 (normal operation) and C1 where speedstep comes in. When throttling, the ghz goes down, not the voltage, because this can still yield enormous hear reduction, try running Intel TAT and create a forced throttle and you will see, I could make my cpu idle at 32C by throttle, while the voltage stayed at 1.120.

I do realize their is head room but I don't expect them to goto the .5, thats why the .85 is what it is speced to actually run at, like how TH's ran at .888-.91 volts. No there are not any p4s running at .8375 but thats like asking if we see any c2d running at .50, the VRM 11's lowest range. Most the p4s with speedstep ran at 1.25, thats about .41 above VRM 10's lowest range. With c2d and VRM 11, there is a .35 difference from what the mobo could supply and what Intel says the CPU can run at, seems pretty reasonable to me.

 

[crow_smiling]

The EHALT state is C1, which if you run Intel TAT you see that they either run at C0 (normal operation) and C1 where speedstep comes in. When throttling, the ghz goes down, not the voltage, because this can still yield enormous hear reduction, try running Intel TAT and create a forced throttle and you will see, I could make my cpu idle at 32C by throttle, while the voltage stayed at 1.120.

What is Intel TAT? Sounds useful.

TM1 allowed only clock throttling, but TM2 supports clock throttling and lower voltage states.

 

[crow_smiling]

so if as you say G965 and P965 are the same then why does your DP965LT not support hardware monitoring and not support the smbus while the 965G does? also i never said one was better than the other i said they are different and are sold to different markets.

What you actually said was,

that sku of that board is not a full featured sku you will want to get a 965G sku which should handle all of the features

You didn’t say get a different board, you said get a different chipset, which implies that you thought that the chipset was the problem.
But the problem with the DP965LT is that it was released with a half finished BIOS. P965 and G965 based boards both typically support hardware monitoring, just about every board does. It’s just a substandard product which has nothing to do with the chipset. I have a P965 board that I’m testing and it supports hardware monitoring etc.

that is why i buy intel boards. they do exactly as advertised all the time. go to frys and buy the intel board

Exactly as advertised, in this case unfortunately that means that it doesn’t do very much.

in regards to the usb part i was not implying that the usb power and the proc power are using the same power plane. what i was saying is that if the usb port is idle with something that is plugged into it the port will constantly try to poll the device which can in turn prevent the cpu from going into the c3 state. not because they use the same power because they dont
but because the CPU cannot go into C3 unless the system is fully idle.

I’m with you now; I thought you’d lost the plot completely there. 8O
How are you monitoring what state the CPU is in?

 

[Black_Knight_MC]

Intel TAT is Thermal Analysis Tool, here is a link:

http://shintai.ambition.cz/files/tat.exe

You can see what states are available and what it is in (at least I think you can see what state its in, i know for sure what state it can perform) and as for c2 and up, I thought those were for mobile and for when desktops enter sleep or standby.

 

[Black_Knight_MC]

Well as far as I know core and core 2's support C1E, C2E, C3E and C4E (check RMclock and you can see them, i told RM to use them all but still no difference.

 

[keeler]

This has been an interesting read, but there may BE something to this sub-1.0 volts deal! In that I've not put my profile in yet:
Asus p5w dh on bios 1101
E6600 wk27 @ 2.9 w/ volts set at 1.36
G-skill 6400 @ 1:1

Sunday morning I came in from work and did a quick set of my memory
timings and wanted to see how 3dmark03 ran. Now note that I've got
speed stepping and all that OFF...

My 3dmark ran fine, Then I looked at AIBOOSTER and saw my CPU temp at 24C (4 deg under room ambient), and cpu volts at .90 (w/ some fluctuation(+- couple of 100's). I've been averaging 34-36C at idle, and I know the rep on aibooster readings, so I looked at coretemp to verify temps.. WELL!! coretemp reported something like 39C, when it normally reports 47-48!!

I figured that I had somehow fat-fingered the cpu volts when I went in to change ram timings, so I reboot and check. CPU was STILL set at 1.36, and all the speed stepping stuff was still off. I let the system come up, and MAGICALLY, I was where I was the night before with 34-36C on CPU;
47-49C via coretemp, and 1.36 on CPU.

Seeing the coretemp readings as I had leads me to believe that there WAS a significant drop in CPU voltage even tho how it GOT there was some kind of bios FLUKE!

I had just wrote the whole thing off to a glitch in the sending units, but know I really wonder!! I would NOT be surprised to see the sub-1. volts
become a reality in speed stepping as these bios' mature!

Rick

 

[Black_Knight_MC]

Well I KNOW it can happen now that I am talking to an Intel tech. Quote, "Thank you for contacting Intel(R) Technical Support.
>
>I understand that you are looking for the proper core voltage of the Intel(R) Core(TM) 2 Duo processor.
>
>The core voltage for the Intel(R) Core(TM) 2 Duo processor is 0.850V-1.3525V"

He said proper voltages for Core 2 Duo processor, reffering to the series. So now I am trying to ask him what I can try to get it working close to .85.

GREAT keeler! I don't know how you got it to run like that with the features off in the BIOS but you still had it running like so, a non ES, infact, my model. Perhaps I will set my CPU voltage TO something in the BIOS, say, 1.325V and go from there. I will reply in a few.

 

[Black_Knight_MC]

Well, made SOME progress. I put my CPU manually, instead of the ATUO option, to 1.3125V. Now when I speedstep I goto 1.05V, instead of 1.120V. It looks to me that the BIOS is not allowing a very big voltage range for speedstep. I am going to go in again and try setting it lower in the BIOS, here we go!

 

[Black_Knight_MC]

Well, I started to lower it further and now it still speedsteps to 6x but now I stay at 1.240-1.248, no matter full load or idle. Oh well. Hey keeler, mind filling us in on what your cpu settings are and your cpu voltages are set to again? You also have the same board that tomshardware had in its review where they got .888V (.9V) so it could just be some setting.

 

[crow_smiling]

Well, made SOME progress. I put my CPU manually, instead of the ATUO option, to 1.3125V. Now when I speedstep I goto 1.05V, instead of 1.120V. It looks to me that the BIOS is not allowing a very big voltage range for speedstep.

The situation with regard the 1.05V that is available when you manually set the VCore in the BIOS is a feature of recent BIOS revisions for the P5W DH Deluxe and the P5B Deluxe; see here for details. It wasn’t available with the P5H DH up to and including 0901 Beta from my experience.
Some motherboards do the opposite type of thing; if you manually set the FSB with the Gigabyte DS3 using BIOS F4, the VCore then becomes fixed at 1.44V even with Speedstep enabled.
There are a lot of quirks in C2D motherboard BIOS’s; early days I guess.

 

[Black_Knight_MC]

Yea, this is looking like a BIOS issue. They need to increase the range of voltages the CPU can drop to so as to save power and lower heat output. I guess we just have to wait until they support what these CPUs can really do.

 

[crow_smiling]

Yea, this is looking like a BIOS issue. They need to increase the range of voltages the CPU can drop to so as to save power and lower heat output. I guess we just have to wait until they support what these CPUs can really do.

The irony is that the Gigabyte DS3 allows you to manually set 0.5V in the BIOS.

 

[Black_Knight_MC]

So does it allow speedstep to go that low? Because that option in the BIOS is for operating V, not the EIST V. So even if you could pick .85V in the BIOS, you shouldn't be able to run windows since it will become unstable trying to run .85 while underload and at 2.40ghz (or whatever multiplier the C2D perocessor is).

 

[crow_smiling]

So does it allow speedstep to go that low.

With the BIOS versions that I tested with, if you set VCore in the BIOS manually, Speedstep used the same voltage.

 

[Black_Knight_MC]

Well that sucks, guess we just have to wait.

 

[crow_smiling]

I received this reply from Intel:

"Thank you for contacting Intel(R) Technical Support.
I understand that you are able to achieve 1.15V, this is lower than 1.3525V and above 0.850V, so this means that the core voltage of your Intel(R) Core(TM) 2 Duo processor is running under specifications. I do not see any issue with this 1.15V."