Say Hello to ""B2"" stepped Gulftown ^^

[ajaidev]

so we all flashing 980's around here?

LOL im tempted to join the club...

Ive had my 980 for a while now..



Ive had both A0 and the 980.

My 980X does 4.4 @ 1.387Vcore with HT on like Pie.

My A0 required @ least 1.42 vcore for that range..
So yes, the B steppings are beasts.

One wierd thing i noticed tho... The A0's like a higher QPI vs the B's unless im seriously doing something wrong... :\

Quite correct i am very surprised with these B1's tough i have a B0 and it has a different mentality than the B1.

I think since even my A0 like higher QPI maybe that is the cause for the higher heat ??

Not the new Q4EG B1 Stepping...cool as a cucumber...only 6 degrees above room temp at idle...never goes above 55C on full 100% load...very nice chip! Hoping for 4.6 24/7....We'll see....It will probably do it, but at voltages I am not comfy with...

[IG]http://i139.photobucket.com/albums/q315/managerman/gulftown1.jpg[/IMG]

Very nice, what are you using ? Chilled water or something, the temps are real nice are you sure the temp sensor is reading the temps correct

Has nobody put these under a single stager yet? I am guessing 5ghz+ easy for 24/7? All I see is air/water/ln2...

I have asked permission to put in under because i dont want to kill it with voltage and then get blamed for it.

[managerman]

=Very nice, what are you using ? Chilled water or something, the temps are real nice are you sure the temp sensor is reading the temps correct .

Here is what UncleWebb said on another forum..

-M


"The new 32nm temperature sensors that Intel are using are not very accurate at idle.

The i7-920 temperature sensors were excellent but Intel has taken a step back to the good old days where the sensors they use can have a lot of slope error. This means that at temperatures a long ways away from the calibration point, these sensors will start to read either too high or too low. The amount they are off is random and varies from one core to the next.

Your load temperatures will be somewhat accurate if you get these sensors up to about 70C where the amount of slope error tends to be minimal. Below 70C it's pointless trying to compare reported temperatures from these CPUs with the original 45nm Core i7-9xx series CPUs because the sensors are not good enough for that.

Intel only designs these sensors to trigger thermal throttling at about 100C and thermal shutdown at about 125C. I find the sensors, even though they are not accurate, tend to be consistent so if a core reads a few degrees low at idle one day it will read the same amount too low 6 months from now.

RealTemp GT reads these new CPUs OK. It doesn't get the name string correct yet since I was originally guessing that Intel was going to call them an i9 when the first ES processors were being tested but MHz and temperatures should be OK. It looks like TJMax for these is set to 101C if you haven't manually changed it. That data is being read directly from a register within each core. Intel actually calls this data TJ target and TJMax is supposed to be somewhere around this number but no one knows how close it really is. It can vary slightly from one CPU to the next and from one core to the next on the same CPU."

[Badfastbusa]

What are you guys getting on air with these 980x chips?

[AllAgainstPaul]

It's nice to see that Intel is finally using a better stock Heatsink, even if it is only available with their flagship CPU. Hopefully they will start shipping it with low and mid-range CPU's in the future.

[Badfastbusa]

It's nice to see that Intel is finally using a better stock Heatsink, even if it is only available with their flagship CPU. Hopefully they will start shipping it with low and mid-range CPU's in the future.

Why? I mean, who uses a stock heatsink on a flagship chip that will overclock to the moon other than people that use them for work? They should completely cater to us overclockers

[ajaidev]

Here is what UncleWebb said on another forum..

-M


"The n...........................ightly from one CPU to the next and from one core to the next on the same CPU."

Yep i knew that the A0 had major issues with temp sensors and well my B0 while better does not actually work as good as nehalems i am surprised that this was not fixed in B1 i was thinking it will be fixed with time.

[Badfastbusa]

Yep i knew that the A0 had major issues with temp sensors and well my B0 while better does not actually work as good as nehalems i am surprised that this was not fixed in B1 i was thinking it will be fixed with time.

Yeah, but from what I have read, the retail 980x's aren't going to overclock anywhere close to these ES chips

[AllAgainstPaul]

Why? I mean, who uses a stock heatsink on a flagship chip that will overclock to the moon other than people that use them for work? They should completely cater to us overclockers

I agree that most 980 will have a better cooling option on them, I just wonder how their new heatsinks compares to aftermarket ones. Perhaps we have are at the beginning of the end for many aftermarket heatsink manufacturers.

[managerman]

Here is my latest results with B1 stepping...

What do you think the highest safe 24/7 Vcore is on these 32nm chips??

-M

[zalbard]

Here is my latest results with B1 stepping...

What do you think the highest safe 24/7 Vcore is on these 32nm chips??

-M

1.45-1.5 tops.

[Badfastbusa]

I agree that most 980 will have a better cooling option on them, I just wonder how their new heatsinks compares to aftermarket ones. Perhaps we have are at the beginning of the end for many aftermarket heatsink manufacturers.

Thats why I went water hehe

[firerx]

I like mine so far.

[Badfastbusa]

I wouldn't know what to do with that much computing power Folding here I would come!

[Tommy_Hewitt]

Here is my latest results with B1 stepping...

What do you think the highest safe 24/7 Vcore is on these 32nm chips??

-M

1,4V is max. according to Intel

[Sam_oslo]

Max 1.4v? that's way under the 1.55v for i7. On could really get tempted to push it over on a such a cool chip.

980x is unlocked and you guys are not pushing the BCLK, but what kind of max-BCLK have you been getting/expecting for 24/7 use?

[zalbard]

Max 1.4v? that's way under the 1.55v for i7. On could really get tempted to push it over on a such a cool chip.

I wouldn't be so sure...

Intel's new 35nm temperature sensors seem to have a lot of error at idle and many of them read way too low.

Intel writes TJ Target information into each core of each CPU but TJ Target isn't always equal to TJMax. Actual TJMax might be 10C or more higher from one core to the next even though they all have the same rating of 105C for this CPU. With undocumented sensors like these, accurate temperatures aren't possible. The 45nm Core i7-920 sensors were some of Intel's best, very consistent. These new sensors are at the other end of the spectrum. The new Core i3 32nm CPUs have similar issues.

Edit: Another problem is that these CPUs are designed to turn individual cores off at idle or when lightly loaded. This can also lead to a lot of temperature variation from core to core.