Which 3.2 Stepping is capable for Lock Free? (LGA775)

[death metal]

Anyone knows if P4 3.2 will have "lock free" on supported mobos? I've seen at Asus sometime ago (and at THG) that P4 3.4 will be supported. Any particular stepping I need to look into for the 3.2? (3.4 isn't available here so .....)

[death metal]

up....

[BigStan]

Somewhere in these two threads over at the OC Forum is a discussion of which 3.2s and 3.4s are unlocked.

http://www.ocforums.com/showthread.php?t=338876

http://www.ocforums.com/showthread.p...8&page=1&pp=30

[DanIdentity]

For LGA775, only the 3.4 and 3.6 are unlockable.

Not sure for 478...

[death metal]

thanks for the infos guys...

bigstan, i'll skim thru the links you gave....it's rather large but i'll try to chew as much as i can...

dan, i know about the 3.4 and 3.6, but i'm just wondering if it works on 3.2 as well since s478 c0 3.2 is unlockable...

[Petr]

For LGA775, only the 3.4 and 3.6 are unlockable.

Not sure for 478...

Not true anymore. Newest E0 revision (model 4, stepping 1) is 3.4 GHz 78A chip - the best Prescott you can get. Not 119A like the older D0 revision (model 3, stepping 4).

[DanIdentity]

Not true anymore. Newest E0 revision (model 4, stepping 1) is 3.4 GHz 78A chip - the best Prescott you can get. Not 119A like the older D0 revision (model 3, stepping 4).

What do you mean? I'm not familiar with those A ratings, i.e. 78A.

[DanIdentity]

Bump for my question.

[Petr]

What do you mean? I'm not familiar with those A ratings, i.e. 78A.

There are two grades of Prescotts.

For s478 those two:
78 Amps max. current
91 Amps max. current

For LGA775 those two:
78 Amps max. current
119 Amps max. current


The higher grade is performance and those CPUs run very hot (max. thermal power is voltage * current). The lower is mainstream. Intel made three public revisions of Prescott:

C0 - 78A is for <= 3.0 GHz
D0 - 78A is for <= 3.2 GHz
E0 - 78A is for <= 3.4 GHz.

The less Amps a CPU takes the better because of less heat, therefore possibility for higher frequencies. Back to topic - only performance class (91A or 119A CPUs) allow for multiplier 14x because that is a safety rule that a performance CPU inserted to non-performance motherboard (eg. motherboard designed for 78A Prescotts) decreases it's frequency to 2.8GHz to reduce Amps, eg. to meet 78A grade. Since mainstream parts meet 78A, they do not have the ability to decrease multiplier. So if you want multiplier 14x you must buy an older hotter and less overclockable CPU (or the highest end offered - like today 3.6 GHz chip).

[DanIdentity]

There are two grades of Prescotts.

For s478 those two:
78 Amps max. current
91 Amps max. current

For LGA775 those two:
78 Amps max. current
119 Amps max. current


The higher grade is performance and those CPUs run very hot (max. thermal power is voltage * current). The lower is mainstream. Intel made three public revisions of Prescott:

C0 - 78A is for <= 3.0 GHz
D0 - 78A is for <= 3.2 GHz
E0 - 78A is for <= 3.4 GHz.

The less Amps a CPU takes the better because of less heat, therefore possibility for higher frequencies. Back to topic - only performance class (91A or 119A CPUs) allow for multiplier 14x because that is a safety rule that a performance CPU inserted to non-performance motherboard (eg. motherboard designed for 78A Prescotts) decreases it's frequency to 2.8GHz to reduce Amps, eg. to meet 78A grade. Since mainstream parts meet 78A, they do not have the ability to decrease multiplier. So if you want multiplier 14x you must buy an older hotter and less overclockable CPU (or the highest end offered - like today 3.6 GHz chip).

Great explanation, thanks a lot. I knew there were two classes of LGA775 Prescotts, but I hadn't heard anyone distinguish between them with an amp rating, only by wattage, i.e. 84W vs. 115W.

Now if I could just find an E0 LGA775 Prescott for sale anywhere I could build my new system...

[death metal]

Petr, I understand the Amps thingy, this is why I am confused as to which 3.2LGA can have CPU lock free. Right now, most LGA775 cpu is 84w rated, meaning, they are above the 78A safety margin.

For example:

(3.2 D0 stepping): http://processorfinder.intel.com/scr...107&CorSpd=ALL

versus

(3.4 E0 stepping: http://processorfinder.intel.com/scr...107&CorSpd=ALL

Both the 3.2 and 3.4 uses 84W rating for the CPU. If we're basing the safety trigger to come up to effect, the 3.2 is also a candidate ....

[DanIdentity]

The 78A and 84W ratings describe the same chip, they're just different measurements.

The chip needs to be a performance class, 119A or 115W, for it to have an unlockable multiplier.

[death metal]

dan, i might have kinda mixed up "A" with "W", but looking at the "rules" (on LGA775 only)

All 3.4+ LGA775 have CPU lock free, nothing below 3.4+ LGA775 is approved for CPU lock free (that's what Asus is saying here http://www.asus.com/news/2004/20040921.htm).

Why? Petr answered it, because of the power rating, i.e. the Performance class.

Now, assuming we compute for the "A" rating of the given 3.2 and 3.4...

3.4 (see link in my previous post) = 84w, max voltage is 1.4. Computing for amps, 60a.
3.2 (see link in my previous post) = 84w, max voltage is 1.4. Computing for amps, 60a.

See, they both fall in the same 60a category, and yet the 3.4 benefits from the CPU lock free and the 3.2 doesn't? Besides, in s478, 3.2 C0 stepping can have CPU lock free.

[death metal]

ok, i think i found the answer, from the same Asus site:
CPU compatible with CPU Lock Free

Prescott LGA775 CPU (Performance/04B)
800MHZ FSB, 3.4GHz, 3.6GHz, 3.8GHz, 4.0GHz+

My general impression was "all" 3.4Ghz+ *lol*...I didn't notice the 04B...

[Petr]

Now, assuming we compute for the "A" rating of the given 3.2 and 3.4...

3.4 (see link in my previous post) = 84w, max voltage is 1.4. Computing for amps, 60a.
3.2 (see link in my previous post) = 84w, max voltage is 1.4. Computing for amps, 60a.

See, they both fall in the same 60a category, and yet the 3.4 benefits from the CPU lock free and the 3.2 doesn't? Besides, in s478, 3.2 C0 stepping can have CPU lock free.

78A 3.4 GHz (E0) is not a performance class therefore it has not the option of 14x multiplier. Only 119A 3.4 GHz (D0) do but those are worse than 78A because they can be much hotter.

You can not divide TDP value by voltage because:
1. TDP value is NOT a maximum power. TDP is typical power under typical circumstances running typical applications and taking into account Thermal Monitor feature that will slow down the CPU when it runs too hot. Maximum thermal powers are 100W for 78A chips, 115W for 91A chips and 151W for 119A chips.
2. P4 uses VR-Down guidelines what means voltage under full load is significantly lower that in idle. A P4 with 1.4V VID (Voltage Identification) should be supplied about Vcc = 1.25V under full load (that depends on how many Amps are supplied, the more the lower Vcc). Some people think those fluctuatins are their motherboards can't handle their CPUs but that is wrong as this behaviour is normal and defined by Intel.

[DanIdentity]

2. P4 uses VR-Down guidelines what means voltage under full load is significantly lower that in idle. A P4 with 1.4V VID (Voltage Identification) should be supplied about Vcc = 1.25V under full load (that depends on how many Amps are supplied, the more the lower Vcc). Some people think those fluctuatins are their motherboards can't handle their CPUs but that is wrong as this behaviour is normal and defined by Intel.

Interesting. What is the purpose of supplying a lower voltage to the chip when it is under load vs idle?

[death metal]

Petr, I agree and know about what you posted. Again, my confusion lies on my "ingrained fact" that all 3.4Ghz are capable of CPU Lock free, while those below this speed have to qualify for something else (i.e. 3.2E C0). Upon looking at Asus' web site, it turned out that there is also qualifications for 3.4Ghz+. The example I used, discounting any other variable, is just to show that virtually there's no difference between the 3.2 and 3.4 flavor (in terms of amp rating). Here's a screenie:


In that screenie, there's two 3.4 P4 550 models ...

Dan,

Interesting. What is the purpose of supplying a lower voltage to the chip when it is under load vs idle?

to reduce heat output, and current draw, thereby reducing power loss and power consumption, and better accoustic for the stock hsf (i.e. more silent)...