Advanced Micro Devices Inc (AMD)

[j3pflynn]

re: VID re-post for spokeshave.

"YB: Elmer and wbmw are wrong...

...regarding VID. It is all in the Intel datasheet if one takes the time to read it thoroughly. I have corrected them numerous times before, yet they continue to stubbornly stick to their misconceptions.

I will try to clear it up. The VID is first defined in section 2.2, which states:

The processor VCC pins must be supplied with the voltage defined by the VID (Voltage ID) pins and the loadline
specifications (see Figure 2-4).

Note the reference to the loadline specifications in Figure 2-4. This will become very important later. However, this statement is pretty clear - the processor Vcc pins must be supplied with the voltage defined by the VID pins. For VID 1.550, the VID pins define a voltage of 1.550 volts. That is pretty straight-forward. But it becomes even clearer as one reads further. Next let's look at section 2.4 - Voltage Identification. In particular:

The voltage set by the VID pins is the maximum voltage allowed by the processor. A minimum voltage is provided in
Table 2-6 and changes with frequency.

Clearly, the VID is the maximum voltage allowedby the processor. This is a bit misleading, though, as will become apparent shortly. First though, just in case there is still doubt about what the maximum voltage might be, let's look at table 2-2. This table shows the Vcc_max for the different VID pins. Clearly, the Vcc_max is 1.550V. There it is in plain black-and-white. It does not say VID_max, it says Vcc_max, in other words, the max voltage for the core.

Now, let's move on the table 2-6. This is the source of all of the confusion. I understand the confusion because in my opinion, Intel is intentionally attempting to obfuscate the maximum parameters of the CPU. So, on to the table. First, note the contradiction. The maximum Vcc found anywhere in the table is 1.425V. This directly contradicts the information referenced in table 2-2 where the Vcc_max is clearly defined as 1.550V. So where does the discrepency come from? Easy. Read the footnotes. In particular, footnote 1, which states:

These voltages are targets only. A variable voltage source should exist on systems in the event that a different voltage is required. See Table 2-2 for more information. The VID bits will set the maximum VCC with the minimum being defined according to current consumption at that voltage.

Ahhhh. There's the rub. The values in table 2-6 are not minimums and maximums (though that is clearly what we are intended to believe). They are targets only. Well. That clears up a few things. Interstingly, after saying that the "maximum" voltages listed are really only "targets", the footnote goes on to reference table 2-2. Recall what is in table 2-2. That's right. The Vcc_max value of 1.550. So, in summary, the footnote clearly says that the "maximum" values in the table are not "maximum" at all but instead are "targets" and if you want the real "maximum" you should see table 2-2. I hope that settles that.

But the real meat and potatoes are in table 2-7 and figure 2-4. These are the "real deal" and are referenced throughout the rest of the document. These are the "load line" data. They represent the true nominal, maximum and minimum operating parameters. Once can see from the load line that the Vcc_max (Vcc_max = VID, remember) is practically never achieved. It can only be approached at very low currents, and under non-typical conditions. So, in fairness, the Vcc_max is only approached or realized at very low currents.

Now, let's look at maximum power which is derived from Vcc_max and Icc_max. Once again, I am rejecting the information in table 2-6 since the footnotes clearly state that those values are "targets". Instead, I will follow the footnotes and refer to table 2-7 and figure 2-4 for actual maximum operating parameters. In that table, and on the load line, the Icc_max is clearly listed as 70A, and the Vcc_max can be gleaned from the table to be 1.550 - 0.133, or 1.417V at that current. Also note that the Vcc decreases as Icc increases. So, for a Vcc_max of 1.417V at an Icc_max of 70A, the maximum power is 99.19W. Now, I recognize that this is not likely to be seen except for the worst of circumstances, but if we are to speak of maximum power, this is the figure that should appropriately be used.

ftp://download.intel.com/design/Pentium4/datashts/29864309.pdf

I believe that Elmer and wbmw still have me on ignore. Can someone repost this so they can see it?"