Liquid3D | 22nd August 2004 03:29 | Sorry, I think my post above was misintepreted. There's absolutely no motherboard power circuitry issues with recent 90nm LGA Prescott on the Socket-T 915/925 boards. The original issue existed between the Socket-478 motherboard, which was designed for a completely different processor architecture, and was therefore unable to meet Socket-478 Prescott power requirements. In fact the Socket-478 Prescott is at best a misnomer, and at least a contradiction. While Socket-478 Prescott was based on lower voltages, it required higher current (e.g higher wattage), and presented an entirely different load to naive Socket-478 power circuitry. This is why in some early Socket-478/ Prescott overclock's, motherboards actually melted from power circuitry temps as high as 93.3C/200F.
It was these issues which originally prompted MSI to lock-out BIOS Vcore adjusts on their "Prescot ready" Socket-478 motherboards. Here's MSI's words in a discussion with [H], about this decision; Quote:
The 865PE Neo2-P Platinum Edition is designed to support the Intel(r) next generation Pentium 4(r) processor: Prescott. However, owing to the different manufacturing process and power management, we've made some voltage limits in the BIOS to protect the Prescott P4 CPU from burning out. Meanwhile, you'll also find that there is no CPU voltage settings to adjust. This is also for system protection.
| While the first half of this statement seems to simultaneosly place blame, under the guise of protecting 478-Prescott for their "voltage limits" it's the latter part of the paragraph which is telling, "This is also for system (read: motherboard power circuits) protection." In another statement between [H] and MSI found in this article, they delve into the D-VID issue discussed in my article here at [M]. Quote:
I discussed the Vcore change question with the board designer over last weekend. Base on my understanding, the reason we didn't offer that when Prescott is populated was because a new Intel VID design on Prescott: "Dynamic VID" (VID becomes variable on different stages). According to the board designer, if M/B (BIOS) overwrite the Vcore, Dynamic VID will become invalid which might damage the CPU (Prescott Core) when power stage changed.
| .
Now, if you've recently purchased an LGA Prescott and Socket-T motherboard and are having overclocking problems, the issue is not related to the motherboard's power circuitry. At least not in-so-far as a design oversight. What Intel has done to prevent overclocking beyond a certain percentage, is to limit voltage to the 915/925 North Bridge. In just about all the LGA Socket-T overclocking scenario's you'll find almost everyone is limited to about the same 227FSB (1:1) maximum. Any excessive heat generated is merely the end-result of frustrated Overclocker's pushing too much Vcore into their CPU's.in the false belief thisis the obsacle tey must overcoe. Nothing could be further from the truth. This is because the same D-VID power circuitry found ion the 90nm Socket-478 Prescott is found in it's younger sister 90nm LGA Prescott. Ergo, when pushing excessive Vcore into these processors they become overheated. First; as the prima facie result of added current. Second; as D-VID begins to malfunction. The trick to acheiving a decent overcock on all LGA Socket-T Prescott platforms, is increasing the voltage to the North Bridge, and actually leaving the Vcore at Deafult or Auto. No other Vcore setting will suffice, because only on the Default or Auto settings in the BIOS is the D-VID feature allowed to operate the way it was intended. Even Manual adjustment to the Vcore where it's set to the minumum required Vcore will essentially Lock-out the D-VID feature, causing power ciruitry conflicts. I found this to be true time and time again, even with my Socket-478 Prescott, and Asus P4C800E-Deluxe, or Abit AI7 combo's. My 478-Prescott 3.0E easily attains 250FSB (1:1)(15x250FSB=3750MHz) on Auto but will struggle at higher Vcore at same speeds. At one time this wouldn't occur in any other scenario, however; even AMD now implements "Cool n Quiet" technology, utilizing such power saving/thermal throttling features. You'll also notice Intel has prevented most (perhaps all) OEM mobo manufacturers from allowing chipset voltage adjustments in their 915/925-chipset BIOS. This is because down the line when DDR-2 attains another speed increase, they'll want to introduce another chipset version such as the 915X or 925X-Pro for example. |