Articles: CPU

Bookmark and Share

Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 ]

Temperature and Power Consumption

The technologies for a lower heat dissipation and power consumption are no less important features implemented in the E0 stepping of the Prescott core. Intel’s talking about two innovations in this respect: Enhanced Halt Mode C1E and Thermal Monitoring 2. These technologies differ from their predecessors in a bigger reduction of the heat dissipation which is achieved by reducing the voltage of the processor core. Particularly, when the CPU load is low and a Halt command is issued, the CPU voltage is dropped to 1.2v, which results in a greatly reduced idle heat dissipation. As for Thermal Monitoring 2, this technology drops the CPU frequency to 2.8GHz after the CPU temperature reaches a certain threshold value. Coupled with the voltage reduction, this produces a fast cooling effect.

But let’s check this out in practice. We assembled the following testbed for our tests:

  • ASUS P5AD2 Premium mainboard (LGA775, i925X Express);
  • 1024MB DDR2-533 SDRAM (OCZ PC2 4300, 2 x 512MB, 4-4-4-8);
  • Sapphire RADEON X800 XT graphics card (PCI-Express x16);
  • Maxtor MaXLine III 250GB hard disk drive (Serial ATA-150).

We used this testbed to check out the temperature mode of LGA755 Pentium 4 processors based on the Prescott core of D0 and E0 steppings. We selected frequencies of 2.8, 3.0, 3.2, 3.4 and 3.6GHz for both steppings of the core, and 3.8GHz for the E0 stepping. The FSB frequency and the core voltage were default. We used a boxed LGA755 cooler in our tests and measured the CPU temperature via the core-integrated sensor. We read the temperature data in two modes: Idle and Burn (the maximum load as created by the special-purpose S&M utility version 1.0.0 alpha).

As we see, the idle temperature of the processor with the new stepping is really lower. That’s the effect of the implementation of Enhanced Halt Mode C1E. As for the Burn temperature, it is similar between the D0 and E0 steppings of the Prescott core. The new core is slightly cooler than the older, but its temperature of 70°C under a load is still very high.

Besides the temperature, we measured the power consumption of the Prescott cores of D0 and E0 steppings by measuring the current in the 12v line the CPU is powered up through.

It’s all even clearer with the power consumption, which equals the heat dissipation, as the law of conservation of energy states. The idle power consumption of processors with the new E0 stepping has diminished in two times. This improvement, however, doesn’t allow it to compete with the Athlon 64 based on the 90nm Winchester core whose idle heat dissipation is less than 14 watts.

Under a load, an E0-stepping Prescott-core processor consumes less power than a D0-stepping processor of the same frequency, but the difference is negligible, as the main power-consumption improvements have been targeted at the idle mode.

This also makes clear why Intel refused to launch faster Pentium 4 processors with frequencies above 3.8GHz. The company’s engineers just couldn’t find a way to considerably decrease the processor’s power consumption under the maximum load in the third (E0) stepping of the Prescott core. A further frequency growth would require supply currents that go out of the FMB 2.0 requirements, and Intel just doesn’t dare to propose another FMB specification that would make necessary new mainboards and cooling systems. That’s why the company’s going to find other ways to improve the performance – those that wouldn’t entail revising the already formulated requirements to the heat dissipation and power consumption.

Thus, Intel’s refusal to release a 4.0GHz or higher-clocked Pentium 4 doesn’t mean the Prescott core has reached its frequency maximum. Of course, this is good news for overclockers who may find the following section of this review interesting.

Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 ]


Comments currently: 9
Discussion started: 12/04/04 01:19:09 PM
Latest comment: 01/04/05 12:07:12 AM

View comments

Add your Comment