Only a few months have passed since "burst went bust": Intel's power hungry Pentium 4 NetBurst architecture left little doubt about AMD's superior efficiency in the desktop space, as the two battled for performance supremacy. At the same time, the Pentium M provided notebook users with some relief, Centrino became a household word, and manufacturers brought forth desktop and mini desktop motherboards to support this platform in the desktop space.
One big obstacle to desktop Pentium M adoption was pricing, as this combination of mobile components in the desktop space carried a price premium. That impediment vanished when Pentium M led to Core 2, and to lower-cost Core 2 Duo desktop processors.
It may look as if AMD missed its chance to "make hay while the sun shines", but a lack of marketing skills hasn't held back technological development. Higher quality yields that are normally reserved for mobile cores are now able to deliver sufficient quantities to bleed over into the desktop space. AMD sees new opportunities as shrinking enclosures demand smaller power supplies and coolers.
Thermal Design Power (TDP) refers to the maximum amount of processor heat a cooling system should be capable of dissipating, and therefore might indicate the maximum power the CPU should be expected to consume. But AMD groups its processors according to core type and voltage, fostering the odd notion that every speed requires the same wattage. (Note that the table below abbreviates Energy Efficient and Small form Factor as EE and SFF, respectively.)
|AMD Athlon 64 X2 Dual Core Processors|
|Processor||Model Number||Clock||L2-Cache||Die||Power (TDP)|
|4800+||ADA4800IAA6CS||2.4 GHz||1 MB + 1 MB||90 nm||89 W|
|4800+ EE||ADO4800IAA6CS||2.4 GHz||1 MB + 1 MB||90 nm||65W|
|4600+||ADA4600IAA5CU||2.4 GHz||512 kB + 512 kB||90 nm||89 W|
|4600+ EE||ADO4600IAA5CU||2.4 GHz||512 kB + 512 kB||90 nm||65W|
|4400+||ADA4400IAA6CS||2.2 GHz||1 MB + 1 MB||90 nm||89 W|
|4400+ EE||ADO4400IAA6CS||2.2 GHz||1 MB + 1 MB||90 nm||65W|
|4200+||ADA4200IAA5CU||2.2 GHz||512 kB + 512 kB||90 nm||89 W|
|4200+ EE||ADO4200IAA5CU||2.2 GHz||512 kB + 512 kB||90 nm||65W|
|4000+||ADA4000IAA6CS||2.0 GHz||1 MB + 1 MB||90 nm||89 W|
|4000+ EE||ADO4000IAA6CS||2.0 GHz||1 MB + 1 MB||90 nm||65W|
|3800+||ADA3800IAA5CU||2.0 GHz||512 kB + 512 kB||90 nm||89 W|
|3800+ EE||ADO3800IAA5CU||2.0 GHz||512 kB + 512 kB||90 nm||65W|
|3800+ EE SFF||ADD3800IAT5CU||2.0 GHz||512 kB + 512 kB||90 nm||35W|
The problem for anyone trying to calculate the actual power consumption of a system is that no two cores are exactly alike. For example, we know that increased clock rates raise temperatures even without a voltage change, so the 89 W Athlon 64 X2 3800+ will typically draw less power than the 89 W Athlon 64 X2 4800+.
The debate about theoretical power maximum versus actual power consumed brings with it another question: might the 89 W Athlon 64 X2 3800+ actually use less power than the 65 W Athlon 64 X2 4600+ Energy Efficient version? We compared them to the 35 W Athlon 64 X2 3800+ Energy Efficient Small Form Factor version, adding the Intel Core 2 Duo E6400 to broaden the field.