A Large L2 Cache Isn't All Roses

By Harald Thon, published on December 16, 2003
Source: Tom's Guide US | Keywords: , , , , , | Themes: Business Notebooks

5. A Large L2 Cache Isn't All Roses

A glance at the theoretical maximum dissipation of both processors running at their top speeds merits a look:

On the Mobile Athlon64 3000+ it's three times larger than that measured on the Pentium-M with a clock speed of 1.7 GHz. At first glance, this fact may surprise since both processors work at almost identical power voltages (1.50 V/1.484 V) and similarly high core frequency speeds (1.8 GHz/1.7 GHz).

Consider the following formula:

This equation over simply describes the connection between a processor's clock speed, supply voltage and overall capacity and the resulting power consumption P to calculate dissipation.

If you consider that the Athlon64 has almost 106 million transistors clamoring for attention - 38% more than in the Pentium-M - it's obvious that where power demands are concerned, the Mobile Athlon64 is relegated to the rear. That's of course because the more transistors there are, the more gates the CPU has.

If we also recall that the L2 cache on the Pentium-M has intelligent switching, it's clear why the maximum possible dissipation on the Mobile Athlon64 is so much higher. On the Pentium-M, a mere 1/32 of the L2 cache is permanently active while the entire L2 cache on the Mobile Athlon64 has to be permanently powered.

Thus the Mobile Athlon64 doesn't really have what it takes to be a top energy-saver among mobile processors. Yet its large L2 cache, and what AMD says is improved branch prediction, should place it among the fastest processors. Only the results will confirm or disprove our hunches.

Comments | Print | Send to a friend

Sponsored links

Comments

Comments are closed on this page.

Sponsored links