Content-Length: 12991 Content-Type: text/html; charset=UTF-8
|
AMD K6-2 Review(back to Reviews and Opinions) AMD K6-2 Review Overview The AMD K6-2 is the latest processor from AMD. It is the fastest alternative x86 processor available, and is very competitive with high-end Pentium II's. The K6-2 is built around the K6 (NexGen 686) core, giving it a strong base to perform well in all areas, including floating point. The core is based on what is known as "RISC86" technology, which is essentially a hybrid RISC/CISC design that is effective in providing the best of what both RISC and CISC have to offer. The main change in the K6-2 over the K6 is the addition of a 3D instruction set known as "3DNow!". 3DNow!, unlike MMX, is a set of instructions to speed up important floating point instructions. A K6-2 using a floating point-intensive application that can make use of 3DNow! can potentially outperform a Pentium II, even though the Pentium II's FPU is pipelined. This is because some of the most important floating point instructions do not take advantage of the FPU pipelining, and therefore the Pentium II's FPU has little advantage for them. Even though 3DNow! can potentially make a K6-2 do certain things faster than the Pentium II, the pipelined FPU of the Pentium II still beats the K6-2 in many areas. The main drawback to 3DNow! is that some programs may have to be reworked to take advantage of it, although many can benefit from 3DNow! via the use of Direct3D or OpenGL. The benefit of 3DNow! is very appearent according to some recent benchmarks, although AMD's claim to it being 30% faster than the equivilant Pentium II seems far fetched. Aside from 3DNow!, it is important to look at what performance benefit the K6-2 will provide in other areas. The integer performance of the K6-2 should be the same as the K6, simply because no modification has been made to anything that effects integer performance. The integer performance, however, will benefit from the higher clock speeds and use of higher bus speeds to achieve them. The K6-2 offers balanced performance like the K6 without the use of 3DNow!. It will definitely outperform anything Cyrix/IBM will release in the immediate future, and it should perform right along with the Pentium II. The K6-2, like the K6, does not use PRatings, rather MHz ratings. The K6-2 is manufactured using the .25 micron process, meaning it should run relatively cool and should be easy to overclock (often dependant on the motherboard used with the chip). The K6-2 will work in a Socket 7 motherboard, however to use it with the 100 MHz bus, it is recommended by AMD to use a Super Socket 7 motherboard. BIOS support should be very good on Super Socket 7 boards, but it is unknown how many standard Socket 7 boards will be able to correctly identify it. As a final note, please take into account that the K6-2 is what was formerly called the K6-3D. The real K6-3D, formerly known as the K6 3D+, will be released late this year. I will post information on the K6-3D shortly. Key Good Points
Key Bad Points
Architectural Specifications
Technical Specifications
My Opinion The K6-2 is an excellent chip. I have used Cyrix and IBM chips for several years, but this is definately the chip that is converting me. 3DNow! is a great idea, and support for the 100 MHz front-side bus is even better. Additional Resources AMD K6-2 Benchmarks AMD K6-2 FAQs |