Fast Hashing on the Pentium

Abstract

With the advent of the Pentium processor parallelization finally bccarne available to Intel based computer systems. One of the design principles of the MD4-family of hash functions (MD4, MD5, SHA-1, FLIPEMD-160) is to be fast on the 32-bit Intel processors. This paper shows that carefully coded implementations of these hash functions are able to exploit the Pentium's superscalar architecture to its maximum effect: the performance with respect to execution on a non-parallel architecture increases by about 60%. This is an important result in view of the recent claims on the limited data bandwidth of these hash functions. Moreover, it is conjectured that these implementations are very close to optimal. It will also be shown that t,he performance penalty incurred by non-cached data and endianness conversion is limited, and in the order of 10% of running time.

Selecting Cryptographic Key Sizes

“…Software data points. In [4] 241, 345, 837, and 1016 Pentium cycles are reported for MD4, MD5, SHA-1, and RIPEMD-160, respectively. This compares with 360-500 cycles for the DES depending on fixed or variable keys, as reported in [2] and [8] (see 2.2.4).…”

IDEA: A Cipher for Multimedia Architectures?

“…In our implementation, 693 instructions are paired into 358 cycles. Excellent pairing (0.517 cycles per instruction) is a little miracle (cf to > 0.56 cycles per instruction got by Bosselaers when implementing hash functions for the Pentium, [Bos97]) and is definitely one of the sources of the effectiveness of our implementation.…”

“…We shall not present a detailed outline of Intel Pentium's architecture (an interested reader may turn to [Int97b] or [BGV96]). …”

Authenticated encryption: how reordering can impact performance

“…The use of universal hash-function families to construct MAC algorithms is due to Carter and Wegman [23,24]. Compared to blockcipher based MACs, such as [9,18,29,41], and cryptographic hash function based MACs, such as [8,21,58,65], universal hashing based MACs lead to faster message authentication [17,38,47,59]. The security of MACs based on universal hashing has been extensively studied (see, e.g., [4,5,39] The speed of a universal hash family based MAC relies mainly on the speed of the used universal hash family.…”