Abstract

Madge Networks commissioned The Tolly Group, as part of its 1994 Industry Benchmark for PCMCIA LAN performance testing, to evaluate the Smart 16/4 PCMCIA Ringnode with the main focus on measuring notebook Token Ring performance versus a desktop ISA-adapter baseline in both ODI and NDIS environments. The project examined how closely a mobile PCMCIA Token Ring adapter could approach desktop-class throughput under Novell NetWare and IBM OS/2 LAN Server conditions, using one-to-one client/server testing over 16Mbit/s Token Ring.  

The November 1994 Technology Spotlight explains that this work was part of Tolly’s first comprehensive performance study of Ethernet and Token Ring PCMCIA LAN adapters. The goal was to determine whether notebook users sacrificed meaningful LAN performance when moving from desktop systems to portable PCs. Tolly compared the Madge PCMCIA Token Ring adapter against a baseline established with ISA-bus client adapters and tested the product with both Open Data-link Interface and Network Driver Interface Specification drivers. The profile identifies support for 16Mbit/s and 4Mbit/s Token Ring over STP or UTP, with ODI driver MADGEODI.COM version 3.35 and NDIS driver SMARTND.DOS version 1.56.  

In the ODI suite shown on page 1, Madge measured throughput at 64, 256, 512, 1,024, 2,048, and 4,096-byte frame sizes with packet burst disabled, plus a maximum-frame packet-burst test. The chart shows Madge results of 0.6, 1.2, 1.7, 2.6, 3.8, and 5.5Mbit/s respectively, compared with baseline results of 0.9, 1.7, 2.4, 3.5, 5.5, and 8.7Mbit/s. With packet burst enabled at the 4,096-byte maximum frame size, Madge reached 7.5Mbit/s versus an 11.9Mbit/s baseline. Tolly states that the adapter was within 36.91% of baseline in that packet-burst case and within about 27% to 36% of baseline at the other tested frame sizes.  

In the NDIS suite shown on page 2, the adapter tracked more closely to baseline. Madge measured 0.5, 0.8, 1.3, 2.0, 2.9, and 6.4Mbit/s at 81, 256, 512, 1,024, 2,048, and 4,200-byte frame sizes respectively, compared with baseline results of 0.6, 0.9, 1.5, 2.3, 3.7, and 6.4Mbit/s. Tolly notes that at the maximum frame size with packet burst, the adapter was 0.34% faster than the baseline, and at other sizes it ran within roughly 9% to 20% of baseline. The test bed used a Toshiba T1910CS notebook with 12MB RAM, Madge card and socket services, NetWare 3.12 or IBM OS/2 LAN Server Advanced 3.0, and a 16Mbit/s Token Ring network instrumented with a Network General Expert Sniffer and Hewlett-Packard Series J2300 analyzer. Overall, the report presents the Madge Smart 16/4 PCMCIA Ringnode as a strong-performing notebook Token Ring adapter, especially in the NDIS environment where it essentially matched desktop performance at the maximum frame size.