Abstract

Proteon commissioned Tolly to evaluate the ProNET 4/16 Plus Token Ring Adapters, with the main focus on comparing competitive Token Ring adapter performance across EISA, ISA, and MCA bus architectures in both server and end-station roles. The project measured maximum throughput and server CPU utilization against leading offerings from IBM, Madge, and Olicom, using NetWare 3.11 and ODI drivers at 16Mbit/s Token Ring speeds.  

The February 1994 report explains that the ProNET 4/16 Plus family was tested on adapters designed for Extended Industry Standard Architecture, Industry Standard Architecture, and Micro Channel Architecture platforms. Proteon’s products were compared with IBM Busmaster EISA, Adapter II, and LANStreamer adapters; Madge Smart Ringnode adapters for EISA, ISA, and MCA; and Olicom 16/4 and 16/4 Server adapters. Tolly states that the adapters were chosen because of their competitive performance ratings across the industry. Page 3 notes that the Proteon family supported EISA, ISA, and MCA buses, ODI and NDIS driver specifications, 16Mbit/s and 4Mbit/s ring speeds, STP and UTP media, and used Texas Instruments TMS380 and Packet Blaster technology with Proteon’s RapidDriver.  

Testing used Novell’s Perform3 application with file sizes of 7, 967, and 4,039 bytes, yielding frame sizes of 64, 1,024, and 4,096 bytes. For server throughput, up to 12 NetWare DOS Requester workstations were attached to maximize traffic through the server adapter, while single station-to-server transfers were used for end-station tests. CPU utilization was taken from the server’s MONITOR.NLM utility. All tests ran over a 16Mbit/s Token Ring built with an IBM 8228 multi-access unit and 10 feet of shielded twisted-pair cabling.  

Results showed that the ProNET series performed competitively, especially in EISA server configurations. The throughput table on page 4 shows Proteon’s EISA server adapter delivering 3.61Mbit/s at 64-byte frames, 14.03Mbit/s at 1,024 bytes, and 15.23Mbit/s at 4,096 bytes. Tolly notes on page 2 that the EISA card was within marginal percentages of the top performer at all packet sizes and had the lowest CPU utilization of the adapters tested. The corresponding CPU table shows Proteon’s EISA server utilization at 94%, 31%, and 12% versus IBM at 72%, 55%, and 31%, Madge at 68%, 32%, and 38%, and Olicom at 79%, 34%, and 20%. For ISA end-station throughput, Proteon reached 0.36Mbit/s at 64 bytes, 3.19Mbit/s at 1,024 bytes, and 6.26Mbit/s at 4,096 bytes, making it the second-fastest in that category, while its large-packet CPU efficiency remained the best of the tested adapters. MCA Proteon results were similar, at 0.48Mbit/s, 3.79Mbit/s, and 6.74Mbit/s.  

Overall, the report presents the Proteon ProNET 4/16 Plus family as a strong Token Ring adapter line that combined high throughput with particularly efficient CPU utilization, making the EISA version well suited for server environments and the ISA and MCA versions solid choices for end-station deployments.  

Solutions tested:

• Proteon ProNET 4/16 Plus — Token Ring adapter family tested across EISA, ISA, and MCA bus architectures in server and end-station configurations.  
• IBM Busmaster EISA / Adapter II / LANStreamer — IBM Token Ring adapters used as competitive comparison products for EISA, ISA, and MCA platforms.  
• Madge Smart Ringnode / AT Smart Ringnode — Madge Token Ring adapters included in the comparative performance testing across multiple bus types.  
• Olicom 16/4 Adapter / 16/4 Server Adapter — Olicom Token Ring adapters tested as competing solutions in the benchmark.  

This document was originally published under the InterLAB brand and republished when InterLAB became The Tolly Group in 1994.

(Ed. note. Proteon was the first vendor to bring commercial Token Ring technology to market.)

[OCR needed.]