One of the design goals of mTCP is to get very good performance without using a tremendous amount of memory for buffers or code.

I measure the performance of mTCP using two different methods. The first method is to use the mTCP FTP client which is a good real world test. FTP involves disk I/O so the results will vary greatly from machine to machine depending on the disk setup. Use these numbers to get an idea of what to expect on your hardware.

The second method is to use a test program that sends and receives data over a TCP/IP socket. The target of the test is a much faster machine that is also used to time the entire test. No disk I/O is done during the test and no processing is done on the data. This makes it a pure test of the TCP/IP code - the numbers from this test show the maximum that one can expect for performance.

All tests are conducted on a private network with minimal other traffic. All code is compiled the same way that you get it when you download mTCP.

Speedtest results
Machine	Adapter	Send Rate (KB/sec)	Recv Rate (KB/sec)	Code Date	OS	Notes
IBM PCjr (NEC V20@4.77Mhz)	Xircom PE3-10BT	42.0	34.5	2011-05-30	PC DOS 3.3	4MB transfer, bi-directional parallel port, no hardware IRQ
IBM PCjr (NEC V20@4.77Mhz)	Xircom PE3-10BT	41.3	39.0	2011-05-30	PC DOS 3.3	4MB transfer, bi-directional parallel port, IRQ 7
PCjr (4.77 NEC V20)	WD8003	108.7	110.8	2011-05-30	PC DOS 3.3	4MB transfer, ISA card running on a ISA bus adapter
PCjr (4.77 NEC V20)	NE1000	62.8	61.7			(not tested recently)
PCjr (4.77 NEC V20)	3Com 3C503	71.7	56.1			(not tested recently)
IBM PC XT (4.77 8088)	Xircom PE3-10BT	40.0	22.9	2011-05-30	PC DOS 3.3	2MB transfer, parallel port set to standard (nibble) mode, IRQ 7
IBM PC XT (4.77 8088)	NE1000	57.0	61.5	2011-05-30	PC DOS 5.02	2MB transfer
IBM PC XT (4.77 8088)	3Com 3C503	90.5	92.0	2011-05-30	PC DOS 3.3	2MB transfer
IBM PS/2 Model 25 (8086@8Mhz)	Xircom PE3-10BT	91.0	53.5	2011-05-30	PC DOS 6.1	8MB transfer, parallel port set to standard (nibble) mode
IBM PS/2 Model 25 (8086@8Mhz)	3Com 3C503	153	157	2011-05-30	PC DOS 6.1	8MB transfer
IBM L40SX (80386SX)	Xircom PE3-10BT	219	120	2011-05-30	PC DOS 6.3	8MB transfer
AMD 80386@40Mhz, 128K L2	NE2000	683	984	2011-05-30	PC DOS 5.02	32MB transfer
Pentium 133	LinkSys LNE100	4900	5913	2011-05-30	PC DOS 6.3	32MB transfer

Some fun observations:

• If you are using an adapter like the Xircom which attaches to the parallel port, having a bi-directional parallel port is very helpful. The IBM PC XT with the standard parallel port receives data at half of the rate that it sends data because of the standard parallel port. The PCjr with the same adapter sends and receives data at about the same rate.
• The PCjr doesn't have ISA slots. I tested the WD8003, NE1000 and the 3C503 on a special PCjr bus adapter. Otherwise, the same packet drivers and software are used.
• Sending is slower than receiving on the faster machines. This might be a flow control problem that I need to work on.
• The PS/2 machines have bi-directional parallel ports, but use a non-standard port address to flip the direction of the port. The Xircom drivers do not recognize this and thus they have to be run in nibble mode.