All work well. The U-Link is the fastest (the J-Link Lite is not as fast as the J-Link full version which is equivalent to the U-Link). However, the U-Link has a potential problem that it is only really supported by Keil (it is not supported by other IDEs that I know). The JLink is more universal since it is supported by everyone.
OSJTAG is probably the slowest but still useable - it is based on a small processor rather than an ASIC which the more powerful ones will use.
For low budgets the Olimex ARM-USB-TINY can be looked at too.
I've posted measurements below into other topic, but it is more appropriate here.
As far as JTAG emulation speed I've measured the following using JTAGjet emulator on Kinetis K60 board.
Chameleon debugger was used to measure it, but similar performance may be achieved using drivers for other debuggers.
10MHz JTAG speed (default):
Write to ram : 522Kbytes/sec
Flashing 512KB: 5.2sec (includes erasing)
30MHz JTAG speed:
Write to ram : 892Kbytes/sec
Flashing 512KB: 4.8sec (includes erasing)
In addition to support for Cortex-M4, all iMX ARM-based processors are supported.
See here for list of supported CPUs: