To expose network characteristics by active/passive measurements, measuring some timing issues such as one-way delay, one-way queuing delay, and inter-packet time is essential, and is conducted by time-stamping for packets passing through an observation point. However, emerging high-speed networks require very high precision of time-stamping, far beyond the precision of conventional software-based time-stamping systems such as 'tcpdump'. For example, the inter-packet time of two consecutive 64-byte length packets on a giga-bit link can be less than 0.001 msec. In this paper, to demonstrate the usefulness and strong necessity of precise time-stamping on high-speed links, experiments of network measurements over a nation-wide IPv6 testbed in Japan have been performed, using a hardware-based time-stamping system that can synchronize to GPS with a high resolution of 0.0001 msec and within a small error of 0.0003 msec. In our experiments, several interesting results are seen, e.g., i) the distribution of one-way queuing delay exhibits a considerable difference depending on the size and the type (UDP/ICMP) of packets; ii) the minimal one-way delays for various sizes of UDP/ICMP packets give an accurate estimate of the transmission delay and the propagation delay; iii) the correlation between interpacket times at the sender and the receiver sides in a sequence of TCP ACK packets clearly shows the degree of ACK compression; iv) the inter-packet time in a UDP stream generated by a DV streaming application shows three dominant sending rates and a very rare peak rate, which might provide crucial information to bandwidth dimensioning; all of which would indicate the usefulness of precise time-stamping.