NTPsec

A-ntpsec-12-hour-stats

Report generated: Thu Sep 29 13:08:39 2022 UTC
Start Time: Thu Sep 29 01:08:39 2022 UTC
End Time: Thu Sep 29 13:08:39 2022 UTC
Report published: Thu Sep 29 06:08:45 2022 PDT
Report Period: 0.5 days

Local Clock Time/Frequency Offsets

local offset plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Local Clock Time Offset -2.290 -1.426 -1.004 0.017 0.778 1.123 2.092 1.782 2.549 0.533 -0.020 µs -4.631 12.87
Local Clock Frequency Offset -56.503 -51.971 -45.059 -2.945 9.094 12.558 13.596 54.153 64.529 18.516 -10.994 ppb -9.568 28.61

The time and frequency offsets between the ntpd calculated time and the local system clock. Showing frequency offset (red, in parts per million, scale on right) and the time offset (blue, in μs, scale on left). Quick changes in time offset will lead to larger frequency offsets.

These are fields 3 (time) and 4 (frequency) from the loopstats log file.



Local RMS Time Jitter

local jitter plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Local RMS Time Jitter 0.263 0.373 0.436 0.678 1.012 1.186 1.397 0.576 0.813 0.176 0.694 µs 35.17 138.2

The RMS Jitter of the local clock offset. In other words, how fast the local clock offset is changing.

Lower is better. An ideal system would be a horizontal line at 0μs.

RMS jitter is field 5 in the loopstats log file.



Local RMS Frequency Jitter

local stability plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Local RMS Frequency Jitter 115.000 139.000 163.000 249.000 361.000 418.000 552.000 198.000 279.000 59.840 253.701 10e-12 44.3 184.3

The RMS Frequency Jitter (aka wander) of the local clock's frequency. In other words, how fast the local clock changes frequency.

Lower is better. An ideal clock would be a horizontal line at 0ppm.

RMS Frequency Jitter is field 6 in the loopstats log file.



Local Clock Time Offset Histogram

local offset histogram plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Local Clock Offset -2.290 -1.426 -1.004 0.017 0.778 1.123 2.092 1.782 2.549 0.533 -0.020 µs -4.631 12.87

The clock offsets of the local clock as a histogram.

The Local Clock Offset is field 3 from the loopstats log file.



Local Temperatures

local temps plot

Local temperatures. These will be site-specific depending upon what temperature sensors you collect data from. Temperature changes affect the local clock crystal frequency and stability. The math of how temperature changes frequency is complex, and also depends on crystal aging. So there is no easy way to correct for it in software. This is the single most important component of frequency drift.

The Local Temperatures are from field 3 from the tempstats log file.



Local Frequency/Temp

local freq temps plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Local Clock Frequency Offset -56.503 -51.971 -45.059 -2.945 9.094 12.558 13.596 54.153 64.529 18.516 -10.994 ppb -9.568 28.61
Temp ZONE0 51.540 51.540 52.078 53.692 53.692 54.230 54.230 1.614 2.690 0.665 53.211 °C

The frequency offsets and temperatures. Showing frequency offset (red, in parts per million, scale on right) and the temperatures.

These are field 4 (frequency) from the loopstats log file, and field 3 from the tempstats log file.



Local GPS

local gps plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
nSats 7.000 7.000 8.000 9.000 11.000 12.000 12.000 3.000 5.000 0.984 9.440 nSat 660.3 5900
TDOP 0.500 0.520 0.570 0.830 1.220 1.330 1.440 0.650 0.810 0.197 0.849 46.71 195.6

Local GPS. The Time Dilution of Precision (TDOP) is plotted in blue. The number of visible satellites (nSat) is plotted in red.

TDOP is field 3, and nSats is field 4, from the gpsd log file. The gpsd log file is created by the ntploggps program.

TDOP is a dimensionless error factor. Smaller numbers are better. TDOP ranges from 1 (ideal), 2 to 5 (good), to greater than 20 (poor). Some GNSS receivers report TDOP less than one which is theoretically impossible.



Server Offsets

peer offsets plot

The offset of all refclocks and servers. This can be useful to see if offset changes are happening in a single clock or all clocks together.

Clock Offset is field 5 in the peerstats log file.



Server Offset 169.229.128.134

peer offset 169.229.128.134 plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Server Offset 169.229.128.134 2.157 2.177 2.306 2.735 3.162 3.421 3.492 0.857 1.244 0.259 2.713 ms 874.2 8544

The offset of a server in seconds. This is useful to see how the measured offset is behaving.

The chart also plots offset±rtt, where rtt is the round trip time to the server. NTP can not really know the offset of a remote chimer, NTP computes it by subtracting rtt/2 from the offset. Plotting the offset±rtt reverses this calculation to more easily see the effects of rtt changes.

Closer to 0s is better. An ideal system would be a horizontal line at 0s. Typical 90% ranges may be: local LAN server 80µs; 90% ranges for WAN server may be 4ms and much larger.

Clock Offset is field 5 in the peerstats log file. The Round Trip Time (rtt) is field 6 in the peerstats log file.



Server Offset 173.11.101.155

peer offset 173.11.101.155 plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Server Offset 173.11.101.155 -1.856 -1.803 -1.307 -0.117 0.823 1.047 1.189 2.131 2.849 0.558 -0.144 ms -6.323 18.66

The offset of a server in seconds. This is useful to see how the measured offset is behaving.

The chart also plots offset±rtt, where rtt is the round trip time to the server. NTP can not really know the offset of a remote chimer, NTP computes it by subtracting rtt/2 from the offset. Plotting the offset±rtt reverses this calculation to more easily see the effects of rtt changes.

Closer to 0s is better. An ideal system would be a horizontal line at 0s. Typical 90% ranges may be: local LAN server 80µs; 90% ranges for WAN server may be 4ms and much larger.

Clock Offset is field 5 in the peerstats log file. The Round Trip Time (rtt) is field 6 in the peerstats log file.



Server Offset 192.168.1.11

peer offset 192.168.1.11 plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Server Offset 192.168.1.11 -311.507 -171.013 -110.208 104.413 308.831 349.362 470.277 419.039 520.375 118.566 105.959 µs -0.4764 3.113

The offset of a server in seconds. This is useful to see how the measured offset is behaving.

The chart also plots offset±rtt, where rtt is the round trip time to the server. NTP can not really know the offset of a remote chimer, NTP computes it by subtracting rtt/2 from the offset. Plotting the offset±rtt reverses this calculation to more easily see the effects of rtt changes.

Closer to 0s is better. An ideal system would be a horizontal line at 0s. Typical 90% ranges may be: local LAN server 80µs; 90% ranges for WAN server may be 4ms and much larger.

Clock Offset is field 5 in the peerstats log file. The Round Trip Time (rtt) is field 6 in the peerstats log file.



Server Offset 209.50.50.228

peer offset 209.50.50.228 plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Server Offset 209.50.50.228 -1.029 -0.917 -0.600 -0.141 3.797 3.982 4.372 4.397 4.899 1.812 0.930 ms -0.805 1.665

The offset of a server in seconds. This is useful to see how the measured offset is behaving.

The chart also plots offset±rtt, where rtt is the round trip time to the server. NTP can not really know the offset of a remote chimer, NTP computes it by subtracting rtt/2 from the offset. Plotting the offset±rtt reverses this calculation to more easily see the effects of rtt changes.

Closer to 0s is better. An ideal system would be a horizontal line at 0s. Typical 90% ranges may be: local LAN server 80µs; 90% ranges for WAN server may be 4ms and much larger.

Clock Offset is field 5 in the peerstats log file. The Round Trip Time (rtt) is field 6 in the peerstats log file.



Server Offset 216.218.254.202

peer offset 216.218.254.202 plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Server Offset 216.218.254.202 1.541 1.658 2.221 2.619 3.201 3.611 4.078 0.980 1.953 0.338 2.641 ms 336 2442

The offset of a server in seconds. This is useful to see how the measured offset is behaving.

The chart also plots offset±rtt, where rtt is the round trip time to the server. NTP can not really know the offset of a remote chimer, NTP computes it by subtracting rtt/2 from the offset. Plotting the offset±rtt reverses this calculation to more easily see the effects of rtt changes.

Closer to 0s is better. An ideal system would be a horizontal line at 0s. Typical 90% ranges may be: local LAN server 80µs; 90% ranges for WAN server may be 4ms and much larger.

Clock Offset is field 5 in the peerstats log file. The Round Trip Time (rtt) is field 6 in the peerstats log file.



Server Offset 66.220.9.122

peer offset 66.220.9.122 plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Server Offset 66.220.9.122 1.715 1.831 2.079 2.499 3.037 3.395 3.529 0.958 1.564 0.290 2.500 ms 468.7 3772

The offset of a server in seconds. This is useful to see how the measured offset is behaving.

The chart also plots offset±rtt, where rtt is the round trip time to the server. NTP can not really know the offset of a remote chimer, NTP computes it by subtracting rtt/2 from the offset. Plotting the offset±rtt reverses this calculation to more easily see the effects of rtt changes.

Closer to 0s is better. An ideal system would be a horizontal line at 0s. Typical 90% ranges may be: local LAN server 80µs; 90% ranges for WAN server may be 4ms and much larger.

Clock Offset is field 5 in the peerstats log file. The Round Trip Time (rtt) is field 6 in the peerstats log file.



Server Offset SHM(0)

peer offset SHM(0) plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Server Offset SHM(0) -134.641 -133.889 -132.526 -129.984 -128.390 -127.981 -126.841 4.136 5.908 1.247 -130.117 ms -1.168e+06 1.231e+08

The offset of a server in seconds. This is useful to see how the measured offset is behaving.

The chart also plots offset±rtt, where rtt is the round trip time to the server. NTP can not really know the offset of a remote chimer, NTP computes it by subtracting rtt/2 from the offset. Plotting the offset±rtt reverses this calculation to more easily see the effects of rtt changes.

Closer to 0s is better. An ideal system would be a horizontal line at 0s. Typical 90% ranges may be: local LAN server 80µs; 90% ranges for WAN server may be 4ms and much larger.

Clock Offset is field 5 in the peerstats log file. The Round Trip Time (rtt) is field 6 in the peerstats log file.



Server Offset SHM(1)

peer offset SHM(1) plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Server Offset SHM(1) -2.291 -1.427 -1.005 0.018 0.779 1.124 2.093 1.784 2.551 0.533 -0.020 µs -4.629 12.86

The offset of a server in seconds. This is useful to see how the measured offset is behaving.

The chart also plots offset±rtt, where rtt is the round trip time to the server. NTP can not really know the offset of a remote chimer, NTP computes it by subtracting rtt/2 from the offset. Plotting the offset±rtt reverses this calculation to more easily see the effects of rtt changes.

Closer to 0s is better. An ideal system would be a horizontal line at 0s. Typical 90% ranges may be: local LAN server 80µs; 90% ranges for WAN server may be 4ms and much larger.

Clock Offset is field 5 in the peerstats log file. The Round Trip Time (rtt) is field 6 in the peerstats log file.



Server Jitters

peer jitters plot

The RMS Jitter of all refclocks and servers. Jitter is the current estimated dispersion, in other words the variation in offset between samples.

Closer to 0s is better. An ideal system would be a horizontal line at 0s.

RMS Jitter is field 8 in the peerstats log file.



Server Jitter 169.229.128.134

peer jitter 169.229.128.134 plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Server Jitter 169.229.128.134 0.163 0.195 0.354 1.236 12.441 15.569 17.588 12.086 15.374 3.979 3.211 ms 1.12 3.828

The RMS Jitter of a server. Jitter is the current estimated dispersion, in other words the variation in offset between samples.

Closer to 0s is better. An ideal system would be a horizontal line at 0s.

RMS Jitter is field 8 in the peerstats log file.



Server Jitter 173.11.101.155

peer jitter 173.11.101.155 plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Server Jitter 173.11.101.155 0.350 0.473 0.963 1.942 13.614 26.777 26.787 12.651 26.304 5.450 4.465 ms 1.823 7.02

The RMS Jitter of a server. Jitter is the current estimated dispersion, in other words the variation in offset between samples.

Closer to 0s is better. An ideal system would be a horizontal line at 0s.

RMS Jitter is field 8 in the peerstats log file.



Server Jitter 192.168.1.11

peer jitter 192.168.1.11 plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Server Jitter 192.168.1.11 0.069 0.090 0.130 0.299 8.035 14.423 20.404 7.905 14.333 2.791 1.163 ms 1.651 9.924

The RMS Jitter of a server. Jitter is the current estimated dispersion, in other words the variation in offset between samples.

Closer to 0s is better. An ideal system would be a horizontal line at 0s.

RMS Jitter is field 8 in the peerstats log file.



Server Jitter 209.50.50.228

peer jitter 209.50.50.228 plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Server Jitter 209.50.50.228 0.277 0.297 0.406 1.176 14.942 21.086 21.340 14.536 20.789 5.101 3.847 ms 0.881 3.437

The RMS Jitter of a server. Jitter is the current estimated dispersion, in other words the variation in offset between samples.

Closer to 0s is better. An ideal system would be a horizontal line at 0s.

RMS Jitter is field 8 in the peerstats log file.



Server Jitter 216.218.254.202

peer jitter 216.218.254.202 plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Server Jitter 216.218.254.202 0.233 0.233 0.350 1.232 14.978 21.664 22.296 14.628 21.431 5.178 3.849 ms 0.9645 3.779

The RMS Jitter of a server. Jitter is the current estimated dispersion, in other words the variation in offset between samples.

Closer to 0s is better. An ideal system would be a horizontal line at 0s.

RMS Jitter is field 8 in the peerstats log file.



Server Jitter 66.220.9.122

peer jitter 66.220.9.122 plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Server Jitter 66.220.9.122 0.186 0.213 0.357 1.178 16.196 25.163 25.227 15.840 24.949 5.614 4.057 ms 1.117 4.618

The RMS Jitter of a server. Jitter is the current estimated dispersion, in other words the variation in offset between samples.

Closer to 0s is better. An ideal system would be a horizontal line at 0s.

RMS Jitter is field 8 in the peerstats log file.



Server Jitter SHM(0)

peer jitter SHM(0) plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Server Jitter SHM(0) 0.036 0.066 0.101 0.279 0.920 1.314 2.399 0.819 1.248 0.272 0.367 ms 2.864 10.39

The RMS Jitter of a server. Jitter is the current estimated dispersion, in other words the variation in offset between samples.

Closer to 0s is better. An ideal system would be a horizontal line at 0s.

RMS Jitter is field 8 in the peerstats log file.



Server Jitter SHM(1)

peer jitter SHM(1) plot

Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Server Jitter SHM(1) 0.105 0.218 0.303 0.643 1.276 1.665 2.743 0.973 1.447 0.310 0.700 µs 7.025 23.89

The RMS Jitter of a server. Jitter is the current estimated dispersion, in other words the variation in offset between samples.

Closer to 0s is better. An ideal system would be a horizontal line at 0s.

RMS Jitter is field 8 in the peerstats log file.



Summary


Percentiles...... Ranges...... Skew- Kurt-
Name Min1%5%50%95% 99%Max   90%98%StdDev  MeanUnits nessosis
Local Clock Frequency Offset -56.503 -51.971 -45.059 -2.945 9.094 12.558 13.596 54.153 64.529 18.516 -10.994 ppb -9.568 28.61
Local Clock Time Offset -2.290 -1.426 -1.004 0.017 0.778 1.123 2.092 1.782 2.549 0.533 -0.020 µs -4.631 12.87
Local RMS Frequency Jitter 115.000 139.000 163.000 249.000 361.000 418.000 552.000 198.000 279.000 59.840 253.701 10e-12 44.3 184.3
Local RMS Time Jitter 0.263 0.373 0.436 0.678 1.012 1.186 1.397 0.576 0.813 0.176 0.694 µs 35.17 138.2
Server Jitter 169.229.128.134 0.163 0.195 0.354 1.236 12.441 15.569 17.588 12.086 15.374 3.979 3.211 ms 1.12 3.828
Server Jitter 173.11.101.155 0.350 0.473 0.963 1.942 13.614 26.777 26.787 12.651 26.304 5.450 4.465 ms 1.823 7.02
Server Jitter 192.168.1.11 0.069 0.090 0.130 0.299 8.035 14.423 20.404 7.905 14.333 2.791 1.163 ms 1.651 9.924
Server Jitter 209.50.50.228 0.277 0.297 0.406 1.176 14.942 21.086 21.340 14.536 20.789 5.101 3.847 ms 0.881 3.437
Server Jitter 216.218.254.202 0.233 0.233 0.350 1.232 14.978 21.664 22.296 14.628 21.431 5.178 3.849 ms 0.9645 3.779
Server Jitter 66.220.9.122 0.186 0.213 0.357 1.178 16.196 25.163 25.227 15.840 24.949 5.614 4.057 ms 1.117 4.618
Server Jitter SHM(0) 0.036 0.066 0.101 0.279 0.920 1.314 2.399 0.819 1.248 0.272 0.367 ms 2.864 10.39
Server Jitter SHM(1) 0.105 0.218 0.303 0.643 1.276 1.665 2.743 0.973 1.447 0.310 0.700 µs 7.025 23.89
Server Offset 169.229.128.134 2.157 2.177 2.306 2.735 3.162 3.421 3.492 0.857 1.244 0.259 2.713 ms 874.2 8544
Server Offset 173.11.101.155 -1.856 -1.803 -1.307 -0.117 0.823 1.047 1.189 2.131 2.849 0.558 -0.144 ms -6.323 18.66
Server Offset 192.168.1.11 -311.507 -171.013 -110.208 104.413 308.831 349.362 470.277 419.039 520.375 118.566 105.959 µs -0.4764 3.113
Server Offset 209.50.50.228 -1.029 -0.917 -0.600 -0.141 3.797 3.982 4.372 4.397 4.899 1.812 0.930 ms -0.805 1.665
Server Offset 216.218.254.202 1.541 1.658 2.221 2.619 3.201 3.611 4.078 0.980 1.953 0.338 2.641 ms 336 2442
Server Offset 66.220.9.122 1.715 1.831 2.079 2.499 3.037 3.395 3.529 0.958 1.564 0.290 2.500 ms 468.7 3772
Server Offset SHM(0) -134.641 -133.889 -132.526 -129.984 -128.390 -127.981 -126.841 4.136 5.908 1.247 -130.117 ms -1.168e+06 1.231e+08
Server Offset SHM(1) -2.291 -1.427 -1.005 0.018 0.779 1.124 2.093 1.784 2.551 0.533 -0.020 µs -4.629 12.86
TDOP 0.500 0.520 0.570 0.830 1.220 1.330 1.440 0.650 0.810 0.197 0.849 46.71 195.6
Temp ZONE0 51.540 51.540 52.078 53.692 53.692 54.230 54.230 1.614 2.690 0.665 53.211 °C
nSats 7.000 7.000 8.000 9.000 11.000 12.000 12.000 3.000 5.000 0.984 9.440 nSat 660.3 5900
Summary as CSV file


Glossary:

frequency offset:
The difference between the ntpd calculated frequency and the local system clock frequency (usually in parts per million, ppm)
jitter, dispersion:
The short term change in a value. NTP measures Local Time Jitter, Refclock Jitter, and Server Jitter in seconds. Local Frequency Jitter is in ppm or ppb.
kurtosis, Kurt:
The kurtosis of a random variable X is the fourth standardized moment and is a dimension-less ratio. ntpviz uses the Pearson's moment coefficient of kurtosis. A normal distribution has a kurtosis of three. NIST describes a kurtosis over three as "heavy tailed" and one under three as "light tailed".
ms, millisecond:
One thousandth of a second = 0.001 seconds, 1e-3 seconds
mu, mean:
The arithmetic mean: the sum of all the values divided by the number of values. The formula for mu is: "mu = (∑xi) / N". Where xi denotes the data points and N is the number of data points.
ns, nanosecond:
One billionth of a second, also one thousandth of a microsecond, 0.000000001 seconds and 1e-9 seconds.
percentile:
The value below which a given percentage of values fall.
ppb, parts per billion:
Ratio between two values. These following are all the same: 1 ppb, one in one billion, 1/1,000,000,000, 0.000,000,001, 1e-9 and 0.000,000,1%
ppm, parts per million:
Ratio between two values. These following are all the same: 1 ppm, one in one million, 1/1,000,000, 0.000,001, and 0.000,1%
‰, parts per thousand:
Ratio between two values. These following are all the same: 1 ‰. one in one thousand, 1/1,000, 0.001, and 0.1%
refclock:
Reference clock, a local GPS module or other local source of time.
remote clock:
Any clock reached over the network, LAN or WAN. Also called a peer or server.
time offset:
The difference between the ntpd calculated time and the local system clock's time. Also called phase offset.
σ, sigma:
Sigma denotes the standard deviation (SD) and is centered on the arithmetic mean of the data set. The SD is simply the square root of the variance of the data set. Two sigma is simply twice the standard deviation. Three sigma is three times sigma. Smaller is better.
The formula for sigma is: "σ = √[ ∑(xi-mu)^2 / N ]". Where xi denotes the data points and N is the number of data points.
skewness, Skew:
The skewness of a random variable X is the third standardized moment and is a dimension-less ratio. ntpviz uses the Pearson's moment coefficient of skewness. Wikipedia describes it best: "The qualitative interpretation of the skew is complicated and unintuitive."
A normal distribution has a skewness of zero.
upstream clock:
Any server or reference clock used as a source of time.
µs, us, microsecond:
One millionth of a second, also one thousandth of a millisecond, 0.000,001 seconds, and 1e-6 seconds.



This page autogenerated by ntpviz, part of the NTPsec project
html 5    Valid CSS!