Date: Fri, 31 Jul 1998 12:53:12 -0600 (MDT) From: chanders@timing.com (Craig Anderson) To: freebsd-hackers@FreeBSD.ORG Subject: Using FreeBSD for data acquisition? (long) Message-ID: <199807311853.MAA04187@count.timing.com>
next in thread | raw e-mail | index | archive | help
My company is considering using FreeBSD for a data acquisition
application. We want to know if an upper bound be placed on the
latency of an application reading data from a data acquisition
board? FreeBSD is being considered first and then commercial RTOS's
will be tried.
A test project is being done. The project uses an ISA data
acquisition board. The board interrupts 500 times a second, an 8
byte data block is read from the board on each interrupt.
There is a driver lkm for the board: a) driver buffers 4k data
blocks. b) driver interrupt function reads 1 data block, stores
data block in buffer, wakes the reader. c) driver read function
waits for at least 1 data block, returns minimum of requested data
blocks and available data blocks.
A test application: a) Loops forever requesting 4096 data blocks.
b) Saves the number of data blocks returned on each read to a file.
Because the interrupts/data blocks arrive at 500 HZ, I am interpreting
the number of data blocks returned on a read as a measure of read
latency of the application. If the read returns 50 data blocks, it
has been .1 second since the last read. (50/500 = .1)
Below are code fragments from the test application. The rtprio()
call is to put the process into RTP_PRIO_REALTIME. Should this
make a difference? Is the code correct?
At the end of the message is histogram data for a test run of about
15 hours. I plot the data with gnuplot, using a logscale for y.
Note that the longest read is 480 (480/500 = .960 seconds). Note
that the most common read is 50 (50/500 = .100 seconds).
I'd like comments from FreeBSD hackers out there. Can an upper
bound be put on the read latency? An upper bound of about .100 seconds
is probably acceptable. Why is a read latency of .100 seconds (50
data blocks) so common? Should RTP_PRIO_REALTIME help? Does FreeBSD
3 have realtime features that will help? Is there a problem with
the code or a better approach?
Thanks in advance for any comments.
Code from test application
--------------------------
rtp.type = RTP_PRIO_REALTIME;
rtp.prio = 0;
istat = rtprio (RTP_SET,
0,
&rtp);
...
fd = open (DEV_PATH, O_RDONLY, 0);
...
istat = ioctl(fd,GT401_START_EVENTS,&arg_ioctl);
...
for (;;)
{
istat = read (fd,(void *)&(ttag_buf[0]),read_req);
read_data_count = istat / sizeof(gt401_read);
printf ("%d\n",read_data_count);
fflush(stdout);
} /* for (;;) */
Histogram: number of data blocks returned on a read
---------------------------------------------------
col1=number of data blocks returned by a read
col2=number of reads that returned this many data blocks
# 1 compute bound, slow interrupt
# Thu Jul 30 17:47:58 MDT 1998
# period=0.002000 freq=500.000000 RTPRIO
# max count=480 total count=28303071
1.000000 134048
2.000000 27911
3.000000 4324
4.000000 2003
5.000000 5564
6.000000 4066
7.000000 3911
8.000000 2695
9.000000 2714
10.000000 5091
11.000000 2536
12.000000 2016
13.000000 298
14.000000 657
15.000000 3808
16.000000 1651
17.000000 1489
18.000000 62
19.000000 7897
20.000000 49226
21.000000 1406
22.000000 1432
23.000000 47
24.000000 404
25.000000 3190
26.000000 1537
27.000000 1346
28.000000 64
29.000000 5054
30.000000 51876
31.000000 1413
32.000000 1254
33.000000 50
34.000000 327
35.000000 2890
36.000000 1454
37.000000 1230
38.000000 44
39.000000 367
40.000000 2960
41.000000 1469
42.000000 1312
43.000000 29
44.000000 397
45.000000 2821
46.000000 1570
47.000000 1123
48.000000 184
49.000000 55950
50.000000 412800
51.000000 5
52.000000 6
53.000000 0
54.000000 2
55.000000 1
56.000000 0
57.000000 0
58.000000 2
59.000000 0
60.000000 1
61.000000 3
62.000000 2
63.000000 6
64.000000 4
65.000000 8
66.000000 19
67.000000 22
68.000000 26
69.000000 29
70.000000 8
71.000000 0
72.000000 1
73.000000 0
74.000000 1
75.000000 0
76.000000 1
77.000000 0
78.000000 1
79.000000 2
80.000000 18
81.000000 12
82.000000 25
83.000000 23
84.000000 0
85.000000 1
86.000000 0
87.000000 0
88.000000 0
89.000000 1
90.000000 0
91.000000 0
92.000000 0
93.000000 0
94.000000 0
95.000000 2
96.000000 2
97.000000 2
98.000000 0
99.000000 0
100.000000 6
101.000000 1
102.000000 0
103.000000 0
104.000000 1
105.000000 1
106.000000 1
107.000000 0
108.000000 0
109.000000 1
110.000000 0
111.000000 0
112.000000 0
113.000000 4
114.000000 5
115.000000 7
116.000000 16
117.000000 24
118.000000 28
119.000000 26
120.000000 11
121.000000 0
122.000000 0
123.000000 0
124.000000 0
125.000000 0
126.000000 0
127.000000 0
128.000000 1
129.000000 3
130.000000 14
131.000000 17
132.000000 23
133.000000 13
134.000000 0
135.000000 0
136.000000 1
137.000000 1
138.000000 0
139.000000 0
140.000000 0
141.000000 0
142.000000 3
143.000000 1
144.000000 5
145.000000 1
146.000000 0
147.000000 1
148.000000 0
149.000000 2
150.000000 4
151.000000 0
152.000000 5
153.000000 0
154.000000 0
155.000000 2
156.000000 0
157.000000 1
158.000000 1
159.000000 0
160.000000 0
161.000000 0
162.000000 0
163.000000 5
164.000000 9
165.000000 6
166.000000 9
167.000000 21
168.000000 12
169.000000 20
170.000000 10
171.000000 0
172.000000 0
173.000000 0
174.000000 0
175.000000 0
176.000000 0
177.000000 1
178.000000 0
179.000000 1
180.000000 12
181.000000 19
182.000000 13
183.000000 16
184.000000 0
185.000000 0
186.000000 1
187.000000 0
188.000000 0
189.000000 0
190.000000 0
191.000000 0
192.000000 0
193.000000 0
194.000000 0
195.000000 0
196.000000 0
197.000000 1
198.000000 0
199.000000 2
200.000000 5
201.000000 1
202.000000 0
203.000000 0
204.000000 0
205.000000 2
206.000000 0
207.000000 0
208.000000 0
209.000000 3
210.000000 0
211.000000 1
212.000000 1
213.000000 3
214.000000 4
215.000000 4
216.000000 9
217.000000 14
218.000000 25
219.000000 19
220.000000 10
221.000000 1
222.000000 0
223.000000 0
224.000000 0
225.000000 0
226.000000 0
227.000000 0
228.000000 0
229.000000 0
230.000000 24
231.000000 26
232.000000 21
233.000000 22
234.000000 0
235.000000 0
236.000000 0
237.000000 1
238.000000 1
239.000000 1
240.000000 0
241.000000 0
242.000000 0
243.000000 0
244.000000 0
245.000000 3
246.000000 0
247.000000 0
248.000000 0
249.000000 1
250.000000 3
251.000000 0
252.000000 0
253.000000 0
254.000000 1
255.000000 0
256.000000 0
257.000000 0
258.000000 0
259.000000 0
260.000000 0
261.000000 2
262.000000 1
263.000000 4
264.000000 6
265.000000 1
266.000000 10
267.000000 23
268.000000 11
269.000000 25
270.000000 10
271.000000 0
272.000000 0
273.000000 0
274.000000 0
275.000000 0
276.000000 0
277.000000 1
278.000000 0
279.000000 0
280.000000 27
281.000000 23
282.000000 21
283.000000 19
284.000000 0
285.000000 0
286.000000 0
287.000000 0
288.000000 0
289.000000 0
290.000000 0
291.000000 0
292.000000 0
293.000000 0
294.000000 0
295.000000 3
296.000000 0
297.000000 2
298.000000 1
299.000000 1
300.000000 1
301.000000 0
302.000000 0
303.000000 0
304.000000 2
305.000000 2
306.000000 0
307.000000 0
308.000000 0
309.000000 0
310.000000 0
311.000000 1
312.000000 1
313.000000 2
314.000000 1
315.000000 5
316.000000 6
317.000000 9
318.000000 19
319.000000 12
320.000000 7
321.000000 0
322.000000 0
323.000000 0
324.000000 0
325.000000 0
326.000000 0
327.000000 0
328.000000 0
329.000000 0
330.000000 19
331.000000 25
332.000000 37
333.000000 25
334.000000 0
335.000000 0
336.000000 0
337.000000 0
338.000000 0
339.000000 0
340.000000 0
341.000000 0
342.000000 0
343.000000 0
344.000000 0
345.000000 1
346.000000 0
347.000000 1
348.000000 0
349.000000 0
350.000000 1
351.000000 0
352.000000 0
353.000000 2
354.000000 2
355.000000 1
356.000000 0
357.000000 0
358.000000 0
359.000000 0
360.000000 0
361.000000 0
362.000000 0
363.000000 1
364.000000 1
365.000000 4
366.000000 6
367.000000 24
368.000000 21
369.000000 11
370.000000 11
371.000000 0
372.000000 0
373.000000 0
374.000000 0
375.000000 0
376.000000 0
377.000000 0
378.000000 0
379.000000 0
380.000000 33
381.000000 33
382.000000 30
383.000000 24
384.000000 1
385.000000 0
386.000000 0
387.000000 0
388.000000 0
389.000000 0
390.000000 0
391.000000 0
392.000000 0
393.000000 0
394.000000 0
395.000000 1
396.000000 0
397.000000 0
398.000000 0
399.000000 0
400.000000 0
401.000000 0
402.000000 0
403.000000 0
404.000000 1
405.000000 1
406.000000 0
407.000000 0
408.000000 0
409.000000 1
410.000000 0
411.000000 0
412.000000 0
413.000000 1
414.000000 0
415.000000 1
416.000000 1
417.000000 19
418.000000 13
419.000000 15
420.000000 34
421.000000 0
422.000000 0
423.000000 0
424.000000 0
425.000000 0
426.000000 1
427.000000 0
428.000000 0
429.000000 0
430.000000 55
431.000000 19
432.000000 21
433.000000 22
434.000000 0
435.000000 0
436.000000 0
437.000000 3
438.000000 0
439.000000 0
440.000000 0
441.000000 0
442.000000 1
443.000000 0
444.000000 0
445.000000 1
446.000000 0
447.000000 0
448.000000 0
449.000000 0
450.000000 82
451.000000 0
452.000000 1
453.000000 0
454.000000 0
455.000000 0
456.000000 0
457.000000 0
458.000000 0
459.000000 0
460.000000 0
461.000000 0
462.000000 0
463.000000 0
464.000000 0
465.000000 0
466.000000 0
467.000000 2
468.000000 0
469.000000 1
470.000000 34
471.000000 1
472.000000 0
473.000000 0
474.000000 0
475.000000 1
476.000000 0
477.000000 0
478.000000 0
479.000000 0
480.000000 31
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199807311853.MAA04187>