vdt=5.4.67 Linux=2.6.36 x86_64=2010/10/05 Usage : vdt Perform i/o tests. basic parameters: -b size alternate i/o buffer size -c filename target file for comparison tests -D deadline run until [[[CC]YY]MMDD]hhmm.ss -e seed random number generator seed -f filename use this file for read/write (compound) tests -I 'comment' (short) information to appear on screen and logs -m number maximum number of operations per test -M number maximum number of passes per test -p percent test area = given % of testsize -P pattern filler for output buffer (0oct|Xhex|Bbin|dec|Rand|Zero) -r filename input(source) file for read-only(comparison) tests -s stripes split test area in given # of stripes -t lifespan [[hh:]mm:]ss = maximum elapsed time per test -w filename use this output file for write-only tests -z testsize test area = just this much -Z bulksize terminate after transfer this much basic tests: --make create & fill test (Bunch_of_Sectors at a time) --poke poke test (same as '--writ --sect') --read read test (B.o.S. at a time) --seek seek test (same as '--read --sect') --writ write test (B.o.S. at a time) compound tests: --aver ratification test (write, read back, compare) --blot update test (read, stain, write back) --comp comparison test (B.o.S. at a time) --cool freshen test (read, write back) --copy copy test (B.o.S. at a time) --stir shuffle test (read, write somewhere else) stress tests: --burn write-only test (1 file, all addressing modes) --cook read-only test (1 file, all addressing modes) --mess read+write test (1 file, all addressing modes) addressing patterns: --incr incrementing addresses (start to end) --decr decrementing addresses (end to start) --conv converging addresses (edges to middle) --scat scattering addresses (middle to edges) --gold 'diamond'-pattern (middle to edges and back) --spot 'X'-pattern (edges to middle and back) --rand random addresses region selection: --edge --head and --tail combined --head select initial portion of test area (default) --core select central portion of test area --tail select final portion of test area --slab --edge and --core combined --even divide test area proportionally --flip invert selection of test area entropy control: --same yields same random numbers across tests --firm yields same random numbers across platforms --fate same as '--same --firm' --whim extra entropy for unpredictable randomness cache illustration: --cloy stops after (default 90) hit limit --glib smooth transition to next round --snag cloggy transition to next round operation flags: -A append to log file -E separate error messages into 'vdt.err' -F force overwrite existing files -T timestamped names for log and error files -V show version info and exit -v verbose and additional details --si use powers of 1000 not 1024 --bias skip (possibly) cached B.o.S. --bang same as -F above --boon allow last test to end cleanly past deadline --daze full buffer random pattern fill (implies --stew) --fast apply faster algorithms --full random seek within B.o.S. --help show this usage help --once same as --uniq --part compute performance using alt. buffer size --raze flatten (truncate) output file --sect i/o just a single (512-Byte) sector at a time --stew applies '-P Rand' for each B.o.S. --sync synchronous (blocking) output --uniq test each and every random address just once --wait request user interaction to finish --wrap wrap around EOF until timeout extra parameters: -L lines override screen's # of lines -C columns ditto for columns