salmi@dicomed.UUCP (salmi) (01/22/86)
i have just recently brought up ultrix 1.1 on a vax 785. after running the recently distributed dhrystone benchmarks on the new system, i was appalled at the outcome, somethint like 50000 passes in 102 seconds! (480 stones/sec?) running vmstat on the system gives the following results, with one user on the system... # vmstat 2 procs memory page disk faults cpu r b w avm fre re at pi po fr de sr r0 x1 x2 x3 in sy cs us sy id 1 0 0 445 7769 0 0 0 0 0 0 0 3 0 0 04994 18 4 1 82 17 1 0 0 445 7769 0 0 0 0 0 0 0 0 0 0 05156 163 5 0 83 17 1 0 0 445 7769 0 0 0 0 0 0 0 0 0 0 05170 107 3 0 83 17 0 0 0 423 7746 0 0 0 0 0 0 0 0 0 0 05178 73 1 0 81 19 0 0 0 423 7746 0 0 0 0 0 0 0 0 0 0 05183 49 0 0 84 16 0 0 0 476 7742 0 0 0 0 0 0 0 2 0 0 05179 35 0 0 81 18 0 0 0 476 7742 0 0 0 0 0 0 0 0 0 0 05182 26 0 1 78 21 0 0 0 476 7742 0 0 0 0 0 0 0 0 0 0 05187 20 0 1 84 15 0 0 0 157 7742 0 0 0 0 0 0 0 0 0 0 05190 16 0 1 82 17 0 0 0 157 7742 0 0 0 0 0 0 0 0 0 0 05192 13 0 0 84 16 0 0 0 375 7742 0 0 0 0 0 0 0 0 0 0 05201 8 0 0 82 18 the interupts per second are averaging more the 5000 on an 'idle' system. our loaded up vax 750 running 4.2 gives the following results from vmstat... # vmstat 2 procs memory page disk faults cpu r b w avm fre re at pi po fr de sr h0 r0 u0 f0 in sy cs us sy id 2 0 0 1139 3207 0 1 0 0 0 0 0 3 0 1 0 4 68 20 18 25 57 1 0 0 911 3185 0 0 0 0 0 0 0 0 0 0 0 3 239 18 6 19 74 1 0 0 911 3185 0 0 0 0 0 0 0 0 0 0 0 2 180 14 3 14 83 1 1 0 868 3176 0 0 0 0 0 0 0 20 1 0 0 10 133 13 5 22 73 1 1 0 868 3176 0 0 0 0 0 0 0 0 0 0 0 6 111 10 4 13 83 0 0 0 629 3173 0 0 0 0 0 0 0 0 0 0 0 4 96 8 4 19 77 0 0 0 629 3173 0 0 0 0 0 0 0 0 0 0 0 2 88 8 3 15 82 0 0 0 629 3173 0 0 0 0 0 0 0 0 0 0 0 2 88 9 6 16 77 all of the ports had been shut off on the 785, thinking that a runaway getty may have been the culprit, but no change was evident. another thought was the DEC ethernet board may be the bad guy. anyone got any clues/ideas? the 785 must be fast, as the response time was fairly fast even with the 5000+ interupts/second running rampant. any/all ideas would be appreciated. please respond directly to me at: ihnp4!dicomed!salmi john salmi dicomed corporation minneapolis, mn 612/885-3000 and, as always, thanks in advance :-)
lrr@princeton.UUCP (Larry Rogers) (01/24/86)
In article <710@dicomed.UUCP> salmi@dicomed.UUCP (salmi) writes: >i have just recently brought up ultrix 1.1 on a vax 785. after running the >recently distributed dhrystone benchmarks on the new system, i was appalled at >the outcome, somethint like 50000 passes in 102 seconds! (480 stones/sec?) > >running vmstat on the system gives the following results, with one user on the >system... > > ># vmstat 2 > procs memory page disk faults cpu > r b w avm fre re at pi po fr de sr r0 x1 x2 x3 in sy cs us sy id > 1 0 0 445 7769 0 0 0 0 0 0 0 3 0 0 04994 18 4 1 82 17 > 1 0 0 445 7769 0 0 0 0 0 0 0 0 0 0 05156 163 5 0 83 17 > 1 0 0 445 7769 0 0 0 0 0 0 0 0 0 0 05170 107 3 0 83 17 > 0 0 0 423 7746 0 0 0 0 0 0 0 0 0 0 05178 73 1 0 81 19 > 0 0 0 423 7746 0 0 0 0 0 0 0 0 0 0 05183 49 0 0 84 16 > 0 0 0 476 7742 0 0 0 0 0 0 0 2 0 0 05179 35 0 0 81 18 > 0 0 0 476 7742 0 0 0 0 0 0 0 0 0 0 05182 26 0 1 78 21 > 0 0 0 476 7742 0 0 0 0 0 0 0 0 0 0 05187 20 0 1 84 15 > 0 0 0 157 7742 0 0 0 0 0 0 0 0 0 0 05190 16 0 1 82 17 > 0 0 0 157 7742 0 0 0 0 0 0 0 0 0 0 05192 13 0 0 84 16 > 0 0 0 375 7742 0 0 0 0 0 0 0 0 0 0 05201 8 0 0 82 18 > >the interupts per second are averaging more the 5000 on an 'idle' system. our >loaded up vax 750 running 4.2 gives the following results from vmstat... > > ># vmstat 2 > procs memory page disk faults cpu > r b w avm fre re at pi po fr de sr h0 r0 u0 f0 in sy cs us sy id > 2 0 0 1139 3207 0 1 0 0 0 0 0 3 0 1 0 4 68 20 18 25 57 > 1 0 0 911 3185 0 0 0 0 0 0 0 0 0 0 0 3 239 18 6 19 74 > 1 0 0 911 3185 0 0 0 0 0 0 0 0 0 0 0 2 180 14 3 14 83 > 1 1 0 868 3176 0 0 0 0 0 0 0 20 1 0 0 10 133 13 5 22 73 > 1 1 0 868 3176 0 0 0 0 0 0 0 0 0 0 0 6 111 10 4 13 83 > 0 0 0 629 3173 0 0 0 0 0 0 0 0 0 0 0 4 96 8 4 19 77 > 0 0 0 629 3173 0 0 0 0 0 0 0 0 0 0 0 2 88 8 3 15 82 > 0 0 0 629 3173 0 0 0 0 0 0 0 0 0 0 0 2 88 9 6 16 77 > >all of the ports had been shut off on the 785, thinking that a runaway getty >may have been the culprit, but no change was evident. another thought was the >DEC ethernet board may be the bad guy. > >anyone got any clues/ideas? the 785 must be fast, as the response time was >fairly fast even with the 5000+ interupts/second running rampant. > >any/all ideas would be appreciated. please respond directly to me at: > >ihnp4!dicomed!salmi > >john salmi >dicomed corporation >minneapolis, mn >612/885-3000 > >and, as always, thanks in advance :-) I also have Ultrix 1.1 running on a VAX-11/785 with the ports disabled and the Deuna enables and I get: procs memory page disk faults cpu r b w avm fre re at pi po fr de sr r0 r1 r2 r3 in sy cs us sy id 2 3 0 66367 58438 0 0 0 0 0 0 0 0 0 1 0 2 7 13 92 2 6 2 3 0 66367 58438 0 2 1 0 0 0 0 0 0 3 0 19 341 29 96 4 0 2 3 0 66367 58438 0 1 0 0 0 0 0 0 0 0 0 15 277 26 96 4 0 2 3 0 66367 58438 0 0 0 0 0 0 0 0 0 0 0 12 226 23 98 2 0 2 0 0 66291 58408 0 0 0 0 0 0 0 0 0 0 0 9 185 21 98 2 0 2 0 0 66291 58408 0 0 0 0 0 0 0 0 0 0 0 7 153 19 97 3 0 2 0 0 66291 58408 0 0 0 0 0 0 0 0 0 2 0 6 129 19 92 8 0 2 0 0 66291 58408 0 0 0 0 0 0 0 0 0 0 0 7 113 19 93 7 0 2 0 0 66291 58408 0 0 0 0 0 0 0 0 0 0 0 8 101 19 95 5 0 2 0 0 66366 58398 0 0 0 0 0 0 0 0 0 0 0 9 97 19 96 4 0 2 0 0 66366 58398 0 0 0 0 0 0 0 0 0 1 0 8 83 19 95 5 0 2 0 0 66366 58398 0 0 0 0 0 0 0 0 0 0 0 8 74 19 94 6 0 2 0 0 66366 58398 0 0 0 0 0 0 0 0 0 1 0 9 68 19 93 7 0 2 0 0 66366 58398 0 0 0 0 0 0 0 0 0 0 0 8 62 19 94 6 0 2 0 0 66344 58394 0 0 0 0 0 0 0 0 0 0 0 9 62 20 98 2 0 2 0 0 66344 58394 0 0 0 0 0 0 0 0 0 0 0 7 55 19 94 6 0 2 0 0 66344 58394 0 0 0 0 0 0 0 0 0 0 0 9 55 19 95 5 0 2 0 0 66344 58394 0 0 0 0 0 0 0 0 0 1 0 10 53 19 95 5 0 Only a normal number of interrupts. If you look at the output carefully, you may get the impression that this machine has (66344+58394)K of memory, roughly 124M. This is indeed correct as we have 128M on this baby. We don't page much. Larry Rogers Princeton University Department of Computer Science Engineering Quadrangle Building, Room C334 Princeton, NJ 08544 UUCP: princeton!lrr CSNET: lrr%princeton@CSnet-relay PHONE: 609 452 6483
wls@astrovax.UUCP (William L. Sebok) (01/25/86)
In article <1202@princeton.UUCP> lrr@princeton.UUCP (Larry Rogers) writes: >... If you look at the output >carefully, you may get the impression that this machine has (66344+58394)K >of memory, roughly 124M. This is indeed correct as we have 128M on this >baby. We don't page much. But when they do page ... wowee! -- Bill Sebok Princeton University, Astrophysics {allegra,akgua,cbosgd,decvax,ihnp4,noao,philabs,princeton,vax135}!astrovax!wls