jbm@uncle.UUCP (John B. Milton) (12/11/89)
Ok, lets try this again. I originally posted to unix-pc.general only, but broadened it to include comp.sys.att this time. I goofed the instructions last time, so let's try this again. The idea is to compare the speed it takes to complete the 1000 random seeks test to the speed in the spec sheet. I expect ALL tests to come out slower than the claimed speed, because of seek verification delay and bad block avoidance delay. Please do the following steps: (<cr> means to press the [Return] key) 1. Shutdown UNIX 2. Boot the machine with the first disk of the Foundation set "DIAGNOSTIC DISK" 3. After the loader screen you should see a list of the memory in the machine. Do what it says, press return 4. At the prompt after the menu, enter "s4test<cr>". You will now see an "expert>" prompt. Don't get carried away, it doesn't know any better. 5. Enter "6,1<cr>" to recal (move the hard disk head to track 0). I goofed this step last time. 6. Get your stop watch ready, then enter "6,5<cr>". Start the stop watch right when you press return. 7. Stop the stopwatch when the "expert>" prompt comes back. With some drives, you'll KNOW when the test stops! 8. Repeat steps 5 to 8 five (5) times, writing down all the times. 9. Send the following information: a. The five times b. The number of cylinders c. The number of heads d. The number of bad blocks e. The drive manufacture and model number, if known You can get 9b-d from a "6,12<cr>" expert mode command, or from an "iv -vt /dev/rfp000" command. If you have two drives on an ICUS HD kit or one of my (rare) HD2 Beta boards, do everything for the other drive with "2," instead of "6,", and the iv commands with "rfp010" instead of "rfp000". Some interesting things are already starting to emerge. Both uncle and oink have CDC-85 drives: 917 cylnders by 9 heads. Our times differed by about 1.3, with his being faster. His drive has 1 bad block to my 16. Hmm. Thanks to the people who already sent me times and to John Campbell for pointing out that I goofed step 5, which will throw off the times. As I mentioned before, do it right away, so I can post the results back to the net. I have received times from Bill Carpenter(bad), John Campbell (good), Brooke Jarrett(bad), Ronald L Fletcher(bad) and Jon Allen(bad) What I hope to get out of this is: Some close-to-real-life measurements of seek time, one of the real killers with lower priced hard disks. The effect of bad blocks on seek time. A table others can use to guess what drive they have without opening the box. Come on folks, you've been wanting to reboot to reset your clock for a long time now... Thanks for contributing, John -- John Bly Milton IV, jbm@uncle.UUCP, n8emr!uncle!jbm@osu-cis.cis.ohio-state.edu (614) h:252-8544, w:469-1990; N8KSN, AMPR: 44.70.0.52; Don't FLAME, inform!