Oracle SQL query count group by timestamp substring

Given a table that has a column of string "timestamps" (yyyyMMddHHmmssSSS format), I want to substring the first 8 characters, and get a count of how many rows have that substring, grouping the results.

Sample data...

TIMESTAMP
20100802123456开发者_高级运维123
20100803123456123
20100803123456123
20100803123456123
20100804123456123
20100805123456123
20100805123456123
20100805123456123
20100805123456123
20100806123456123
20100807123456123
20100807123456123

...and expected results...

SUBSTRING, COUNT
20100802, 1
20100803, 3
20100804, 1
20100805, 4
20100806, 1
20100807, 2

I know this should be easy, but I'm not having any luck at the moment.

I don't have a database to test with, but it seems like you are looking for

select
  substr(timestamp, 1, 8),
  count(*)
from
  my_table
group by
  substr(timestamp, 1, 8);