Simple Performance Help

I am looking to take unique key value from some data
Check if that key value ( or record ) exists within a table
If it already exists then

update the row with my data based on the unique key
else

insert the data as a new row


for example ..........

DECLARE
@myNewID integer
@Result integer
@myNewValue VARCHAR(10)

@myNewId = 1234
@myNewValue = 'abcdefg'

SELECT @Result = SELECT count(*) FROM mytable WHERE keyID = @myNewID

if @Result > 0

UPDATE mytable SET colValue = @myNewValue
WHERE @KeyID = @myNewID
else

INSERT into mytable ( keyID, colValue ) values ( @myNewID, @myNewValue )


My question is, is there a better, faster way of doing this. I have to do this check around 100 times per minute 24x7x365
In reality the actual table I am checking is approx 2.5 million rows and growing fast. I'd like to eliminate the SELECT COUNT(*) step if possible, I know I could use SELECT 1, or an EXISTS type approach.

I'm thinking would it be better to just go ahead and INSERT and then check for a primary key constraint error and if found perform an Update. I was also thinking about INSERT and/or UPDATE triggers.

Any thoughts ?

KD