Can someone explain something to me?

[Ben90]

K; first off i want to start with these graphs from AT
http://www.anandtech.com/memor...owdoc.aspx?i=3589&p=12

Im basically looking at minimum performance only here...

Ok it was my assumption that DDR response time was directly proportional to timings/clock rate (read timings over clock rate). I believe that DDR 1000@5s has for all intended purposes the same data response time as DDR 2000@10s

Now in those charts im basically comparing 1200@5s and 1600@6s.

Notice how in all graphs except the last one, 1200@5s delivers a higher minimum frame rate then 1600@6s despite the 1600 having a ~10% faster response time and more bandwidth available.


Can i attribute this to the tests not being 100% repeatable, or is there a basic flaw in my way of thinking?

 

[schenley101]

I generally think of the clock rate and timings as similar to top speed and acceleration/handling in a car. the clock is the top speed and the timings are the acceleration/handling. this is not a perfect description but it usually helps. keeping with the car analogy, a muscle car might beat a Porsche on straightaways but when it comes to turns, the situation reverses itself.

 

[Ben90]

Im starting to think that Gary has very aggressive subtimings on the 1200@5, but relaxed on the 1600@6; your scenario makes perfect sense but 1600@6 should have more accel and top speed

Im not finding that article conclusive of anything until i see:
1200@5-5-5-18-1-8-60-4-509-18-43-18-4-7-7-0-14-8-8-8-7-6-4-7-7-4-50 ect