by karog » Tue Sep 18, 2012 3:32 pm
$this->bbcode_second_pass_quote('WarheadsSE', 'Y')ou know MIPS is subjective, yes? The Pro is dual-issue .. and .. anyways.. you're doing the math wrong because those numbers are wrong in the first place from the Pro. No, I am not saying that you're wrong in what they present, I am saying they are wrong in how they are calculated in the first place.
I didn't think BogoMIPS were actual MIPS but I had thought (wrongly) that they were comparable.
I wrote a little timed Fibonacci program to compare V2 vs V3. Crude but interesting. I ran fib(40)
V2 (1200MHz) 4860 ms
V3 ( 700MHz) 5022 ms
V3 ( 850MHz) 4140 ms
Note the V3's are dual core. On the V3 I ran the test twice at the same time and got similar ms for both. I then ran 3 at a time and the times were longer but the 3 times added up to triple the single time as expected.
So for a single thread, the V3 is about 3-4% slower than the V2 at stock 700 MHz. Overclocking to 850 MHz provides the expected speed increase.
I am not claiming this test is in any sense thorough. But it is in a gross sense illustrative.