Official Superpi Mod Ranking Thread

Rocko- Nice voltage at 3.6ghz!

Jet- It's a 45nm chip, 1.5 on air is fine especially with only 1 core going. As long as temps are in check, the voltage doesn't matter too much. Now if you had a 32nm chip, I'd say 1.45 max on air. I've taken my I7's to 1.55v on air with no harm.
 
You can run 1 core if you want. And actually I would suggest going down to 2 cores, to help to keep the temps down. Lower temps=higher overclock achieved on the same voltage.
 
jasonn20 said:
Not sure why but a dual core at higher clocks does score better than this.... :confused:
Click to expand...

SuperPi isn't multithreaded, so it only utilises 1 thread. You will get better scores with higher clocks and better architecture than you will with extra cores
 
Oh, I was wondering why he said that LOL. I didn't realize that he didn't know this was a single threaded bench.
 
I errored .. what I meant to say is a dual core at higher clocks does NOT score higher than with more cores.... some how they help.... ;)
 
No, more cores do not help. You are probably comparing CPU's of different architecture. You can't compare core 2 duo to the I3/I5/I7, the core 2 duo will be slower clock for clock.
 
Dude just get the I3, higher multiplier and you can turn HT off to get even higher clocks. You won't regret it.
 
jasonn20 said:
I errored .. what I meant to say is a dual core at higher clocks does NOT score higher than with more cores.... some how they help.... ;)
Click to expand...

only thing I could possibly think is that the processing of background tasks was shared more over the other cores, or that you changed memory speeds to a lower frequency on the dual core run. What you are saying is the complete opposite of what everyone else who runs the benchmark sees, and completely against how the program was written
 
Aastii said:
only thing I could possibly think is that the processing of background tasks was shared more over the other cores, or that you changed memory speeds to a lower frequency on the dual core run. What you are saying is the complete opposite of what everyone else who runs the benchmark sees, and completely against how the program was written
Click to expand...

Even on a single threaded application the load is spread across the cores so this might be why it helps...

for example an AMD quad will run 25% cpu usage on this single threaded application but will run the load on more than 1 core...
 
25% spread over 4 cores or 100% on one core makes zero difference for this bench. Most of the fastest Super Pi scores are on 1 core (the rest being deactivated to achieve the highest overclock possible).
 
jasonn20 said:
Even on a single threaded application the load is spread across the cores so this might be why it helps...

for example an AMD quad will run 25% cpu usage on this single threaded application but will run the load on more than 1 core...
Click to expand...

That would be slower if anything, because the processes to split the calculations between the different threads, then to put them back together, and perform the necessary checks to ensure that they are in the right order would, for each clock cycle, add on a fraction of a second, which would add up for the end result.

As 87dtna says, you will see better results from higher clocks than from more cores
 
New bench, I take 2nd place from myself...lol

SPII7950475ghz.jpg
 
You must log in or register to reply here.
Back
Top