Disable cores for higher O.C.?

Solution
It makes logical sense that if your cpu has less cores it will run cooler and consume less power so I would assume that you would be able to achieve a higher overclock.

A quad core @ 3.9GHz should be just fine. I would rather run 4 cores at 3.9GHz that 3 cores at 4.2 - 4.3GHz.

Why not just check yourself? Disable 1 core and bump up the multiplier or bus speed and see what happens. I have been wondering about this too...

cromedome

Distinguished
Dec 26, 2009
563
0
19,010
It makes logical sense that if your cpu has less cores it will run cooler and consume less power so I would assume that you would be able to achieve a higher overclock.

A quad core @ 3.9GHz should be just fine. I would rather run 4 cores at 3.9GHz that 3 cores at 4.2 - 4.3GHz.

Why not just check yourself? Disable 1 core and bump up the multiplier or bus speed and see what happens. I have been wondering about this too...
 
Solution
I could do that but i wanted to save the hastle.If someone knew the direct answer it would easiest.

Power/cooling isn't really an issue it's just the limiations of this processor.I don't see the point in going from 1.425v to 1.5v just to get it over the 4ghz hump/wall.
 

cromedome

Distinguished
Dec 26, 2009
563
0
19,010
Yea, I also don't feel like disabling a core on my i5 just to test.

Most of the time when a cpu hits an overclock wall it is just one of the cores that can't handle it. For example if core #2 is the weakest core and cannot go further than 3.9GHz and then you disable core #4 you will still be limited by core #2 and will be stuck at 3.9Ghz.

However, if core #4 is the weakest and you disable it then you should be able to go over 4.0GHz with core #1, #2 and #3.

This is all assuming that if you disable a core, core #4 is always the core that will be disabled.
 

trihedral

Distinguished
Jan 20, 2008
481
0
18,810
disabling cores just for the sake of overclocking is pointless.

rather stay at 4 cores than to go over the 4ghz barrier.

If your chip can't hit 4.0ghz with four cores, then just give up and stay at 3.9.

it's like a box of juice, you ran out of juice to suck, you can't just add water in and expect it to taste as good. It will be more full though, i can assure you that. (higher oc wise).
 

thedooce

Distinguished
Apr 14, 2010
15
0
18,510
This is interesting. I'm running my Q6600 at 3.55GHz and have had it as high as 3.9GHz (all cores). I don't know if you were at all interested in gaming performance, or if you were just simply going for the highest clock on your CPU, but I'll do a simple Crysis cpu benchmark (low resolution and maxed physics settings) with my current settings, and with a higher o/c with 2 cores disabled and report back. I'm excited! :D
 

thedooce

Distinguished
Apr 14, 2010
15
0
18,510
Here we go.

I used the Crysis Benchmarking utility set on "benchmark_cpu" (the one with a lot of frags and rockets destroying buildings). I used settings that I felt would remove any GPU bottleneck so that differences in CPU speed would be more apparent.

Settings:
-DX10 32bit
-All settings at "medium" with particle and physics quality to "very high"
-1280x1024 resolution

Q6600 @ 3.55Ghz, all cores enabled
-Minimum FPS=51.76
-Maximum FPS=102.12
-Average=79.34

Q6600 @ 3.75Ghz, running dual-core (2 cores disabled in BIOS)
-Minimum FPS=31.97
-Maximum FPS=109
-Average=63.36

All values are an average of 3 benchmark runs.

I'm convinced that these numbers show that there is no way, with my setup, that I could get 2 cores to perform better than 4, even if I went for a 4.0Ghz clock (which I may try for the heck of it =P. It may be worth testing with disabled cores for a maximum overclock, but for raw performance, I'd keep everything enabled.

Here are my 4.0Ghz (2-cores) results:
-Minimum FPS=32.59
-Maximum FPS=112.76
-Average=65.7

Thats less than a 2 FPS (Avg.) gain over 2 cores at 3.75Ghz and still 14 FPS shy of my 4-core performance at 3.55Ghz.

I can't say my results weren't expected, but I love running tests and seeing the numbers. As an added note, I had to bump up my Vcore to 1.43V (from 1.32) to run the tests at 4Ghz. Extra voltage and heat might even make o/c testing a waste of time and potentially dangerous to your CPU. Keep a close eye on your temps if you decide to really push things (even if it is only on half of your cores).

 

Mordecai Walfish

Distinguished
Mar 12, 2011
11
0
18,520
I know this thread is a couple months old but I thought that I would add some thoughts.

The conclusion you came to is generally correct for this scenario. Leaving your system @ 4 cores will give you optimal performance for the vast majority of programs/apps/games you use.

That being said, there are some benefits to disabling cores to achieve a higher overclock, but it is dependent on the software/game being used. Some (even recent) games were not coded with multi-core utilization in mind and are limited to using say 1 or 2 cores total of the 4 you have. Oblivion, for example, came out about 5 years ago and will only use at most 2 cores, but is a *very* processor intensive game, especially with all the additions and mods that have enriched it over the years. For a game like this, you *will* see a more dramatic improvement in performance by improving the clock speed on the processor, but the game will just ignore any cores past the 2 it's using. I personally have an AMD Phenom 2 B55 (quad core, unlocked 555BE) and use a 3.6GHZ 4-core profile for pretty much everything I do, but I set up an additional profile in my mobo's bios (it has 3 presets) to only have 2 cores active if I choose and clock them @ 4ghz. This does give Oblivion and other 'non-optimized-for-3+core' programs a significant boost, very noticeable in-game. On the other hand, games like GTA4 and the Crysis series run remarkably slower because they are optimized to take advantage of 3+cores.


Remember if you are planning on making a separate profile like this, be sure to test it for stability *THOROUGHLY*. Extensive testing is the only way to be comfortable with a setup like this. 12 hours of prime95, 12 hours memtestx86 (or 64),windows memory diagnostic extended test (at least 2 passes), LINX, OCCT, etc..

Don't fool yourself into thinking that just because you are able to boot into windows and things are running well that your work is done. Test, test, test. And then test a little more just to be safe =P. Catastrophic failure could result if you don't. Just my 2cents.

/M.Walfish
 

AnshulBains

Distinguished
May 22, 2011
24
0
18,520
although it would help u overclock but on longer run it wont make any sense !
cause most of the games today run on minimum of 2-3 cores and optimum on four cores so chances of performance drop increases so my personal advice u dont need to overclock on the cost of dissabling your 1or 2 cores
 

rio747

Honorable
Jun 26, 2012
34
0
10,540
So here's my question, I have a quad core that has core number 3 as the weak link. With an equal load it runs about 8-9 degrees hotter than the other three cores. I would like to disable it but keep core number 4 running. So far all the ways I have found to disable cores dont allow you to choose a specific core.
 

Joeyob32

Honorable
Mar 28, 2013
65
0
10,640
I know this thread is old, but I'd appreciate if anyone could riddle-me-this...
What about if you have an 8-core, and want to disable 4 of them to achieve higher clocks for gaming? yay/nay?