PDA

View Full Version : G74Sx 3DMark findings



Spook50
01-21-2013, 05:25 AM
I know there's already a 3DMark thread, but since I ran so many tests under different conditions I figured this should be given its own thread. All tests were done using 3DMark 11 Basic Edition so that there would be consistency throughout the testing.

My rig is a G74Sx
Intel Core i7 2630QM
14GB DDR3
GeForce GTX 560M 3GB
650GB 5400 RPM HDD (the real bottleneck when it comes to overall system benchmarking)

So here's a basic breakdown of the different results I got, with notes pertaining to each test following the scores

Test 1) P2283 - Factory Win7 64bit install from ASUS (bloatware present)

Test 2) P2296 - Same settings as above, but with ThrottleStop active

Test 3) P2298 - Clean Win7 non-ASUS install W/ current drivers as of 20 Jan 2013

Test 4) P2300 - Same settings as above, with ThrottleStop active

Test 5) P2269 - Same as test 3, with Intel's TurboBoost disabled. Interesting how big of a hit the system took.

Test 6) P2663 - Clean Win7 install, current drivers, GPU OC'd to 902 MHz GPU, 1625 MHz memory using MSI Afterburner 2.3.0

Test 7) P2664 - Same settings as above, with ThrottleStop active (winner winner chicken dinner)

Test 8) P2655 - Same as Test 7, after installing SP1 and all Windows and DirectX updates

Test 9) P2662 - Same as Test 8, after running Defraggler

Notes:
I found it interesting that activating ThrottleStop made very little difference with 3DMark (despite the fact that it tests the CPU alone along with "both at once" testing of the GPU and the CPU). The biggest difference (damn near 300 points) came from overclocking my GPU using MSI Afterburner. The memory clock slider was maxed out, but it allowed for the GPU itself to be clocked much higher. From my research this looked like the highest stable speed for the GPU, so I didn't bother going any higher. The max temp my GPU hit was 128F throughout all the tests (obviously reached during the overclocked tests), and 135F for my CPU, which I didn't bother overclocking.
The biggest high to performance came from disabling TurboBoost. I figured it would take somewhat of a hit, but was very surprised at how much it actually did. Even more surprising is that preventing the CPU from using TurboBoost cause such a big drop in the score, but activating ThrottleStop (with TurboBoost re-enabled of course) hardly made any difference at all.
Something else that surprised me was that there was a 7 point gain after defragging my HDD. Not a considerable difference, but enough to get my attention. Makes me very curious what my results will be by repeating Test 7 with a SSD installed.
For the sake of making sure all memory was cleared and the system was "fresh" I also rebooted just prior to each test I ran.

Spook50
06-04-2013, 06:49 AM
Out of curiosity after reading another user's post stating their settings, I downloaded NVidia Inspector and used that to set my 560M to 900/1800/1800, turned on ThrottleStop and ran another test (on a fresh defrag using Vopt 9.1, rebooting immediately after).

New test score: P2675, a new winner (for my system).

I haven't done a long term stress test in Skyrim yet, but it looks like everything will be alright, since my GPU read 51C and my CPU 48C immediately after running 3DMark.

I also used the same version as in previous tests. Just for the sake of avoiding too many variables that might affect the score.

Spook50
06-08-2013, 11:17 PM
Yet another interesting find. My settings from the last test proved to be unstable in Skyrim (gave a lot of shadow artifacts and glitching), so I backed it off to 915MHz GPU core and 1625MHz memory using Afterburner. Knowing this would give a lower score than my last test I didn't bother running one, but after setting up a 10GB RAM disk using Primo RAMdisk (out of the 14GB I have installed), I ran a test using the 915/1625 settings with 3DMark installed on my RAM disk.

New score: P2692. The highest yet for my system by far, even with my GPU dialed back from the previous test.

For comparison sake, I ran CrystalDiskMark 3.02 on a couple different RAM disks made via different programs.

The first test is on my 5400RPM HDD, just for a baseline. the second is on my RAM disk that was made using IMDisk.
http://i4.photobucket.com/albums/y104/Spook50/IMDiskTest_zpsba3af8d5.png (http://s4.photobucket.com/user/Spook50/media/IMDiskTest_zpsba3af8d5.png.html)

This test was on the RAM disk created by Primo RAMdisk. I looked like my own avatar for a few seconds after seeing this one.
http://i4.photobucket.com/albums/y104/Spook50/PrimoRAMdiskTest_zpsd7d96d31.png (http://s4.photobucket.com/user/Spook50/media/PrimoRAMdiskTest_zpsd7d96d31.png.html)

And Skyrim runs BEAUTIFULLY off a RAM disk. Load times? What load times? :D

EDIT: Just for the sake of throwing more information out there, I played with some of the options in Throttlestop a bit and duplicated my previous test (which used the RAM disk I created using Primo RAMdisk BTW), resulting in a new high of P2695 for my system.

Kind of interesting seeing how my system went from a first-time score of P2283 to P2695 after reformatting and various tweaks (and seeing how much of a difference each tweak made).

Very curious to see what I end up with once I upgrade my RAM and CPU down the road. No immediate plans to do either yet though. Money's tight and I'm not struggling on performance with anything I run.