Sign in with
Sign up | Sign in
Your question

Stress testing vs 'real life' gaming ?

Last response: in Overclocking
Share
November 8, 2007 7:44:46 AM

Hi Guys,

I apologise in advance for this noob question!

I'm running a Q6600 at stock with the Intel HSF and decided to look at giving the CPU a 'tweak' to help gain a few more FPS in Crysis. I knew I'd probably not get far, using the standard Intel HSF but thought I’d check it out.

Using the BIOS I up'd the FSB from 266, in stages, to 311, to get to the 3.0GHz speed. I know it's not the most 'efficient' way to overclock, but I set the voltage control to 'auto' so that BIOS worked out what volts I'd need to handle the extra o/c.
I ran a stress test using Prime 95 and, very soon, the CPU was registering 66-67 degrees (viewed via CoreTemp) so I thought this was a bit toasty, and perhaps a bit too near the 71C safe limit of the 'G0' version I have, so I shut the test down. I also ran a test at 2.8GHz and this ran at about 64-65C, so still hot.

I notice when I run Crysis, using an o/c to 3.0GHz, that CoreTemp shows the game running at about 57/58C which is perhaps acceptable. I measured this during a particularly ‘intensive’ part of the game, when the CPU was (hopefully) being pushed a bit.

I don't do a huge amount of multi-tasking - i.e. if I'm playing a game then I'm unlikely to be doing a virus scan or downloading huge files in the background, etc. I want to dedicate as much resource as possible to the game in hand…

…. With this in mind I’m just wondering how valid (to me, personally) the stress tests are? Just how ‘stressful’ for the CPU, etc, are (modern ‘high end’) PC games when compared to doing a test using the likes of Prime 95 or Orthos?
I know I haven’t o/c’d the system in the most efficient way (i.e. I understand that, if set to ‘auto voltage control’, the BIOS overcompensates on the voltage it requires and thus more heat is produced) but am I ‘safe’ in doing this? (PS. I’ve set PCI-e to ‘100Mhz’ to safeguard the GPU and a few other settings as recommended in the forums).

If any Crysis fans are interested, I was playing the game on Very High settings for everything (including Post Processing) but set Shadows to Medium (as they look pretty much the same as if on Very High), on 1024x768. The game ran at 31 FPS on average at stock 2.4Ghz speeds. Raising the CPU to 2.8GHz gave a 4 FPS gain. Raising to 3.0Ghz gave a 6 FPS gain. Not a huge jump I know but, when you are playing at the c30FPS level, the extra 6 FPS is a nice little performance jump and worth it.

Any help really appreciated! Cheers guys.




November 8, 2007 8:42:03 AM

From what I've heard the Crysis demo isn't optimised for Quad core CPU's?(anyone correct me if I'm wrong) Not all games stress the CPU to it's fullest whereas the stress test programs will stress your cpu to 100%.That's probably why your temperatures during Prime95 is much higher compared to running Crysis.You could benefit from using an aftermarket heatsink/fan if you plan on keeping your CPU overclocked.
November 8, 2007 8:47:23 AM

cfvh600 said:
From what I've heard the Crysis demo isn't optimised for Quad core CPU's?(anyone correct me if I'm wrong) Not all games stress the CPU to it's fullest whereas the stress test programs will stress your cpu to 100%.That's probably why your temperatures during Prime95 is much higher compared to running Crysis.You could benefit from using an aftermarket heatsink/fan if you plan on keeping your CPU overclocked.


Ah, yes, I have heard that too, now you come to mention it. Oh well, that may be handy as I may not need to o/c after all. But, then again, it's human nature to want just that little bit more so I'll probably end up getting a good HSF again.

This was my first PC build and installing the Intel HSF was probably the most challening thing about it. I was so worried I was going to crack the CPU die but all went OK. Had a DOA mobo so had to return but used ArctiClean and Arctic Silver 5 paste for the new one.

Are the likes of the Tuniq Tower, etc, easier to install than Intel's stock HSF?
Related resources
November 8, 2007 6:57:00 PM

Tuniq / Ultra are easier to install. Both use a back mounting bracket and screws. Can't be afraid to be rough (being smart about it tho.. if that makes sense).

Ultras beat the Tuniq out on Quads.

I'd say 80-90% of games/programs won't ever max out your cores to 100% on a quad core (until later in 08). The reason we use Prime95 is so we can see the max temp we will EVER see. So we know we are under the Thermal Specs.
November 8, 2007 7:36:00 PM

Of coarse it maxes them, if it just uses 10% then you are basically using like 300mhz. It'll use 100% a lo of the time to do it faster.

Its unlikely it'll max all 4 at the same time without stress testing, but ive seen my 3ghz quad maxed on 1core plenty of times.
!