Far Cry 2 – Single to Quad Core Scaling

Overall Score

Introduction:

With the onset of quad cores becoming mainstream for gaming and everday average user computers — not just enthusiasts — we decided to see how much benefit one actually gains using a multi-core CPU in a popular current game title that claims to make use of more than a single or dual core. We look to do an analysis of Grand Theft Auto IV once this reviewer acquires a copy of the game.

 

The main result we wanted to see was whether or not the game scaled well at normal playable settings. Naturally we tested at the lowest graphical settings the game had to offer as well, to eliminate any bottleneck in the graphics cards. We are certainly more interested in finding out if (for example) upgrading from a dual core to a quad core will give you any noticeable gains during your normal gameplay in a game such as Far Cry 2. All testing was done on a regular install of Windows 7 Beta x64, the same install you might encounter on your own system.

 

Test Setup:

  • Intel QX9770 @ 4.0Ghz (400 x 10) CPU on water
  • Asus Maximus II Formula P45 Motherboard
  • 8GB G.Skill DDR2-1000 @ 800Mhz 4-4-4-12
  • 2 x Asus HD 4850’s in CrossFire @ 725Mhz Core, 1125Mhz Memory
  • OCZ GameXstream 700w PSU
  • Western Digital 74GB Raptor (OS & Bench Tool Drive)
  • Seagate 750GB 7200.10 (Game Drive)
  • Windows 7 x64 Beta Build 7000
  • Catalyst 8.12 Drivers w/ Hot Fix

 

All overclocked settings were set to the speeds above, below the normal 24/7 gaming clocks for this system to reflect what an average user might be able to accomplish with air cooling on the CPU and non-volt modded graphics cards. For anyone that isn’t sure how we are able to run 1, 2, 3, and 4 cores on the same CPU, here is a quick explanation. Though the method differs slightly based on which OS you are running, there is an option somewhere withing the command “msconfig” to change your boot settings. One of these “advanced” settings is known as “/numproc” flag and the number following “/numproc” results in the number of cores your OS will recognize and make use of. Most commonly this is used for extreme benchmarking for a few tests that give better results with a higher clockspeed (/numproc = 1 will generally give you a higher core clock stable within windows) rather than extra cores or threads.

 

Game Settings Used:

All tests were completed using the very good built in Far Cry 2 Benchmarking Tool version 1.0.0.1. Three runs of 5 loops on test “Ranch Small” were run, with the median score accepted as the representative sample for each test. No results stood outside the self-imposed 1.5% margin of error, this is a very consistent benchmarking tool indeed.

1680 x 1050 settings:

  • Post FX – High
  • Ambient – High
  • HDR Enabled
  • Bloom Enabled
  • All Other Settings – Very High
  • Direct X 10 Enabled

 

640 x 480 settings:

  • All settings – Low
  • Direct X 9 Enabled

Just to provide an example of the output graph that the benchmarking tool provides, here is the test results graph with all four cores enabled at:

 

1680 x 1050 Settings:

[img]4%20cores.png[/img]

 

 

640 x 480 Settings:

[img]4%20cores%20640.PNG[/img]

 

As you can see, the graphs result in very uniform lows and highs, even with the drastic difference in settings. Results are also output in numerical values of minimum, maximum and overall average FPS across all 5 loops. Let us take a look at the 1680 x 1050 results.

[img]Average%201680.png[/img]

 

What we see here is virtually no improvement going from 2 to 3 or even 4 cores using our “actual gameplay” settings. Gaining less than 2 FPS (4%) going from a dual core to a quad core is a slight disapointment. As expected there is a significant loss in FPS when only using a single core, but anyone hoping to play current generation games at 1680 x 1050 with settings at this level probably should have more than a single core by now anyway.

 

Have a look at the maximum FPS acheived during 5 loops of the Ranch Small test.

[img]Max%201680.png[/img]

 

Similar to what we saw with the average FPS, the maximum FPS acheived showed only a small (5%) gain going from dual to quad core. We do have a slight gain going from dual to triple core, something not seen in the overall average. As seen above, single core takes a heavy hit in this test.

 

Now for the minimum FPS test results.

[img]Min%201680.png[/img]

 

Impressed with the consistency of the results, we again see a 4% improvement going from dual to quad core in the minimum FPS test.

 

 

 

 

To find out if the game is even making legitimate use of extra cores, we decided to test at the lowest possible graphical settings. Here are average FPS results at this ridiculously low resolution.

[img]Average%20640.png[/img]

 

We can see that Far Cry 2 is in fact making use of extra cores. Going from dual to quad results in a 23% gain in FPS once the graphics bottleneck is removed. Even going from dual to triple core gave a 9% gain in average FPS. Both of you AMD Tricore users out there will appreciate knowing your CPU will outperform an equally clocked dual core CPU from the same architecture in Far Cry 2.

 

Maximum FPS test results at low resolution.

[img]Max%20640.png[/img]

 

In terms of maximum FPS, going dual to quad resulted in a 25% gain. Adding one core and going dual to triple gave us a small 6% gain in FPS. Single core has by now officially staked it’s claim at the bottom of the charts, with nearly half the maximum FPS as the dual core.

 

Minimum FPS results.

[img]Min%20640.png[/img]

 

Keeping with the surprisingly consistent results, we see a 25% gain when we double the cores from two to four. Three cores gives a 17% gain over dual and again, single core turns in a minimum FPS of nearly half of the dual core.

Well, the results were surprisingly consistent across all tests. But if you hoped that upgrading from a dual to a quad would give you a noteable increase in FPS for normal gameplay, you will be disappointed by these results. We are confident that Grand Theft Auto IV will provide better scaling, but we are disappointed by the results turned in by one of the latest “multithreaded” games.

 

In conclusion, having a dual core for Far Cry 2 will give you everything you need for normal gameplay. One thing to think about here is that for an average gamer’s setup and budget, a dual core will hit consistently higher clocks (for the same price point) for 24/7 gaming and that alone will probably more than cover the small gains seen going from dual to quad in this particular title. We were disappointed to see that at an “average gamer’s” resolution of 1680 x 1050, even relatively highly clocked HD 4850’s were GPU bottlenecked, unable to make use of extra cores, even at 4Ghz. So a combination of GPU bottlenecking and relatively poor scaling results in a game title that is unable to invoke a feeling of “need” to make the move from dual to quad. Hopefully other games are more able to make quad core CPUs an advantageous factor for current and future generation gamers.

 

[img]Far-Cry-2-1422.jpg[/img]

 

Discuss this article here. Feel free to ask any questions you may have, we welcome your feedback!

SHARE THIS POST

  • Facebook
  • Twitter
  • Myspace
  • Google Buzz
  • Reddit
  • Stumnleupon
  • Delicious
  • Digg
  • Technorati

Leave A Response