Techware Labs Header
RSS
Home | Reviews | Articles | Downloads | Guides | Staff | Advertising | Links
Mainboards | Networking | Video | Cases | Storage | Other

RAID and Gaming Performance


Author:  Jason Jacobs
Date:  2008.04.01
Topic:  Hardware
Provider:  HighPoint
Manufacturer:  Samsung






Raid and Gaming Performance

 

Gaming Performance :

Our Gaming performance results are based upon testing in the following games:

Crysis
Team Fortress 2
Fear X2
Enemy Territory Quake Wars

We selected these games due to their recent release, large nature of the maps and textures in game, as well as the relatively heavy graphical nature of the games. We measured in game frames per second with Fraps as well as load times under a single drive non-raid (1), two drive(2) , and four drive (4) raid configuration. We did not test beyond 4 drives for gaming as we feel that gamers are unlikly to purchase more than 4 drives for a Raid configuration.

While the RocketRaid 3520 is an 8 port card with 256MB of DDR2 memory, HighPoint does have the RocketRaid 3510 card available which is a 4 port card at a lower price point. There is no difference between the 3520 and the 3510 other than the number of ports on the card. We would suggest the 3510 for all consumers interested in configuring a 4 drive raid array for their gaming needs.

We configured the single, dual, and four drive array using windows default segment size of 512b.

Each configuration was loaded on fresh drives, no image was used, and in each case the array was also the boot partition.

Crysis: Level - Core

 
Single Drive
Raid 0 - 2 Drives (1TB Array)
Raid 0 - 4 Drives (2TB Array)
Level Load Time
31 Seconds
24 Seconds
21 Seconds
 
Single Drive
Raid 0 - 2 Drives (1TB Array)
Raid 0 - 4 Drives (2TB Array)
Game Load Time
15 Seconds
10 Seconds
7 Seconds


Team Fortress 2

 
Single Drive
Raid 0 - 2 Drives (1TB Array)
Raid 0 - 4 Drives (2TB Array)
Level Load Time
22 Seconds
18 Seconds
15 Seconds
 
Single Drive
Raid 0 - 2 Drives (1TB Array)
Raid 0 - 4 Drives (2TB Array)
Game Load Time
12 Seconds
7 Seconds
4 Seconds

 

Fear X2

 
Single Drive
Raid 0 - 2 Drives (1TB Array)
Raid 0 - 4 Drives (2TB Array)
Level Load Time
24 Seconds
20 Seconds
17 Seconds
 
Single Drive
Raid 0 - 2 Drives (1TB Array)
Raid 0 - 4 Drives (2TB Array)
Game Load Time
10 Seconds
8 Seconds
7 Seconds


Enemy Territory Quake Wars - Africa

 
Single Drive
Raid 0 - 2 Drives (1TB Array)
Raid 0 - 4 Drives (2TB Array)
Level Load Time
27 Seconds
21 Seconds
18 Seconds
 
Single Drive
Raid 0 - 2 Drives (1TB Array)
Raid 0 - 4 Drives (2TB Array)
Game Load Time
9 Seconds
6 Seconds
4 Seconds

 

In none of the configurations was there an increase in FPS due to the introduction of a Raid array. FPS remained consistently within1-4% between all configurations.We did not see an increase or benefit of Raid to frames per second.

As you can see there is anywhere between 10 seconds and 7 seconds difference loading time between a Raid 0 and single drive configuration. What cannot be measured but was experienced was a big reduction in in game lag due to hard drive access. Any occasional stutters were momentary pauses as opposed to elongated stutters which can be experienced with single drive configurations.

Overall gameplay comes down to your video card and CPU more than it does the hard drive. While game load times show here are for games with huge maps and a lot of textures your average games should load faster and you will be waiting less and playing more with a raid configuration. You may want to weigh this against the cost of a 2 to 4 drive configuration plus the raid controller card. We cannot stress enough the large difference between onboard and stand along raid controllers CPU consumption and would suggest a stand alone card for your best performance. See our initial assessment of the HighPoint RocketRaid 3520 here.

 



« Server Testing Continued
Analysis of Results »