| RAID and Gaming Performance |
|
Raid and Gaming Performance
Introduction to Raid:
Raid which stands for Redundant Array of Independant Disks is a technology that has been used for years to enhance performance and provide fault tolerance to the server and workstation market segments. Recently Raid has become more and more commonly found in the desktop and consumer markets as the prices of controller cards and drives has dropped. This trend will likely continue as storage and competition increases to capture this growing market. For a full explanation of raid and its intended purposes see our article here. For now though understand that there are several implementations of raid which can be used but the most common are Raid 0 and Raid 1. Since this article focuses upon the performance enhancements gained by Raid we shall be using a Raid 0 array for our tests.
The Theory:
The theory behind creating a raid array to benefit gaming is that since many currently available games load a huge amount of textures and information from the hard drive into video and system memory that utilizing a raid array can lead to significantly reduced load times and lag while playing the game. The load times come into play during map and level transitions as well as during gameplay when movie sequences or transitions occur. These transitions can be very frustrating when your game pauses to load information from the hard drive into memory and you are faced with a stuck or stuttering frame while it loads.
Synthetic vs Real World Testing:
Thanks to the combined efforts of HighPoint and Samsung we are bringing you real world testing to prove or disprove the currently held opinions about Raid and gaming. First we tested the HighPoint Raid controller 3520 in an enterprise server setup to demonstrate the common uses of this equipment. We will bring you these resultgs and then show you the desktop gaming application testing.
Gaming Test Rig:
Intel E8500 3.16Ghz
EVGA 780i Motherboard
ATI Sapphire 3870 X2 Video Card
2GB Crucial Ballistics Tracer DDR2
4 x Samsung 500GB Hard Drives Raid 0
HighPoint RocketRaid 3520
Windows Vista Ultimate
Server Test Rig:
Dual AMD Opteron 2220 Processors
Tyan S2927 Motherboard
4GB Kingston DDR2 667 ECC
Western Digital Raptor 74GB
8 x 1TB Samsung
Windows Vista Ultimate
Windows Server 2003 Standard
Windows Server 2008 Beta
HighPoint RocketRaid 3520
The Samsung 1TB and 500GB drives were chosen for several reasons:
Server Testing (Samsung F1 HD103UJ- 1000GB)
- Large Storage Size
- SATA II Interface (300Gb/s)
- Large Onboard Cache (32MB)
- Relatively low cost per MB
Game Testing: (Samsung Spinpoint HD501LJ - 500GB)
- Relatively Low Cost to purchase 2 to 4 drives.
- SATA II Interface (300Gb/s)
- Large Resulting Raid 0 Storage Space - 2TB
We would like to thank each of the manufacturers for their co-operation in helping this testing happen. This is especially true of HighPoint and Samsung who each went out of their way to assist with getting us hardware for these tests.
It is important to note the performance of the Samsung drives used in testing as it will affect performance.
|
|
Drive Configuration
|
Capacity |
500 GB |
|
Interface |
Serial ATA 3.0 Gbps |
|
Buffer DRAM Size |
16 MB |
|
Byte per Sector |
512 |
| Rotational Speed |
7200 RPM |
Performance Specifications
|
Seek time - track to track |
0.8 ms |
|
Seek time - average |
8.9 ms |
|
Seek time - full stroke |
18 ms |
|
Average Latency |
4.17 ms |
|
Rotational Speed |
7200 rpm |
|
Drive Reday Time |
10 sec |
Reliability Specifications
|
Non-recoverable Read Error |
1 sector in 10^14 bits |
|
Start/Stop Cycles |
50,000 |
|
Component Design Life |
5 years |
Acoustics
|
Idle |
2.7 bel |
|
Random Read |
2.9 bel |
|
Random Write |
2.9 bel |
Environmental Specifications
|
Operating Temperature |
0 to 60 °C |
|
Non-operatingTemperature |
-40 to 70 °C |
|
Operating Humidity |
5 to 90 % |
|
Non-operationg Humidity |
5 to 95 % |
|
Operating Liner Shock |
63 G |
|
Non-operating Liner Shock |
300 G |
|
Operating Altitude |
-1,000 to 10,000 ft |
|
Non-operating Altitude |
-1,000 to 40,000 ft |
Power Requirenents
|
Voltage |
+12V±10% |
|
Spin-up Current |
2,000 mA |
|
Seek |
10.6 W |
|
Read/Write |
10.0 W |
|
Idle |
8.2 W |
|
Standy |
0.7 W |
|
Sleep |
0.7 W |
Physical Dimension
|
Height (Max.) |
25.4 mm |
|
Width |
101.6 mm |
|
Length |
146 mm |
|
Weight |
639 g |
|
Drive Configuration
| Capacity |
1 TB |
| Interface |
Serial ATA 3.0 Gbps |
| Buffer DRAM Size |
32 MB |
| Byte per Sector |
512 bytes |
| Rotational Speed |
7200 RPM |
Performance Specifications
| Average Seek time(typical) |
8.9 ms |
| Average Latency |
4.17 ms |
| Data Transfer Rate / Media to/from Buffer(Max.) |
175 MB/sec |
| Data Transfer Rate / Buffer to/from Host(Max.) |
300 MB/sec |
| Drive Reday Time(typical) |
12 sec |
Reliability Specifications
| Non-recoverable Read Error |
1 sector in 10^15 bits |
| Start/Stop Cycles |
50,000 |
Acoustics
| Idle |
2.70 Bel |
| Performance Seek |
2.90 Bel |
Environmental Specifications
| Temperature / Operating |
0 ~ 60 °C |
| Temperature / Non-operating |
-40 ~ 70 °C |
| Humidity(non-condensing) / Operating |
5 ~ 90 % |
| Humidity(non-condensing) / Non-operating |
5 ~ 95 % |
| Liner Shock(1/2 sine pulse) / Operating |
70 G |
| Liner Shock(1/2 sine pulse) / Non-operating |
350 G |
| Altitude(relative to sea level) / Operating |
-300 to 3,000 m |
| Altitude(relative to sea level) / Non-operating |
-300 to 12,000 m |
Power Requirements
| Voltage |
+5V±5%, +12V±10% |
| Spin-up Current (Max.) |
2.4 A |
| Seek (typical) |
8.6 W |
| Read/Write (typical) |
7.9 W |
| Idle (typical) |
7.7 W |
| Standby (typical) |
0.8/1.2 W |
| Sleep (typical) |
0.8/1.2 W |
Physical Dimension
| Height (Max.) |
26.1 mm |
| Width |
101.5 mm |
| Length |
147.0 mm |
| Weight (avg.) |
640 g |
|
