![[H]ardOCP](/images/blank.gif){kind=small}
- Date:
- Friday , August 01, 2003
- Author:
- Keith Dugger
- Editor:
- Kyle Bennett
- Google +1
EPoX 8RDA3+
EPoX has released their nForce2 Ultra 400-powered AMD motherboard during highly competitive times. Will the 8RDA3+ perform well and maintain the EPoX reputation of an enthusiast manufacturer?
Subsystem Testing
Audio – CPU Utilization
We use Ziff Davis’ Audio Winbench to test the CPU utilization of audio subsystems. When designed and implemented correctly, the audio subsystem should have minimal impact on the CPU under use. The less CPU used for sound frees cycles for racking up frags in your favorite FPS.
The audio subsystem EPoX implemented on the 8RDA3+ is from C-Media and it performed flawlessly. Any result less than 5% is acceptable and the C-Media chip performed well below 3% in most tests and only once rose even close to the 5% mark. These are superb results and will have almost zero impact on system performance. Please be aware that these results and analysis are based on a synthetic benchmark and may not represent what you would see on your own system.
Audio – Subjective Listening
Performance numbers offered up by a synthetic test can only measure a portion of the audio subsystem. A large part of the enjoyment of a sound card comes from the way it sounds. I listened to MC Hammer's track U Can’t Touch This. I chose this song because it includes a chest-pounding baseline and speaker-stressing highs that fully exercise the limits of the C-Media sound subsystem.
There were no audio defects at any volume. The performance of this sound subsystem is perfect.
Audio – In Game Testing
One of the other factors of sound subsystem quality is accurate sound reproduction during game play. Our very own Brent Justice provided his custom-designed benchmarking script based on recorded BOT matches to adequately test sound quality during game play. The benchmark runs through three loops of each demo, one with no sound, one with normal 3d sound, and, finally, one with EAX 3d sound.
The 8RDA3+, at low quality, performed as expected, with performance being more affected during this test. The effect was really minimal as the difference between 80 and 90fps will be hardly noticeable. In the high quality test, the sound subsystem did not adversely affect performance at all as the scores were equal across all three loops.
A part of this test is subjective as well and I turned up the volume, but did not hear any sound impurities during any of the tests.
USB 2.0 / IEEE 1394
Pairing a USB 2.0 / IEEE 1394 external drive from ACOMDATA HD060U2FE-72-USB 2.0/FireWire HDD with TCD Labs’ HDTach program gives us the ability to evaluate the capabilities of the onboard USB 2.0 and IEEE 1394 controllers. The first test is with the drive connected to the USB 2.0 port and the second is with the drive connected to the IEEE 1494 port.
A RAT (Random Access Time) of 19.5ms is decent for a drive connected to a USB2.0 port as is the average write speed of roughly 14.5 MB/s. The read speed of just under 21 MB/s is not perfect, but acceptable. The CPU utilization of 16% is really good for USB2.0.
A RAT of 19.5ms falls in line with that of the USB2.0 test, but IEEE 1394 really shows its performance capabilities with an average read speed of 30 MB/s. While the write speed of 14 MB/s is not terrific, the average CPU utilization of 5.7% will only partially affect system performance.
IDE/ATA Performance
The performance of the CPU, RAM, and hard drives affect overall usability of a system. Using the same TCD Labs’ HDTach, I tested the IDE performance of the 8RDA3+. My test bench has Maxtor 40Gb ATA 133 model 6E040L0 hard drives on the IDE controllers and Seagate 80 Gb Barracuda SATA hard drives for SATA testing.
Testing was conducted with an IDE drive connected in a primary slave configuration, single SATA drive, SATA RAID 0, and, finally, SATA RAID 1.
The parallel IDE controller on this EPoX board put up some good numbers. Read speed was excellent at 51 MB/s and the RBS (Read Burst Speed) of 112.5 MB/s is a terrific score. The average write speed of almost 32 MB/s is also an admirable result. However, the CPU utilization is a high 6.0% and, if combined with other system activities, may slightly impact overall performance when in use.
The stand alone SATA test resulted in typical scores. Read/write averages and RBS numbers were good, but the RAT was a few tenths of a millisecond off. It probably won’t be noticeable, but 13ms is what I consider minimum for these drives. The CPU utilization is a little higher than I have seen on other boards, but is only perhaps a minor issue to performance.
As expected the RAID 0 with a block size of 16K tests performed better than at 128K or RAID 1. RAID 0/16K posted almost 60 MB/s read average mated with an impressive 35 MB/s write average. Combine those results with an adequate RBS of over 100 MB/s and this controller will transfer files at good speeds. The RAID 0/128K raised the write average to 36.5 MB/s. It is normal to see the 128K block size suffer in read performance as a larger block size is better suited for write operations.
The RAID 1 test performed as well as the stand alone SATA drive as should be the case. The average write speed was minimally impacted by this setup. RAID 1 is not designed for performance, but instead offers terrific redundancy, so I can accept these results.
Network Utilization Tests
I connected my home machine via an Intel Gigabit NIC (configured at 100 Mbps/Full Duplex) and cross-over cable to the EPoX 8RDA3+ to adequately test the network performance of the two on-board NICs. These tests are conducted by performing two series of tests. First, transferring approximately 600MB of various-sized MP3 audio files both upload and download and, second, transferring a single file of approximately 600MB both upload and download. To measure performance I used both Window Task Manager to get CPU utilization and Hagel Technologies’ DU Meter to measure the performance of the network subsystem itself.
The 8RDA3+ comes with two Realtek 10/100 Mbps on-board LAN connections. One directly controlled by the nForce2 MCP (noted as nForce2) and the other sitting on the PCI bus (noted as Realtek).
The average for the nForce2 NIC during the multiple, small files tests was normal. Almost 10 MB/s is decent performance for a 100 Mbps connection. What really surprised me was the ability this controller had at keeping the CPU utilization below 5%. That is the lowest I have seen to date.
The large file upload and download tests were similar to the previous tests. The average CPU utilization was still below 5%. I am really impressed at this number.
The Realtek (PCI) out-performed the nForce2 version slightly, but had difficulties keeping the CPU average below 10%. I shouldn’t complain about this, though; 10% is still an outstanding number for any NIC.
The large files tests still show the Realtek (PCI) is faster at transferring bits around, but again posted results with higher CPU utilization averages that the nForce2.
Both of these controllers will meet your 100 Mbps needs and then some. EPoX did a terrific job pairing these controllers with this board and chipset.
Test Systems
The following system configurations were used for the Sandra memory benchmark graph, as well as all graphs listed under the Application and Gaming Benchmark sections:
EPoX 8RDA3+ (nForce2 Ultra 400) – AMD AlthonXP 2500+ CPU (clocked at 11x166), and an AMD AthlonXP 3200+ CPU (clocked at 11x200) - 2 x 256Mb Corsair XMS3200 - ATI Radeon 9700 Pro w/ ATI Catalyst 2.3 drivers - 40 GB Maxtor ATA133 HDD - Allied 400w PSU - WindowsXP w/SP1
NOTE: For all benchmark tests done on the EPoX 8RDA3+ motherboard, the optimized default BIOS settings were loaded for the benchmark tests. The following BIOS settings were also engaged during the tests: CAS Latency Time = 2; (T)RD = 2; (T)RCD = 2; (T)RAS = 6.
Asus A7N8X (nForce2 Ultra 400) - AMD AthlonXP 3200+ 2.2GHz CPU (clocked at 11x200), 2 x 256MB Corsair XMS3200 @ DCDDR400 2-2-2 - ATI Radeon 9700 Pro w/ ATI Catalyst 2.3 drivers - 40 GB Maxtor ATA133 HDD - Allied 400w PSU - WindowsXP w/SP1
ASUS A7N8X Barton (nForce2): AMD AthlonXP 3000+ CPU (clocked at 13x166); 2 x 256MB Corsair XMS3200; ATI Radeon 9700 Pro, onboard nForce2 10/100 NIC; 40 GB Maxtor ATA133 HDD, Zalman 300w PSU. WindowsXP w/SP1, ATi Catalyst 2.3.
ASUS P4C800 Deluxe (i875P) - Intel Pentium 4 3.0 GHz CPU (clocked at 15x200) - 2 x 256Mb Corsair XMS3200 - ATI Radeon 9700 Pro w/ ATI Catalyst 2.3 drivers - 40 GB Maxtor ATA133 HDD - Allied 400w PSU - WindowsXP w/SP1
ABIT IS7-G (i875P) - Intel Pentium 4 3.0 GHz CPU (clocked at 12x250) - 2 x 256Mb Corsair XMS3200 - ATI Radeon 9700 Pro w/ ATI Catalyst 2.3 drivers - 40 GB Maxtor ATA133 HDD - Allied 400w PSU - WindowsXP w/SP1
Graphs are labeled as follows: Motherboard - CPU Clock - FSB Clock - Memory Clock
SiSoft Sandra Memory Bandwidth Benchmark
Note that all results above were obtained running in Dual Channel mode.
The 8RDA3+ presents a fair showing in the synthetic memory benchmark. The memory scores are where they should be for this chipset and they trade leads with the other nForce2 board.