Overclocking

ASUS M2N32-SLI Deluxe
Overclocking Testbed
Processor: AM2 4800+ (X2, 2.4GHz, 1MB Cache per core)
CPU Voltage: 1.525V (default 1.4V)
Cooling: AMD Stock Heatpipe FX62 Cooler
Power Supply: OCZ Power Stream 520W
Memory: Corsair Twin2x2048-PC2-8500C5 (2x1GB)
(Micron Memory Chips)
Hard Drive Hitachi 250GB 7200RPM SATA2 16MB Cache
Maximum OC:
(Standard Ratio)
260x12 (5x HT, 3-3-3-Auto)
3120MHz (+30%)
Maximum FSB:
(Lower Ratio)
350 x 9 (4x HT, 3-3-3-Auto)
(3042MHz, 2 DIMMs in DC mode)
(+75% Bus Overclock)

ASUS told us in meetings that the M2N32-SLI Deluxe was capable of overclocking to a 350 CPU clock speed. At 1.525V on the CPU we reached a highest stable speed of 350x9. This is a 75% bus overclock and the highest speed we have ever reached in any of the AM2 motherboard tests. The M2N32-SLI Deluxe was extremely stable at this speed, and we may have reached even higher speeds if we had dropped the multiplier further, increased the voltage more, or used better CPU cooling. We stayed at the 9x multiplier for consistency with other test results.

At the stock 12x multiplier the system reached 260x12, 3.12GHz, at the same voltage. This is again the highest this CPU has achieved. The ASUS 8-phase design appears to work very well in maximizing stability in overclocking. The passive cooling system was also very effective even at these high overclocks. Again, you may be able to achieve even better results with some active cooling on the chipsets. The SPP, in particular, seems to get hotter than the new 590 MCP and it may benefit from active cooling.

It was very satisfying to reach a stable 350 MHz with a system with passive cooling. The only active fan on the board was the CPU cooler fan, and even that was a stock AMD cooling fan with a heatpipe.

Memory Testing
Optimum tRAS

DDR2 memory behaves quite differently in ASUS tRAS testing than we have seen on other AM2 boards. As you can see from the standard chart below, a 2GB kit of Corsair 8500 (DDR2-1066) experienced the SAME bandwidth no matter what tRAS setting was used. ASUS is apparently controlling and optimizing tRAS timings internally.

Memtest86 Bandwidth
ASUS M2N32-SLI Deluxe with AM2 4800+
6 tRAS 3138
7 tRAS 3138
8 tRAS 3138
9 tRAS 3138
10 tRAS 3138
11 tRAS 3138
12 tRAS 3138
13 tRAS 3138
14 tRAS 3138
15 tRAS 3138
16 tRAS 3138
17 tRAS 3138
18 tRAS 3138
AUTO
(CPUZ tRAS 18 Reported)
3248

To further test this idea, we also ran tRAS at the AUTO setting, which yielded the highest memory bandwidth in the tRAS tests. To determine the tRAS that was being set by AUTO we looked at memory timings with CPUZ. That utility reported tRAS of 18, which is the highest available setting. We do not really know whether 19 is being set by auto or whether AUTO merely reports the highest setting but then dynamically adjusts tRAS.

Whatever the mechanism, a tRAS setting of AUTO produced the highest memory bandwidth. All stock benchmarking was performed with Corsair 8500 settings of DDR2-800 at 3-3-3-Auto settings at 2.225V.

Memory Stress Testing

The wide range of voltage controls on the ASUS M2N32-SLI Deluxe help users to get the most out of overclocking and the best performance from memory modules. Most DDR2-800 memory is rated at conservative 5-5-5-15 timings, but that's also assuming you only run with the default 1.8 V. We have found in our DDR2 testing that many modules are easily able to run 4-4-4-12 timings at 2.0 V, and the Corsair modules we are testing with (rated at 5-5-5-18 DDR2-1066) are generally able to run with 3-3-3-9 timings with approximately 2.2 V.

ASUS M2N32-SLI Deluxe
DDR2-800 Timings - 2 DIMMs
(2/4 slots populated - 1 Dual-Channel Bank)
Clock Speed: 200MHz
Timing Mode: 800MHz - Default
CAS Latency: 3
RAS to CAS Delay: 3
RAS Precharge: 3
RAS Cycle Time: Auto (18 Reported)
Command Rate: 1T
Voltage: 2.225V

With two DIMMs installed, testing was completely stable at 3-3-3-Auto 1T timings at DDR2-800. This is very rare for two reasons. First, AMD does not really support 1T timings with DDR2-800 in the current on-chip memory controller, and Gigabyte is the only other board we have tested that can also operate at 1T timings. All of our utilities report operation is at 1T also.

Several benchmarks were run to compare performance at 1T and 2T timings, but we could not find any real-world performance improvements with the 1T timings compared to 2T. AMD is said to be readying a revision to the AM2 memory controller that will officially support 1T Command Rate. Perhaps that new design with support for 1T will make a more significant improvement in performance.

ASUS M2N32-SLI Deluxe
DDR2-800 Timings - 4 DIMMs
(4/4 slots populated - 2 Dual-Channel Banks)
Clock Speed: 200MHz
Timing Mode: 800MHz - Default
CAS Latency: 3
RAS to CAS Delay: 3
RAS Precharge: 3
RAS Cycle Time: 10
Command Rate: 2T
Voltage: 2.225V

The ASUS M2N32-SLI also exhibited remarkable performance with four 1GB DIMMs. Installing four DIMMs stresses the memory subsystem further, but the ASUS was still stable at the same 3-3-3-Auto timings that worked best with 2 DIMMs. As expected, we did have to drop command rate to 2T but we doubt you can measure that impact on a memory controller designed for 2T operation.

The ASUS is the only AM2 motherboard we have tested so far that was able to run four 1GB DIMMs at 3-3-3 timings. That performance speaks very well for the stability of the ASUS design, since the memory controller in all our motherboard tests is the same initial release AM2 CPU.

ASUS: Features & Board Layout Epox: Features & Board Layout
Comments Locked

23 Comments

View All Comments

  • Missing Ghost - Wednesday, June 28, 2006 - link

    yep the choice of I/O ports on the back panel is pretty poor
  • Larso - Wednesday, June 28, 2006 - link

    Perhaps someone should clarify this for me. I have been noticing how there is a growing interest in how the motherboard makers have done the power conversion for the CPU. Why has this become an issue to investigate?

    I don't think there was a similar focus on the power converters for the netburst chips, which I believe soaked a lot more juice than these AMD chips. I believe they managed to deliver stable power to the netbursts without needing an 8 phase converter cooled by heatpipes??

    I'm fearing that the motherboard producers will start to create extravagant and foolishly designed converters to please the reviewers. I believe there is no good reason to go for an 8 phase design, when a 4 phase would do the job, considering the money that can be spend on each phase. And that ASUS need to cool the converter by heatpipe seems to indicate that the convertion have a bad effeciency, is this really a step forward?

    I really enjoy reading more about the technical solutions on the motherboards, but it just seems to me that the power converter should be less of an issue now, than with the power hungry netburts?
  • Operandi - Friday, June 30, 2006 - link

    A 8 phase design is more efficient then a 4 phase, not less.
  • Missing Ghost - Wednesday, June 28, 2006 - link

    It seems to me that the heatpipes are more for cooling the chipset in fact.
  • erwos - Wednesday, June 28, 2006 - link

    What WiFi chipset does it use? "It has WiFi" is not terribly precise.

    -Erwos
  • Wesley Fink - Wednesday, June 28, 2006 - link

    The WiFi chip used on the Asus WiFi module is Realtek RTL8187L. The Features have been updated to reflect this.
  • Wesley Fink - Wednesday, June 28, 2006 - link

    Asus refers to the Wi-Fi in specifications as "WiFi Home USB wireless module supporting IEE 802.11 b/g". There is no mention at all of the supporting chip. The module is attached to the motherboard and the chip is loacated behind heatpipes. We can't read the model number, but the logo is clearly Realtek.
  • highlnder69 - Wednesday, June 28, 2006 - link

    On page 8 under Half Life 2 - SLI Gaming Performance, I think that the Asus Single/SLI results are labeled incorrectly. It's currently showing the Single card configuration with the highest FPS results and the SLI with the lowest.
  • Wesley Fink - Wednesday, June 28, 2006 - link

    You are correct, the labels were reversed. They are now corrected.
  • DigitalFreak - Wednesday, June 28, 2006 - link

    Someone needs to teach the idiots at Asus about how to design the proper motherboard layout. With an SLI setup with dual width cards, there's no PCI-E slots available.

Log in

Don't have an account? Sign up now