Articles: Memory
 

Bookmark and Share

(9) 
Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 ]

Testing Participants Technical Specifications

Before we move on to the actual benchmark results of our today’s DDR2 SDRAM kits, let’s sum up all the nominal specifications of our today’s testing participants side by side:

Corsair Dominator
TWIN2X2048-9136C5D

Corsair Dominator
TWIN2X2048-10000C5DF

OCZ DDR2 PC2-9200
FlexXLC Edition

Capacity

2 x 1GB

2 x 1GB

2 x 1GB

Frequency

1142 MHz

1250 MHz

1150 MHz

Timings

5-5-5-15-2T

5-5-5-18-2T

5-5-5-18-2T

Voltage

2.1 V

2.4 V

2.3 V

Chips

Micron D9GMH
(B6-3)

Micron D9GMH
(B6-25E)

Micron D9GMH
(B6-3)

Cooling
system

DHX heat-spreaders

DHX heat-spreaders
and
Dominator AirFlow Fan

FlexXLS heat-spreaders
May use liquid-cooling

Overclocking

Now that we have discussed the default specifications of the high-speed and at the same time pretty expensive memory kits, we get very curious to find out how well they can do during overclocking. For our overclocking experiments I put together a testbed with an overclocked Intel Core 2 Extreme processor and eVGA 122-CK-NF67-A1 mainboard (Designed by Nvidia) based on the most suitable for DDR2 SDRAM overclocking Nvidia nForce 680i SLI core logic set. So, our test platform looked as follows:

  • Intel Core 2 Extreme X6800 CPU working at 3.2GHz (8 x 400MHz);
  • eVGA 122-CK-NF67-A1 mainboard (LGA775, NVIDIA nForce 680i SLI);
  • PowerColor X1900 XTX 512MB graphics card (PCI-E x16);
  • Western Digital WD1500AHFD HDD (SATA150);
  • Microsoft Windows Vista x86 OS.

Since Nvidia nForce 680i SLI chipset allows pseudo-synchronous memory clocking, the FSB frequency remained unchanged during our memory overclocking experiments. To ensure the system was running stably during DDR2 SDRAM overclocking, we used Memtest86, S&M and SP2004/ORTHOS utilities.

First of all I decided to find the maximum frequencies for the most widely spread timing sets when the memory modules can work stably and reliably. All the memory kits were tested at their nominal voltage. The first batch of tests was run with Command Rate set to 2T. This allows the memory modules to hit higher working frequencies on the one hand and to project the obtained results to systems on Intel P965 chipset, where Command Rate parameter is locked strictly at 2T and cannot be changed.

So, you can see the maximum frequencies for each set of timings on the chart below. These are the frequencies when the memory remained stable and reliable.

With “easy’ timings of 5-5-5-15, which are practically nominal for all the modules participating, the best results belong to Corsair Dominator TWIN2X2048-10000C5DF, which managed to overclock to 1257MHz. This result is quite logical, because this memory is designed for the highest frequency, even according to the specifications. OCZ DDR2 PC2-9200 FlexXLC Edition and Corsair Dominator TWIN2X2048-9136C5D memory modules hit the same frequency of 1155MHz, which is also not surprising at all, as they both use identical chips. I would like to stress that these results indicate unambiguously that all high-speed memory modules are running almost at the utmost of their potential. They are very uneager to overclock beyond the specified frequencies. This once again proves that the frequency potential of the DDR2 SDRAM chips has been fully exhausted.

You can push the frequency a little bit higher, at least on DDR2-1150 SDRAM modules, by raising the voltage. For example, if you set the voltage of Corsair Dominator TWIN2X2048-9136C5D modules to 2.3V, you can push their frequency to 1164MHz with 5-5-5-15 timings and without losing operational stability. You can achieve the same with OCZ DDR2 PC2-9200 FlexXLC Edition, however in this case you will have to raise the voltage to 2.35V and use liquid-cooling for better heat dissipation.

By the way, since we came to speak about the option of using liquid-cooling systems with OCZ DDR2 PC2-9200 FlexXLC Edition memory modules. I have to say that the use of liquid-cooling solution hardly improves the overclocking potential of these modules. Although I have to admit that the operational temperature in this case gets considerably lower.

As for the opportunity to use the tested solutions with more aggressive timings, Corsair Dominator TWIN2X2048-10000C5DF continues to break all overclocking records here. The especially impressive result can be obtained with timing set to 4-4-4-12, as the memory in this case can run at frequencies up to 1143MHz. The memory modules on Micron D9GMH (B6-3) chips can hardly reach 1GHz frequency with the same timing setting.

We have every right to say the same about our testing participants running with 3-3-3-10 timings. Only the fastest Corsair Dominator TWIN2X2048-10000C5DF can overcome the psychologically important frequency level of 800MHz.

Besides the tests with 2T Command rate, we would also like to see the results with Command rate set to 1T. In this case the memory subsystem latency gets significantly lower. However, the test session demonstrated that the overclocking potential of DDR2 SDRAM drops dramatically with Command Rate set to 1T.

From the peak frequency standpoint, OCZ DDR2 PC2-9200 FlexXLC Edition appears the leader with 1T Command Rate setting. Corsair Dominator TWIN2X2048-9136C5D can also boast good functionality at 800MHz with 4-4-4-12 timings. However the DDR2-1250 Corsair Dominator TWIN2X2048-10000C5DF kit that was ahead of the others in the test session with 2T Command Rate, could only work stably at 800MHz with the timings set to 5-5-5-15.

However, you shouldn’t forget that using high-frequency memory with 1T Command Rate setting actually contradicts the mere idea of the high-speed memory: its ability to hit ultra high frequencies. You should keep in mind that overclocker memory manufacturers usually offer other specialized solutions intended to work with lower latencies.

 
Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 ]

Discussion

Comments currently: 9
Discussion started: 05/05/07 03:34:11 AM
Latest comment: 01/05/10 11:01:49 PM

View comments

Add your Comment