Random Phrase: Home sweet home...


Read Me First!

This is the Spode's Abode archive. The old articles and forum have been kept here purely for historical purposes.

This section of the site is not kept up to date and there is no guarantee that all portions of the site transferred to this sub-domain intact.

Please visit the new site!



Guide to Memory Performance, Technology and Purchasing
Written by Spode (24/Nov/04)
Page 5 of 8

Untitled Document

Memory Performance Analysis

Memory performance is not down to the memory alone. In a lot of cases, it is restricted by the chipset. It is this key point that helped differentiate one chipset from another, as memory performance affects overall system performance considerably.

With the Athlon 64 architecture, this is no longer an issue. The memory controller is now integrated onto the CPU die and can bypass the north bridge entirely. This means that performance from chipset to chipset is almost negligible. Price and features are now where it matters.

Intel on the other hand, still rely on the north bridge, so there is still a difference from chipset to chipset.

In order to understand the impact of memory performance, we have taken a fairly scientific approach to showing difference between low and high latency memory, DDR2 and DDR, dual channel memory and of course different frequencies.

In order to do this, we used the incredibly capable Asus P5GDC-V Deluxe motherboard, coupled with a 3.6GHz Prescott processor. This LGA775 motherboard has the 915G chipset and has support for both DDR and DDR-II memory. This makes it an excellent test bed for our memory comparisons. We used an nVidia reference 6800GT (PCI-E) clocked at 400/1100 which is the same speed as a 6800 Ultra. This helps remove any possible bottlenecks. The latest drivers were used, including the latest BIOS from Asus.

Aquamark and 3DMark were all left at default settings. Doom 3 Low setting represents 640x480 with no AA or AF, at a “High” texture level. Doom 3 High setting represents 1280x1024 with 2x AA and 2x AF, at a “High” texture level. UT2004 Soft, represents UT2004 running in software rendering mode at 1600x1200 which is not 3D accelerated – this puts a lot of strain on the CPU and Memory. UT2004 High represents 1280x1024 with 3D acceleration.

During all these testings DDR was kept at X-4-4-8 and the DDR-II was kept at X-4-4-12, where X is the CAS latency..

DDR vs DDR-II

400MHz DDR, C3, Dual Channel, 2 x 512MB

400MHz DDR-II, C3, Dual Channel, 2 x 512MB

533MHz DDR-II, C4, Dual Channel, 2 x 512MB

Aquamark Overall

60,131

60,001

61,191

Aquamark CPU

9,494

9,454

9,999

3DMark 03

11,115

11,094

11,194

3DMark 01

20,359

20,329

21,013

Doom 3 Low

77.3

76.2

79.6

Doom 3 High

69.0

68.7

69.8

UT2004 Soft

17.05

16.97

17.24

UT2004 High

132.87

133.23

137.13

When DDR was originally released, there was initially very little improvement over SDR. It was only until the processors got faster that the benefit was seen. As you can see, when compared like for like and with the same latency, the differences are hardly noticeable with 1fps being the biggest difference in Doom 3 High. We can only predict this is a similar situation.

However, DDR-II memory comes in at 533MHz C4 as standard, so a like for like comparison is hardly fair. When compared at these speeds, we can see DDR-II offers a benefit, but very minimal – 4% at most.


<< Back | 1 | | 2 | | 3 | | 4 | | 5 | | 6 | | 7 | | 8 | Next >>



Copyright Andrew Miller
Please read our disclaimer

Search the site:

Random