OCZ EL DDR PC-4200 Dual Channel

Memory by KeithSuppe @ 2004-01-17

Has this kit the potential to provide tight timings at lower FSB? While also providing head-room for the continuous rise in memory speeds? Find out in this review.

  • next

Introduction

Madshrimps (c)


Introduction:

OCZ Technology, has been making some remarkable strides in the semi-conductor field. Their high speed DDR, has been labeled the "fastest memory available" by AnandTech and has been received warmly by just about every review done. Currently IC production for 550MHz speed DDR, is non-existent, and what makers are limited to, would be DDR400 capable of DDR533. This in itself is remarkable, and while all the credit can't go to the Fabs, themselves, it does say a lot for Samsung and Hynix. It's the latter OCZ chose, and at some cost. Since the Tariff's were enforced Hynix IC's are somewhat costly. Most of the PC4000 to PC4200 found today derived from either Hynix Rev. B, or Samsung TCCC. In this case, however; I believe OCZ is using something different. The memory has been given the moniker "universal" DDR because it's ability to run CAS-2 at 400MHz, as well as CAS-2.5 at 533Mhz. I ran this memory up to 460MHz (230FSB 1:1) at CAS-2-3-3-6. Pretty impressive for memory, which also climbed as high as 570MHz(285FSB 1:1) at 2.5-4-4-8, and in both cases the VDIMM was just 2.85V, considering it really wants 3.0V. The memory remains under warranty up to 2.9V, and I cannot say I've done this memory justice, as it would surely would have performed much better with more VDIMM. I may have seen 300FSB or 600MHz DDR, at 3.0V. Given the IC's flexibility, and high speed I believe OCZ may be using Hynix HY5DU56822CT-D Rev. C, Oct. 2003, instead of the usual Rev.B HY5DU56822BT-D Aug. 2003.

In so far as (adequate) VDIMM is concerned, unfortunately I fried my Epox 4PCA3+ when I replaced the NB/HS after lapping both. I mention this to forewarn others to leave the chip itself alone. Irregardless, I'm restricted to 2.85V using my Asus P4C800E-Deluxe, an excellent platform, but certainly not without its faults. Its Vcore is often much higher then the BIOS setting, and the VDIMM could be somewhat higher. It perplexes (or frustrates) me that Asus would allow the Vcore adjustable to 1.90V (which is high even for phase-change) and leave so much to be desired where VDIMM is concerned. I ran several different benchmarks, to get a feel for the memory's overall effect on system performance. Rather then restricting the benchmarks to Sandra, and Aida, I ran UT2003, CodeCreatures and GunMetal. The system listed below was used for testing throughout

Liquid3D's Test Setup
CPU P4 2.4C L310A735-0101 (with IHS removed) AS5 TIM
Mainboard Asus P4C800E-Deluxe (rev.1 BIOS 1013)
Cooling Danger Den RBX & Z-chipset blocks, Hydor L30 (320Gph), Black Ice Extreme, 2x120mm/92CFM (push/pull) Sunon.
Memory OCZ EL4200 Dual Channel
Video Asus V9950 (FX5900/128MB Detonator 53.03)


Let's see how this memory performs ->
  • next