Date: November 18, 2005
Author(s): Matthew Harris
Just because a memory kit is ‘value’ doesn’t mean it’s not worthy of your attention. Today we are taking a look at a fair priced 2GB kit from Crucial, which rolls in at DDR400 speeds. Let’s see how it stacks up, and see if we can squeeze any overclocking out of it.
The Crucial CT12864Z40B.M16FD kit consists of 2 * 1024MB sticks of some very unusual (for DDR1) ram. It uses BGA chips as opposed to TSOP. This is unusual for DDR, but pretty standard for DDR2, which at first had me wondering if I’d received a DDR2 set by mistake!
But I’m getting ahead of myself…
The ram arrived in a nice little box instead of a blister pack like some makers use.
Inside the box, there were two sticks of ram individually packaged in anti-static bags along with a multi-lingual installation guide that covers every type of ram package from EDO-SIMM to SODIMM to DIMM.
But upon pulling the ram out of it’s anti-static bag I was greeted with this, which was quite a bit different:
The ram is rated at 3-3-3-8 at 200mhz which is fairly standard for value ram. As the speeds of the ram are decreased, the CAS latencies are tightened by SPD to 2.5-3-3-7 at 166mhz to a fairly impressive 2-2-2-6 at 133mhz. This means that if you have an older AXP system riding on the NF-2 chipset (which supports ram of this size), you can get some low latencies for a decent price. The same can be said of the P4 – B series which uses the 533mhz bus since the FSB of the "B" series is 133mhz. A 3.06Ghz HT P4 will be able to benefit from low latencies and is compatible with the 865/875 chipsets for improved bandwidth afforded by dual channel DDR… and everyone that is an Intel faithful knows the P4 craves bandwidth.
Pentium 4 2.4C
Chaintech 9CJS Zenith
Antec 480W PSU
Crucial PC3200 2 * 1024
2 x Western Digital SATA WD1600JD hard drives
Onboard VIA Envy24 sound
eVGA GeForce 6 6800
Unlocked to 16P 6V with the core @ 385mhz, Ram @ 810mhz
Windows XP Professional with SP2
After some work, I managed to uncover the highest stable speed for the Crucial CT12864Z40B.M16FD, which was 214mhz at 2.5-4-4-8 2.89Vdimm. For overclocking, I’m limited to running a divider of 5:4 to keep the ram stable. Oddly enough the CAS# Latency has nothing to do with the stability issue, as it errors at 2.5 and at 3 at the same point, so the issue lies with either RAS# to CAS#, RAS# Precharge or the TRAS#. I just thought I would point this out just in case you were thinking that maybe by raising the CAS# Latency, you’d be able to squeeze more speed from these sticks. That just isn’t the case.
As we can see from Everest at default speed with the CAS settings of 2.5-4-4-8, the Crucial ram has a bit of a lead over the OCZ PC4200 with a latency of 3-4-4-8. I tried booting the OCZ with a comparable latency setting but sadly it won’t boot at anything but 3-4-4-8.
As the FSB speeds increase, we see that the read speed increases and the latency decreases. Oddly enough at 214 1:1 the write speeds are unchanged.
At a FSB of 267 with the divider at 5:4 for a ram speed (again) of 214 you’ll note that the reads have really pepped up with an increase of 481MB/s over the same ram speed at 1:1 and a decrease in overall latency of 10.6ns but the write speeds only see an 89MB/s boost but compared to the OCZ at 200mhz FSB 1:1 you’ll note that it’s still a respectable boost.
In SANDRA, we find that the OCZ and the Crucial are pretty close to being tied at 200FSB 1:1. Aat 214 FSB the performance has picked up close to 225MB/s across the board with the Float becoming the fastest calculation that the ram performs. With the FSB at 267 and a 5:4 divider, this evens up considerably with the Integer once again gaining the lead, but by a very narrow margin. It does however gain well in excess of 600MB/s over the 214FSB 1:1 which so far is the most impressive performance increase we’ve seen.
Just for laughs I decided to do a little gaming to see what difference running the ram at 250FSB with a 5:4 divider versus 267 FSB with the 5:4 divider.
First we’ll begin with the 267FSB readings:
With 250FSB it comes out like so:
You’ll see that in most cases the average framerate suffers somewhat, although this is a cumulative effect of both the lowered FSB and lowered ram bandwidth. As much as I’d like to be able to lower the multiplier on my CPU to allow a CPU speed that is as close as possible in both cases, I’m unable to due to running the P4.
The Need For Speed; Most Wanted demo and HL2 both buck the trend, although these are probably due to inconsistencies in my gaming rather than actual impact from hardware, since in both cases I was unable to accurately duplicate what had transpired in the first test. This is what happens when you play a real game as opposed to running a static benchmark though :)
All in all, since I was previously using 512MB of ram at 1:1 with a FSB of 272, I can say that even with the lower clock speed and decreased bandwidth, the performance increase seen by having 1.5 gigs more ram to work with more than makes up for the the loss in both clockspeed and bandwidth.
This ram is great for those people that are willing to compromise bandwidth for sheer ram density per dollar, for people with older AXP systems that will be able to utilize it at it’s tighter timings and people with higher clocked P4 systems that don’t have a burning desire to overclock the living crap out of their PC’s. That said, this isn’t ram for avid overclockers that are looking to wring the maximum performance from every part since it simply isn’t designed for that. This is a no excuses value part that delivers the biggest amount of ram that you’re going get for a very reasonable price backed with a no excuses warranty from a very top notch organization.
For the value and the commitment of the company backing it I give this ram a 9/10 and an Editors Choice award.
Feel free to comment on this review, or ask questions, in our related thread. You do not have to register in order to post in our content threads.
Copyright © 2005-2018 Techgage Networks Inc. - All Rights Reserved.