Date: April 5th, 2003
Article by: Nathan Glentworth
(Owner, Head Editor & Hardware Reviewer)
Product was donated by: OCZ Technology
<--Shop for this item
SiSoft SANDRA Memory Bandwidth Module (Settings: CAS2 and Beyond)
Ok, now we bump up the CAS latency to CAS2 and see what comes about.
Now, to show the reader where the OCZ really starts to out perform, I
tightened up it's memory settings as stated below. The OCZ memory was
the only memory in this review that was able to handle the tweaking.
The OCZ tweaked settings were:
- DRAM Settings set from By SPD to Manual
- SDRAM CAS Latency was manually set at CAS2. (Computer would not boot
at CAS1.5 setting.)
- The bank interleave was set to 4-bank from default settings. (Highest
- The Precharge to Active to Active (Trp) was set from 3T to 2T. (Highest
- The Active to Precharge (Tras) was set from 7T to 6T. (Highest Available
- The Active to CMD (Trcd) was set from 3T to 2T. (Highest Available Setting)
The result was:
The increase noted wasn't overly dramatic, but it was an improvement.
The performance increase of about 20MB/sec would not be really noticed
in everyday computer usage, but a free increase in performance is always
good whether big or small. One theory I am putting forward is that the
OCZ memory's onboard EPROM may already have the memory settings close
to maximum in its "By SPD" configuration.
Another real good point I have to state is that even with the memory
timing tightened up quite a bit, the memory never had one single problem.
prove the stability, I fired up Unreal Tournament 2003 and sat down
with a couple cups of coffee and fragged my butt off for two whole hours.
Everything ran smoothly and not one hiccup noted.
Ok, now you are probably waiting for the Mushkin performance results.
Well, I would give you some if it could even boot up at CAS2. That's right,
black screen, nadda, zilch, nothing. With the memory voltage maxed at
2.8 volts, the Mushkin didn't want to come out and play. After grumbling
to myself and having to completely clear my bios, I popped in the Crucial
sample and set it to CAS2. Now, it's stock value is CAS2.5, but it booted
up fine at the higher setting. But, upping the CAS Latency to CAS2 is
the only thing I could muster out of it.
And the result was:
As stated in my Crucial DDR333 ram review HERE,
I am indeed impressed by the sample's 10% increase in overall performance
when upping only the CAS Latency. I only wish I could tweak it even more.