| about us | advertise | careers | links |

[an error occurred while processing this directive]


A Tale Of Two Mushkins: PC133 CAS3 VS. PC100 CAS2

There's nothing better than a good fight and SDRAM makes for particularly good ammunition. In automotive circles, Chevy vs. Ford always makes the interested parties' blood boil, but what about a Chevy vs. Chevy dukem out? How does Mushkin stand up against itself? For that matter, what is CAS and what does 2 or 3 designate? For this comparison, I've nabbed two 128 MB strips of Mushkin's DIMM type RAM and benchmarked the results for your perusal.

Factory Specifications
The Strips

PC133 CAS3

  • 7.5 nS Nanya Chips
  • 16 x 8MB (128MB)
  • Lifetime Mushkin Warranty

PC100 CAS2

  • 8 nS Infineon Chips
  • 16 x 8MB (128MB)
  • Lifetime Mushkin Warranty
The Case Of CAS

CAS literally means column address strobe, which is fancy terminology for how fast the RAM in your precious box can process temporary information. The common designations for SDRAM has been 2 and 3. The 2 signifies two cycles and the 3 signifies 3 cycles. Because more cycles takes more time, the lower your CAS setting, the quicker your RAM is at a particular FSB.

PC "XXX" Speeds

PC100 theoretically gives rise to a timing spec of 10 nS. A nS is a short period indeed, being 1E-9 or 0.000000001 seconds. Fast, eh? An easy method of calculating the nS required to meet a certain MHz requirement is to use the 1000 (MHz-nS) rule. For example: 1000 (MHz-nS) / 7.5 nS = 133 MHz (rounded from 133.33). So if somebody tells you they have 5 nS RAM, that would imply a speed of 200 MHz. The vast majority of the time, this one or two digit nS rating can be easily found by closely examining the chip at hand.

The Test Rig

The motherboard of choice was the Abit BE6-II combined with Maxtor's excellent 30.7GB ATA66 HD. Alas, all I had on hand, other than the ubiquitous TNT2 Ultra AGP card, was a tired old 1 MB PCI video card. This served to stabilize the setup and offset the very unstable AGP overclocking speeds of over 91 MHz, which the TNT2 Ultra definitely did not like. At 137 MHz, the PCI speed is only 2.7% out of spec, which is usually well within manufacturer's tolerances.

Mushkin PC133 CAS3

Since this is rated for 133 MHz FSB, it should be stable at least at that frequency. As the benchmarks revealed, this Nanya chipped strip does an excellent job all the way up to the 650E Coppermine's limit of 137 MHz FSB. As I suspected even before AnandTech's PC133 roundup was published, this strip had more to give. Under the right conditions, it tops out at 146 MHz FSB if and only if your video card and processor chip are fully stable at that high speed. Interestingly enough, this CAS3 sample would run at CAS2; however, a lowering of the FSB all the way down to 99 MHz was required. At 100 MHz FSB, the computer wouldn't even boot let alone run stably. Whoever Nanya is, they make a solid, stable, high performing SDRAM chip.

Mushkin PC100 CAS2

At the time of purchase, this CAS2 sample cost slightly more than the faster FSB rated Nanya type. Is Mushkin pulling a fast one here? Not at all. Upon closer inspection, I spied the 8 nS rating, which would indicate that this sample would be stable at 125 MHz although not necessarily at CAS2. The tests show full stability at CAS2 at up to a 123 MHz FSB. CAS3 tests reached (again) the limits of the 650E at 137 MHz. This was much more performance than I ever expected from a PC100 rated strip.






Web Target PC




 

[an error occurred while processing this directive]

Contact us | About us | Advertise
Copyright © 1999-2007 TargetPC.com. All rights reserved. Privacy information.


targetpc