Here's a rundown... btw [link lost in import] has a link to a page
defining RAM terms + pros/cons that's slightly out of date, but covers
everything but the newest stuff.
> Hi everyone. I'm a big-time lurker here and occasionally make a post
>worth reading <g>. Have a tech question that I'm kinda unsure about. Most
>of you seem to know your stuff when it comes to the innards of 'puters.
> 1. What is the difference between the different types of RAM: DRAM, SDRAM,
>etc.
Dynamic RAM, Synchronous Dynamic RAM, EDO=Extended Data Out, FPRAM=Fast-page
mode RAM, VRAM=Video RAM, many more. Lot's of differences, but the bottom
line is they employ different design features, caching, and lookup
algorithms that provide different levels of performance dependant on their
use. SDRAM is nice because it communicates with the memory bus at system
speeds limiting wait-states (broadly, the amount of time the rest of the
system waits for a response from memory). EDO and FPRAM aren't 'synchronous'
or aren't linked with the rest of the system speeds, their speed rating is
independant of the system bus. In both cases, they employ different lookup
routines to enhance their response time. VRAM is fast, expensive, and used
mostly in video cards. WRAM is (afaik) Matrox-specific windows RAM (not MS
Windows windows) that is faster than your average VRAM (so they say, but
their cards are nice).
SIMM=Single inline memory module, DIMM=Dual etc. Performance wise? Other
factors determine theoretical comparisons, but SDRAM (which you'll find in
the average new PC, Macs, and workstations) comes only in DIMM packaging
(again, afaik). SDRAM is the current fastest memory you'll find **outside
of special applications**
Visually, the difference between the two is you'll see chips on one side of
the packaging (PCB board the chips are attached to) in a SIMM. The chips
are attached to both sides of a DIMM. SIMM and DIMM refer only to how the
chips are aligned/attached on the board (for a broad definition). With
SIMMs on many systems, you had to upgrade memory in pairs of boards, with a
DIMM, the 'pair' is all on one board.
TX and BX are apples and oranges. TX is a Socket-7 chipset, and is for
Pentium and Pentium-MMX class processors, like AMD K6, Pentium 200MMX, etc.
BX is a Slot-One chipset for Pentium II processors (though you can setup
Celerons on a BX, if you know what you're doing). In other words, go for
the Pentium II class chipsets - I like the BX, fwiw. Latest K6-2 come real
close, but for the money, Celeron A series (especially, 300A) are hot *if*
you can overclock successfully on a BX board (mine runs solid at 450mhz and
cost me $140). Standard overclocking disclaimers apply, i.e., doesn't
always work, and heat can kill your system - dead, unless you take
appropriate measures and know what you're doing.
PII and Celeron (don't forget the Celeron "A") are better performing (newer,
faster, more expensive - usually) than the K6 and Cyrix chips (M2?).
Differences include smaller L1 (internal) cache on non-Intel processors
(except for Celeron A, but this is mostly overcome for other design
reasons), slower FPU (floating-point unit on the processor), etc. On the
plus side for AMD k6-2+ are a number of integer related performance
improvements, and a hot multimedia instruction set that, if developers and
OS's take advantage of it, should rock.
Rendition based boards can be had just for the asking (and less than $100).
If you play only games that take advantage of Redline, you'll do fine. 3Dfx
produces the Voodoo family of chipsets that are hot for 3D games, and Riva
TNT (while fighting driver-related issues) will cost you, but in many cases
give the Voodoo a run for the money (or beat it).
You're probably referring to Voodoo card setups. Pop the card in another
PCI slot, plug the monitor into the card, and you 'regular' video card into
the Voodoo (with the included 'pass-through' cable), and the Voodoo will
handle you're Direct3D, OpenGL (sort of) and Glide (3Dfx- proprietary)
games. Windows and non-3D enhanced titles run through your regular video
card, hence the term "pass-through cable". To further confuse, you can
actually have two Voodoo 2 based cards, and one (or more, under Windows 98)
2D or 2D/3D combo card (like the Rendition 2200, Riva TNT or Matrox G200).
If you have the slots, that is. But then you're talking about $400-$500 and
one *hell* of a 3D *** system!. Enjoy.
>Jimmy McKinley
>USA Series Director
>http://www.kjsl.com/~jmckin12/usa/index.htm
>Holeshot Engineering Chevrolet #12
>ICQ # - 20750220
Hi everyone. I'm a big-time lurker here and occasionally make a post
worth reading <g>. Have a tech question that I'm kinda unsure about. Most
of you seem to know your stuff when it comes to the innards of 'puters.
1. What is the difference between the different types of RAM: DRAM, SDRAM,
etc.
2. What about the difference between SIMMs and DIMMs performance-wise? How
do I know which is which by looking at it?
3. I want to build a new computer. I have the housing, but would like
suggestions as to the different types of mother boards and CPUs, i.e., what is
the performance difference between a TX and BX chip set? What about
performance differences between P-II, P-II Celeron, AMD K6, and Cyrix?
4. Video cards? I'll most definitely be running N2, N2/1999, and ICR2. But,
right now, I don't have a Rendition card (I have a Matrox Mystique 2Mb). I
don't want to spend an outrageous amount of money either. I'll be running
these in DOS, unless the wife lets me get onto TEN....<g>
5. What is the purpose of having two video cards? I haven't figured out that
one yet. Is one the main card and the other to support the Rendition graphics
or 3Dfx? How do you go about putting two cards in the machine?
Jimmy McKinley
USA Series Director
http://www.kjsl.com/~jmckin12/usa/index.htm
Holeshot Engineering Chevrolet #12
ICQ # - 20750220
