|
Introduction
The
GeForce2 MX chipset has been a very dominating
force in the value market because of it's low
cost but good performance. NVIDIA now targets
all areas of the market and they have products
from the entry level market up to the performance
market. NVIDIA has been very much successful with
their aggressive product cycle...new products
coming up shortly after the previous one.
The
GeForce2 series comes in a lot of flavors...the
GTS, the Pro, the Ultra and the MX. Now NVIDIA
even added more flavors to the highly-successful
GeForce2 MX: the MX400 and MX200. It is a little
odd why this move had to be made but perhaps this
move was done to finally phase out the TNT2 line
of cards which have been covering some parts of
the market.
 
On
the table above is the common and probably standard
configuration of NVIDIA's GeForce2 MX line. The
production of the original MX has been slowed
down to give way to its successor: the MX400.
The MX400 sports a 200MHz core clock and can come
in either 128-bit SDR or 64-bit DDR memory bus.
The MX400 comes in 64MB variants compared to the
usual 32MB on the original MX. The MX200 however
runs at the same core and memory speeds as the
original MX except that the MX200 uses only a
64-bit SDR memory bus which cuts the memory bandwidth
in half. Why this? It is a well known fact that
OEMs sometimes bundle with their systems a TNT2
M64 vid card. A common OEM system that we see
around is composed of a Pentium III/4, 128MB of
memory, a 20GB HDD and a TNT2 M64. Since OEMs
have customers who aren't really familiar with
computers, all they care about is the processor
and memory. The video card doesn't really matter
because most of OEM buyers only use their computers
for office and internet usage.
Why
the MX400 and MX200?
The
GeForce2 MX has already been highly successful...why
make more flavors? It is pretty much easily to
be fooled by specs without understanding what
they mean. The MX400 is a very attractive card
when viewed from a less-knowledgeable point of
view. Look at the fill rate: 800MTexels/sec, runs
at 200MHz and has 64MB of memory. The common way
of viewing things is that: higher is always better.
Since the MX400 runs at 200MHz, people would immediatley
think that this card will leave the original MX
behind and even be comparable to the GTS since
they both run at 200MHz. Next is the presence
of 64MB of memory over 32MB of memory. Again,
more means better but there is a problem with
the MX400: memory bandwidth. As proven almost
everwhere, the MX gains more speed if the memory
clock is increased without even touching the core
speed. In effect, the additional fill rate of
the MX400 would be wasted since it still runs
at the same memory speed as that of the original
MX. The MX400 costs more than the MX but in theory,
it should pretty much perform at the same level,
still far away from the higher end cards.
The
MX200 on the other hand already makes itself known
through the 64-bit SDR memory bus. The MX200 now
banks on carrying the GeForce2 name. Everybody
knows that the GeForce2 line has T&L, and
that boosts game performance. It is also known
that the GeForce2 is a GPU and the MX200 falls
under the GeForce2 line. Non-enthusiasts would
most likely be deceived by the MX200 for the simple
reason that: It's a GPU, it should be good. However,
the problem again lies in the memory bus of the
card which would really hit the performance. I
expect the MX200 to be performing even slower
than a TNT2 Ultra.
Now
why make them? NVIDIA is probably the most dominant
market right now and producing products like these
would again raise their income. The original MX
is being killed to make way for the MX400/MX200.
The MX400 costs significantly more but performs
at pretty much the same level. Therefore, if they
are able to sell the MX400, they would gain more
money by just producing a similar level of performance
as that of the original MX without justifying
the price increase. Same goes for the MX200: people
would think it's a high performance card when
it isn't really that good. Priced higher, unjustified
performance: poor customer.
Abit
Siluro T400/T200
Abit
has always been synonymous with overclockers thanks
to their excellent overclocking motheroboards.
It wasn't too long ago when Abit started producing
other products like sound cards, graphics cards,
speakers and other things. Since the Siluro bears
the name of Abit, it should pretty much also carry
the same quality and goodness that we see on Abit
motherboards. While it may not have SoftMenu III,
it still separates itself from the pack by simply
carrying the name Abit.
Left
to Right: Abit Siluro T400 and Abit Siluro T200
The
Siluro is really a striking card: Black PCB. The
Abit Siluro T400 is Abit's card based on the MX400
chipset and the T200 is based on the MX200 chipset.
Both of them follow the reference design quite
closely and both of them have tv-out. The MX chip
doesn't really run hot so even a passive heatsink
will be enough to cool the chip. Both cards use
EliteMT 6ns memory chips, the T400 has 64MB compared
to 32MB on the T200.
Strange
Discovery
The
review samples sent over to me seem to be pretty
much different from what other sites have received.
The Siluro T400 sample here runs at 64-bit DDR
instead of a full 128-bit SDR. While the memory
bandwidth may be the same, the latencies and other
issues of DDR will spell the performance difference.
The MX200 uses the standard 64-bit SDR. This is
something to watch out for but identifying a 64-bit
card from a 128-bit card isn't hard: just cound
the number of pins on the memory chips. If it's
a total of 64, then the card is 64-bit and 128-bit
if it has 128 pins. Axis Global Technologies (the
local distributor of Abit in our country) only
received samples of the card and they sent it
to me. I suppose this is not the production model
because the box clearly states that the T400 is
supposed to be 128-bit SDR and other sites also
verify that 128-bit SDR model.
Test
System |
