|
DFI
AD77 Infinity KT400 - First Look
|
|
Author : Wayne
Date : 4th September 2002
|
BIOS :
The BIOS is another area where DFI have brought
significant improvements that clearly signal their intent to
target this board at the enthusiast sector. All the features
are there we've come to expect plus one or two interesting new
ones. To save time and space I'll only cover the main areas
likely to be of interest to most.
DRAM Clock/Drive Control :
Here we see the usual suspects required to fine
tune and tweak your memory performance including the option
to alter DRAM drive strength.
DDR DRAM Clock :
As you can see there's the option to run memory
at either 266, 333 or 400MHz asynchronously though as mentioned
earlier your particular brand of memory may or may not operate
suitable at 400MHz due to the lack of an official standard.
DRAM Timing :
For those with less confidence when it comes to
manually tweaking memory performance there's the usual array
of preset memory performance options.
Talking
of DRAM timing an interesting option appears under the "CAS
Latency" category. I found the option of CAS3 to be a little
odd but DFI made up for this with a CAS1.5 option. I did see
if I could run at this setting but failed :(
FSB Frequency :
Depending on the dip position FSB frequency is
adjustable in 1MHz increments from 100 to 133MHz or from 133MHz
right the way up to 250MHz.
DIMM Voltage Adjust :
DIMM voltage is adjustable from 2.50 to 2.90volts
which is possibly not high enough for some but it's probably
the safest limit to make directly available though the BIOS.
AGP Voltage Adjust :
AGP voltage adjustment doesn't seem to be a particularly
useful feature in the scale of things, it's certainly never
helped me with a graphics card hampered overclock but it's better
there than not..
Chipset Voltage Adjust :
An interesting and new feature is the ability
to directly control the chipset voltage. I haven't had a lot
of time to play with this feature so I can't comment on its
real world benefits, which are certain to be linked to overclocking
stability, but I have to admit I'd prefer to see the North Bridge
actively cooled if this option is going to be offered.
CPU Vcore Adjust :
Vcore adjustment is available from 1.1v to 2.0v.
CPU Ratio Adjust :
The first real disappointment is the 10.5x upper
limit for manually setting the multiplier. I hope this increases
slightly with future BIOS revisions, not because it hampers
overclocking which will usually involve low multipliers and
high FSB but because it means you can't manually configure the
BIOS to run newer and perhaps unsupported CPUs until DFI release
a BIOS update.
VLink 8x Support :
I mentioned on page 1 that the KT400 now supports
8X V-Link (533MB/s) high bandwidth North/South Bridge interconnect.
Presumably for stability reasons you cab disable this feature
and revert back to 4x V-Link (266MB/s). I should mention that
this is a total figure in both direction, 266MB/s V-Link is
actually 133MB/s in each direction between the North and South
Bridge.
Performance :
DFI made it quite clear that this board and BIOS
weren't ready for benchmarking yet but of course I couldn't
resist the chance to see where the KT400 chipset is up to. I
know from other manufacturers that performance at this stage
isn't great but I wasn't expecting it to be quite so bad as
it was. I didn't run a lot of benchmarks (actually I did but
I'm not going to publish them yet) as it wouldn't be fair at
this early stage.
It's important that you don't use these results
to judge either the chipset or the motherboard/BIOS at this
stage simply because KT400 is very much a work in progress despite
the fact that one or two manufacturers have decided to release
official products to an unknowing public. When DFI revise either
the hardware or the BIOS I'll bring you the news, good or bad
and if there's no significant improvement in performance I'll
revise this review and alter its content and conclusion before
re-releasing it. Right now it's a first look at an early sample
that's being worked hard on behind the scenes and we should
treat it as such.
SiSoft Sandra Cache and Memory :
As I hope you can see there's a bit of a strange
thing happening here. For data block sizes up to and including
256k the KT400 with its memory running at 200MHz (400MHz DDR)
is actually slightly faster. Where things go badly wrong is
when the data blocks become too big to be cached by the CPU
(512k and above) at which point the KT400 really takes a serious
dip in performance dropping to the low 800MB/sec against KT333's
low 1000MB/sec. Memory controller problems using 400MHz DDR?
Kind of looks that way.
SiSoft Sandra Memory Bandwidth :
Epox 8K3A+, Memory @ 133MHz (266MHz DDR) CAS2.0
|
AD77 Infinity, Memory @ 166MHz (333MGz DDR) CAS2.0
|
AD77 Infinity, Memory @ 200MHz (400MHz DDR) CAS2.5 |
Here again we see a graphic demonstration of how
poor DDR400 support is with bandwidth for DDR400 actually trailing
that of DDR333 although it was running CAS2.5 as opposed to
CAS2.0. As I said earlier the system simply wouldn't boot with
our Corsair XMS3200 CAS2 actually set at CAS2 which is annoying
when the Epox 8K3A+ can manage it! Even so both results trail
the KT333 at this early stage.
Overclocking :
Not much point spending too long overclocking
a board that may be significantly different either in BIOS or
hardware when it's officially released but I did have a quick
check at how it coped with a bit of an FSB tweak. Actually it
did pretty well running at 140MHz FSB with no voltage increase
at all but the memory timings had to be lowered to CAS2.5. This
also brought a bit of a boost to the ailing memory performance.
Conclusion :
There's a lot to like about the AD77 Infinity
and if past boards are anything to go by price will be one of
the biggest things to like.
BIOS options are perfectly suitable for just about
everyone and despite a few layout niggles and the presence of
numerous jumpers the board is otherwise quite tidy if unspectacular
looking. Fortunately once the jumpers are set almost all of
the regularly accessed options are set through the BIOS. Also,
by opting for 5 rather than 6 PCI slots DFI have been able to
avoid the old "remove video card to add/remove RAM"
problem. I could live without the CNR slot but it is a shared
slot so you're not sacrificing anything to have it there. The
RAID IDE connector isn't best placed and would be better fitted
at 90 degrees to its current position between the PCI slot lines
or perhaps lay on its side facing out the side of the board
at a push.
Another nice touch is the inclusion of an additional
two USB2.0/1.1 ports through the expansion brackets (totaling
six with the four hardwired ports) and two IEEE 1394a ports
through a further expansion bracket. I'm nit-picking but there's
room for all four ports on a single bracket rather than loose
two expansion slots. On the audio front I hope the expansion
bracket required for 6 channel sound, S/PDIF and MIDI outputs/inputs
are made easily available for those who want them or better
yet include it in the box.
Stability on the AD77 was 100% through two days
of tweaking, pushing, pulling and generally being quite unkind
to it so if DFI can boost the performance and keep the price
as keen as they have in the past (which looks likely if their
estimated price of £100 inc. VAT holds up) this board
should be a great success for them, that is unless we see nForce2
any time soon and it lives up it's potential at a reasonable
price.
No ratings or final summary at this point, I'll
wait until DFI have achieved all they believe they can before
I do that. I will however say that DFI are making all the right
moves and that the AD77 Infinity deserves to be a big seller
for them. The name may not carry the prestige that the likes
of Abit or ASUS do but that doesn't make them any less able
to put out a solid product, and no doubt one that puts a few
coins in your pocket to go towards that stick of DDR400 you'll
be needing!
<<<
Back to Review Index >>>
Home
