MSI K7T Turbo-R
Let's make a bet! I bet that most people don't know which four mainboard manufacturers produce the third
of all motherboards available worldwide. Many would probably for guess small, "secondary" companies barely having 1-2%
share, trying to live side by side with the big names. And although a few OEMs do take a great portion of the
whole market, users buying complete systems from companies like HP/IBM/Compaq/etc generally have little idea about the make
of the mainboard that resides therein. Therefore, being loyal to a certain company's certain product makes absolutely no
sense in this case..
We will get back to all this, but before that, please take the time and allow us to introduce the newest guest in our humble
lab (drum rolling in a distance.. :)).. ..the MSI K7T Turbo-R! Here it is, full sized, all in its glory, which is only
strange because the board shouldn't have existed when it first knocked on our door: MSI announced the K7T Turbo-R only a few
days later. Of course, most of the big hardware sites already had their review online. Nevertheless, this is a good start.
For us :).
While most of the boards that we have tested were carrying the VIA KT133 chipset, the K7T Turbo was equipped with the newer
and shinier KT133A chip. To put it short, this boils down to being able to run the FSB (Front Side Bus) at 133MHz (266MHz
DDR) and not only the memory (using the asynchronous memory clock option) as it was the case with the now "obsolete" KT133.
The KT133A allows the RAM and the EV6 system bus to share the same clock (yes, the EV6 bus uses a DDR signaling, but the
input clock actually remains at 133MHz, with data being transfered at both rising and falling edges), boosting
performance. As you will see from our upcoming GeForce2 Ultra article, FSB plays a vital role in global performance..
Back to DDR: one would expect that the KT133A chipset, with only SDR (Single Data Rate) memory could significantly fall
behind the existing DDR solutions (such as the AMD-760 and the ALi MAGiK1 chipset) in terms of performance. This is
not the case, you don't necessarily need 266MHz DDR SDRAM to accompany the 266MHz DDR EV6 bus. At least, the
Athlon/Duron might not be able to take full advantage of the extra bandwidth (but certainly more than the Pentium III).
Our Iwill KA266-R review shows that the migration to DDR RAM didn't bring any dramatic differences. Not yet. It will, it
should. Otherwise, we will once again face strange and mixed reactions as we did with Intel's brand new i815 chipset. It was
new, it was hyped, but the "rusty-ol'" BX chipset, which should have been history by the time of the i815 debut, still
performed better.. so, until the winds of change (and some CHEAP DDR memory) the KT133A is VIA's grand weapon and as
such, the K7T Turbo could be the board to have..
I promised to get back to the little bet I made at beginning of our article. Any guesses so far for the four (say that
very fast 10 times :)) companies? We were talking about big players, others "being seated" only in the second row,
but haven't mentioned any names. Well, here they are in alphabetical order: ASUS, ECS, Gigabyte and MSI. And here they
are in the order of market shares: ASUS, Gigabyte, MSI and finally ECS. But this is not all. ECS and ASUS had average income
growth of 21%/26% compared to last year, which is far behind that of Gigabyte's 48%, and even further behind MSI's phenomenal
58% (source: Taiwan Digitimes, 2001.01.10.)! Of course, many of you might remember that MSI had a huge, 100%
increase in sales in '98. Please don't be skeptical as the situation is very different now, trying to double an existing 7%
share of the market is not trivial by all means, that +58% is big numbers. One thing for sure however: if MSI is able to
maintain their current rates, ASUS will be trumped for good. This already happened, but only for a month (this January),
when MSI surpassed ASUS in sales by over 10%, producing about 980.000 motherboards.
So, this means that MSI (Micro-Star International) is one of the biggest players in the mainboard industry. But in fact, MSI
is a lot more. MSI is a main innovator of motherboard features. Just think of "magic words" like Fuzzy Logic III, Live BIOS,
PC2PC, D-Led and D-Bracket. And these are all a feature bonus to the MSI boards that already have the stability of their ASUS
counterparts and the "tweakability" of the famous/infamous Abit boards! Don't forget the reasonable price, as MSI is
determined to deliver prices 15% better than ASUS.
If any of you have had an MSI mainboard in the past then the above mentioned "magic words" should sound very familiar. To
kick off with D-Led, forget the sound and look at the light, as the leds will blink if there is a boot problem. Should the
boot process be flawless, the leds will light up informing us that certain parts of the POST (Power-On Self Test) just
finished normally. Nice feature. D-Bracket is a similar, but more elegant approach, as the leds are located on a slot-unit
back panel, so if you have any spare slots (take AMR/CNR :)), you can just screw the D-Bracket in place and get to see
the led signals at the back of the PC.
PC2PC is a lot more "practical" though. It will be incorporated in every newer MSI
motherboard and its purpose is to be able to connect 2 MSI boards (2 PC's with MSI boards) through their USB ports
to provide general file transfer/sharing, LAN emulation, etc. This move is a bit "MSI-discriminative" :), but I guess no other
manufacturer supports a similar construct, so MSI had no choice. And anyway, looking at the company's current rate of growth,
everyone will have MSI boards in the near future, all we need is some patience :). USB 1.1 supports up to 8.6MBps transfer
rates and PC2PC runs with all major operating systems (with the exception of Linux - big minus for MSI there!). We wouldn't be
very suprised :), if PC2PC eventually supported USB 2.0 as well, as the next-gen port is here knocking on our doors (and
the newest mainboards) and is wowed to have a 40X transfer rate increase over its predecessor.
With Live BIOS we can
update our BIOS fast and painless through the internet, although the stability of a Windows 9x system can eventually cause
us some pain through big blue screens.. anyhow, the option of flashing the BIOS through DOS is still there, for the careful
user. Fuzzy Logic promises no more than to manage the overclocking of our machine all by itself. It raises the FSB, tests
everything, raises the FSB again, etc. If needed, it runs down the nearest Fry's shop and gets us a FOP38 cooler, doesn't
forget to apply some thermal paste to the CPU core, in fact it might make us a watercooler, while we are updating our BIOS
or doing our regular pizza orders and online shopping.. oh, this is not actually true, as these features will only be
supported by Fuzzy Logic 4, but this is all NDA info anyway, so we will have to quit now.. :)
MSI is an "oldie" in the industry, it began its activities sometime in 1986 and although its private history is sort of
hidden in the www.msi.com.tw page, it is still a good source for some forward-looking thoughts. It is known however, that
the company bears the ISO-9001/2/4 quality assurrance and that 120 highly qualified engineers are designing the newer and
newer motherboards so that the 99.6% "hit rate" of the produced boards can be maintained. This means that only 0.4% of the
manufactured boards are faulty, which is really not much. On the other hand, if I base my calculations on the 980.000 boards
sold in January, that means 3920 pieces, which is on twentieth of the complete output of SHUTTLE.. I guess doing it seriously in
volume does make a difference..
It is an interesting thing that MSI is concentrating heavily on the European region. In 1998, MSI sold about one and a half
times as much boards here than in the Far-Eastern markets and around double of what in the US. It's a general tendency that
Taiwanese manufacturers are leaving the "continent" and setting up branch offices in other regions, we could already observe
this with Iwill, so we needn't go very far for examples..
Of course, MSI is not only the manufacturer of mainboards, but graphics cards, bare bone systems, CD-ROMs and a whole lot
more, but it is their motherboards that made them famous.
To remain so famous, one must not slow down. Perhaps this is one of the reasons why we could witness FIVE new mainboard
releases in the last 10 days from MSI! During these 10 days, they also cashed a "Symbol of Excellence Award" - the most
respected prize in Taiwan, standing for the highest quality - for its K7T Pro2 board and an AMD Certificate. Shouldn't this
be enough, the Starforce 822 video card with nVidia's GeForce3 chip (call it GPU, if you must :)) was also announced.
This article promised to be a K7T Turbo-R review, not an MSI introductory, so we will stop praising
MSI now and get back to the subject, the K7T Turbo-R, as he ought to be standing in the spotlights.
As we noted earlier, the K7T Turbo (MS6330 v3.0) is built around VIA's KT133A chipset. The VT8363A North Bridge and the
VT82C686B South Bridge, to be precise. This is VIA's last trump for the SDR platform and we can't help it but agree with
them, no matter where we look at. Take the MSI board, or take Iwill's offering (KK266) and you definitely won't regret it,
as both of them are reportedly cruising quite well with insane FSB speeds - up to 166MHz (333MHz DDR). We could only verify
this up to 145MHz (our CPU maxed out at that speed, but the board was rock stable). More on this later, but first let's dig
into the most important specifications.
|MSI K7T Turbo-R
||Socket462 (Socket A)
||EV6, 200/266MHz DDR
||VIA KT133A (8363A/686B)
||100-132MHz, 133-166MHz, 100/133MHz default FSB selection via jumper
||Multiplier adjustments, VCore/VIO adjustments, FSB tuning (1MHz stepping!) all from the BIOS
||3db 164 pin DIMM slots for SDR SDRAM
|Expansion slots (AGP/ PCI/ ISA/ AMR|CNR)
||ATA-100 RAID controller Promise 20265R
||Award BIOS v6.00PG
|Related online material
||"Msi.com", BIOS, manual (PDF)
The K7T Turbo also carries a RAID capable IDE controller, the Promise 20265R. Compared to the AMI chip found on Iwill's
offering, Promise offers a bit less, but manages RAID 0/1 just as well. This means that mirroring (for double safety) or
striping (for double speed) is possible, but not both at the same time (RAID 0+1, or RAID 5 operation). Actually, IDE RAID
is definitely cheap and fun to play with, but given the fact that the IDE controller can only address one device on the
bus at a time, it might not be the wisest choice. If you are very serious about RAID, vote for SCSI. On the other hand, if
you are seeking for a reasonable transfer speed increase, IDE RAID is much-much cheaper. Anyhow, hats off to MSI for
including the option.
We used a 600MHz Duron (first series, shipped with unlocked multiplier) for our test, which unfortunately maxed out before
the K7T Turbo did, at 870MHz. I am saying unfortunately, because we were really kicking for high FSBs. To make sure RAM
wasn't the limiting factor, 2 sticks of Crucial PC133 CAS/2 (2-2-2 timing Micron chips) SDRAM were seated in the DIMM slots. As the MSI
board, unlike the Soltek SL-75KAV-X, did not support a 5.5X multiplier (with which we could have lowered the multiplier
and raised the FSB even more..), our config gave up at anything above 870MHz (6 x 145MHz). That is still very
respectable, at 145MHz FSB the board was operating with the most agressive RAM settings, and have not crashed a single
time! Soltek's version didn't even post at 139MHz, so MSI is a clear winner here. General concensus is that the K7T Turbo
runs flawlessly at a 150MHz FSB, which is pretty much the last sensible FSB to use anyway, as PCI (with 1/4 multiplier)
and AGP (with 1/2 multiplier - no matter what Sandra says, as even Sandra 2001 does NOT recognize the KT133A chipset
properly) speeds above 150MHz start to get unhealthy.
Not all is perfect, however. We did encounter a few tricks during the two week testing period. At first, the K7T Turbo
gave us some headache, as well as some VCore overvoltage, 0.14V more than what was set in the BIOS. OK-OK, that
should be no problem, we lowered our setting to get the needed results. Freezing the system at 600MHz was unacceptable
though. How? Why? Something was not right, every second-third boot gave us fatal errors somewhere and we could barely get
to the BIOS to check the settings. Once we got there, it was a shock to see that the CPU was set to operate at 8.5 x 140MHz
(1190MHz). Right :). MSI once again came to the rescue, all we needed is to drop them a note that something is wrong and
within a few hours, a fresh beta BIOS was awaiting in the mailbox. We immediately proceeded with the flash and were happy
to see that all of our problems went away. Well, almost all. The voltage regulator still overvolted the CPU no matter what
we did, but since we were overclocking anyway, this actually came in handy :) (just kidding, I know this still was a
problem. Eighter the voltage regulator or the hardware monitor was off a bit, but we could not determine for sure. As
everything was running silky smooth and the CPU wasn't too hot, we suspect that the hardware monitor chip gave
inaccurate readings.). All in all, the K7T was exceptionally stable, just make sure you use the latest available BIOS.
We were a bit unhappy not to see any ISA slots. Surely CNR is a good thing, but we can't stuff our Gravis UltraSound P&P
sound card in a CNR slot, it won't fit :). And although the days of ISA are already numbered, there are still some old
SCSI and network cards lying around that could be put to perfect use. Thinking of Gravis: the onboard AC'97 CODEC is VERY
basic, it is mainly a feature for OEMs, if you do intend to listen to music/play games, turn it off for your sake. It has
a pretty high CPU usage anyway.
Ergonomy. Not the strongest point of this particular model. Nothing too serious, but... It all began with the power cable
connector.. I decided to
take the system apart for some reason. So, pull the [power] plug, right? Not exactly. The space between the inserted SDRAM modules
and the power connector was very little. So, take the RAM out first, right? Not exactly. The space between the SDRAM slot
levers and the inserted VisionTek GeForce2 Ultra was very little. So, take the AGP card out first, right? Right! But don't
forget to open your AGP retension mechanism, then take out the AGP card, then the SDRAMs and finally, pull the plug :).
Of course, taking the whole system apart is by no way a "regular" occasion, but a little better arrangement of the
components would have been appreciated.
Another little nuissance was the inconsequency of the BIOS and the mainboard manual. One part where they differed was the
already mentioned multiplier adjustments and the other being a misterious 66MHz DRAM Clock option (mentioned in the manual).
Well, it wasn't available. It wouldn't make sense anyway, so forget the whole thing, we just thought it might be worth
mentioning it, for the sake of completeness.
As with any board based on VIA's newer chipsets, like the KT133A, one must not forget to install the VIA 4-in-1 driver
pack (preferrably v4.28a), else the global performance will decrease somewhat. We first forget about the AGP GART driver
and it did take some time before we realized that our GeForce2 Ultra should be scoring around 130FPS instead of 90... so,
for the sake of 30-40% performance, don't skip this step.
Here it is: the good ol' SiSoft Sandra 2000 memory benchmarks. Instead of measuring Quake3 FPS results, which tell us a
lot about our video card, we voted for Sandra. Besides general CPU stability, memory stability and performance is a key
differentiator amongst motherboards. A well-built board should perform well and be able to run the memory at the most
agressive timings at high FSBs. The K7T Turbo-R excelled in all these areas, if you don't believe us, check out the
diagram below. MSI's offering only falls short of the Iwill KA266 DDR mainboard, but that shouldn't suprise us. The
difference in performance is far less however, than the difference in the prices.
The time has come to "sum it all up". MSI's mainboard opted for one of the newest members of the VIA chipset family,
the KT133A and we can't argue with the decision. We were able to run the board at a 145MHz (290MHz DDR) FSB with the
fastest possible RAM settings, without suffering a single crash. Although the word on the 'net is that competing
boards can go even higher, we have yet to see the evidence. At first the K7T Turbo-R did show us a few trick, but a
new BIOS from MSI solved everything and after the flash there was nothing that could stop us [the K7T Turbo-R, to be
precise]. Extended usage and day long stress tests didn't cause the slightest hickups, our CPU ran flawlessly even at
60 degrees C. Everything could be set from the BIOS, including 1MHz FSB steppings, VCore and VIO voltages and the CPU
multiplier. The only thing we had to decide in advance was the default FSB speed, 100 or 133MHz, which had to be set
Although this is only our second English review at PLUSABIT, we have tested many Slot/Socket A mainboards and from what
we have seen so far, MSI takes the crown. We have not given awards in the past and it would be a fair thing to at least
let Iwill's and ASUS's KT133A offering join the arena, we still feel that the K7T Turbo-R deserves something. Therefore,
we will now award it with PLUSABIT's PLUS A LOT Golden Award. Until a worthy
competitor arrives the award will stay right where it belongs to, that's for sure.
The K7T Turbo-R was sent to us by MSI through their authorized distributor, CHS Hungary. We thank both companies for
(Márton Balog & Dávid Szőts)