ABIT KT7-RAID: the best of his class
Using RAID in the KT7-RAID
If he decides to form the discs in RAID, as soon as the badge was proved it turns out to be very advisable to go to the ABIT web site to update the BIOS of the motherboard (what will update also that of the checker RAID), in addition to obtaining the last drivers. It is not anything essential... but it is the best moment to do it.
Since we have said, the ABIT KT7-RAID 1 supports the ways RAID 0, 1 and 0 +. If a system RAID is going to settle, unless a RAID 1 is going to be done (mirror), the normal thing (and the easiest and advisable thing in any case) is to use empty discs; care: to do a RAID 0 erases the whole content of the discs! They get connected to the channels of the HPT370 (if there are only 2, better in separated channels), it goes away to the BIOS of the checker (pulsating Control+H on having started) and there are chosen the way and the discs to be used.
A note on the topic of the discs to be used in RAID: we know for other reports than there are no importance limitations in the size, type or mark of the discs to be used, but we recommend to be discs exactly of the same model, not to introduce phase lags and imbalances for speed differences that they can affect to the yield (and because if you can make it easy: why to complicate you?). In any case, following certain rules (equal or major size for the disc mirror in RAID 1, for example) to mix discs is perfectly possible.
The configuration in the BIOS is simple and it is relatively well explained (in English) in the manual; next we will have to spend to the part software and particionarlos with FDISK from a starter diskette. We will see the discs like the only one, if we have a RAID 0 of the sum of the capacities; in case the size overcomes 64 GB, it seems that FDISK does not recognize them well (Windows yes, but it is still not installed!), so we will have to use another particionador (one of Linux, for example). Done this, it is formatted by FORMAT and installs the operating system to him.
Simple, truth? Since not, really NOT. For someone with some experience with the hardware and knowledge of English yes it it is, but if the word "BIOS" causes him headache and he has never installed a hard disk it can spend a bad moment; although whole, if the discs are empty, little damage can do...
And as the topic deserves it, there goes slightly more of information (in English, skylight):
Tests and yield
The tests in our KT7-RAID have not been a prank thing, fundamentally for the added work of trying with and without RAID (and to format, to install Windows...). As microprocessor we have chosen an AMD Duron 700 MHz, a mike that offers an exceptional yield with a cost simply ridiculously, as we will see soon in an article dedicated to the same one.
The rest of the material was quite standard: 128 MB of PC133, graphic card TNT2 Ultra and a pair of discs Seagate ST38422A, ATA/66 of 8 GB, to prove the characteristics RAID. Unfortunately, economic and temporary limitations have forced that the above mentioned discs were "normal", that is to say, of 5.400 rpm, but they have served to show the yield increase in RAID 0.
First, the most important thing: the stability. Be for the fan of the chipset, for the electrical design, for the quality or for other motives, our badge KT7-RAID worked perfectly stably at all times.
... Eye, ours, because this always of recommending is dangerous, in any production process better and worse series exist. And also it is true that just in case we update the BIOS and the drivers of chipset, checker and graphic card, because unfortunately to obtain a raised stability has always him to work a little to the user.
As for the general yield, let's say that it was "correct", since at present the speed differences between badges with the same chipset are small and slightly conclusive. The same happens with the RAID 1: simply it worked, did a copy of a disc in other, which is useful but slightly exciting.
The case of the RAID 0 had more "illness", for verifying if the speed was significantly major. According to Sandra 2000 and Winstone 99, yes it is: an increase of yield of 23 % (like average, Sandra is a synthetic benchmark and his values are not reflected so much in the reality) and a time of greatly minor load, although in other tests simply it is not appreciated (Quake loads the levels slightly more rapid, but one sees exactly at the same speed).
As for the overclocking, there were no negative surprises: he supported everything without problems, the factors limitantes were the mike and the memory, as it is habitual; there were reached easily 805 MHz (fitting the bus of the mike to 115 MHz), and raising the voltages a tenth we obtain 35 additional MHz. The stability to 805 MHz was almost perfect, what is not badly in an overclocking: of 15 %, neither more nor less!
Right now we keep on "pressing" the Duron in this badge, in the next article we will see if there are innovations on this matter.