барракуды выпускаются в двух вариантах: либо 100Гб (маленькая ёмкость) либо 200Гб (немаленькая цена, это ещё мягко сказано...).

На сайтах-продавцах написано что там типа нет передного моста IDE>Serial ATA

The nVidia controller will give you better performance because it is directly in the chipset, where the Silicon Image controller is on the PCI bus.

where are the proves that nVidia controller is directly in the chipset?

Errmmmmmmmmm..........you mean besides the signals coming off the chip?

There's a block diagram somewhere showing it. I can't seem to find it on nVidia's site. That being said, the implementation of nVidia's onboard controller doesn't seem to be quite as good, performance-wise, as Intel's.

мы говорим об упоре скорости в массивах raid0 на всех контроллерах..

Errmmmmmmmmm..........you mean besides the signals coming off the chip?

There's a block diagram somewhere showing it. I can't seem to find it on nVidia's site. That being said, the implementation of nVidia's onboard controller doesn't seem to be quite as good, performance-wise, as Intel's.

it easily can be located on PCI bus and all will say that it's not true...

it's only on the paper... in fact - all devices are on PCI bus

than it's 100% on PCI bus, cause all intel's SATA/ATA controllers are on PCI bus

than it's 100% on PCI bus, cause all intel's SATA/ATA controllers are on PCI bus

Errmmmmmmmmmm............ok, can you prove that? Every block diagram I've ever seen indicates otherwise. In fact page 50 of this Intel document clearly shows that the PCI bus, the IDE bus, the SATA bus, the Lan controller, and pretty much everything else all come off the internal chip bus. This is, as near as I can tell, an extension of the ICH Hub Interface, which is (I believe) a 100 MHz, Quad Pumped, 12-bit bus capable of about 600 MB/second of data throughput.

In addition, numerous benchmarks have shown the Intel SATA implementation to be fully capable of supporting the full 300 MB/s burst data rate of a dual SATA implementation. If they were all sitting on the 33 MHz PCI bus, the absolute BEST you'd see would be 133 MB/s, and the actual sustained throughput would top out at about 100 MB/s or so. In fact, nVidia's implementation has been benchmarked at over 200 MB/s bursts, IIRC.

take 2 HDD's WD raptor and make raid0 array

you'll get same perfomance on ICH5 SATA controller and any PCI SATA controller

it's only on paper... but in fact all mobo makers ignore this

where i can see this benchmarks?

interesting... why they are measuring burst rate? and not linear read/write speed?

for example - 2*WD raptor on ICH5 shows same performance as 2*WD200JB on TX2000 on pic bus... it's linear speed, but it should be higher on raptor, right?