I read the Sandisk did not require undue force to insert the cards, which was the deciding factor up against the Lexar alternative.

The unit reads and writes to CF type I and II cards and is reportedly based on the Oxford OXFW912 1394B (firewire 800) to ATA7 (IDE) bridge chip which supports UMDA6 mode (80+ MB/s).

The SDDRX4-CF-904 model sports a "45 MB/s" claim sticker on the packaging - I think it's an update to the previous SDDRX4-CF-902 version which specifies 40 MB/s in the specs. The device supports FW800 "connectivity" but is only operating at FW400 speeds since FW800 supports 98 MB/s. According to Wikpedia the CF ver 4.0 standards supports 133 MB/s transfer rates.

Using Apple Activity Monitor I saw 42 MB/s transfer rates with this unit using a Transcend 16 MB CF. This is curious as my 133x speed CF card should only be capable of 20 MB/s according to Wikpedia. The same card on a USB2 reader (built-in to Dell 2405 monitor) achieved about 11 MB/s read and 7 MB/s write.

Strangely while all the retailers list the new part no. the page for new item seems to be missing from most Sandisk sites as of Aug 09.

The unit is powered by the firewire bus, has one 9-pin FireWire 800 port (1394b) and includes a 9-pin to 9-pin cable and a 9-pin to 6-pin Firewire cable. These cables are useful, as I can use the latter to connect the reader to a FW400 equipped Mac or connect the Sony MRC-1K to the FW800 Macs as additional or back-up to the Sandisk.

One term is the bandwidth of the bus; another the throughput (maximum data transfer rates) of a determined device.

In multi-drive configuration through daisy chain, the maximum sustained data rate in a 1394b-bus can reach to 90-98MB/sec, when multiple drives are connected to a single host.

In a single-drive configuration, the sustained data rate of host port connection is the same as that of the device connected on it, thus, the bandwidth of the 1394b host connection is not fully occupied.

In the case of the Sandisk-firewire-reader up to 45MB/s with a 300x CF. The throughput obtained with a fw400 controller is up to 41 MB/s max. which is near to the limit of a single 300x CF device. The situation change if we connect two devices to the same bus working simultaneouly... Another interesting configuration would be having more than one 1394 controller in a system...

The real performance (FOR EXAMPLE) of a SATA HDD is up to 64MB/sec (or something else with the lastest products). Then we can obtain its total throughput connecting it through a external 1394b enclosure. But if you connect several modern hdd's working simultaneouly in RAID configuration, it`s not possible to get de maximun data transfer through a 1394b-bus and you will need an eSATA 1500 or 3000mbps. (or an unreal 1394c bus!!! pitiablely)
With the modern SSD disks I had watch read performance up to 200Mb/s (and up...) in one disk configuration but are expensive yet...

For example, in the web page:
eBay Österreich: PHOTOFAST 366X Compact Flash 16 GB CF Extreme 16GB (Artikel 140337011247 endet 08.08.09 02:32:10 MESZ) (http://cgi.ebay.at/PHOTOFAST-366X-Compact-Flash-16-GB-CF-Extreme-16GB_W0QQitemZ140337011247QQcmdZViewItemQQptZDE_Computer_Elektronik_Foto_Camcorder_Speicherkarten_PM ?hash=item20acbcda2f)

it`s sold a CF 366x card, and the vendor has included a benchmark test result conneting the card to a Lexar firewire reader.
Sequencial Read = 58.10 MB/s
Sequencial Write = 25.47 MB/s
Random 512K Read = 58.13 MB/s
Random 512K Write = 2.802 MB/s
Random 4K Read = 14.73 MB/s
Random 4k Write = 0,026 MB/s
This was got through a 1394b controller because in a 1394a controller the max.read result would be 40,39 or 38 MB/s.

We can notice in the test that the CF card (or the Lexar reader) is not a good (pretty) writer.

With the 133x CF card that you had mentioned,it would necessary to get a bench test to know its sustained performance, because through the Activity Monitor (a very good tool) you has rendered the system resources with ist buffers, memory-cache,...

I was reading about the performance of flash card to use it as external Cache devices on windowsXP.
And it would be a great help for me if somebody could get a bench test to the Sandisk firewire reader with a Sandisk extreme iv CF (or similar) to know ist random writes behavior. I know its behaviour in sequencial transfers but i'm concerned in random writes.

To remember here that normal Hard disks are worst (more slow) managing small (a few KB's) files than flash memories and SSD HD's. The items in which I was concerned would be "Random 512k write" and "Random 4k read" achieved with a Sandisk extreme iv CF (or similar) with the Sandisk firewire reader, these are the render-parameters involved to use it as ReadyBoost Performance Requirements.

Minimum performance requirements:
Random 4KB reads 2.5MB/s
Random 512KB writes 1.75MB/s

Higher performance requirements: (or better)
Random 4KB reads 5MB/s
Random 512KB writes 3MB/s

If you could connect the drive in a windows system with a firewire connection, an easy and fast benchmark program to measure this parameters is Crystal Disk Info:
Crystal Dew World - Software - CrystalDiskMark (http://crystalmark.info/software/CrystalDiskMark/index-e.html)
Although in other OS plataforms surely will be similar softwares: xBench,or http://www.aja.com/ajashare/AJA_System_Test_v601.zip
Thanks.

Note: In general, the flash memories are slow writers but I suspect that sandisk has cached the writes in someway (ESP technology). This is my doubt.
Other option will be using easyCo MFT technology by software, or a CF to SATA adapter.


Note: The PC laptops with firewire controller usually have 4pin connector-fw400 (which is powerless) to avoid laptop-board accidental damages. Fw800 specification has too power for a laptop (anti power-safe). I think that was one important reason for the low firewire difussion on Pc-world. Another was that the windows systems suport fw800 in some determined soft-configurations only.


Note: Also I has the doubt about the difference between SDDRX4-CF-901 902 903 904 models...

Our Sandisk Extreme Reader drops out of the Device list when a disk is removed and fails to ident when a new disk is inserted. The only solution is to reboot. Has anyone had a similar problem and what was the solution. The disk is connected 9-pin/9-pin into a three firewire connector PCI card - all sockets of which work fine with other 1394 gear.

Actually, the firewire card wasn't as fine as I thought. Eventually we traced the problem to the card. However, what we did discover was that, unlike USB card readers which show in Device Manager4 as a drive regardless of whether there's a card in them, the firewire card only idents as a drive if there's a card in it. Removing a read card knocks the drive out of DM, but inserting the next card re-idents the card in DM. Odd but it works well.

And the read speed is awesome. It's turned what was a day's job plus into a few hours, including the time Avid takes to import the files.