Promise FastTrak TX4000 Raid Controller 18th May 2003

Feedback from our recent review of the Promise SX4000 RAID Controller card indicates that HDTach graphs are a little too confusing for most our readers who would like to see simpler graphs. We still believe HDTach provides the most comprehensive analysis of hard drives across their entire surface area but instead of detailed graphs we will now be providing sustained read and write scores (ignoring the effects of burst read and writes as well as random read/writes - we've never encountered a hard drive fragmented badly enough to come close to random read/write figures).

The card also has a lot of exciting features. Here's what you get.

As usual, Promise include 4 IDIE cables and 2 power splitter leads along with the drivers on a floppy and tools on a CD. Those with an aversion to flat cables or those wanting improved airflow will probably want to purchase rounded cables (especially if you have a small case and are using 4 drives.

The card itself is quite small like the TX2000 but main feature of this product is the four independent channels to allow 4 drives to operate as "Master".

Here's a screen showing the configuration of the test system:

Here are the 4 drives we managed to pull together showing the ease of connectivity to the card:

Instead of many smaller graphs we are going to test 2 main configurations (2-disk RAID0 and 4-disk RAID0)and tabulate these in the form of 2 graphs that are (hopefully) easy to understand. For comparison we used a Highpoint HPT372 which has only 2 channels so we will be able to see clearly not only the comparison between the two controllers but also the benefits of having a 4-disk array on 4 channels instead of 2 Master and 2 Slaver drives. We'll also see which stripe size is best for performance and generally try and dissect as much information as possible from the graphs. The TX4000 is also capable of mirroring (RAID1 - you will need 2 or 4 drives) and both (RAID0+1 - you will need 4 drives).

Starting with the 2-drive Arrays:

While we were expecting the TX4000 to be a little faster we weren't expecting gains of up to 30%! The HPT372 is not a bad controller - it's just outclassed by the TX4000. Interestingly we see that read performance for both controllers is consistent across stripe sizes so we are not encountering any PCI contention problems (remember the PCI bus can only manage about 120MB/s - 133 less about 10% for overheads). We have some way to go before hitting this barrier. The write results show that a stripe size of 16K is best in this case.

Now for the 4-drive Arrays:

The gap has narrowed not because the HPT372 is better with 4 drives (it actually suffers because it has only 2 channels) but because we are hitting the ceiling of what the PCI bus can handle. If we had a 64-bit PCI port available we would see the same proportions as the first graph. We have not done this because few people have the luxury of 64-bit PCI slots outside of the server environment and this card is aimed at the home user so we used a standard PCI slot. Both read and write performance has reached a plateau on the HPT372 and changing the stripe size makes no difference. The TX4000 now shows better results with larger stripe sizes because it is hitting the PCI ceiling. The reason for this is that using a 64K stripe size means 4 times less packets of data sent over the bus and therefore less overhead allowing a bit more performance.

Tips for Improving Performance

Some thought needs to go into planning a RAID array. Small stripe sizes favor large numbers of files being transferred while large sizes favor activities such as video editing etc. Generally a 16K stripe size works best for everyday use. To get the best performance it is necessary to format the Array using a cluster size that is a whole multiple of the stripe size. For example if the stripe size is 16K then it is best to use 16K or 32K as the cluster size. The reason for this is that Windows sends/requests data in blocks that are made up of the cluster size. The RAID controller allocates these to the first free drive in the array in sizes of the stripe size. So a 32K cluster would be split into 2 16K blocks and sent to 2 disks and this would be optimal for 2 or 4 disk Arrays. 4K clusters would have to be accumulated until 16K was ready and then sent to the first disk while the other(s) were waiting for data. In practice this is not too bad as the cache on modern drives compensates for this but the greater the number of drives in an array the greater the need to take such factors into consideration.

Conclusion

We are reaching a transition point from Parallel to Serial interfaces for hard drives and increasingly this is the focus of many review sites. It is often overlooked that Serial drives are still hard to get hold of and will not be generally available until the end of the year. Most home users do not buy 2 or 4 drives in one go to build a RAID but gradually add more drives to their systems as their needs expand. Onboard RAID solutions are limited (often cut-down versions of mainstream products) and unlikely to satisfy users wanting performance and/or future expandability.

The TX4000 represents excellent value for money and superb performance. While a PCI ceiling may be in effect on 4-drive Arrays it will not be very noticeable in everyday use. The long-term solution is Serial ATA but is still in the distance and requires spending money on new drives and today the TX4000 is the cheapest way to have a 4-channel RAID Array supporting the fastest ATA drives available. For a few years now RAID products have been entering realm of the home user. 4-channel controllers like the TX4000 are now as good as it can get with current technology so consumers can purchase a TX4000 safe in the knowledge that the PCI bus will be obsolete before their RAID controller.

We give the Promise FastTrak TX4000 our Silver Award.

We would like to thank Promise Technology Inc. for the review sample.

All trademarks are the property of their respective owners.