Our first picture is of the actual upper surface of the drive. Looks plain Jane, no special bells or whistles; well really, what can you do to a hard disk? Performance is what matters here.
The belly of the Cheetah
Here we see the underside of the drive. This is where all the important stuff goes. Here we see the DSP controller chip that allows access to the drive at 320MB/s. This chip is the same for both Ultra160 and 320 drives, only the command and control chip is changed, which we will look at next.
Command and Control chip
Here we see Seagate's own Agere 5900D4 controller chipset. This chip determines the speed of the drive's interface with the SCSI controller card. The D4 is 160MB/s and the D5 is 320MB/s. So you know what to look for if you want Ultra320 speeds; but be warned, it will cost you.
The Cheetah 36ES series does not skimp on the cache. 4MB (2x2MB) Hyundai 6ns cache is supplied onboard to cope with heavy read/write loads. 4MB is the optimal amount for access speeds up to 200MB/s. the Ultra160 is well within this limit, the Ultra320 uses the same amount of cache, so some speed looks to go unused. Neither Seagate's website nor their information documents supplied with this drive told us the cache clock speeds. We can only guess its the maximum of 166MHz.
The Connectors and Jumpers
Here we get a view of the drive from the back. It's from this view that we can see where it differs from IDE. The connector interface is a standard Ultra wide 68-pin SCSI connector, so this drive will work on any ultra wide SCSI controller card. You just need the card to support Ultra160 speeds if you want to gain the full potential.
Next to this is the SCSI-ID jumpers. This allows you to set the drives ID for the SCSI channel. Unlike IDE which uses only two modes, master and slave, SCSI can allows up to 15 drives and devices on the SCSI bus. And to set this up, you need to have a jumper block to set what order it is in the daisy-chain. SCSI has always been a pain to setup for me, as I prefer IDE's simple plug and play system.
Now that we have looked at the specifics, lets have a look at a couple of benchmarks.