Introduction

My Atlas 10K, humming along at a cool 29.2 C after hours of operation.

If the high price and temperatures of 10,000 rpm Ultra 2 SCSI drives have kept you from upgrading your Mac's hard drive performance, I've got good news. Competition and the free market have finally delivered a cheaper, cooler alternative, Quantum's new top-of-the-line Atlas 10K Ultra 2 LVD 160/m drive. The Atlas 10K is available in the familiar 9.1, 18.2, and 36.4 capacities, but price and temperature aren't its only advantages over Seagate's Cheetah. My testing with MacBench 5.0 and ATTO's benchmark utility show the Atlas is also slightly faster than the Cheetah. My results are shown below, along with a short tutorial on some neat gadgets to help you monitor temperature inside your computer and drive enclosures and achieve drive operating temperatures below 30 C with the Atlas 10K for about the same price as a bare Cheetah.

Finding a Cool Deal

The price difference between the Cheetah and the 10K is truly amazing. I found my 9.1 Gig Atlas at Hyper Microsystems for $419, ninety dollars less than the best price I could find on the Cheetah, $509, also at Hyper.$419 is only about 77% of what some Mac vendors charge for a bare Cheetah! I used my savings to take advantage of the "make external" option on Hyper's drive price list page. For just $130, Hyper will install any drive you order in an external case that comes complete with an LVD cable and a multimode terminator with an LED. The drive arrives all set to hook up to your Ultra2 SCSI card - it couldn't be simpler. The "make external" option is a good value. The cable and terminator are not the highest quality available, of course, but they work well. The case is excellent quality. The drive mounts in the middle of an aluminum compartment that allows heat transfer by conduction to the sides of the compartment and radiation from the top and bottom of the drive as well as all four of the compartment's interior surfaces. A small fan at the rear of the case provides convection cooling for the drive and the power supply. Without modification, the Hyper enclosure allowed the Atlas to run at temperatures of 40-41 C, well within the Quantum's 0-50 C operating specs and about as cool as the Cheetah ran in tests with dedicated drive cooling systems cited in a June review of the Atlas 10K by Storage Review, a PC oriented webzine. With the same cooling systems, as we'll see, the Atlas will operate below 30 C in Hyper's external enclosure. You probably won't be able to find the components Hyper delivers in its "make external" option for less then $130 at your local PC parts store - I couldn't. Since their option also saves you the time and effort of assembling the enclosure, "make external" is a great way to go if you want an external drive. I had never heard of Hyper but stumbled across them thanks to a link from Deal-mac, and Hyper certainly delivers good deals. They are mainly PC oriented, but they do sell the same MacAlly RAID enclosures as popular Mac vendors. In fact, you can purchase the enclosure, cabling, and termination for a two drive raid for about $170, roughly a third of what other vendors charge. Hyper's customer service is also top notch. I canceled an order for a $289 Atlas IV, which Storage Review's June issue says beats a Barracuda, the day the drive arrived in favor of a second external 10K. Hyper handled the change with no sweat, and no restocking fee.

The components of Hyper's "make external" option. The wire is a temperature lead, not part of the external package. The compartment that houses the drive slides forward to reveal power and SCSI connectors. The enclosure's small fan draws warm air from the drive and power supply compartments. The aluminum interior of the sliding disk compartment conducts heat away from the drive and radiates it into the compartment, where air flow can remove it.

Performance Testing - MacBench 5.0 and ATTO Express Pro Tools 2.2

I tested the Atlas 10K with an Initio Miles 2 Ultra2 card in my Power Mac 8500 with 336M of physical ram. The 8500 has a Bottom Line RailGun G3 running at 313 MHz and a 1M L2 running at 1:1 on a system bus speed of 45 MHz. Virtual memory was off, of course, and I used an extension set I use for A/V capture which drops out file sharing and desktop printing from the standard MacOS 8.1 extensions. Appletalk was inactive and disk cache was set to minimum - 96K - for the MacBench tests. In Storage Review's June review, the Atlas bested the Cheetah by as much as 15% in some Windows and NT benchmarks. Since I don't have a Cheetah, I can't compare the results of the two drives with the ATTO benchmark, but my Atlas delivered MacBench 5.0 disk test scores of 2117 compared to the 1955 Mike listed for the Cheetah/Miles2 combination in a 400MHz machine on XLR8yourmac, more than an 8% increase in performance. In the publishing disk test the Atlas was 6% better, posting a 1767 compared to the Cheetah's 1660. MacBench tests system performance, consequently it is sensitive not only to the raw characteristics of the drive and interface but also to system variables like processor speed, so it seems likely the Atlas might have beaten the Cheetah by a wider margin had it been tested on the same 400 MHz machine. The Atlas' operating temperature is noted in the MacBench graph, a cool 28.4 C! For tests with ATTO tools, settings were the same except the system disk cache was set to its maximum, 7680K. Results with and without the disk cache are tabulated below, along with a screen shot from the 8 Meg file test, the slowest sustained read and write performance. Screen shots of all the ATTO Tools performance tests are here. Disk cache did not significantly effect sustained read or write rates at any file size - the size of the file was the most important factor. The system disk cache did effect peak rates, with the effect of cache on peak transfer rate increasing as the file size decreased. [Note: PR=Peak Read, SR=Sustained Read, PW= Peak Write, SW=Sustained Write -Mike] ATTO ExpressPro Tools 2.2 all figures in MB/s File Size Without Disk Cache With Disk Cache 512K PR 69.76 PW 23.61 PR 108.23 PW 83.26 SR 68.36 SW 22.22 SR 69.08 SW 23.05 2M PR 72.72 PW 24.53 PR 73.07 PW 25.35 SR 37.85 SW 23.47 SR 42.22 SW 23.58 8M PR 71.30 PW 25.00 PR 72.88 PW 25.46 SR 27.96 SW 24.01 SR 28.14 SW 24.20