Those dissatisfied with the performance of mechanical storage solutions can tap solid-state storage devices that substitute silicon for spinning platters. Such devices shed the mechanical shackles that limit hard drive performance, but they’ve hardly been affordable options for most users. Then Gigabyte unveiled the i-RAM, a $150 solid state-storage device that plugs directly into a motherboard’s Serial ATA port, accommodates up to four run-of-the-mill DDR SDRAM modules, and behaves like a normal hard drive without the need for additional drivers or software.
Gigabyte first demoed the i-RAM at Computex last summer, and cards have finally made their way to the North American market. One has also made its way to our labs, where it’s been packed with high-density DIMMs and run through our usual suite of storage tests. Read on for more on how the i-RAM works, what its limitations are, and how its performance compares with a collection of single hard drives and multi-disk arrays.
The i-RAM’s greatest asset is easily its simplicity. Just populate the card with memory, plug it into an available PCI slot, attach a Serial ATA cable to your motherboard, and you’ve got yourself a solid-state hard drive. There’s no need for drivers, extra software, or even Windows—the i-RAM is detected by a motherboard BIOS as a standard hard drive, so it should work with any operating system. In fact, because the i-RAM behaves like a standard hard drive, you can even combine multiple i-RAMs together in RAID arrays.
Gigabyte equips the i-RAM with four DIMM slots, each of which can accommodate up to 1GB of unbuffered memory. The card is sold without DIMMs, giving users some flexibility in how it’s configured. However, most will probably want to shoot for that 4GB maximum. After all, if you’re going to have a solid-state hard drive, you want it to be as big as possible.
Be careful when adding memory, though. The i-RAM’s DIMM slots are mounted on an angle to ensure that the card doesn’t interfere with adjacent PCI slots, and there isn’t enough room for DIMMs with thicker heat spreaders—at least not if you’re planning on packing the card with four memory modules.
While tight DIMM spacing limits compatibility with thicker heat spreaders, it’s not a major concern, because it’s unlikely you’ll want to waste high-end memory on the i-RAM. You see, the i-RAM’s Serial ATA controller is limited to 150MB/s transfer rates, creating a bottleneck that will constrain performance long before memory speeds or latencies enter the picture. In fact, even DDR200 memory has ample bandwidth to saturate the i-RAM’s Serial ATA interface.
Translating Serial ATA requests for a bank of four DIMM slots is no small task, but Gigabyte gets the job done with a Xilinx Spartan-3 field programmable gate array (FPGA) chip. The Spartan-3 is programmed to act as the i-RAM’s memory controller, Serial ATA controller, and the link between the two, accomplishing three tasks with one piece of silicon. The single-chip solution is elegant, but it’s also the source of the i-RAM’s biggest limitations. For example, the memory controller doesn’t support ECC memory or 2GB DIMMs, both of which would be useful. And then there’s the Serial ATA controller’s lack of support for 300MB/s transfer rates, which will probably be the card’s most serious performance impediment.
Since it relies on volatile memory chips for storage, the i-RAM will lose data if the power is cut. Fortunately, the card can draw enough juice from a motherboard’s PCI slot to keep its four DIMM slots powered, even when the system is turned off. The system does have to be plugged in and its power supply turned on, though.
To allow users to unplug their systems for periods of time and to protect against data loss due to a power failure, Gigabyte also equips the i-RAM with a rechargeable lithium ion battery that packs 1600 milliamp-hours of power. The battery charges while the system is plugged in, and according to Gigabyte, it can keep four 1GB DIMMs powered for more than ten hours. Battery life will vary depending on the i-RAM’s memory module configuration, though. It’s probably a good thing to back up anything you actually store on the drive, just in case.
Today we’ll be comparing the i-RAM’s performance with that of a handful of the fastest Serial ATA drives on the market and a couple of SATA RAID configurations ripped from our recent chipset Serial ATA RAID comparison. Even against a four-drive RAID 0 array, we expect the i-RAM to clean up. However, it will be interesting to see by how much.
We’ll be subjecting the i-RAM to our standard gauntlet of storage tests, but we’ve had to cut a few apps to accommodate the i-RAM’s capacity constraints. Even packed with 4GB of memory, the i-RAM doesn’t offer enough storage to complete the WorldBench suite or File Copy Test’s partition-to-partition file copy test. The i-RAM’s limited capacity also impacts our iPEAK multitasking tests, as we’ll explain further in a moment.
Our testing methods
All tests were run three times, and their results were averaged, using the following test systems.
|Processor||Pentium 4 Extreme Edition 3.4GHz|
|System bus||800MHz (200MHz quad-pumped)|
|Motherboard||Asus P5WD2 Premium|
|North bridge||Intel 955X MCH|
|South bridge||Intel ICH7R|
|Chipset drivers||Chipset 126.96.36.1993
|Memory size||1GB (2 DIMMs)|
|Memory type||Micron DDR2 SDRAM at 533MHz|
|CAS latency (CL)||3|
|RAS to CAS delay (tRCD)||3|
|RAS precharge (tRP)||3|
|Cycle time (tRAS)||8|
|Graphics||Radeon X700 Pro 256MB with CATALYST 5.7 drivers|
|Hard drives|| Maxtor DiamondMax 10 300GB SATA
Western Digital Caviar SE16 250GB SATA
Western Digital Raptor WD740GD 74GB SATA
Hitachi 7K500 500GB SATA
Western Digital Caviar RE2 400GB SATA
Seagate Barracuda 7200.9 160GB SATA
Seagate Barracuda 7200.9 500GB SATA
Gigabyte i-RAM 4GB SATA
|OS||Windows XP Professional|
|OS updates||Service Pack 2|
We packed our i-RAM with four 1GB Value Series DDR400 modules from OCZ. The DIMMs in question are among the least expensive 1GB modules around, making them perfect for the i-RAM.
We used the following versions of our test applications:
- Intel IOMeter v2004.07.30
- Xbit Labs File Copy Test v1.0 beta 13
- TCD Labs HD Tach v3.01
- Far Cry v1.3
- DOOM 3
- Intel iPEAK Storage Performance Toolkit 3.0
The test systems’ Windows desktop was set at 1280×1024 in 32-bit color at an 85Hz screen refresh rate. Vertical refresh sync (vsync) was disabled for all tests.
All the tests and methods we employed are publicly available and reproducible. If you have questions about our methods, hit our forums to talk with us about them.
Boot and load times
To test system boot and game level load times, we busted out our trusty stopwatch. The entire system partition was housed on the i-RAM during the system boot test, but there was only room for each game’s install files for our level load tests.
The i-RAM allowed our test system to boot faster than any other configuration, but only by a few seconds. The RAID arrays look a little slow here because of the extra time it takes for the motherboard to initialize them during the boot process.
In-game level load times benefit more from the i-RAM than the Windows XP boot process, but to be honest, we were expecting a little more. Clearly, the storage subsystem isn’t the only bottleneck that constrains game level load times.
File Copy Test is a pseudo-real-world benchmark that times how long it takes to create, read, and copy files in various test patterns. Scores are presented in MB/s.
Now that’s more like it. The i-RAM rips through FC-Test’s file creation tests two to three times faster than any other single drive. Only our four-drive RAID 0 array exceeds the i-RAM’s performance here, and then only with a couple of test patterns.
File read performance isn’t even close; the i-RAM tears through this test, offering better performance across the board.
The i-RAM doesn’t slow down in the copy test, either. It’s more than four times faster than some of our single-drive configs, and easily much faster than any of our multi-drive RAID arrays.
We recently developed a series of disk-intensive multitasking tests to highlight the impact of command queuing on hard drive performance. You can get the low-down on these iPEAK-based tests here. The mean service time of each drive is reported in milliseconds, with lower values representing better performance.
Although iPEAK would run on our 4GB i-RAM partition, the app did warn us that tests would have to be wrapped due to the drive’s small size. This means that any I/O requests that would have referenced areas of the drive beyond 4GB would be wrapped around to the beginning of the drive.
The i-RAM continues to rip up the field, taking top honors in all but one of our first round of iPEAK multitasking tests.
Our second round of iPEAK tests proves just as fruitful for the i-RAM, which has little problem outclassing the rest of the field.
The i-RAM is in another league in IOMeter, offering transaction rates that are an order of magnitude higher than its closest competition. It doesn’t take long for the i-RAM to get revved up, either. The card hits its peak transaction rate with just two simultaneous I/O requests.
The i-RAM’s IOMeter response times are as low as we’ve ever seen, by a long shot. Not even our four-drive RAID arrays are in the same ballpark.
The i-RAM’s blistering transaction rates and seemingly impossibly-low response times do come at a price. CPU utilization is much higher than our single-drive configs and RAID arrays. The i-RAM still pushes more I/Os per CPU cycle than any other configuration, though.
We tested HD Tach with the benchmark’s full variable zone size setting.
The i-RAM doesn’t quite make it to 150MB/s in HD Tach’s sustained read and write speed tests, but it’s faster than any other configuration.
Unfortunately, a lack of support for 300MB/s Serial ATA transfer rates keep the i-RAM out of the running in HD Tach’s burst speed test. It is the fastest drive among those that only support 150MB/s transfer rates, but that’s not good enough to catch several competitors.
Since the i-RAM uses random access memory, it’s no surprise that its random access time is significantly faster than our hard drives and RAID arrays.
CPU utilization results are just within HD Tach’s +/- 2% margin of error in this test. Given these results, it seems likely that the high CPU utilization we observed in IOMeter was more a factor of the i-RAM’s incredibly high transaction rates than any inherent CPU utilization penalty associated with using the device.
The i-RAM is silent, so it doesn’t add any noise to a system—well, at least not unless you have some funky active DIMM cooling.
Power consumption was measured for the entire system, sans monitor, at the outlet. We used the same idle and load environments as the noise level tests.
Given its performance, the i-RAM’s power consumption is certainly reasonable. It doesn’t really consume less power than your average Serial ATA hard drive, but it’s much more frugal than multi-drive RAID arrays, and those are the only configurations with a chance of even coming close to matching its performance.
To be honest, we didn’t actually expect Gigabyte to turn the i-RAM into an actual end-user product, much less make it available in North America. But they have, and at $150 online, the i-RAM is actually pretty affordable, all things considered. With the price of 1GB DDR modules is hovering around $80, it’s possible to build a 4GB i-RAM drive for under $500. That’s a horrific cost per gigabyte for a hard drive or RAID array, but it’s pretty good for a solid-state storage device with this kind of performance.
Of course, the i-RAM isn’t without limitations. Performance is undoubtedly constrained by the 150MB/s Serial ATA interface, and I shudder to think how much faster the i-RAM could be if it supported 300MB/s transfer rates. Size is an issue, as well. With only four DIMM slots and no support for 2GB modules, the i-RAM hits a capacity ceiling at 4GB. That might be enough storage for certain applications, but it leaves us wanting more. We’d gladly accept a double-wide design if it allowed for a greater number of DIMM slots and a larger overall capacity. As it stands, you’ll have to rig up multiple i-RAM drives in RAID to breach the 4GB barrier.
While we’re griping, it’s tempting to suggest that Gigabyte skip Serial ATA altogether and build an i-RAM that taps the bandwidth of multiple PCI Express lanes. Such a card could offer considerably more throughput than even a 300MB/s Serial ATA interface, but it would require drivers, if not additional software, and that would ruin some of the i-RAM’s elegance. As it stands, the i-RAM should work in any system with a Serial ATA port and PCI slot, regardless of the operating system.
Although the i-RAM’s cost and limitations ultimately constrain its appeal, they don’t take away from the fact that it’s significantly faster than any other storage solution we’ve tested. Performance oscillates between impressive and awe-inspiring, and for those niche markets that demand blistering I/O, the i-RAM may be just the ticket.