USB Flash Drive Roundup - 10/2005
- Posted in
- Memory
How We Tested
Originally, when we set out to compare the drives in this roundup, we wanted to do so with 100% real world tests that we would put together ourselves, which we ended up doing. We devised three separate tests consisting of 500MB of small files, 500MB of medium-sized files and one 500MB file. We performed read and write tests using each of the file groups individually, and recorded the time that it took for those tests to complete.
Alongside our own home-brew tests, we also took a look at SiSoft Sandra 2005's built-in Removable Storage/flash Benchmark, which measured performance of the drives at various file sizes. To be more specific, Sandra's test looks at performance of 512-byte, 32KB, 256KB, 2MB and 64MB file sizes. Much to our surprise, Sandra's results almost entirely echoed our own, all while providing more points of reference. So, we scrapped our tests and moved entirely to Sandra; not only were the tests easier to run, but they can also be run by our readers as well. Obviously, the test system does impact (to a certain extent) the numbers, but it is better than not having anything to which you can compare at all.
Since the vast majority of the Sandra tests focus on smaller-file write performance, we found that many manufacturers' claims weren't even remotely close to being true. While we expected that small file read/write performance would be significantly slower, we also seemed to surprise a few manufacturers in how poorly their drives did at some of the medium-sized file tests. It seems that most manufacturers tend to only pay attention to large file transfers, which are useful, but more users tend to copy lots of smaller files rather than one or two massive files onto these drives.
Case in point is something that we discovered when running our own real world tests. We copied about 2000 files, totaling 500MB in size, to some of these USB drives. Most of these files were digital camera photos, while others were small text files. The fastest drive completed the copy in just under two minutes, but the majority of drives took more than 8 minutes to complete the task! Keep this scenario in mind when you evaluate the usefulness of a drive that performs well at small file sizes as well as large file sizes.
We also looked at the impact that AMD vs. Intel systems had on the devices, but most drives exhibited platform-independent performance.
Test Hardware
Originally, when we set out to compare the drives in this roundup, we wanted to do so with 100% real world tests that we would put together ourselves, which we ended up doing. We devised three separate tests consisting of 500MB of small files, 500MB of medium-sized files and one 500MB file. We performed read and write tests using each of the file groups individually, and recorded the time that it took for those tests to complete.
Alongside our own home-brew tests, we also took a look at SiSoft Sandra 2005's built-in Removable Storage/flash Benchmark, which measured performance of the drives at various file sizes. To be more specific, Sandra's test looks at performance of 512-byte, 32KB, 256KB, 2MB and 64MB file sizes. Much to our surprise, Sandra's results almost entirely echoed our own, all while providing more points of reference. So, we scrapped our tests and moved entirely to Sandra; not only were the tests easier to run, but they can also be run by our readers as well. Obviously, the test system does impact (to a certain extent) the numbers, but it is better than not having anything to which you can compare at all.
Since the vast majority of the Sandra tests focus on smaller-file write performance, we found that many manufacturers' claims weren't even remotely close to being true. While we expected that small file read/write performance would be significantly slower, we also seemed to surprise a few manufacturers in how poorly their drives did at some of the medium-sized file tests. It seems that most manufacturers tend to only pay attention to large file transfers, which are useful, but more users tend to copy lots of smaller files rather than one or two massive files onto these drives.
Case in point is something that we discovered when running our own real world tests. We copied about 2000 files, totaling 500MB in size, to some of these USB drives. Most of these files were digital camera photos, while others were small text files. The fastest drive completed the copy in just under two minutes, but the majority of drives took more than 8 minutes to complete the task! Keep this scenario in mind when you evaluate the usefulness of a drive that performs well at small file sizes as well as large file sizes.
We also looked at the impact that AMD vs. Intel systems had on the devices, but most drives exhibited platform-independent performance.
Test Hardware
| CPU: | AMD Athlon 64 X2 4200+ (2.2GHz/512KBx2) Intel Pentium 4 530 (3.0GHz) |
| Motherboard: | ASUS A8N-SLI Deluxe Intel D955XCSLKR |
| Motherboard BIOS: | ASUS: Version 1013 Dated 08/10/2005 |
| Chipset: | NVIDIA nForce4 SLI Intel 955X |
| Chipset Drivers: | nForce4 6.66 Intel 7.0.0.1025 |
| Memory: | OCZ PC3500 DDR 2-2-2-7 Crucial DDR2-667 5-5-5-15 |
| Video Card: | NVIDIA GeForce 7800GTX |
| Video Drivers: | NVIDIA ForceWare 78.01 |
| Desktop Resolution: | 1280 x 1024 - 32-bit @ 75Hz |
| OS: | Windows XP Professional SP2 |
| Benchmarking Software: | SiSoft Sandra 2005 SR2a-1060 |
Originally published at https://www.anandtech.com/show/1807 (Wayback Machine).