The Setup
While Micron recommends testing the cache drive with a 7200 RPM desktop drive, the decision was made to do something a bit more unorthodox. One of the cases procured over the last year, came with a 2×2.5” bracket. Since most cases include 2.5” brackets now, it was decided to test this compared to a 7200 RPM laptop drive. Laptops are abysmally slow. It has nothing to do with the lower cpu clocks or limited RAM, it it totally related to the storage speed. A 7200 RPM 2.5” drive is a cross roads, still slow, but closer to desktop speed and servers as the crossover point between desktop and laptop systems.
The test system is the same used in our more recent RAID testing. An AM3 setup with 8 GB of RAM, a discrete and an integrated GPU are initialized although only discrete GPU is used in testing.
Hardware
- AMD Thuban X6 1090T
- 2x4GB Corsair Vengeance 1600 C9
- Gigabyte GA-890GPA-UD3H
- Western Digital Scorpio Black 250GB
- Patriot Pyro 256 GB SSD
- OCZ SLI Gamer Xtream 700W PSU
- ASUS 6770 Silent Video card
Software
- ATTO Disk Benchmark
- Crystal Disk Benchmark x64
- Passmark Performance Test 7 Storage
- PCMark Vantage Storage Test
- BootRacer (where applicable)
BootRacer is a OS boot device timing software and as such was only used in specific controlled circumstances. Laptop OS drive only and cached only. Not SSD specific as such the individual SSD results were not included in the Adrenaline timings for that test. It is not meant to be a boot drive and was not tested thusly.
All tests were done on the HDD, the SSD and then done with the SSD in caching mode. SSD solo tests are done with bare drive SSD, HDD results are a desktop OS experience and the same full bodied installation the average reviewer, gamer, desktop jockey might incur with only 250 GB of space to play with. AIDA write tests were not performed as the OS would be lost. BootRacer for solo SSD are also not included as the drive was only tested in factory condition.
Results
ATTO Disk Benchmark
One expects the naked SSD to far and away exceed either HDD or Cached mode and one would be wrong. In fact the caching software works so well that it actually outperforms the naked SSD in 4KB read tests and at some larger file sizes. Very impressive.
Crystal disk Benchmark x64
Crystal Disk testing was done with 0 fill test data, but it makes a different telling. Cache and naked SSD results were even throughout each individual test. Both drives completely blow the 7200 RPM laptop drive out of the water. Anywhere from twice the speed in sequential 1024K writes, fifteen times the 512K reads, to 100 times faster in 4K writes (and 4K 32QD writes).
Passmark PT7 Storage
Just like Crystal results Passmark storage tests puts cache performance at almost equal to naked SSD results. There is one exception though random seek results are only slightly better than the hard drive. Cache performance does still have an increase of about 5% which is unusual as the SSD is 20% the size of the HDD.
PCMark Vantage Storage Test
PCMark Vantage has become more problematic at producing valid results recently, and faith in Futuremark’s ability to keep pace with modern technology is reaching skeptic levels. While Dataplex software moves information to the SSD the more it is accessed, and this was easily shown in most every other test, Vantage results were far from expected results. the cache performance varies from usually barely an improvement to at most a 30% improvement in storage test scores. Naked SSD performance varies from a 3 to 20 times better performance. So at least for benching Vantage, there is no substitute for an SSD.
BootRacer Results
Most benchmark statistics can not adequately represent “desktop feel.” BootRacer however is one such application that benchmarks the time that it takes to boot inwdows and is a perfect “desktop feel” statistic, especially when the time to boot to a useable desktop is halved like in this case. The difference between logon to desktop remains the same at 7 seconds, this is due to the PCs capability to process the startup applications and is indicative of that alone. The time to logon is hugely improved and really showcases the SSD drive performance.
1 Comment
Good article. I’ve been doing some research on SSD caching. On the first page of the article, you mention that the Adrenaline SSD with Dataplex software will work with AMD, Intel or Nvidia chipsets. This is what caught my attention because Nvelo says their Dataplex software does not work with Nvidia storage controllers. As you might have guessed, I have a Nvidia chip set and I’m trying to find a caching solution that has been proven to work. When I chatted with Crucial, they first said it would work and then they said no after I pointed them to the Nvelo limitation with Nvidia. They did say that I could buy it and return it if it didn’t work. Before I do that, I wanted to check if anyone knows for certain that this will work with Nvidia. I appreciate your help.
Thanks!