The most interesting thing in these results, to my eyes, is the OCZ Vector's results in BootRacer. They've probably been in past reviews, but this is the first time I saw them.
Single-task benchmarks like sequential and random read and write clearly don't model the boot process well. That's a mix of random and sequential reads and writes. The drives are black boxes containing cache and logic; probe them with such simplistic tests, and it's reasonably easy for the logic to detect the test, and optimize cache for that use case. Assuming the logic can do this, all the benchmarks are testing then is the raw speed of the underlying hardware. The smartness of the logic itself is not being tested.
For example, in a read benchmark, no cache need be reserved for buffering writes; and vice versa.
But real-world use cases have a mixed set of reads and writes, of varying sizes. That's where the real test of the logic comes in. The logic needs to make hard decisions about how much cache to keep spare, and which to spend on optimizing reads vs writes. And it's likely that one drive which is best for boot times won't be as good as the best drive for launching apps, and that won't be the best one for a database cache.
We need better benchmarks. These benchmarks are too easy to game, and risk rewarding market to inferior products.
Originally Posted by [USRF Obiwan]So much money for a 3cm x 6cm strip of electronics inside a large empty box....
And yet it's competitvely priced. I remember when the first round of SSDs came out in 2008/9 and they were worth close to their weight in gold. Mental, but then look at CPUs! almost nothing to them, but some are worth Thousands. its what it can do you for in the that you pay for, not what it is.
Whoa whoa whoa, wait...what TRIM "command" is the author referring to on the second page? I thought TRIM ran somewhat automatically, cued up when files are permanently deleted...some clarification would be greatly appreciated!
Originally Posted by toolio20 Whoa whoa whoa, wait...what TRIM "command" is the author referring to on the second page? I thought TRIM ran somewhat automatically, cued up when files are permanently deleted...some clarification would be greatly appreciated!
you can force windows to run TRIM, just as using SSDTweaker can disable it -which is exactly what was written on page 2
Not to be a grammar Nazi or anything, but the sentence:
"first allowing five restarts to allow Windows 7 to get it's caching together"
under the BootRacer section, should read
"first allowing five restarts to allow Windows 7 to get its caching together."
Otherwise, sound review. Glad bit-tech take benchmarking so seriously.
Originally Posted by Acanuck Not to be a grammar Nazi or anything, but the sentence:
"first allowing five restarts to allow Windows 7 to get it's caching together"
under the BootRacer section, should read
"first allowing five restarts to allow Windows 7 to get its caching together."
Comments 1 to 12 of 12
ReplySingle-task benchmarks like sequential and random read and write clearly don't model the boot process well. That's a mix of random and sequential reads and writes. The drives are black boxes containing cache and logic; probe them with such simplistic tests, and it's reasonably easy for the logic to detect the test, and optimize cache for that use case. Assuming the logic can do this, all the benchmarks are testing then is the raw speed of the underlying hardware. The smartness of the logic itself is not being tested.
For example, in a read benchmark, no cache need be reserved for buffering writes; and vice versa.
But real-world use cases have a mixed set of reads and writes, of varying sizes. That's where the real test of the logic comes in. The logic needs to make hard decisions about how much cache to keep spare, and which to spend on optimizing reads vs writes. And it's likely that one drive which is best for boot times won't be as good as the best drive for launching apps, and that won't be the best one for a database cache.
We need better benchmarks. These benchmarks are too easy to game, and risk rewarding market to inferior products.
And yet it's competitvely priced. I remember when the first round of SSDs came out in 2008/9 and they were worth close to their weight in gold. Mental, but then look at CPUs! almost nothing to them, but some are worth Thousands. its what it can do you for in the that you pay for, not what it is.
That and it would be cool if you could plug this guy directly too your motherboard.
thats what i was thinking , for what it is its tiny ... runs off to open up my 60 gig ssd to see if its the same
you can force windows to run TRIM, just as using SSDTweaker can disable it -which is exactly what was written on page 2
"first allowing five restarts to allow Windows 7 to get it's caching together"
under the BootRacer section, should read
"first allowing five restarts to allow Windows 7 to get its caching together."
Otherwise, sound review. Glad bit-tech take benchmarking so seriously.
Good spot! Fixed now, cheers.
Ahhhhh, page 2...yessss, I see now. I jumped straight to the "TRIM" section on the summary page. Thanks for the asked-for clarification!
-
« Previous
-
1
-
Next »
Discuss in the forums