Corsair Neutron & Neutron GTX: All Capacities Tested
by Kristian Vättö on December 19, 2012 1:10 PM ESTAnandTech Storage Bench 2011—Light Workload
Our new light workload actually has more write operations than read operations. The split is as follows: 372,630 reads and 459,709 writes. The relatively close read/write ratio does better mimic a typical light workload (although even lighter workloads would be far more read centric).
The I/O breakdown is similar to the heavy workload at small IOs, however you'll notice that there are far fewer large IO transfers:
| AnandTech Storage Bench 2011—Light Workload IO Breakdown | ||||
| IO Size | % of Total | |||
| 4KB | 27% | |||
| 16KB | 8% | |||
| 32KB | 6% | |||
| 64KB | 5% | |||
As our Light suite is more read centric, the benefit of more and faster NAND is smaller. All except the 120GB Neutron perform almost equally and even the 120GB Neutron is only ~16% behind whereas in our Heavy suite it was as much as 30% slower than the larger capacities.
Facebook
Twitter
RSS
44 Comments
View All Comments
Oxford Guy - Sunday, December 23, 2012 - link
Ad hominem much?Plenty of buyers would be interested in knowing that the 830, for instance, tops the charts in terms of power usage under load, particularly given the fact that Samsung's "full specs" advertised number is impossibly low.
People have been tricked by this, which is exactly why Samsung publishes that low number.
Ever heard of laptop battery life? What about heat? I suppose not.
Kristian Vättö - Saturday, December 22, 2012 - link
The figures Samsung reports are with Device Initiated Power Management (DIPM) enabled. That's a feature that is usually only found on laptops but it can be added to desktop systems as well.With DIPM disabled, Samsung rates the idle power at 0.349W, which supports our figures (we got 0.31W).
The same goes for active power, Samsung rates it at 3.55W (sequential write) and 2.87W (4KB random write QD32). The 0.069W figure comes from the average power draw using Mobile Mark 2007, which is something we don't use.
Oxford Guy - Sunday, December 23, 2012 - link
So, in a laptop, the load power for the 830 amazingly plummets from, what 5+ watts, to .13 watts?That's really amazing. I guess the next thing to ask is why these amazing results aren't part of the published charts.
Cold Fussion - Saturday, December 22, 2012 - link
I think the power consumption tests are particularly useless. How come you don't test power consumption under some typical workload and heavy workload so we can see how much energy they use?Kristian Vättö - Sunday, December 23, 2012 - link
Because we don't have the equipment for that. With a standard multimeter we can only record the average peak current, so we have to use an IOmeter test for each number (recording the peak while running e.g. Heavy suite would be useless).Good power measurement equipment can cost thousands of dollars. Ultimately the decision is up to Anand but I don't think he is willing to spend that much money on just one test, especially when it can somewhat be tested with a standard multimeter. Besides, desktop users don't usually care about the power consumption at all, so that is another reason why such investment might not be the most worthwhile.
Oxford Guy - Sunday, December 23, 2012 - link
And we know only desktop users buy SSDs. No one ever buys them for laptops.lmcd - Monday, December 24, 2012 - link
Howabout you buy the equipment for them, if it's such a great investment?Cold Fussion - Tuesday, December 25, 2012 - link
That line of thinking is flawed. If you're only catering to desktop users, why even present the power consumption figures at all? The 3-5w maximum power consumption of an SSD which will largely be idle is not at all significant compared to the 75 watts the cpu is pulling while gaming or the 150watts the gpu is pulling.The tests as they are server no real purpose. It would be like trying to measure power-efficiency of a cpu purely by it's maximum power consumption. I don't believe a basic datalogger is going to run into the 1000s.
Kristian Vättö - Tuesday, December 25, 2012 - link
I didn't say we only cater desktop users, but the fact is that some of our readers are desktop users and hence don't care about the power consumption tests. It's harder to justify buying expensive equipment when some will not be interested in the tests.Don't get me wrong, I would buy the equipment in a heartbeat if someone gave me the money. However, I'm not the one pulling the strings on that. If you have suggestions on affordable dataloggers, feel free to post them. All I know is that the tool that was used in the Clover Trail efficiency article costs around $3000.
Cold Fussion - Tuesday, December 25, 2012 - link
But it doesn't cater to mobile users because the data provided is simply not of any real use. I can go to my local retail electronics store and buy a data-logging multimeter for $150-$250 AUD, I am almost certain that you can purchase one cheaper than that in the US from a retail outlet or online.