Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Do the 2010 MacBook Pro's have 6 GB SATA?
- Thread starter steffi
- Start date
- Sort by reaction score
There is absolutely no need for 6 Gb/s SATA(at least not in a MBP) since there are no hard drives that come even remotely close to filling the bandwidth that 3Gb/s SATA II provides.
The only real use for 6 Gb/s SATA would be in some high end RAID setup that I can't even fathom...
The only real use for 6 Gb/s SATA would be in some high end RAID setup that I can't even fathom...
There is absolutely no need for 6 Gb/s SATA(at least not in a MBP) since there are no hard drives that come even remotely close to filling the bandwidth that 3Gb/s SATA II provides.
The only real use for 6 Gb/s SATA would be in some high end RAID setup that I can't even fathom...
Not true. There's a few users on these boards running these in their MacBook Pros already. They fully saturate the bandwidth of SATA II during sequential reads and come close during sequential writes.
There is absolutely no need for 6 Gb/s SATA(at least not in a MBP) since there are no hard drives that come even remotely close to filling the bandwidth that 3Gb/s SATA II provides.
The only real use for 6 Gb/s SATA would be in some high end RAID setup that I can't even fathom...
SATA II is enough for hard drives but many SSDs are achieving the limits of SATA II, so to get more out of SSDs, you need SATA III.
SATA III isn't very common yet, only some desktops has it and that costs nice extra too
Ok..riddle me this.
What application works like that for a meaningful period of time where no CPU is required to do anything with the data..outside of a synthetic benchmark?
Well, 90% of beachballs happen when CPU has to wait for HD...
Majority of people don't need i5, but they still want one
Hate to break the bad news, but CPUs wait for HDD IO 99.999% (or more) of the time.
Im bettin the beachball thing...aint waiting for drive IO.
Im not a software expert, and with that analysis, Id bet you aren't either.
But I am a qualified enterprise storage performance expert, and speeds & feeds on the outside of the box only impress people that dont know how things really work.
As earlier stated, interface speeds only start to matter in larger shared infrastructures where intelligent IO not present in desktop/consumer designs makes it more advantageous. Any amount of random IO destroys the idea of a spinning media drive saturating SATA-2 speeds in a real world workload.
Apply a little queuing theory, then plug in what you can find about bus interrupt, drive rotational and track-to-track latency numbers within SATA, and you'll find a lot of empty bandwidth in the path.
It can take a significant number of FCAL/SAS HDD's to saturate a 4Gpbs FCAL loop with reasonably normal real-world IO data rates and patterns out 'in the world' as well.
But it all comes down to defining the unique dataset, and the unique workload to be applied against it. Is it realistic..benchmark, or synthetic to prove a point?
I'd bet, that if someone had a specific dataset, with a specific workload, that had specific IO requirements, wouldn't be relying on internal storage to get the job done anyhow...also earlier stated.
I agree that sequential reads are a pretty silly way to measure SSD performance, but your assertion that no drives available today max out the SATA II bandwidth is unequivocally wrong. I was just pointing that out.
That said, I don't care at all that we don't have SATA III in these new MBPs. It is definitely not something I would personally benefit from having in my usage.
The 15" have now intel chipsets and Intel SATA controllers. It is still SATA 2 but faster than the old nvidia controllers.
Only the highest End SSDs benefit from SATA 3 and even than not by much as only STR increases and how do you want to achieve it with no input output link that is fast enough other than copying to and from ram.
Only the highest End SSDs benefit from SATA 3 and even than not by much as only STR increases and how do you want to achieve it with no input output link that is fast enough other than copying to and from ram.
I read the whole article. I still don't see why it's needed right now. Certainly it'll eventually come but I agree with the other guy that in real world applications of today (and for quite a while) it's more of a theoretical advantage and marketing point than anything else.