Read more.Wanting SATA 6Gb connectivity on your existing board? HighPoint has a solution.
Read more.Wanting SATA 6Gb connectivity on your existing board? HighPoint has a solution.
Correct me if I'm wrong, but there's no way the PCI-e bandwidth available to those cards can handle even one of those ports on full whack, let alone two.
You're wrong
PCI-E 16 lane slot:
* v1.x: 4 GB/s
* v2.0: 8 GB/s
* v3.0: 16 GB/s
Scan list the following, Don't know how good it is
Oooh, so close. But unfortunately those would be PCIe x1 connectors (rather than PCIe x16) on the cards, which means you need these figures instead:Actually, it gets interesting when you look at storage evaluation in Hexus's motherboard reviews. Most SATA II controllers average around 90MB/s, but can hit well in excess of 200MB/s in bursts. That meanas that 2 SATA II drives in full flow could just about max out the PCIe x1 bandwidth. So, in terms of sustained data rate the cards can probably (just) cope, but they'll almost certainly be bus-bottlenecked for peak bandwidth. Now, if they knock up a PCIe x4 version with 4 ports on it, you'd be good to goCapacity Per lane:
v1.x: 250 MB/s
v2.0: 500 MB/s
v3.0: 1 GB/s
But when does burst speed actually matter? I tend to disregard it completely unless it is very similar to the sustained speed (that happened to me once, basically my Raptor was running at 10MB/sec if that due to a compatibility issue with a controller).
I'd assume that if the burst speed gets bottlenecked it has some effect on the average speed as well, but I'll happily admit that I'm no expert on storage evaluation I was just working on the principle that any time the interface bottlenecks the device it connects to you're in a sub-optimal place...
I'm no expert either, but that sounds a sensible aproach to viewing the problem. I suspect however that you would rarely reach the burst speed available either because the data is fragmented on the drive - introducing momentary seek delays - or because the volume of data being read is too small to fully occupy the burst capability.
Personally I think that having a version with a PCIe x4 connector would be advantageous, but in its current form I doubt that it would be too limiting unless you were running some SSDs in RAID or something like that.
I would also like to pose the question: how does the occupying of some PCIe lanes affect graphics card performance etc if you are pinching a few to run HDDs on?
"unless you were running some SSDs in RAID or something like that" First off RAIDing SSDs doesn't mean they need SATA 6gbps, RAID doesn't affect that. Secondly, what's the point in having a SATA 6Gbps-capable card if it's only capable of SATA2 performance? Yes it's true, you could use one port to its full potential at any one time.. but by the logic of "I doubt it'd be too limiting" you may as well just stick with a SATA 2 as its cheaper and just as effective!
As for lanes - not likely to be a concern. Supposing a 'typical' mobo has 32 lanes - which might be one x16 slot, then either another x16 or two more x8 slots, you might theoretically see a difference in benchmarks if you had two ultra-expensive graphics cards in SLI + a controller card, as the second card would be 'limited' to x8 bandwidth (4Gbyte/s in each direction simultaneously, IIRC) which is not going to be limiting the card in most if not all cases. When PCIe2.0 came out, x16 (today's x8) was still perfectly adequate for the high-end cards.
Basically, I could be wrong, but I don't think it would matter.
6Gb/s is ~ 715MiB/s
One lane of PCIE 2.0 is 500MiB/s link speed, which with 8B/10B encoding gives you 400MiB/s before protocol overheads of the various layers.
So yeah, a lane of PCI-E ain't enough for a full burst of SATA 6Gb. It should just be enough for a 3Gb/s burst.
Still, there are bigger bottlenecks in the storage architecture than a PCI-E lane.
Biscuit (02-11-2009)
There are currently 1 users browsing this thread. (0 members and 1 guests)