Printable View
Another top board from ASUS for P55, has SATA III and other interesting features(PLX chip??:))...but personally I still prefer other boards...here it is, enjoy:
http://www.techpowerup.com/img/09-08-19/106a.jpg
More:
http://www.techpowerup.com/101925/AS..._SATA_III.html
Other pictures:
http://translate.google.it/translate...hl=it&ie=UTF-8
What that PLX chip does is in that link.
According to the article, it makes sure the Marvell 88SE9123 controller doesn't provide a bottleneck for the SATA III ports.
Now look at those pwms? Gigabyte doesn't have to feel ashamed anymore. :rofl:
Definitely.:DQuote:
Originally Posted by RPGWiZaRD
Its the essence of competition, even though useless at times(I didn't said totally, don't shoot me.).:yepp:
PWMs are like transistors. They get smaller and smaller, and everyone wants more, MORE of them! :DQuote:
Originally Posted by RPGWiZaRD
MSI slogan is: less is better! ;)
yeah but how? 0_oQuote:
Originally Posted by SKYMTL
its not connected to the cpus pciE lanes... then how can it improve bandwidth for the marvell chip... wutt? this doesnt make any sense? :confused:
and those extra chokes... oh god... asus and gigabyte are caught up in a "i can put more chokes on my board" competition :shakes:
Arggh.
Such a nice board with good (perhaps overkill) features, utterly ruined by the crap PWM heatsinks and cheesy SB heatsink! This is supposed to be the top model (not counting ROG), one would think they could be a little bit more mature in their designs. :(
Quote:
Originally Posted by zalbard
;)Quote:
Originally Posted by Kean
I had some doubts of my own and called hipro5 about it. He said that all these extra chokes and multi-PWMs/capacitors on the motherboards are pure marketing, they do nothing of value. Less and better quality is always better. Generates less heat - produces better results.
ASUS and GIGA only seem to want to impress people based on the "american muscle" philosophy, but this is not a car :ROTF:
My thoughts exactly, I thought that the Marvell SATA3 controller was not yet ready for mass production because it is still under beta tests...They have absolutely no reason for putting it on their mobo right now, they only do it in order to impress the buyers.Quote:
Originally Posted by saaya
From what I knew, the Marvell controller would be 100% ready somewhere between October and November, when JMicron would begin making their own for a Q4 release...If only we had a SATA-3 drive in order to test it...I believe ASUS would have a lot to explain if the performance was erratic...
And all this about the PLX chip :shrug::confused: What does it have to do with the SATA-3? How can it produce greater b/w for the Marvell chip? This really doesn't make any sense....
Good to know certain people are intent on simply trolling a specific company's product without even informing themselves about it's key features... all stated in the linked article which they surely could've read?
How does the PLX help mitigate the PCIe 2.5Gb/s > SATA 6Gb/s limitation? It takes four 2.5 Gbps 1.1 lanes from the PCH and bonds them into a single 5 Gbps 2.0 lane:
http://pic.xfastest.com/MB/ASUS/P7P5...Premium-18.jpg
I still don't get what this slide suggests - if the PCIe has a bandwidth of 2.5 Gb/s and LIMITS the SATA controller, why doesn't this happen to all our current motherboards? Or does it happen but we don't notice it? Does the Revolution or the SuperComputer have a similar PLX-design for their SAS controller?
It seems weird to me, linking the Marvell chip or any other SATA controller on the PCIe lanes and their bandwidth...
And you know what? If this is the case indeed and the SATA-3 controller gets limited by the PCIe lane, then this thing that ASUS did is purely immature, and saaya is right to "troll" at them.
You know why? Because I know that the Marvell chip wasn't supposed to be out at the time being, but they implemented it purely for marketing reasons. If the SATA-3 needs a PCI-Express 3.0 lane in order to work at full bandwidth, then they should have waited and not make a quick fix via the PLX chip (and call it a revolution!!). Heck, we don't even have the SATA-3 HDDs/SSDs, it is a simple move of "domination" over the other brands.
Somebody ban this n00b troll please.Quote:
Originally Posted by gznp
your kidding right?Quote:
Originally Posted by sholvaco
what they basically say is that there is no hdd that will use more than 600MB/s so 500MB/s is "good enough"... well then why do we need SATA 6GBps in the first place? this is better than hooking up the controller via 1 pciE 1.1 lane, i guess... im curious to see some benchmarks, but i doubt it really makes a diference... id say if you do 6Gbps either do it right or dont do it and offer a raid controller or some extra sata 2 ports instead...
in the end who we really should blame is intel by giving 1156 cpus only 16 pciE 2.0 lanes...
Am I missing something or are they being really silly here? :confused:Quote:
PLX8613 (the bridge) offers a PCIe Gen2 x1 lane of 500MB/s, which can FULLY support 600MB/s of SATA6Gb/s
Quote:
Originally Posted by SKYMTL
Some misreading there. Here are two quotes from the article:Quote:
Originally Posted by saaya
Marvell 88SE9123 is a PCI-E 2.0 x1 device. So yes, it's getting a 500 MB/s connection.Quote:
ASUS used an interesting (and expensive) method of making sure the controller works to its full potential.
There you go. :)Quote:
One of these ports, configured as PCI-E 2.0 x1, connects to the Marvell 88SE9123 (which is a PCI-E 2.0 x1 device) to make sure it is provided with a 500 MB/s interconnect, reducing its bottleneck compared to most other motherboards that simply connect one of the four PCI-E 1.1 lanes from the PCH to it.
hahaha :rofl::rofl:Quote:
Originally Posted by HKPolice
Thanks for clarifying exactly what I was asking...
Let me state it yet again:
1) Does my Marvell SAS Controller on my ASUS Revolution board get limited by it's link with the PCIe 2.0 x1 lane, as the SATA-3 one does at the P55 board?
2) Why didn't they wait for PCI Express 3.0, which will deliver a 1 GB/sec transfer rate, but prefered to combine 4x PCIe 1.1 x1 (250MB/sec each) lanes to a 1x PCIe 2.0 x1 lane of 500MB/sec, not only limiting the SATA3 controller but also doing something useless, since there is no SATA3 drive on the market yet?
edit: I just saw btarunr's post - as I recall, Marvell had PCI-Express 3.0 in their early specs for this controller. So, if they really made an early PCI-Express 2.0 one (people in the industry told me that the controllers are not fully ready yet), it seems like they rushed things a bit. It is lame to give the customers a SATA3 controller which limits the SATA3 bandwidth...
The 9123 is a pcie 2.0 x1 device, with that in mind more lanes wouldn't make a difference anymore than putting a x1 card in a x4 slot.
http://img36.imageshack.us/img36/7876/maxc.jpg
Have a 400 MB/s SATA III SSD, will run. That's the objective here. It's like how the first SATA controllers used PCI and were capped at PCI bandwidths. ASUS' solution reduces the bottleneck, not eliminates it. And no, I don't think PCI-E 3.0 was ever part of 88SE9123's specs. The bus itself is far from ready.
then the limitation is actually on marvells side? it only has a single lane interface? :stick:
ok i take it all back, the blame clearly lies with marvell...
thanks btarnr for clearing that up! :toast:
and its nice from asus to go the extra step and offering the bridge chip...
though i have to wonder if a sata2 raid chip with a 2 or 4 lane interface wouldnt make more sense and be faster... and cost less...
Yep. That's the cheapest SATA 3 controller out there, and has very few pins (so it's the easiest way for motherboard makers to boast of SATA 3 support). How board makers implement it, is their headache. Sure, it will never run a drive at >500 MB/s (PCI-E 2.0 x1 theoritical max.), but at least it makes sure a 300~450 MB/s SATA III SSD will run just fine.Quote:
Originally Posted by saaya
yeah... 1... but how about 2? :DQuote:
Originally Posted by btarunr
so there ARE marvell sata3 controllers with more than 1 pciE lane?
then why dont they use those instead of using the celeron marvell controller with a plx bridge chip? how complicated lol... and plx bridge chipset arent cheap... this plx chip is probably around 5-10$? i doubt the marvell controller with more lanes, if it does exist, would come at a bigger premium than that...?
Oh, so you "know that the Marvell chip wasn't supposed to be out at the time being?" Has it occured to you that this SKU maybe hasn't even began being stocked yet? Unlike the boards already circulating into retail channels, that were supposed to feature this, but would have failed to meet the deadline by waiting on Marvel to pull their dancing bananas together.Quote:
Originally Posted by gznp
Just marketing, huh? So going by your logic they shouldn't be allowed to integrate a newly available feature on a highly specialized product aimed specifically at people who are either (going to) need it or are simply interested in the tech? They have other boards in their line up that skip on this feature if you are so bothered by it, so why even get caught up on something like this?
Bottom line: Marvell made this use the x1 interface, not ASUS. Intel limited the PCH PCIe lanes to 2.5 Gbps, not ASUS. They've actually bothered to provide a daisy chain solution where others have clearly not.
No offense but rereading your first post it didn't sound like you even took those "measly" 500MB/s into account.Quote:
Originally Posted by saaya
Intel could have done that and not cripple the PCH connectivity on top or just provide the damn native SATA 3.0 in the first place. But they didn't so this is what the board engineers have to work with. In this case they actually went through the lengths of rigging this whole setup instead of just tacking the controller on and calling it a day. The later would have been a pure marketing stunt, this actually has some real world value.
1) Yours and every other onboard integrated Marvell SAS conroller is 3Gbps (6Gbps came out in february this year).Quote:
Originally Posted by gznp
2) PCIe 3.0 is lagging well behind. It's specs aren't even going to get finalized until mid next year, with products available maybe a year after that!
Marvell on the other hand could have saved everyone the trouble by making the controller use an x4 2.5 Gbps interface. Not every chipset out there has the 2.0 lane redundancy of the X58.
Actually four. 88SE912 supports two SATA 3 ports, and one IDE channel (for two devices), so it's a big incentive for mobo designers dealing with lack of IDE from recent Intel chipsets (they don't have to buy another IDE controller, if they're providing IDE). Maybe this PLX chip + 88SE912 solution is the cheaper route for ASUS (compared to a beefy SATA 3 controller with a many-lane PCI-E interface). All they end up with is a shiny "SATA 3 support" badge on the package (which lives up, to an extant), and implemented better than by others.Quote:
Originally Posted by saaya
Thats why MSI sucks. ;)Quote:
Originally Posted by Kean