We have been hearing some odd rumblings in the world of motherboards in the last few days centered on a particular item that most motherboard vendors had been hoping would help drive the desire to upgrade: integration of the first SATA 6.0 Gb/s controllers. We first found out about the information over the weekend but wanted to wait until we had some more information from other vendors before posting the news.
While cruising along Computex in Taipei this past June we saw more than a handful of new motherboards claiming support for the new storage standard. We saw boards from Gigabyte with it:
We also saw some ASUS products with SATA 6.0 Gb/s (SATA 6G) support as well as MSI offerings. And not just on motherboards for the upcoming Lynnfield processor either: Gigabyte stated they were planning an entire refresh of their Intel and AMD lineup of motherboards in late summer that added a few features, one of which was the support for SATA 6G storage components.
At the heart of all this SATA 6G madness was a single chip as far as we can tell: the Marvell 88SE9123 controller. The big three (ASUS, Gigabyte and MSI) were all using this controller on their motherboards to support SATA 6G speeds when those hard drives and SSDs are finally made available.
But a problem has apparently crept up. ASUS told us recently that the P7P55D EVO, based on the P55 chipset for the upcoming Lynnfield processors (and previewed by PC Perspective earlier this week) would not have the Marvell 88SE9123 chip on-board when it ships. The engineering sample that we have been playing with does have the chip and connections on it for the SATA 6G tech, but without a SATA 6G compatible devices we haven’t been able to test it anyway.
What’s more, ASUS is saying that the chip is being pulled off nearly ALL of their P55 motherboard lineup and only the super-high-end motherboard, the P55 Premium. Why? The only information we have gotten from anyone revolves around some hardware AND software issues that are preventing the SATA 6G speeds from actually reaching 6.0 Gb/s. Obviously that’s an important part of what customers will want on their boards when the hardware is finally ready. This is a significant issue if a company like ASUS is willing to complete redo its motherboard lineup at the last minute before a major product launch.
The Marvell 88SE9123 on our ASUS P7P55D EVO pre-release board
Gigabyte is also apparently taking the same actions: they are pulling the Marvell 88SE9123 SATA 6G chips from all of their P55 motherboards. In fact, many Gigabyte offerings were to have TWO Marvell chips on them supporting 4 channels of SATA 6G – obviously now that is out the window. Initially when contacting the motherboard vendor, they seemed optimistic that it would be able to fix the problem before the P55 launch window. That apparently has changed in the last 48 hours and Gigabyte made the decision to pull the feature.
One other interesting note from Gigabyte is that they are planning on integrating SATA 6G on the upcoming refresh of their X58 lineup coming out in late August. The hardware issue should be fixed by then, though by who (either Gigabyte or Marvell) is in question. Then, so we are told, the company will “gauge adoption” to see when they should reintegrate the technology on their P55 product line.
This leaves us with a lot of questions unfortunately. What exactly is wrong with the Marvell 88SE9123 chip? Is it a hardware problem or something that will be fixable with a firmware update or driver change down the road? It is likely the former as companies like ASUS and Gigabyte would be able to save their current designs and simply release updates once the SATA 6G products begin to hit the scene. If two of the big three motherboard companies are willing to simply drop the feature, the issue is more likely a hardware or implementation defect.
In discussions about what could possibly be wrong here with other PCPer staffers, ideas like inadequate testing for signaling issues on cramped motherboard designs were the most popular. Since these motherboards are possibly all built on a 6 layer PCB designs, with the exception of the ASUS P55 Premium offering (and other high end boards, like Gigabyte’s X58), that seems like a possible scenario. The extra room engineers would have to run traces on an 8 layer board could help alleviate the hardware malfunctions. It’s also possible that the speed of 6 Gb/s over a single differential pair may be causing some interference issues of its own. Higher-end boards may be able to retain the chip with additional heatsinks (layers of large block traces to help dissipate heat along the PCB) along the SATA trace paths, acting as extra shielding. At this point, it’s all speculation though.
I am sure a lot of readers will simply shrug their shoulders on this news and mutter to themselves, “what does it matter, there aren’t any SATA 6G hard drives out anyway?” – and that’s a valid point. It’s the classic chicken and egg scenario – would you rather have a motherboard with SATA 6G support and now hardware to run at that speed or buy a SATA 6G SSD early in 2010 but not have the hardware to support it in your system yet? The obvious best-case scenario is the former and we were hoping that this would be the case for the large amount of early adopters that will surely fall in love with the upcoming Lynnfield-based platform.
We will keep you updated as more information comes in on this situation – stay tuned to PC Perspective!
UPDATE 7/15/09 – The folks at Marvell got in contact with us to touch base about this issue and the information they provided is in fact, very interesting. Here it is:
Recent news of delays of SATA 6Gb/s product launches is due to a hardware issue related to PATA (legacy Parallel ATA, or IDE) compatibility on the Marvell 88SE9123. This Marvell controller chip supports both 6Gb/s SATA as well as PATA on the same chip. It is the legacy PATA functionality, not the SATA, which is the root cause of the product delays. Marvell has already identified a solution to the PATA issue and a new chip revision to resolve the issue will be introduced quickly. To date, as far as Marvell is aware, no SATA (6Gb/s, 3Gb/s, or 1.5Gb/s) issues with the 88SE9123 have been reported by Marvell customers or partners, or discovered via extensive Marvell internal qualification testing. Marvell continues to work with its customers and partners on development, as well as on compatibility and performance testing in preparation for the 6Gb/s SATA industry introduction, which remains on track for a 2009 launch.
What this essentially says is that the problem on the 88SE9123 chip didn’t lie with the SATA 6G implementation but rather with the legacy PATA support they built into the chip. Going back and looking at the spy shots we have for most of the P55 motherboards we have seen, they ALL do in fact have an IDE channel on them – and since the P55 chipset doesn’t offer support for it, we can assume that all of these motherboard vendors were using the Marvell chipset in that fashion. That would explain why all of the boards would need to be changed drastically at the last minute.
There is still NO explanation for the problem or why a technology that has been around longer than I have been alive is suddenly more difficult to do and test for than SATA 6G performance. It also doesn’t explain the answers we were hearing from motherboard vendors claiming things like “interference pattern issues” and “speed problems” with the SATA connections. It could be that the PATA functionality on the chip was the “root cause” of the problem as Marvell states above, but that the issue propogates throughout the controller and affects SATA speeds as well.
Either way, we now have confirmation that a new spin of the controller is required to fix the problem. What we don’t have is a time table for product re-introduction. All we know is that Marvell says that SATA 6G will reach our hands sometime in 2009.
Join the discussion in our forums about this topic.