Pogoplug 'BASIC' - Clarification on 'B01' Hardware

[dscambik]

A quick recap of something odd that happened to me last night and some accompanying quicknotes/comments/questions about this whole Pogoplug hardware transition:

Early last month, I caught the $39 sale on Buy.com for what I thought (assumed?) was the Pinkv2 (labeled 'Classic, Pink') and, though I knew the black Pro/'v3' with the OXNAS platform was the 'new thing' to keep an eye out for, $40 (really just $30, actually, after all my cash-back) just seemed like a steal for the legacy platform with its extra RAM, so I went ahead and ordered two of them. A week later, before I'd even picked those up from my mail service, bought a Pro on a Buy.com promo for $49 ($40 ACB) and started fooling around with it before I ever planned on opening the Pinks. After installing my eSATA cable in the Pro, testing the native Pogoplug functionality, and reading up a bit, I finally went ahead last night and completed a successful run of the OXNAS install on a flash drive and, seeing the ease of it all, took out my Pinks to give the ALA Marvel/ARMv5 install a try right out of the box.

Out of pure, unadulterated luck, I was watching a movie on another monitor and my ADD had me mindlessly opening up the device - honestly do not even know why (maybe subconsciously curious to see how the internals visually compared to those of the Pro?...) - but just as I realized the pointlessness of it and started reverting back, I gave a quick glance through the crack I'd pried open and caught a glimpse of the red SATA port. Finished prying it open, saw it was blantantly a v3 device, and checked out all of my labels/order records, confirming I'd in fact ordered two 'B01' devices, or the 'Pogoplug BASIC', as indicated on the bottom-of-the-box 'model number'. Saved myself a whole headache of doing the wrong install on what I was sure had to be older / 'Pink/v2' devices.

So I guess I'm kind of scratching my head now and wondering just wth is going on over at CE, and/or what to make of having two v3's instead of v2's:

1.) I've noticed that there's a lot of mention on the ALA site warning about the new pink versions of the OXNAS-based hardware, but for some reason, a lot of the references in terms of the product numbering are only "B03/B04" - why is that? Were there problems with the B01 & B02 versions that were reason enough for vendors to slash prices on them, even though they were built off the new platform? Are those B01/B02 editions of the 'Pink v3' basically hardware and firmware equivalents to the B03/B04 models, or am I missing something that's not covered in the warnings and my pink devices are effectively 'discontinued' models of the new v3 platform?

2.) Should I be happy or disappointed with getting new devices rather than the Kirkwood-based models with the additional 128MB of RAM? With the lackluster SATA throughput performance I've both read about and tested firsthand (topping out less than 10MB/s above max USB 2.0 transfer rates?) and equally puzzling lack of application performance boosts I've read about in other posts, it just seems like it'd make sense to straddle my hardware set between the new platform and the highly-developed Marvell version that translates so well with other devices and plug distros (but I suppose that also depends on the answer to question 3). I ask this only because the Grey (definitely v2 - E02G) models are $35 ($27 ACB) on Buy this week and I don't want to miss that opportunity before it's gone.

3.) Can somebody provide /point me in the direction of a bit more background on this greater transition away from ARMv5 to ARMv6 that CE is doing? Somebody mentioned in another thread that CE is "done with" Marvell entirely, so I'm wondering if that trend extends to the greater 'plug computing' world or if its specific to just CE and the Pogoplug line. Related to that, I don't understand why all of these new boards have SATA ports that are going completely unused by CE in the casing design - is keeping the port on the board just some ploy to bait people into voiding their warranties or something? I wouldn't ask that question were it not for the (suspicious) accompanying reduction in RAM and the noted effects it's been suggested that this has had on realizing a performance boost from the dual-core design without setting up a SATA-based swap partition.

4.) Does anybody have a handle on the future direction (either rumored or publicly announced) of CE product development in the wake of the Video recall? Is CE going to focus on a re-launch of the Video or just scrap it and continue to re-number the v3 products to give the illusion of progress until they manage to launch something with USB 3.0?

5.) Just a suggestion, but given the confusing pricing and accompanying 'Classic' misnomer, it seems the warnings about the discrepancy could be better-noted in both the install section as well as in the device pages with an image of the Pink device directly on the 'v3' page (seeing as how its just as relevant as the black one now, given that the 'Video' is out) and putting the word 'BASIC' in the slashed naming of the v3 platform, considering that that's the model reference on the UPC/Device sticker label on the bottom of the product box.

[WarheadsSE]

The B03/4 are just the first people to get them and come in moaning, but B0x is a better fit (B01/B02/B03/B04) and covers the entire line that appears to just be regional/firmware revision stuff.

As for the SATA port, I believe they left it because they use it to load the NAND at the factory, and the fact that it does not support hotplug, or multiplexing, it's better that they didn't put it in the outside of the case. I can't speak to the specific reason, as I am not associated with CE in any way, and they tend to deflect my questions.

[dscambik]

My last post was much longer than it needed to be, my apologies:

Main point was that the 'Classic' label attached to this model, at least as it's listed on Buy.com, along with the identical Pink coloring as was used on the E02 model, could be a bigger source of confusion than it needs to be for people trying their first ALA install and 'bricking' their 'B0x' device with the wrong installation script. After finally realizing that the B01/B02/B03/B04 series was labeled with a 'B' for 'Basic', it struck me as a pretty simple solution to refer to it as the the 'v3-Basic' site-wide to distinguish it from both the E02 as well as from other v3 versions better than just 'the Pink edition of the v3'. Trying to help ease the load on the lead developers / admins and make a suggestion for trimming the fat - that's all.

And the other main point was just to get any insight on potential new directions for plug development specifically from the commercial end of the market, e.g./i.e. over at Cloud Engines. Rumors, leaks, SWAGs... just on timeframes. If I go and get some half-height cards to mod my Pro's and then go through all the trouble of getting them to work the way I want them to, it would be quite poor if CE came out with new upgraded/ added-feature models resembling the same or similar functionlities 5 months from now and make it feel all for naught. Re: the transition to PLX from Marvell, re-read some of the smallcloudbuilder reviews and saw a note suggesting that despite the huge success of the Kirkwood platform, Marvell simply wasn't expressing enough intent on continuing to produce new ARM CPUs for CE to stick with them before launching their next product line, and thus the (apparently) best available explanation for the switch in the v3 Basics' hardware without material improvement in native specs & performance over the v2 - Pink v3's just for the sake of their pinkness, apparently...

Fully realized long-before that post that Warheads wasn't affiliated with CE nor with any upstream vendor - was just trying to open a convo with the broader group on what they thought would make sense going forward as far as what to expect, and would still welcome any thoughts.

[kmihelich]

I don't own any Pogoplugs, so I'm just contributing as a developer for the project who has watched as we try to shine some light into the black hole that has been created with the V3 devices.

I designed the new site you see, and because of these newer PLX versions popping up in the same colors and on mis-labeled sites, I explicitly added in not only the platform description but also the platform specs and the very first thing you see on the installation page that these are not Kirkwood plugs. Now, there is only so much you can do to try to fully inform everyone that visits the site. If you read around these forums, and if you were a patron of the old wiki site, you should have realized by now that no matter how much we say something, how much we bold it, make the text bigger, colorize it, make click-through "agreements" on disclaimers, there will always be someone that thinks they know more than we do and breezes through every important bit that we've tried to impart. For the sake of preserving a professional looking site, we chose to put days and weeks of work into hand-crafting each installation page to be simple, to the point, and cover everything in a step-by-step manner. If someone chooses to not follow these clear instructions, that's not our problem. If someone decides they don't want to read the intentionally separate warning to double-check their model number to ensure they're following the correct steps, that's not our problem. If someone chooses to follow another site's instructions, or a Google cached version of our instructions from the old wiki site, again: not our problem. We do our best to facilitate people, but it's not ours to insult the vast majority that are wholly capable of following instructions.

On to whatever decision Pogo made now. Marvell has had their Armada line of chips in the pipeline all the way back when companies were putting new devices on the market based on Kirkwood. Buffalo's recent Cloudstor is one of the first to market, and not all that far behind the new Pogo devices. I don't believe that they made any conscious choice to move away from Marvell because it didn't fit with their release schedule, and this is why: Their first device out the door based on PLX was the Video, and I'll quote one of the first gems you'll find looking up that chip, "It is the ideal solution for media enabled consumer storage devices and supports multiple HD quality media files, remote access, back-up, security (including AES-128/256 bit encryption and hashing), file sharing, and much more." They wanted to push video better, faster. The whole idea as I saw it was to do to your video collection what they did to your documents and pictures, but a Kirkwood wouldn't be handling that load. Then I thought, OK, they want to do video transcoding on the fly.. why did they pick the most obscure company to supply them a decoding card that no one could actually buy otherwise? What of CrystalHD cards that are the standard in powering playback in netbooks? To me, the whole thing felt like either an engineer was given too much leeway, or someone in power owned a too-large stake in that fledgling graphics company. What they are doing is very much their own as well, the other players are sticking with what works best: Marvell.

On the core processor itself, we already know it does better than a Kirkwood when both cores are fired up. We also know that two processors are better than one when it comes to multitasking and responsiveness. We also know that almost all applications operate in a single thread of execution, and that one core of the PLX is slower than a Kirkwood. So, you can do more at once, possibly faster, but probably not. Then there's the issue of kernel support. Kirkwood is solid in the kernel, and adding new boards is simplistic. The 7820 does not exist in the mainline kernel, in someone's private scm repo, or even has source that is accessible outside of an NDA or sifting through GPL-mandated source releases from manufacturers. The 7820 also follows what many ARM processor manufacturers do, which is create a brand new chipset, (apparently) arbitrarily choose a favorite kernel version, write in support to make that kernel work, then stop. Completely. This is also part of what has Linus enraged about the ARM community for the kernel. People keep trying to get brand new chipset frameworks pulled in to support their device because it doesn't play by any of the foundations laid by anyone else. Either that or no one is willing to do the legwork (fingerwork?) to make the code not be completely on its own.

Which brings me squarely to the kernel. We have said more than enough times on this site that if you have an issue with these specific plugs, it's likely going to lead back to the kernel in some way, which we simply aren't equipped to fix. The PLX-issued kernel at 2.6.31, which Pogo is more than happy to use with their years-old software versions for their firmware images, has just become old enough right now that all of our new stuff starts breaking on it. Arch Linux is a rolling release, they very much do not stay still, and as a derivative using the same sources, neither can we. If we fall behind a week in packages, it could take us another week to catch back up assuming we could hold up against everything new coming in that second week. We're doing what we can to keep the hardware usable, and WarheadsSE is alone trying to patch in support to a newer kernel so the devices' life support can be sustained for a while longer. By the way, none of us are kernel programmers.

Going forward, from my point of view, these devices make up a small portion of the ARMv5 base. There is also nothing that is going to "get better" for v5. The future is truly in ARMv7 territory, with Cortex-A8/A9 and Tegra2 chipsets. While the price point is a bit higher, you get nearly the same proportional processing increase that you get with the price increase.. and you stay at the exact same power requirements, if not better. So, more cost, but far more power, while using less electricity. Or another way: on-chip HD decoding, dual 1080p HDMI/DVI output, dual cores, six times as much RAM, less power consumption than a plug computer, and cheaper than the top-shelf Pogo offerings. Losing interest in Cloud Engines yet?

It's not a surprise that CE has shifted their hardware to a second-place standing against their new software offerings. Their hardware isn't exactly exciting. But I don't mean to degrade your purchase decision, I have two Dockstars and two GoFlex Nets running myself, and they exceed exceptionally in their duties of being mini-servers. But unless we can get the kernel improved, the situation does not look good going forward for the oxnas crowd.

[WarheadsSE]

*nods*