Fusion-io ioDrive 2 1.2TB Reference Page

[acquacow]

ASICs were developed and ready to go. It wasn't nvme, but basically the same difference, but optimized for our architecture. Had bios level support and the card worked without drivers at a basic level.

As for host CPU/memory usage, that's the fastest core and memory in the system. Beats a 600MHz asic any day back then. It also let us scale performance liberally from a single card to 100 cards. NVMe doesn't scale like that.

 

[Amphiseum]

Got a bit of a weird one for you @acquacow . Have you ever tried to get an ioDrive working with UWP Windows apps? Specifically, running those apps directly off of the ioDrive. UWP apps can see the drive fine for saving/loading files, but for whatever reason Windows just has a fit when you try to actually run the app itself off of it. It's mostly an issue if I want to use the drive for game storage for Xbox Game Pass stuff, which obviously wasn't the intended market for the drive when it was $10k+

Basically, I just wanna know if I'm overcomplicating this for myself and there's an obvious thing I'm doing wrong, or if it's probably just that the drivers were never tested in this scenario so I'm stuck waiting for someone at WD to randomly decide to update an old enterprise-only product to do something no enterprise client would ever care about. The drive works amazingly for everything else I want it to do otherwise.

It's a 3.2TB ioDrive 2, if that matters.

 

[acquacow]

I've never messed with any UWP apps that I know of... I install lots of other apps to ioDrives though just fine. For some apps that don't let me specify the install path, I've used junctions to move them over to the ioDrives w/o any issues.

 

[Starbomba]

I've been using my ioDrive 2 1.2 TB for a while now. Been totally liking it, however, as that one got into my personal PC (unlike the SX350 which got into my server), i've been having a bit of cooling issue on it. So far it works, and it stays at like 55c, but i had to jump a couple hoops which made that setup a bit cumbersome.

Seeing that drive had removable NAND modules, i got an ioFX (420 GB) for dirt cheap and thought i'd just swap the NAND modules, however, for some reason it did not work, and i couldn't make the ioFX work with the ioDrive NAND modules.

Second thought i had, was just to replace PCBs and just steal the fan off the ioFX, but after tearing the ioFX down, i noticed the PCBs are slightly different, and i'm not THAT knowledgeable to go just poking around, nor am i handy enough with a soldering iron to just solder the fan somewhere on the ioDrive PCB.

Is there some way i can make the ioFX recognize the 1.2 TB NAND chips, or is there a way to cross flash the ioFX into an ioDrive 2? Maybe something you have an idea about, @acquacow ?

 

[acquacow]

There probably are ways at the firmware level, but they are different products with different # of pci-e lanes/etc.

I'd just run your 1.2 and find a way in your case to get some airflow over the heatsink. Either a 60mm noctua, or a larger 80mm ~2000rpm fan that you can zip tie to something and use to get airflow to the card.

I used to use one of these exclusively for iodrive cooling in my desktop:

Amazon.com: Antec SpotCool System Cooler: Computers & Accessories

Buy Antec SpotCool System Cooler: Case Fans - Amazon.com ✓ FREE DELIVERY possible on eligible purchases

www.amazon.com

 

[waith]

Are there any 3rd party/enthusiast efforts to get the Fusion IO2 working in VMWare 7.0?

 

[gb00s]

... I'd just run your 1.2 and find a way in your case to get some airflow over the heatsink. Either a 60mm noctua, or a larger 80mm ~2000rpm fan that you can zip tie to something and use to get airflow to the card ...

Doesn't help these cards a lot to just fix a fan on it with a zip tie. The heatsink is 'closed' on one side and blocks airflow. I saw almost zero effect with a fan directly attached to it. Instead :

in 4U >> io-Drive2-cooler-4-U or 2U >> io-Drive2-cooler-2-U ... Temps approved by 8-12C down to ~38-40C

 

[acquacow]

io-Drive2-cooler-2-U ... Temps approved by 8-12C down to ~38-40C

This is the way I usually do it in my cases. I just put a fan in front of all the pci-e cards and blow some air over them.. Haven't had any issues. I do the same with that antec fan I linked as well.

 

[Oddworld]

What's the maximum safe temperature? I'm showing 52C idle and 65C under load. Do I need a fan?

 

[acquacow]

What's the maximum safe temperature? I'm showing 52C idle and 65C under load. Do I need a fan?

All of the ioDrive II cards use an industrial xilinx FPGA that is good for 100C.

I think we start logging info messages and throttling writes around 86C and finally park all data and take the device offline around 96C.

 

[Starbomba]

There probably are ways at the firmware level, but they are different products with different # of pci-e lanes/etc.

I'd just run your 1.2 and find a way in your case to get some airflow over the heatsink. Either a 60mm noctua, or a larger 80mm ~2000rpm fan that you can zip tie to something and use to get airflow to the card.

I used to use one of these exclusively for iodrive cooling in my desktop:

Amazon.com: Antec SpotCool System Cooler: Computers & Accessories

Buy Antec SpotCool System Cooler: Case Fans - Amazon.com ✓ FREE DELIVERY possible on eligible purchases

www.amazon.com

Yeah, i thought i could do that as the PCBs are like 95% similar. The only thing visually different between both is the supplementary power connector on the ioDrive, and the fan connector and LEDs on the ioFX.

So far, what i've done as i kinda don't need the ioFX (most of my storage needs are now above 400 GB, but i still have it in my parts cabinet), i took the ioFX shroud and fan, bolted them onto my ioDrive, then wired the fan to a normal 3-pin plug running on my motherboard. I was even able to downvolt it a little bit so it isn't as noisy (it is a 30-40mm fan spinning at 4k RPM), yet still averaging 60-65c when querying fio-status, which should be completely acceptable.

 

[gb00s]

... I was yet still averaging 60-65c when querying fio-status, which should be completely acceptable ...

I never experienced such temps even in max load. Did you enable the power overdrive and set it to persistent?

Edit: Maybe 365Gb and 785Gb are running cooler. But my 785Gb runs way cooler than all of my 365Gb drives.

 

[Patrick]

Since we lost this question from the backup restore earlier today:

So I just got 2x 785GB MLC iodrive2 half heights. I noticed there are 2 identical modules on each, with a plastic spacer in a third slot. These go up to 1.2TB and minimum 385GB, so the thought was to take a module off one and add it to the other, giving me a 1.2TB and a 365GB iodrive2.

Obviously once I removed a module, the card assumes it's died suddenly and runs in minimal mode, so I need to change the model in the firmware somehow? I tried looking around before asking, but I came to the conclusion that I'm the only madman trying this, and that this thread seems to be the only place that these cards are discussed. I've updated the firmware to the latest version via command prompt on a windows 10 desktop pc.

Any ideas? Thanks in advance.

 

[acquacow]

No easy way to convert a card from one config to another unfortunately. I think there's a config in part of the FPGA that doesn't get flashed on a firmware update. Also, the LEB map won't really work, you'd need one from a larger card...

It's on my list of things to hack up sometime.

 

[HERB]

Looks like my registration was perfectly timed to get evaporated with the backup fiasco, but here I am again.

No easy way to convert a card from one config to another unfortunately. I think there's a config in part of the FPGA that doesn't get flashed on a firmware update. Also, the LEB map won't really work, you'd need one from a larger card...

It's on my list of things to hack up sometime.

I see. It'll be on my list too then. There's some kind of port that I don't recognize right off the back of the fpga, maybe that's something. Reverse engineering isn't my strong suite yet.

 

[acquacow]

Looks like my registration was perfectly timed to get evaporated with the backup fiasco, but here I am again.

I see. It'll be on my list too then. There's some kind of port that I don't recognize right off the back of the fpga, maybe that's something. Reverse engineering isn't my strong suite yet.

That's the JTAG port. It's what you'd use with an FPGA programmer for the virtex FPGAs to read/write to them.

 

[HERB]

That's the JTAG port. It's what you'd use with an FPGA programmer for the virtex FPGAs to read/write to them.

That's what I thought. Then that'd be where you'd read and edit the ROM and grab LEB maps, provided you had the likely proprietary software to do so?

And it looks like this has been asked before, but has there been any progress on booting from the iodrive2? I've installed the uefi module to the card like mentioned earlier in this thread, but the motherboard uefi says something like [change support for boot device] no matter what boot options I change. Motherboard is a sabertooth 990FXA r2.0. I could try some other motherboards too but it sounds like the UEFI support is only half baked. It also adds a whole extra 20 seconds to boot times while the iodrive overlays it's command sequence over the regular boot animation.

I've got plenty of time and spare computers to try to make this work, but I'm missing the know how.

 

[kapone]

Got two of the SX350 3.2TB a few days ago and only started setting them up today. Woot! Almost no data written!

c:\Program Files\Common Files\VSL Utils>fio-status.exe -a

Found 1 VSL driver package:
   4.3.7 build 1205 Storport Driver: loaded

Found 1 ioMemory device in this system

Adapter: ioMono  (driver 4.3.7)
        ioMemory SX350-3200, Product Number:HJ7HN, SN:US0HJ7HN7605466R0021A00
        ioMemory Adapter Controller, PN:NVV4D
        Product UUID:7d2a787d-0ebd-55a8-b59c-356922870508
        PCIe Bus voltage: avg 12.20V
        PCIe Bus current: avg 0.63A
        PCIe Bus power: avg 7.65W
        PCIe Power limit threshold: 24.75W
        PCIe slot available power: unavailable
        Connected ioMemory modules:
          fct0: 02:00.0,        Product Number:HJ7HN, SN:US0HJ7HN7605466R0021A00

fct0    Detached
        ioMemory Adapter Controller, Product Number:HJ7HN, SN:1623D04BF
        ioMemory Adapter Controller, PN:NVV4D
        Microcode Versions: App:0.0.15.0
        Powerloss protection: protected
        PCI:02:00.0
        Vendor:1aed, Device:3002, Sub vendor:1028, Sub device:1fbf
        Firmware v8.9.9, rev 20200113 Public
        Geometry and capacity information not available.
        Format: not low-level formatted
        PCIe slot available power: 25.00W
        PCIe negotiated link: 8 lanes at 5.0 Gt/sec each, 4000.00 MBytes/sec total
        Internal temperature: 43.80 degC, max 50.69 degC
        Internal voltage: avg 1.01V, max 1.01V
        Aux voltage: avg 1.79V, max 1.80V
        Rated PBW: 11.00 PB, 100.00% remaining
        Lifetime data volumes:
           Physical bytes written: 1,900,544
           Physical bytes read   : 592,281,600
        RAM usage:
           Current: 34,240,000 bytes
           Peak   : 34,240,000 bytes

 

[HERB]

I mean the uefi.rom file is in the firmware file, and there are definitions in the INFO file that show how to apply it, but it was kinda experimental for some OEM workstation stuff. It was setup for our ioXtreme cards so that they could be bootable in workstations, but the FPGAs should all accept that rom.

When updating the cards, you can fio-update-iodrive --enable-uefi [...]

I've gotten this far trying to boot windows off an ioDrive2 785gb with uefi rom file installed. I can see the drive and its partitions and begin a windows 10 install from a usb, and everything works fine until it needs to restart to complete the install. When attempting to boot, I get the error

fioerr HP 785GB MLC PCIe ioDrive2 for ProLiant Servers (7,7): read failed - size
fioerr HP 785GB MLC PCIe ioDrive2 for ProLiant Servers (7,7): request read -57b1
_

Where do I go from here? Thanks in advance.

 

[Felipe Leonardo]

No, you can't boot from it w/o a TON of UEFI hackery that I haven't had time to look into.

I have a Mac Pro 2009 and I intend to keep on OSX 10.9 Mavericks for the long term. I study the possibility of installing an IoScale 3.2TB for storing video and photo footage, but I would love to get a bootable disk.

Currently, there is a very interesting project called OpenCore It simulates EFI loading before the system boot, offering the possibility to load drivers (kexts), such as video boot for graphics cards.

I think it would be the ideal environment for hack work that aquacow cited as necessary to be able to boot from Fusion disks. Any idea which path would be needed?