I’m a Windows and a MAC User. It really chaps my knickers that I went out and bought a brand new Promise R4/R6 RAID Array only to find out I can’t use it with Bootcamp. (Having an SSD in my Macbook Pro is great, but space is at a premium).
Apple doesn’t want to help, nor does promise regarding getting it working on Bootcamp. Yes, you can kind-of use Parallels, but the reality is you’re going to get completely horrible speed. This really isn’t a solution anyways, because it just adds tons of overhead to a system.
So I decided I’d start noodling around and figure out the problem. So here are the known factors:
Thunderbolt will work with Thunderbolt devices on the MAC and sometimes on the PC (depending on the device).
The Pegasys R4/R6 WILL WORK on Bootcamp (with a bit of trickery), because Apple doesn’t want it to work with Bootcamp. Just another way to move Windows users over to MAC OSX completely, although there are some of us that need the R4/R6 and it’s bandwidth capabilities over Thunderbolt on Windows (me included).
A Thunderbolt device will work on Bootcamp as long as it’s connected during startup, but daisy-chaining (is that even a word!?!!) to the primary device will allow hot plug Thunderbolt devices to work on secondary+ devices (if supported by Bootcamp).
I’m a smart duckie (at least that’s what I’m told).
First things first.. The evaluation or the problem. Plugging a Pegasus into the Thunderbolt port and running apple updates (both apple and bootcamp side) gives us tons of updates (Mostly 2 EFI updates and the Thunderbolt drivers).
I booted up into bootcamp only to find out that device manager shows there’s a RAID controller made by Promise without the drivers. So I get to thinking.. What other RAID controllers out there use the same controller chip? PMC-Sierra’s PM8011. Well, seems like there are quite a few and after mucking around with all the big brand name drivers, none would install. Now obviously, this is a problem.
Knowing what I know from the Linux world, drivers detect chipsets and that’s basically it. So I figured I would do some hacking on my own. After 2 days of trial and error, I come across a press release stating that PMC and Promise are building controllers that use this exact chip! Woohoo! That was the first major step in the right direction.
I figured out that the SuperTrak models use the PM8011 and simply downloaded the windows drivers. You’d think it would be a straight-forward install right? Wrong. It didn’t work. So poking around in the inf files, I realize that the PCI vendor and device IDs are missing for the RAID controller.
I simply modified the .inf file in the drivers for the SuperTrak and added a line as follows under the [Promise.ntamd64] location and did a driver update from device manager and what do you know! It works!
This is the RAID controller PCIVendor ID for Promise: 105A and the Device: 8760 the windows Bootcamp version of the Raid controller.
I saw the controller and the Array pop-up in my device manager and immediately went to Disk Manager to see if it appeared. Well, guess what? It does! Now, we’re not done.
You will need to boot into the MAC side of things and run the Pegasus R6 configuration utility to configure the array, but that’s not a big deal. I set mine to RAID-5.
I decided I wanted to do something special and have access on both the MAC and Windows side, so I erased the disk and made it one big honkin HFS+ partition.
I went to Paragon Software and purchased HFS+ for windows + NTFS for mac and installed it. Just because I like access to all my files without rebooting into one OS or another. This is optional as you can partition all you want with the Disk Administrator in Windows.
Create Logical Drives, Format Them, do whatever you want. Next comes the speed test. Could I really achieve Thunderbolt like speeds from Bootcamp with my little hack? Only a test will tell. So I did the following:
Copied a 1.6GB RAR file from and to the array. Avg speed? 311 MB/s. Did this three times and to and from an SSD drive, I’d say It did pretty well. I’m running a 512 GB Crucial CT512M4SSD2CCA M4 2.5″ 512GB Solid State Drive rated at
|Sequential Access – Read: Up to 415 MB/s (SATA 6Gb/s)|
|Sequential Access – Write: Up to 260 MB/s (SATA 6Gb/s)|
All the extra crap I’ve been running in the background makes it come pretty damn close to the limits on not Thunderbolt, but the SSD. So I’d say I have some pretty good numbers. The true test though is to test whether or not I’m going to get true Thunderbolt numbers that are a little more scientific.
So I installed a 2GB RAMdrive using DATARAM RAMDISK free version and did some performance testing on the ramdisk using Performance Test 7.0 Evaluation version. Here are the results. Please keep in mind my machine was transferring data from my expresscard 3.0 WD Mybook 2TB Drive to the Pegasys Array at the time, but still, the numbers are pretty impressive:
RamDisk: 2GB File 2GB Disk
Now I decided to transfer the same 1.6GB file from the RAMDISK to the R6 Array while doing the USB 3.0 Transfer from expresscard / 34 adapter moving at 97.4 MB. I was a little surprised when I pulled 1.09 GB/s coming from RAM to the Pegasys and on the way back, I was pulling around 900 MB/s.
So I’d have to say the Thunderbolt Hack was a success… Well, there’s one more test to complete, but I’ll have to get another SSD and stripe it in my macbook so I can truly see how this performs.
If you find this post important and extremely useful, please donate. It helps support the site.
Thank You Friends.
My next project will be to get the ATI RADEON stuff (APP / Stream) working on the Macbook Pro.
UPDATE: I was going to buy a second SSD to Stripe within the Macbook pro to test throughput (and because I wanted more space), but apparently, after doing some research, it turns out Apple Macbook Pros (Early 2011) have an issue with SATA III in the optical drive bay. They don’t function reliably at SATA III, but function at SATA II. Kind of a problem for a RAID-0 Stripe. One SATA III + ONE SATA II = DISASTER and lots of lost time and data. So, I ordered a 16GB RAM Kit to get myself some more performance. (and a larger RAM Drive) for testing to / from the R4/R6.
After a few days of testing with the R6 on Windows 7, I realized that running HFS on it was NOT a good Idea. So I just reformatted the partition using Disk Manager as an NTFS partition and all is well. No data corruption, lightning fast speeds, and lots of safety for my data.
Note: This may sound kind of funny, but now that I’ve got the bandwidth to do some serious data transfer at extremely high speeds, I’m annoyed that my SSD is the bottleneck in the equation. A full 400 GB Bootcamp Partition via Acronis takes 56 minutes. (due to the compression overhead).
**UPDATE: So It’s been about a week or two since I’ve had my Promise R6 running in Bootcamp natively and I have to say it’s been very stable. I’ve maxed the RAM on the machine out to 16 GB and added a 4GB RAM Drive amongst other things. [see my other posts] I managed to eek out another ~15-20% in performance, but I’m still annoyed with the total speed of the whole machine. I was going to stripe a pair of SSDs until research revealed that the secondary SATA port isn’t stable at 6Gb/s. So that idea went out the window. I did, however, move my pagefile configuration around and put a minimal pagefile on the C drive while using the Promise and RAM Disk to help speed things up. I modified my browsers to use RAM instead of disk for caching. It’s definitely brought the machine up to a new level of peppiness.
If you’re really having problems and need a step by step tutorial with screenshots, I have written an ebook on this.
Step by Step Installation Instructions $10.00