Final Progeny

I'm considering getting some additional ODDs to speed up transfer of data from discs to my PC. I haven't had any luck trying to find a 5.25" to 2x Slimline adapter (I'm in the UK by the way). Do these products even exist?

I'm looking to build a HTPC in the near future, but would like it to support playback of 4K video from the likes of Netflix and Amazon Prime. Note that any gaming done on it would be console-like, nothing serious. Are the AMD APUs capable of supporting this? There doesn't seem to be any recent information since 2018 regarding this. If not, what low budget dedicated GPU could I get instead? I am looking at an ITX form factor.

So when viewing photos in Windows Photos, about 1/2 a second after it has loaded, the image becomes a lot darker. See below the two images, one just before it completes loading, and one after. I'm running Windows 10 Build 1809 Version 17763.316, viewing .ARW photos While loading: After loading: Any ideas? Thanks

That is unfortunately what I'm already doing. Accessing the server via \\10.0.0.1 vs \\192.168.0.1 still shares the bandwidth between both adapters.

Ok, thanks

All in the title really, can a RX 580 run 3 4K monitors simultaneously? This is for video production, NOT gaming.

Yes, both PCs are using a Gigabit connection. The server is just running Windows 10 for network shares. The Work PC is also Windows 10.

Hi there, So I'm in a bit of a pickle with regards to my particular setup. I'm running two computers, one acting as my backup and storage server, and another I use for my work. My two PCs are connected as such: External Network (standard 192.168.0.x/24 address) (100Mbps) Work PC -Ethernet- Router -WiFi- Server The original Ethernet connection that was put in was only a 100Mb link, so when transferring 10's of GBs of data, it can take hours. To circumvent this, we put in a direct wired connection using Gigabit Ethernet. Internal Only Network (10.0.0.x/8 address) (1000Mbps) Work PC - Ethernet - Server So my problem is that, even though the Internal network is faster, when using network shares to the server, Windows seems to be aggregating the two links, which are both capable of reaching the Server, which means that while it is a speed increase over the maximum Megabit speeds, it is running about 45% the maximum speed of Gigabit. If I disable the External Network on the Work PC, the transfer speeds immediately shoot up to maximum Gigabit speeds, but this means that I lose internet connectivity, as I'm no longer connected to the router. The reason my Server is also connected to the router is due to the wifi, which allows me to access it from other devices in my home, as well as across the internet, so disabling the internet connection there is also not an option. I was under the impression that Windows would see the faster link, and just use that, especially as I explicitly state to use a UNC path of \\10.0.0.x to access the server, and not a hostname. Is there a way that I can ensure that traffic specifically directed to / from the 10.0.0.x network is not aggregated across the slower link? Do I need to mess with the route tables on both machines, or can I setup a firewall on the server to block traffic from a certain IP on a particular Physical Ethernet port? Any help appreciated, thanks.

I'm looking for a cheap graphics card that is capable of streaming 4k Netflix and the like. I'm not bothered about gaming performance, just that it can adequately handle streaming 4k content. I'm aware an AMD card would likely be cheapest, but can't find specifically which cards can support and are capable of what I require. Thanks for any help.

I had a Corsair CX550M = Coil whine. This is apparently common across the CX series. It wasn't good. The fan bearing also didn't last very long, so started making noises too (both of these occurred within a month or so of purchase). Corsair's solution was to turn the PSU the other way up so the fan was face up and not down. I decided to get a TX550M instead (Tier 2). The fan is the loudest thing in my system, (which is relatively quiet), but it's a much better quality overall. It is more money, but you won't have annoying / disconcerting noises. You should check out the PSU Tier List here.

Cheers for the insight, I remade the array using your suggested method, and updated my post up top.

Interesting. Now that you've said that, I found a lot more articles on RAID 10. Do you know of any good guides for setting this up? (I assume Powershell?)

As far as I'm aware, you can't stripe existing pools on Storage Spaces (at least on Windows 10,and not Windows Server). Unless there is some convoluted way or doing it.

I decided not to use FreeNAS for a few of reasons: I'm using Hyper-V VMs, which I migrated from VirtualBox, the reasoning being I wanted dynamic memory allocation, as well as auto-start and stop features, and secondly, I'm not really familiar with Linux at all, so if there were any kind of problems, I'd be stuck without a clue, and be at the mercy of the internet (although I know there are some great forums out there). Lastly, In the event of a failure, or hardware change, I know I can easily swap the disks between other Windows machines. Also, isn't FreeNAS also software RAID? The OS is built for it, but still, it's software controlled. Ultimately, I think they're much of a muchness, and whatever you are more comfortable with is the better choice. I have researched FreeNAS, because so many people do use it, but I decided against it.

**** EDIT **** See original post below. In light of what @Electronics Wizardy suggested, I deleted the RAID10 volume, and did the whole thing again but through Powershell, including the striping. This is still a RAID 10, so 2 two-way mirrors, but the -NumberOfColumns is set to 2 (a stripe per RAID 1 array). In what may be needless to say, performance is a lot better. Here are the numbers: CrystalDiskMark In this benchmark, we see scores which are roughly equal to the Storage Spaces + Disk Management Hybrid RAID10, but 4KiB Q8T8 and 4KiB Q32T1 did actually come out worse for Reads. Speed copying Hyper-V VM RAID10 (PS): 246 - 279MB/s AVG: 262.5MB/s This is significantly quicker than the hybrid solution I first used, with an increase on the average of 19%! As well as the speed bump, the fluctuations in speed are much less drastic. Read / Write test for Windows 10 ISO RAID10 (PS): 8.9s Write, 14.1s Read Again, much better than the hybrid solution, clocking around ~3 seconds faster for writes and ~2 seconds faster for reads, likely attributed to the decrease in speed fluctuations. Average Response Times RAID10 (PS): 476.3ms This is another big change. Gone from ~2 seconds of response time, down to around a quarter of it, even slightly faster than the per-disk response times. This again, is likely a contributing factor to the speed increases. To do this through PowerShell, I found this guide here. Note that instead of "Storage Spaces*" I had to use "Windows Storage*". To find out what you need to use, run the Get-StorageSubSystem command. Obviously I need to do some reading into some documentation for Storages Spaces. RTFM people. ------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Hi there! TL;DR: Software RAID 10 in Windows works about as expected, in my experience. Skip to the end for the RAID 10 numbers and results! So, I was looking, and couldn't find any information on people's experience, or results of any kind of RAID 10 adventure on Windows, so, decided I'd do it myself. This post is pretty much my test results, and performance. My setup for this is a PC running as a central storage unit for my home, so it's not running anything spectacular, just some parts I had lying around: Hardware PSU: Corsair CX550M CPU: AMD A10-6800K @ 4.1GHz RAM: 16GB 2466MHz Corsair Vengeance MOBO: Asus A88XM-Plus OS: Crucial M4 SSD 4x WD Red 2TB NAS 5400RPM drives (This is what I RAIDed) Software Edition: Windows 10 Pro N Version: 1803 Build: 17134.228 PREAMBLE So, as above, there seems to be no information on what a RAID 10 is like on Windows, especially a software RAID, which is what I did; So I thought I'd share my results and experience. Setting up the RAID 10 was quick and simple, just set up 2 two-may mirrors in Storage Spaces (RAID 1), and then stripe the two mirrors together using Disk Management (RAID 0) to make the RAID 10. Before setup, I ran some fairly basic testing and a CrystalDisk benchmark on all drives before starting the process, to get a baseline. REFERENCE TEST Just as a quick comparison, I also ran CrystalDiskMark on the OS SSD, so you can see the performance of that. I didn't run any other tests. STANDALONE DRIVES First, I ran CrystalDisk Mark version 6.0.1 x64 on all four drives: As you can see, they're all relatively comparable, with Drive 2 doing better overall, and Drive 4 performing worst. Next I transferred a Hyper-V VM file to each of the disks. I used this test to get a real-world example of write speeds. I found the following: Drive 1: 132 - 147MB/s AVG: 139.5 Drive 2: 141 - 154MB/s AVG: 147.5 Drive 3: 137 - 145MB/s AVG: 141 Drive 4: 122 - 144MB/s AVG: 133 As you can see, the max speeds roughly match the max speeds from the CrystalDisk Mark test, so looking good so far! Next, using a Windows 10 ISO file (3.46GB) on the OS SSD, I copied the file 3 times, to, and then from each drive, to get a rough estimation of the time taken to read and write files. Each test was performed 3 times, and the below value is an average for each: Drive 1: 15.2s Write, 13.7s Read Drive 2: 14.4s Write, 13.2s Read Drive 3: 14.8s Write, 18.8s Read Drive 4: 16.2s Write, 13.8s Read Finally, again using the Windows 10 ISO file, I looked at the average response time in Task Manager when copying the file across, again 3 times, with the average result below: Drive 1: 488.3ms Drive 2: 484.0ms Drive 3: 490.3ms Drive 4: 487.3ms MIRRORED DRIVES Next up, I mirrored the four drives into 2 two-way mirrors (Drive 1 & 2, and Drive 3 & 4), and ran all the same tests, so i'll be lighter this time: CrystalDiskMark Mirror 1 seems to have the high and low values of the two drives for the Sequential test. Mirror 2 seems to follow this pattern. For both mirrors 4KiB Q8T8 test, 4KiB Q32T1 test seemed improved over the single drive performance numbers. For 4KiB Q1T1, only the Write seemed marginally improved, while the Read suffered in both cases. Speed copying Hyper-V VM Mirror 1: 131 - 144MB/s AVG: 137.5 (Drive 1 & 2 speeds from earlier average = 143.5) Mirror 2: 121 - 140MB/s AVG: 130.5 (Drive 3 & 4 speeds from earlier average = 137) Mirror 1 performs better, and Mirror 2 lags behind a bit in performance, but it does have the slower drive in it's mirror. Read / Write test for Windows 10 ISO Mirror 1: 15.9s Write, 13.4s Read Mirror 2: 16.1s Write, 14.8s Read So again, not far off the scores from the original standalone tests. Average Response Times Mirror 1: 488.0ms Mirror 2: 498.3ms Again, a comparable score. Overall, my experience from using Storage spaces to mirror the drives is that there is very little overhead, and the mirror performs in a comparable fashion to the disks it is made up of. SOFTWARE RAID 10 So we're finally here! The reasoning for all the previous test was to establish a baseline in performance, from disk to disk, to mirror to mirror, to see what kind of impact putting the two mirrors in a striped array would have. CrystalDiskMark In Sequential tests, we're seeing about the coupled performance of Mirror 1 and Mirror 2 from the previous set of results, great! For 4KiB Q8T8 and 4KiB Q32T1 Reads, were seeing slightly less than double the performance, but it's still and increase, but Writes seem to cope better here. 4KiB Q1T1 had the worst performance for Reads, barely increasing from the speed in either of the mirrors, but for Writes, not far off double again. Speed copying Hyper-V VM RAID10: 170 - 270MB/s AVG: 220 Quite a large fluctuation in this test,but, compared to all other results, even the lower speed is higher than the maximum speed achieved, and the maximum is not quite double that of the highest score. However, 220MB/s is still 72.5MB/s more than the fastest average speed previously seen. Read / Write test for Windows 10 ISO RAID10: 11.7s Write, 16.0s Read By far the fastest write speed we've seen so far, so the RAID 10 is coming into it's own. Obviously it's not half, but, faster is faster right? Compared to the fastest score of 14.4s from a standalone disk, this performs 23% faster. However, read speeds were nothing to write home about. Maybe this will be explained next... Average Response Times RAID10: 2411.7ms Now this is interesting. I'm in no way an expert, but this response time is MASSIVE in comparison to the ~500ms Max we were seeing beforehand. Maybe this is a consequence of the striping, or in particular because of the RAID 10. Perhaps this is contributing to the slow Read speeds observed in the previous result. However, our speeds aren't too bad, so maybe it's nothing to be concerned about, so long as you're only using it as storage, and not using it to run anything off of! OVERALL Overall, I think it's safe to say that, if you're not concerned about super speed (I mean, you're likely using slow HDDs!), but want a RAID which gives you redundancy, as well as a slight speed increase, then this is a viable option for you, which doesn't involve any tweaking or hacking, messing around in the UEFI (Or BIOS if it's that old!) (also, FAKE RAID!, seems to be a popular term), or involve buying an expensive Hardware card (which do have their benefits, but not so much on the cheap end). Perhaps running this with a newer CPU and chipset which isn't 5+ years old from an era where AMD were lagging behind in performance, maybe our numbers would have been better. But for my purposes, it works, and I'm happy with it. SOFTWARE RAID 10 RESILIENCY Now this last part is an added bonus, because I wanted to share my experience, and I hope that people will see this post and it be useful, as I'm trying to fill a gap in the internet with regards to this specific topic! I knew how RAID 10 worked, in theory, but I wanted to test it out. So, to that end, I put my Hyper-V VM on the RAID 10 array, shutdown the PC, and pulled Drive 1. Upon restart, the Logical volume was still there in Windows Explorer, and the VM on it ran just fine, no corruption! However, there are NO prompts by Storage Spaces that a drive has failed or is missing, so be aware of that. I took a look at Storage Spaces, and it did recognise the missing drive, and warned of decreased resiliency. Next, I shut it down again, and pulled Drive 3. This means I have pulled one drive from both parts of the RAID 1 array, leaving a single disk in each, which are RAIDed together, bascially running in RAID 0 now. Again, the drive showed up fine, and the VM ran fine. Next, I kind of knew what to expect, but I pulled Drive 4, and plugged Drive 1 back in, meaning I had 1 Mirror, but missing the 2nd Mirror. This is where it was borked. The drive failed to show in Windows Explorer. So i checked Disk Management, and it showed the Mirror as being missing, and the Dynamic disks as being offline. Now, I didn't go through the trouble of reformatting the unplugged drives, and populating them back in after this, so this next bit IS NOT representative, but I plugged Drive 3 and 4 back in, and started up, and the RAID 10 drive was back. As far as I'm aware, in case of a real drive failure, you would need to "remove" the failed drive from Storage Spaces first, then replace the physical drive, then add it back to the pool of the Mirror. I can't confirm this, as I did not test it. FINAL THOUGHTS I hope this is useful to someone, and I hope someone else can fill me in on their experience of Software RAID 10 in Windows. Cheers!