Jump to content

MSMSMSM

Member
  • Posts

    54
  • Joined

  • Last visited

Awards

Profile Information

  • Location
    Flavourtown, Indiana

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

  1. I have a very absurd problem. I decided to switch from BOINC to F@H for a while for my RX480 and while it seemed things were chugging along, I checked the "Advanced Control" to find that my GPU folding slot has been marked as failed. Here's the system info from FAHControl (Note: F@H reports my install as Windows 10 Enterprise, it's actually Windows 11 Pro for Workstations 22H2) And there's nothing in the logs elaborating on the failure I decided to check my drivers to see perhaps an update had dropped but oddly enough, there doesn't seem to be a PRO update since Oct '22 Oddest of all, while doing some troubleshooting steps as suggested by the internet, I learned that my OpenCL drivers are completely unsigned... they're the only things that are unsigned The simple solution would be just to switch to the regular, Adrenalin drivers and call it a day but I've got two questions. I use PRO drivers for stability and that's worked out so far, BOINC (PrimeGrid) had no problems doing number crunching using the GPU so why does F@H have a problem (if it has a problem with that at all) Why are there unsigned OpenCL drivers on my computer? How'd they get there? I've never in the history of the system install allowed for the installation of unsigned drivers (i.e. I've kept driver signature enforcement enabled). I have no hardware that needs unsigned drivers.
  2. Spy Kids for the family and Everything Everywhere All At Once for couples' night
  3. Are there any reasons why not to use APC? I use APC because it was affordable, communicated over USB to the computer and had Linux support. Eaton was my first choice but... buying them requires me to go to distributors who take the utter piss, who themselves are available only on classified sites who, the moment you give them your phone number, expect the spam calls and scam texts to moon like some meme crypto that Elon shouts out when he needs to dump some coins. For a RAID5/RAID6, it's suicidal :3. The Arch Linux wiki is your friend... https://wiki.archlinux.org/title/btrfs. You'll need to go out of your way to do what Synology does and setup LVM2 and then layer your filesystem on top of that, effectively making RAID handled by the LVM2 instead of Btrfs, where the problems lie. I'd say that TrueNAS's sane defaults are exactly that, sane defaults, unless you're using it in active workloads where every last bit of optimization matters, I'd wager they're fine as they are. I don't see the problem with it, EXT4 doesn't do volume management and RAID anyways, so the responsibility has to be picked up by something a layer below. Makes the lack of integrity/consistency checking a bit smaller pill to swallow as you'd have to use LVM2 anyways (which does have data integrity enforcement, though you do need to manually enable it). But as someone who went that route out of morbid curiosity, do yourself a favour, use ZFS.
  4. Thanks for letting me know that, I'll keep that in mind when considering if I ever do business with Synology again. On one hand, I understand, they're a consumer-facing brand and they want to go the "with the Synology NAS buy the Synology drives and Synology caching SSDs and maybe if you're a baller, the Synology network card and the Synology GPU" which is supposed to make the customer feel at ease and remove the anxiety that purchasers who may have the cash but not the technical knowledge have when they need to source their own parts (and also make some dollar, which I'm not against as long as you do it ethically) But to someone like me, this feels very "razor and blades"y except the blades don't even subsidize the razor, and not like something I expect something from a brand that I give stacks of G's to. I want a QVL and I want you to give it the same treatment as you give your own drives. I trust Seagate to be there for me when the drives fail within their three year period, can Synology say the same? Probably not. I'm sure for folks across the globe, you can do a s/Seagate/BrandNameHere/g and it'll remain true. Disappointing but I guess I should've seen this coming. They did move some of their surveillance software from being license-by-hardware to license-by-subscription. Here's to hoping my array doesn't suffer from the same fate as yours. Then I guess the RAID didn't do its job in maintaining uptime, I'm sorry. I guess having a backup pays off --- Small question, do you use a UPS? You seem to be well researched and probably do but I'm still curious what kinda UPS do you use, if you have one.
  5. Synology is unfortunately known to do some really stupid shit, like making the disabling "advanced data integrity" by default (Seriously? What the fuck?) And, IIRC, the default filesystem is EXT4, not Btrfs (which is more in the league of something like ZFS than say, NTFS). Yes, Btrfs has a write-hole bug for RAID5/6 configurations but Synology sets up the Btrfs volume over an LVM2 array, bypassing the problem altogether, though also, limiting the full extent to which Btrfs can be leveraged. Synology doesn't sell their own drives, do you mean drives from their QVL? Because I have some Seagate IronWolf NAS drives and they work beautifully on my DS920+ --- The first comment is right, RAID is not a backup, only a means of uptime and it seems like you're treating RAID like that, so by that metric, the drive going read-only seems very much in line with what RAID is supposed to do, stay online. --- TrueNAS CORE is the successor to FreeNAS and TrueNAS SCALE is the Linux-based ZFS distribution from the creators of TrueNAS. If you want something with the ability of Diskstation OS (or just want to run Docker containers), then SCALE is the go-to pick. If you want it to remain dumb but reliable (not to imply SCALE is unreliable, it's just not something ixSystems recommends for enterprise deployments), then TrueNAS CORE is the choice you can make. Though, like unRAID, TrueNAS will demand a larger chassis (I'd sell you by Jonsbo N1 if I could) and if you're concerned about USB drives, use an internal header, get one of these things from Amazon (see below), plug your USB device and tuck it in (if you're using unRAID, TrueNAS cautions against hosting the OS on a flash drive) Also, ZFS is a memory-hog, so make sure you have a good budget cause you'll need a decent amount of memory.
  6. Saw the video upload and immediately came here, I've started to notice more Cloudflare captchas and longer verification before I can even use the site. What's going on?
  7. Sweet! Can you also tell me the all the companies that'll go kaput so I can short them, all the companies that'll boom so I can long them, if society will be further polarized so I can decide if I should simply check out of it all and will he ever come back so I have even more reasons to toss and turn before sleeping due to exhaustion? Speculating on concrete prices for products without any leaks to go off, is as magic-crystal-ball-ing as you can go. There are general principles like price deprecation, newer products making the prices of older products slashed proportional to the price efficiency and parity (or improvement) in performance of the replacements which should be kept in mind when making a purchase decision. So I can't give you a specific dollar value but I can give you Backblaze's predictions for cost/GB, which seem to be falling. Though, a cost/GB (which we'll call c/GB now) number doesn't mean it'll apply uniformly throughout every manufacturer or translate into savings from every SKU (don't expect the c/GB of a 4TB drive to match a 2 TB drive regardless of technological advancement as there's a base cost regardless of capacity that is to be factored in). Like the first post said, they're not going anywhere. They're still the dominant means for readily accessible large capacity non-volatile storage. Tape drives can store even large capacities at a lower c/GB at the expense of accessibility (high access latency, practically non-existent random IO), making them suitable for archival but not editing files off. Just as hard drives are alright for storing games that aren't random IO intensive, large and/or not accessed frequently, storing your Downloads but not suitable for storing your OS, which benefits from lower latency and significantly better random IO performance. They all have their niche, it's just a matter of what you need.
  8. There are a variety of factors which imo may have resulted in reduced interest over time Crypto. At least some of the contributors have invested significant computational resources into these projects because of projects like GridCoin (which have long usernames) and CureCoin (which require team membership) Source: https://gridcoin.us/guides/foldingathome.htm Source: https://curecoin.net/knowledge-base/folding-for-curecoin/how-do-i-start-folding-for-curecoin-quick/ These projects are no longer what they used to be and the market speaks for itself (CoinMarketCap price graphs for CureCoin and GridCoin) not to mention the issues with GridCoin that were reported (I don't know if they were truly resolved, I'm not too intimately familiar with either codebases) I suspect some of the more profiteering members came and went. Rising cost of living. An increased cost of consumer electronics (esp. "competitive" hardware) and geopolitical flux mean that, you have the same number on your cheque even though your currency value has sunken (or even gone kaput entirely if you're really unfortunate to be in certain parts of the globe), while costs for everything else go up. Not as much disposable income, not as much to put into hobbies (though this could be utter nonsense, I mean, have you seen what some mechanical keyboard sets cost?). This makes small and medium contributors less likely to get involved as they upgrade to more budget options as it just isn't feasible to. Current international conflict and its effects on energy prices are just the most recent reason why people wouldn't find the idea of keeping their computer on for more than they have to kosher. Competition. The points and ranking system is supposed to make people compete and feel good about the contributions they make but... it can also have the opposite effect, discouraging people from bothering because, they feel like they're just putting a drop in the ocean. It took me months, if not a year of my laptop to reach 30M points in F@H (my desktop would've taken years). Two weeks with an RTX4070? Now I'm at 60M. That card isn't cheap and more importantly, it isn't mine. It is part of a build that I'm doing as a gift and I had it folding after I ran out of synthetic stress tests to run until it was the day to go and deliver it to them (sans any BOINC, F@H or other distributed computer software on it). I will let them know I folded on it. Stuff like that can make your 50k from 12h of CPU contribution feel like a fraction of a drop in the ocean, which is demoralizing. The folks at BOINC have started to realize it and are working on Science United (see their UI/UX goals) Making this stuff actually work. F@H's client is based on Python 2.x, which has been long deprecated. RHEL 9 doesn't come with Python 2 and community repos don't have it either. I tried installing it through various means but eventually found this Snapcraft package which works beautifully with NVIDIA CUDA but has problems with AMD ROCm OpenCL. Three days I've been trying to get this stuff to work to little avail. But that's just it working, the UI/UX is primitive, calling it an advanced control doesn't excuse not even giving it even minor scrutiny from someone who does UI/UX design, the web UI for F@H more than makes up for it (also big ups this extension that makes it even better) but the beginner mode for BOINC is so atrocious that I run to the advanced mode, which I don't expect Joe Joneson Average to figure out. If it's not plug and play, not unobtrusive, most people won't bother. For example, most people are used to OAuth based sign-ins (even though I hate it and never use a site that doesn't also offer the traditional email+password login method). Where's that? Normalization The number of people who sign up don't equal to the number of daily active users on the platform. Maybe we are normalizing and should re-calibrate how much engagement is too little engagement.
  9. I mean, if installed correctly such that each dual channel pair contains the same make and model of RAM, I don't see the problem.
  10. Also, here, the brand isn't just the same stuff with a different coat of paint, NAS drives have qualities that are omitted from consumer drives, better anti-vibration characteristics (usually with a rating of up to how many drives worth of vibrations it can tolerate), higher MTBF (mean time between failure) and more robust after sales service. Of course, if you always have doubt, consult the datasheet for that Seagate IronWolf, WD Red you're eyeing up
  11. Aside from the other comments urging you to tame your expectations about storage device lifespan (which I would urge you take heed too), here's a few things you can check out Check out Backblaze's posts about the longevity of the drives they deploy in their datacenter, here's the latest yearly report Read the datasheets about your drives and make sure they're CMR, not SMR, SMR drives are not advisable for NAS usage (e.g. Seagate's webpage clarifying their product line). Understand the 3-2-1 rule, 3 copies on 2 different mediums with 1 off-site (check out Seagate's writing on that) Research about the after-sales of the hard drive manufacturers available, this is very location-dependent as some brands may have better networks and contractors in your local area than others, other brands may merely sell but offer no after-sales in your country (often because they're imported). Do some reading about RAID and various RAID configurations (RAID1 = mirroring, RAID10 = mirroring + striping, RAID6 = parity), Synology has a table that more or less explains what is offered by their product line. Understand that having solid reliable drives that outlast their peers are the exception, not the norm and that eventually, all drives fail, some sooner than others. That's why we use RAID to keep uptime and alongside that, have a backup. Because RAID is not a backup, it's a means of keeping uptime.
  12. I did some reading and it appears that their kinda-portable edition, named Chamelon has been discontinued, I was under the impression that PortableApps have it but they don't, I seem to have confused their offering of Spybot - Search & Destroy with Malwarebytes, sorry about that. If you're running Windows XP (just bare, no updates, no service packs) and have forwarded every port to be exposed to the internet, then sure, malware hunting for new targets will probably catch your machine during a routine search of the IPv4 address range (which... isn't as infeasible as it used to be) But you're probably not, you're probably using a not-EoL not-esoteric operating system and don't expose ports over the internet (I hope I'm not jinxing it for someone), which means malware doesn't just walk in through the front door, they need an egress point (even a printer exposed to the internet is good enough to enter a network of otherwise unexposed machines, for example). Granted, some machines are easier to pwn than others but they're not that easy. You still need to download and run something, plug in an infected flash drive and run something (there's a reason we don't do autorun anymore), run that totally legit premium video downloader from the hentaisex.uwu ad or if you're journalist reporting in the Middle East, Asia or frankly anywhere at this point, receive carefully crafted non-executable files that exploit bugs in the logic used to parse and render that format to get your foot in the door*. There's a reason browsers need so many processes and take up so many resources, it's for isolation, IPC and sandboxing. You no longer can pwn a system with just one bug, you need an exploit chain, from sandbox escape to privilege escalation, to actually pwn machines. We've come a long way in security (but so have malware authors, so no slacking, m8). You need to convince the user to actually open that executable, showing them a carefully crafted webpage isn't enough anymore. Getting your foot in the door is the hardest part, privilege escalation second**, persistence third***. * - these more complex exploit chains are reserved for mobile operating systems like i(Pad)OS and Android, which have even stricter security policies and aren't as common, though not uncommon, on desktop platforms like macOS, Windows and Linux. ** - unless you're on Windows, privilege escalation bugs seem to be plenty *** - uhh... about that, https://www.theregister.com/2023/03/01/blacklotus_malware_eset/
  13. Interesting choice of data point aside, it's because of a variety of reasons: ISPs can be engaging in traffic shaping and prioritizing speed tests (or throttling P2P connections) The speed test server can be geographically proximate to your current location (versus your P2P peer) As a general rule of thumb, speed is determined by the slowest link in the chain (i.e. the bottleneck), if your ISP-to-local machine link doesn't have the bottleneck, then it can be limited by elsewhere in the chain Also, make sure you're comparing equivalent units, MB/s =/= Mb/s, 1 MB (megabyte) = 8 Mb (megabit)
  14. Why are you using Clover? Are you attempting to build a Hackintosh? If so, why are you attempting to boot the Windows installer through Clover?
×