G4te_Keep3r

new system now or is it a bad time to upgrade with the new CPUs coming out soon that have a better upgrade paths and/or will make current prices fall. I was considering getting a used 30** or 40** (iirc from last week when was considering that, it was between 3080ti/3090 or 3080/4070 as the 2 best performance options at 2 price points (~$1k, ~$500)). If the $999 (new) 4080 super cards come back in stock, that would come into question too (4080super for $1k new vs 3080ti/3090 for ~$800 to ~$900 used vs 3080/4070 for ~$500 used). But then started thinking it might be time to go ahead and upgrade my whole system. My current psu is an rm1000, so that might remove the 3090 as iirc with the power spike issues it required a 1200 or 1600 psu...unless that was just early drivers or something. Budget (including currency): $1500 to ~3000, could probably push it to the 4000/4500 range (usd). Was also looking at upgrading my server ups, but that craigslist listing sold (~$3500 used but ~$8k new)...which keeps me at ~30-45 min server battery time but less limiting on desktop funds. Country: usa Games, programs or workloads that it will be used for: Main game is destiny2 and right now less time is going into it (just give us a bigger vault...even as a paid option), but use cases range from running virtual machine(s) locally, compiling software (sometimes in the VMs), CAD (probably simulations too for an upcoming project), programming/development (including GPU accelerated), video editing, video recording (sometimes while streaming), a handful of planned projects using neural nets and/or LLMs, some other gpu accelerated projects that i have barely started the research into, plus normal daily use stuff (chrome/firefox, excel, email, ...) Other details Current main desktop is i7-8086k & 1080ti (w10). There are times that the comparative lack of power is really felt, and sometimes i do run really low on cpu and/or ram especially when also running VM(s) (when possibly i run them on my server). The ram is maxed out at 64g (per cpu & mobo max) with normal lighter use hovering between 20g and 40g and medium to heavy use leading to paging (good thing it pages to ssd and not hdd). THE CURRENT GPUs STORY/ PROBLEMS In 2018 i bought 2 1080ti ftw3 off ebay (1 in summer 1 in fall). Generally they worked great (rarely sli as i would assign different workloads to different cards). Also bought a pny 1080ti blower in 2019 for my server but thats a whole different problem (still mad i didn't sell 2 of these cards when they were going for ~1k+ each). Occasionally there was this issue where you could see a sharp spike & return to previous value on gpu in task manager, which would frequently crash explorer (and sometimes the chain reaction of freeze or crash that chrome and/or firefox would do and sometimes obs & putty & spotify & sometimes a few others). With sli bridge the spike was seen on both cards, but iirc without bridge with both cards in pc the spike was still shown on both graphs. Over time the frequency got worse and i figured out which card did it much more often and took it out. And then good for a while (3ish years) with it just happening on occasion. But it has gotten more frequent and has accelerated recently to the point of sometimes only an hour or 2 between spikes, at least every day or 2...and instead of sometimes causing issues it now almost always causes issues. Current issues are explorer crashes basically every time, most the time chrome crash (more of freeze and go unresponsive or white windows and unresponsive with rare hard crash/close), obs almost always becomes a nonreactive nonresponsive window (not unresponsive as can't interact with it to become unresponsive), putty half the time doing similar to obs. These are the spikes i know about by witnessing it or seeing the aftermath...as sometimes it spikes and nothing was impacted (how most of the spikes were early on and rare type now). Sticking with my keyboard/mouse/headset/stream deck. Current monitors are 3440x1440, 1920x1080, and 2 x 1680x1050 (both diff diag inches and one runs at 59Hz). Eventually will replace the top 1680x1050 with another ultrawide, but that will probably require a new monitor arm setup to to be engineered later. Probably would stick to 1440p, but eventually might go 4k. I would love to have remote management on the mobo and ability to pass a gpu or other pcie device to a vm (in w10 host mainly, but with more powerful system and pass thru, will switch to linux host and some things in win vm...pending compatibility). I know those are generally only on server/workstation board and have a higher premium that isn't really worth a large premium until a possible future of my desktop migrating to my server rack (after i can move from apartment to house and get proper electrical wiring run). Would be amazing, but not required (pcie pass thru is the big item, management port just a nice to have in this use). When started writing this, the list showed ~$3100, and has since gone up to ~$3500 because multiple sellers have changed on what ppp shows (gpu alone went from 999 to 1450 (all the 999's sold out)). https://pcpartpicker.com/list/CysxCd It complains about the case, but that case is listed as eatx on all 3 sellers. Indifferent on cooler, but probably aio. Mobo is up there in price (without stupid expensive), but the 2 usb4/thunderbolt is part of why leaning its way. Indifferent on ssd, but 2tb is probably the right size for current workflows. Indifferent on case, that one was just cheaper, explicitly eatx, seems to have good airflow, and io being at an angle should put less stress on cables than being on top of case (least important in list but was factored in). On psu went with a name brand and tried to give enough headroom that i could add a second gpu in the future. bad time to build and so go gpu upgrade, or as good a time as any. Advice on gpu upgrade, new system, and/or fixes to problems in current system (especially if you think gpu is not the cause)?

Main issue seems to be a power one. When there is no load on the card (including as soon as booted to desktop), card is at power limit. Increasing power limit didn't help. Neither did voltage or lowering core and memory. Set them all back to defaults as any overclock doesnt matter when performance is crap.

I might need to replace the thermal pads too. Did take the plastic shroud off and am going to blow it out for a test since with fry's and radio shack gone I think the only place I can get replacement pads is online so will have to wait for shipping.

My pny 1080ti performs horrible (PNY GTX 1080 Ti Blower Gaming OC). Somewhere i have a folder of results from other benchmarks when i started looking into this, but since i have no idea where that folder is i can run other benchmarks if needed (windows 10 and ubuntu currently installed). Mining daggerhashimoto it gets ~8.5 mh (my 2 evga ftw3 1080ti each get ~32mh). About a year ago i pulled this card out of my server to troubleshoot but behavior is same on other computers are it was in server (got it for video transcoding mainly...but that is part of a whole different story). Attached are 2 superposition runs (performance went down after geforce driver update). For sanity check i took one of the evga cards and put it on this pc and score was ~16,000 (forgot to screenshot it). Next logical step would be to update the bios/firmware, but i haven't had good luck finding verified versions for this card. According to nvidias tool, it is on the latest version so reinstalling that version should be the step to take but that is not an option their tool gave me. I know i should have sold the card back when everything was crazy high...it was on my todo list but just never happened. Considering how much dead for parts cards were going for, one that works but not as well as it should would have been a decent payday (and gone into buying more exos drives). I dont think it is a thermal issue, but i can go back and run a long furmark again if someone wants to know what it maxes out at. That being said, i have not replaced the thermal paste, but since it hasn't seemed to be overheating that didnt seems like a useful step.

*** Might not be in right place, but it is planning the implementation of new addition *** I have a r720XD running unRAID (2x12tb parity, 5x12tb + 3x6tb array, 3 cache, 1 unassigned). A chunk of the storage is "vault/archive" type files/folders/projects, but also the stacks of external harddrives. I bought a 4246 with 24 x 4tb drives (and the cable and a 9202-16e) which should be delivered friday. When including everything shipping/tax/warranties/cable/lsi/das/drives, total was ~$2k but will make future upgrades easier (and still cheaper $/tb than 12tb exos). After preclearing the drives, I could add most of them to my array, but hoping to do better. You know how you have cache-->array, I'm trying to concoct some way to do (cache-->array)-->vault/archive. So when I search for a file or [plenty of other examples] I just go into the share it would be in and it will show up. I get that unRAID does not have 3 layers. Thinking give the 24 x 4tb drives to a VM running an actual RAID config (figure out number later) and have it mounted on the unRAID side. This would require some custom "mover" script to create the links for files to show up on the share when they are in the VM array. The more I think about this method the more it seems like it will crash and burn hard. Maybe the solution is to run a basic unRAID server using the max number of drives and just have the other 4 (or # somewhere in there) drives ready to hot swap in (they are used sas drives). I am open to other ideas too, worst case I have main storage and vault/archive storage (still run from the VM), and they are separate network entities.

"RAID (Redundant Array of Inexpensive Disks[1] or Drives, or Redundant Array of Independent Disks)" (lazy so copied the definition from wiki) RAID isnt backup, combining local drives into 1 storage volume and redundancy (speed depending on number of drives and config). Data drive fails, insert new data drive and rebuild what was there from the parity and other data drives (or something like that depending on config). True backup follows the 3-2-1 rule, 3 copies on 2 mediums with 1 offsite, reasonable large storage should be ok with raid redundancy and offsite copy. No idea what the industry rules are that google and all them use tho, but probably way out of most any consumers budget. They talked about the "right" ways and the hack here.

My Unraid server is basically at limits again (poor cache drives). Down to 1 drive bay free in my r720XD (not shown is the unassigned devices 500 GB SSD for VMs), and then there is the almost 30 TB on external drives (which includes an Unraid server (trial license there) of 250 GB drives...not powered on unless needed). Remember the iJustine tour, imagine that but mix of internal and external drives (and tmp folders borrowing storage on other peoples computers too (with permission FYI)). My youtube has been in "I need to edit" mode for a few years as the multiple projects folders have moved between drives and new projects created and filmed. Well 1.5 vlogmas (1 and part of another that was cut short because it was a boring December) that were done all on my phone was uploaded, but Premiere is so much better than mobile editing and all the other problems with using just 1 phone. Right now most of the projects are together in a YT share on my server but random stuff like "gopro dump 05" and "goggles from event"..., so working on sorting/relabeling and hopefully to start editing (when not working on research project to graduate). At 6:30 in video, NAS building collabs were mentioned, so figured it couldn't hurt to post. Come down to Texas @LinusTech, have a Whataburger (being a Texas only fast food), and fix my server/setup before I finish my masters degree this summer. Ignore the uptime, had a power outage just over 2 weeks ago and I still do not have a UPS yet.

With video you wont see much space saved if any. On my systems I do search for text iso vdisk files and compress them as they compress well. Have not analyzed a premier project folder but the amount of non video space is probably negligible to the point of not being worth it at all.