Jump to content
Search In
  • More options...
Find results that contain...
Find results in...

StackUnderflow

Member
  • Content Count

    299
  • Joined

  • Last visited

Awards


This user doesn't have any awards

1 Follower

About StackUnderflow

  • Title
    Member

Profile Information

  • Gender
    Not Telling

System

  • CPU
    Intel i7-6700K
  • Motherboard
    MSI Z170I AC
  • RAM
    Corsair Vengence LPX 32GB (16GB sticks)
  • GPU
    EVGA GTX 980 SC
  • Case
    Fractal Designs Core 500
  • Storage
    Samsung 850 Evo 250GB
  • PSU
    Cooler Master VS 550W
  • Display(s)
    LG 24in 1080P
  • Cooling
    Noctua NH-D14
  • Keyboard
    Unsure yet
  • Mouse
    Logitech G9x
  • Sound
    Onboard
  • Operating System
    Windows 10 Pro
  • PCPartPicker URL

Recent Profile Visitors

826 profile views
  1. Try Rainway or Parsec, discover how shitty it is, then give up on it.
  2. This is weird but it fixed itself... I did nothing and Prime95 (all the options smallest FFT to blend) runs fine for at least a minute and no shutdowns... I'll run MemTest85 and if it doesn't come up with anything I'm just going to chalk this up to unclean power and hope it doesn't happen again.
  3. It's as if someone pulled the power plug. It does turn back on by itself I think I have that on the BIOS. Crashing like if someone pulled the power plug. No blue screens no errors, everything's fine then it's suddenly off. I've also tried running furmark and that doesn't seem to crash it, at least not yet. Machine specs: CPU Intel i7-6700K Motherboard MSI Z170I AC RAM Corsair Vengence LPX 32GB (16GB sticks) GPU EVGA GTX 980 SC Case Fractal Designs Core 500 Storage Samsung 850 Evo 250GB PSU Cooler Master VS 550W
  4. Hi, I have a machine that was stable for 5 years, recently it began shutting off by itself. I started with running Prime95 and after a few seconds it'll just shut off. I tried underclocking and undervolting from 4.2GHz to 3GHz and 1.3V to 1V and it ran Prime95 for longer (a minute or so) before shutting itself off. Is this indicative of PSU malfunction or is something else broken? Thanks, P.S. I ran HWinfo along with Prime95, thermals are fine.
  5. I don't think so, to be honest I only tried the 4 corners and found that one to be the least intrusive. It also has the added benefit that I can see what the screen is showing sitting behind the TV since the screen is right above the TV. The middle left could also be a potentially good spot that I can see that doesn't cover up the 2nd CPU heat sink, though it could be weird because it will cover up different components based on if you're looking at it from above or below. But I haven't experimented with that yet. My dream for that display would be for it to be a transparent LCD back lit by the RGB in the case (like this but not bulk https://www.alibaba.com/product-detail/TFT-transparent-LCD-panel-display-for_60698827885.html?spm=a2700.7724857.normalList.52.34d03978W5OBhQ). Though that is too expensive for "non essential" hardware.
  6. Thanks! I'm kinda jibed that the screen is covering the 2nd CPU cooler, but my choices are either to cover up the CPU cooler or the GPU and I have 2 CPUs but only 1 GPU... Thanks! I thought about epoxying the case to the arcylic case but that's a bit too permanent for my tastes right now, and the tape holds it up OK, though it's a bit jank.
  7. This build is currently my everything-machine. I've built this over a period of 2 years, upgrading parts and adding in things as I went along. It currently hosts 10 LXC containers and 4 KVM VMs. This machine runs: pfSense router Web/SQL Hosts for various projects Minecraft host NAS with Plex DVR Work VM for when the work cluster is too slow Gaming VM with USB3.1 Gen2 and GPU PCIe passthrough (couch gaming) With all that being said, here's my parts list, with prices that I bought at, prices new (at launch, approximated for some) and current price if you were to buy this stuff now. Item New Price Current Price Price I bought at Xeon E5-2667v2 2057 175 180 Xeon E5-2667v2 2057 175 180 EP2C602-4L/D16 530 350 330 64GB DDR3-1333 ECC 200 64 100 2TB Seagate 82 85 82 256GB SK Hynix NVMe SSD 200 35 85 512GB WD Black NVMe SSD 170 110 150 960GB Corsair MP510 NVMe 130 130 130 ASUS Hyper M.2 X16 65 55 65 CM Hyper 212 Evo 35 35 35 CM Hyper 212 Evo 35 35 35 Dell Quad NIC 40 25 25 Athereos Wifi card 10 8 10 USB 3.1 Gen 2 card w/ Type C 25 25 25 EVGA GTX 980 SC 550 150 350 EVGA 650W GQ 100 100 81 Kingspec 32GB SATA SSD 20 20 20 Total 6306 1577 1883 I have used up all 5 of the PCIe ports that came with the motherboard. Here are some pics of this janky setup: This is my setup, display output to 4k60, accessories are keyboard with Cherry Blues and G502. Side view This little display I picked up from Amazon for $40 with the case. It was made for Raspberry Pi. It's actually insanely useful because the mobo only outputs to VGA, so I have to break out a VGA monitor to update or change the BIOS, this little display takes care of that. It's powered by a USB->Molex->SATA converter setup. Closer look at the window mounting mechanism (it's just tape) With the side panel open, I tried to make this as clean as I can. The only 2 visible cables are the power going to the USB card and GPU (as well as the leads for the tiny screen). Note the wifi card being held together by twist ties... Back side (Extension from top to bottom: Wifi, 4xM.2 x16 cards, USB Card, quad NIC, GPU, empty slot to get VGA output back into the case for the little display) Todo: Reinstall the PSU basement, have to reroute a bunch of Mobo connections to not use the bottom grommot RGB? Edit: forgot to mention this but tip for anyone using the Enthoo Pro and SSI-EEB motherboards: Remove the grommots to the right of the motherboard. They will raise your board and make it uneven and your board can be warped. I noticed after screwing in every mobo screw that the top right of the board is way above the standoff. I fixed with washers for a time until I figured out that the grommots are the problem.
  8. I'm using KVM through Proxmox 5.4 There are 10 openVZ containers and 3 KVM VMs on the box total.
  9. Hmm I had thought virtualization overhead is actually quite low these days with VTx and VTd. I don't think modern day VMs emulate everything in software, but I can look into the overhead. I thought it was CPU limited since the only thing that changes between playing 4k@30 vs 4k@60 is CPU utilization. In any case, even if it's inefficient, the extra 50% performance from the new processor I think should help. Unless it's because the older Xeons are missing QuickSync or other hardware decoders, in which case they wont. Even without the 4k video argument I think the other points still stands. It just feels like the CPU at 2GHz is the slowest part.
  10. Yeah, 4k@30 works well, 4k@60 stutters, 8k is unplayable. I assigned the VM 16 cores and 32GB RAM with the disk and GPU passed through. Most games work well too. I tried Witcher 3 at 1440p, Destiny 2 at 4k, Eve at 4K and RuneScape (I'm not actually sure how resolution is counted here). RuneScape stutters and drops below 40fps, the other ones are fine if I turn off all post processing and set some settings to medium.
  11. Then I don't understand why it's stuttering. I also tried to play a 8k60fps video and GPU was at 34%, disk at 2-3% and CPU at 67%. That one stutters way worse than the 4k.
  12. So it was using the same drive, GPU, and network as the system before and that played it without stuttering. I checked usage in task manager the disk was at 3%, GPU at 23% and CPU at 50% when it was playing (I'm guessing this is because of hyper threading although I'm not sure why it's not at 100%).
  13. So recently my main "work/gaming" computer stopped working and I'm trying to fix that. I have another server that has 2xE5 2650 v1s on a EP2C602-4L/D16 with 64GB RAM. This used to be my plex, NAS, scientific calculations (parallel CPU heavy), game servers that never goes off. For now, I've migrated the main computer's disks and graphics card and set myself up a VM with a PCIe passthrough SSD, GPU and USB controller. So the E5-2650v1 has a base clock of 2.0GHz and boosts to 2.8. This is OK for most things like 1080p video and most gaming, but 4K video at 60fps stutters a couple frames every 3-4 seconds (30fps is usually fine) and some games wont give 60fps (RuneScape). I want to game on 4K and I'm OK with >50 fps (TV can only do 60 anyways), so the CPU is less important than the GPU for this purpose (which I should also upgrade). For work stuff, the workload is highly parallel anyways (running multiple trials as separate threads and compare results after, so no issues at all with locking or waiting for calculations), so higher core counts help. So I went ahead and bought 2xE5-2667v2s for $350 today which has a base clock of 3.3GHz and boosts to 4.0. I'm just wondering if I made the right decision here. I plan on selling the original PC without a graphics card or drives for ~$400-600 (i7 6700K with 32GB RAM). But then I thought to myself, if I'm spending this much on old CPUs anyways, it's not that far off of a i9-9900K or a Ryzen 2950X or 1950X or something that's more modern. The 2950X beats my setup by 3k passmarks and 1950X is comparable in price and performance. And the i9-9900K is about the same passmarks with 2x less cores. Some other benchmarking sites have the modern CPUs double in single threaded performance. I'll have to invest in a new motherboard and more RAM but you'd have to spend that money eventually for when this isn't good enough when 8K video/TVs becomes the norm anyways. The other option is to spend a whole lot more ($1500) for a new dual Xeon silver system with a modern Mobo + RAM, and have it be upgradable down the line to either old hardware (Xeon platinums when those are in the >$1000 range) or new stuff. On the other hand, the dual Xeon setup is twice as cheap as a 2950X and 4x as cheap as a new Xeon setup. I don't have to invest in a new motherboard or RAM. I doubt DDR4 prices are going to calm down until DDR5 happens anyways. In any case the server RAM pricing is insane. A 32GB stick of DDR3 costs $60, while a 32GB stick of DDR4 costs $300. I originally went with Sandy Bridge because of the RAM pricing (as server Haswell started using DDR4), I thought they would have calmed down by now but apparently not. While the modern processors do have better benchmarking scores, I don't know how much that translates into actual performance. On the consumer side, I need to watch videos and play games in 4K. Both of which could benefit from a GPU rather than a CPU upgrade. On the work side, none of the stuff we do uses AVX2 or AVX512 instruction sets anyways. I suppose I can get screwed with H265 video becomes a thing though. And with the old system, I can upgrade the RAM to 512GB eventually using 32GB DDR3 sticks (once they reach <$1/GB) so even if I can't upgrade the CPU I can still have a lot of RAM and find a use for this thing later as a fileserver or something. I guess I'm trying to rationalize an impulse buy, I'm wondering if I should cancel before its too late.
×