Close
Page 3 of 3 FirstFirst 1 2 3
Showing results 21 to 27 of 27
  1. #21

    Default

    3 display port cables run into the back of the gpus, the angle of the picture probably makes it seem a lot shorter (gap) than it is. Now if you meant how do the cables run out of the 'case'? Behind the PSU in the first picture, I have outlined a section to be cut out in order to run any cables outside -- also giving ventilation for the PSU to push out any heat it generates.

    Everything else (usb, sound etc) is extended and runs to a faceplate on the front of the 'case' -- along with power, reset, temperature/fan controller display etc.

    In a few months (waaaaaaaaay after Christmas) and I do the real thing, I can post better pictures and even the CAD drawings if you wanted.

    *edit*
    it's actually quite pretty with all the LED lights and the UV dye xD
    *second edit*
    The temperature of everything running, is awesome. I had the three GPUS being stress tested via furmark for 50 minutes and they didn't go above 39c lol
    The CPU is a beast.. I'm playing 15 accounts (again) and the damn thing stays in low power mode (1200/2000mhz) -- I am still getting 60 fps (isboxer limit) on the main window in Ashran...

    Ahh crap. You guys were right, 6GB isn't enough to run them all.

    It's fine if they are all on Low settings, but if I bump the main one up, the other screens stutter like crazy haha.
    Last edited by Svpernova09 : 02-13-2016 at 05:15 PM

  2. #22
    Multiboxologist MiRai's Avatar
    Join Date
    Apr 2009
    Location
    Winter Is Coming
    Posts
    6815

    Default

    Quote Originally Posted by Kayley View Post
    Ahh crap. You guys were right, 6GB isn't enough to run them all.

    It's fine if they are all on Low settings, but if I bump the main one up, the other screens stutter like crazy haha.
    1440p x 5 @ low settings eats up > 6GB? That doesn't seem right. When I gave an estimate on of VRAM back on the first page I was measuring VRAM when running on ultra settings, and here's an image that I took a few months back with five game clients running 2560x1080 on ultra (no SSAO) with MSAA 4x and a 125% render scale, and I think a ton of browser tabs running in the background. You can see on the overlay I'm only eating up 7.2GB, but that's in Elwynn Forest, so I'm not entirely sure where you're hitting that number. However, I can try to do another test sometime later tonight to see.

    Also, you're not running your monitors in Surround mode are you? Last time I used Surround mode while multiboxing (even in a partial windowed mode), the VRAM usage was so bloated it was unreal.
    Do not send me a PM if what you want to talk about isn't absolutely private.
    Ask your questions on the forum where others can also benefit from the information.

    Author of the almost unknown and heavily neglected blog: Multiboxology

  3. #23

    Default

    1440p x 15
    I'm all over the place with posting sometimes, on a previous page I think we (whoever it was) discussed VRAM with 15 characters (more so in regards to the FURY X).
    Not using SLI, not using Surround mode, just 'Activate All Displays' - with Gsync.

    I can run them all on low np, if I bump the settings up on the main to even fair, gpu goes to 100% and the others stutter and lag out like crazy.
    But i'm also having problems with W10 I think, Can't select DX11 in WoW, doesn't recognize my cards as being shader model 5+.

    I have to wait a few days for the SLI bridge to come in, then I can try 'Tri-SLI' to see if that helps at all.

    Process of elimination!

    Turns out the GPU drivers were corrupted on install, simple reinstall has fixed everything. \o/
    The end \o/

    It started bothering me, performance wasn't as amazing as I wanted (using only one GPU). Everytime I tried SLI it would only let me use two monitors unless I used that SURROUND gaming config (all monitors combined to one resolution), which I didn't like. So I have been huffing and puffing until today in the shower I thought 'jesus, what if I plug the monitors into the first GPU and then run SLI?' ...... Obviously I am now running TRI-SLI with all my monitors acting independantly now. My GPU was the bottleneck, GPU memory was always 100% while idling, GPU core was usually 90+.. Now under a medium load it's 70% on both. Meaning I get my 60fps in all situations (minus the initial dips when a lot happens suddenly). Gsync seems to work properly too now, it's so gorgeous.

    All I need to do now is set up a better CPU strategy to alleviate the sudden burst of information that occurs during a random 50 group jumping on me.
    Last edited by Svpernova09 : 02-13-2016 at 05:13 PM

  4. #24

    Default

    Well I was wrong about something the other day so I started sulking-- decided to build a new PC case and spruce up my system. The new intel CPUs are a huge letdown so I'll still use most of my current hardware.



    The front is going to look like this. I'll have some really thick foam cut out into the square (well diamond with how it is orientated) shape and 'probably' attach it with magnets for easy removal/clean. The front 'panel' will have the fan attached to it, that too will be held in place with magnets or a basic hinge depending on how it works out.



    The back of the system. The red square outline is just a representation of the I/O shield and space being used up by the PC itself. I'm going to add a 120mm fan in the middle-back area sucking air out, along with two little 60mm fans on the left and right side respectively. Powersupply neatly stacked on top. I've also decided to add two small wheels at the back of the case; near the bottom. So I cart it around relatively easy. Sort of like a suitcase where you lean it back on the wheels and pull it along. I thought about trying to make it stable enough to have wheels on the actual bottom but it's not necessary and prone to tipping (boo).



    The radiators and reservoirs are located outside the case. Covered with some mesh like THIS. I'm not concerned with aggressively overclocking so the fan-speed will be relatively low so as to keep noise to a minimum. As it stands right now with my current setup; I goofed hard and destroyed my current fans.. PC is being passively cooled with just the radiators 'fine'. Fine-being the GPUs under load stay around 48-52c and the cpu hangs around 72c under 100% sustained load. Not the best hence my choice to scrap the 'table build' I have now and make something better.

    I need to add 2 of these and one of these not just for looks but I crammed 30+ fans onto one fan pwm originally which in hindsight was stupid. More to follow as I machine it all / assemble it all this weekend. I'm not sold on which vinyl wrap to use..


    Prototype up and running! Bit messy in my outside workshop but npnp.
    Attached Images Attached Images
    Last edited by Kayley : 10-15-2018 at 10:26 AM

  5. #25

    Default

    So.

    I lost my temper and was a little bit rough.. pulled the pcie connector (gpu-top slot) off the motherboard and uhhhh... the pins remained on the board. DID NOT REALISE. Pushed the cards back down (they are connected to each other via sli bridge/WC loop) and felt the familiar click.. turned the PC and it shut-off; smelled that lovely smokey flavour and.. dead. I have a replacement board here on Friday (hopefully) to check to see if I have blown up the CPU/RAM/Video cards. I de-soldered the pins to remove any short, still a no-go. Siiiiiiiiigh.

    if it's all dead I'll buy the current overpriced stuff. 7980xe/9980xe, 2080tis and 128gbRAM just to box a dumb amount of characters.

    Such a shame. First time I've goofed with physical hardware like this.

  6. #26
    Member Ughmahedhurtz's Avatar
    Join Date
    Jul 2007
    Location
    North of The Wall, South of The Line
    Posts
    7169

    Default

    I'd be somewhat surprised if it killed anything other than the MB and maybe PSU. That said, I've had...spotty success with Asus PC components so I don't use them anymore. Not sure how well protected their circuitry is against this sort of thing.

    I can remember the days when you could accidentally unplug a running system's PCI audio card, plug in a replacement, and have the system actually recognize it, enumerate it, and run like nothing happened. /nostalgia
    Now playing: WoW (Garona)

  7. #27

    Default

    Well. The RAM survived. Two of the Three GPU's survived (RIP).

    I purchased a 2950x as a replacement CPU. Which has been pretty impressive so far for multiboxing.

    The motherboard and CPU came in Apple-esque packaging. Nicely presented, plenty of extra goodies to make you feel like you are opening something amazing. Impressed with the build quality of the motherboard to be honest. Really impressed.

    This whole rebuild took a stupidly long time (17 working hours) due to the weird way it detects/populates the sockets.
    The official documentation on how to use the 'Bios' USB slot quotes from the x99 series which is completely different. For example in the past you couldn't have the CPU/RAM installed in order for the 'BIOS' USB slot to work. With the x399 you NEED it all populated for it to work.
    Also requires a USB 2.0 drive which I didn't see... had to go through about 30 USB drives before I found an old one hah.
    Only had to mess around with it as the Threadripper 2s aren't detected on older bios-- at least they have this autoflash feature. Would suck to have to buy an older CPU in order to get it to post, flashing the old way.

    Had all the RAM slots filled... aaaannnnddd it kept throwing error codes at me. I would take the RAM out, re-seat it, it would work, then stop working. By working I mean it wouldn't be detected in the BIOS. I then found out about the weird quirks.. on the first post it would notice 2 sticks, next post it would notice 4.. until it was all detected. Weird AF.

    Motherboard kept giving GPU error codes. I had to drain the loop and install the cards ONE AT A TIME for it to post. Install one card, boot it up, let the bios see ONE CARD.. save and exit. Install a second, into bios, rinse repeat until all Three were seen by the motherboard. However when I got into Windows only two are recognised (siigh).

    All in all not so bad.

    --overclocking tomorrow! I'm hoping for at least 4.4ghz on all cores.

    Ok so overclocking on this chipset is a little different. Here are some results with Cinebench to show what I mean.
    I metered the mboard tracks and my PSU output.. it was able to push through a stupid amount of power to the CPU. The Ryzen Master program tells me the chip can take 700w... How batshit crazy is that.

    So far I have found that I can OC to 4.5ghz and it's benchmark stable.. however the results are terrible. I then noticed when leaving it at stock and having the Auto-boost (Precision Boost Overdrive) manage everything, it was able to beat my Manual OC (4ghz) with it's boosted frequency of 3.85ghz (Cinebench shows it at 3.50). Poked around to find out why and it turns out this particular chip wants LESS voltage. So dialed back up the 4.1ghz and lowered the Vcore/Vcore SOC and it benched a lot higher. I did think it wasn't getting enough voltage and it was throttling originally-- which is why I metered the connections to double check.

    It continues!

    Another example of the weird Voltage thing with the CPU.
    Someone else who OC'd to 4.1ghz was achieving 3628 cinebench scores consistently. Trying to find out what his voltages were like.
    Last edited by Kayley : 11-04-2018 at 01:05 AM

Posting Rules

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •