Help me find my bottleneck (if any)



  • My home-built desktop is getting rather long in the tooth. I've noticed increased stuttering in reasonably recent games played at 1080p and am considering my options. It's mainly used for games, but hobby VS C# development also happens.

    Here are the current specs:

    • i5-4590 CPU
    • XFX Radeon R9 280X @ 3GB (I think) GPU
    • MSI Z87-G43 MB
    • 2x4GB (unknown speed) RAM
    • Crucial M500 512GB SSD (system disk)
    • WD Blue 1TB HDD (old and probably on its last legs, but I need the storage)
    • Thermaltake 650W PSU (80+ Bronze)
    • No clue what brand/model of case. Mid-tower, bottom PSU. Cramped workspace with more drive bays than I possibly need.

    Am I at the point where it's worth upgrading anything? Or should I start budgeting for a new system?



  • @benjamin-hall said in Help me find my bottleneck (if any):

    I've noticed increased stuttering in reasonably recent games played at 1080p and am considering my options.

    Have you gotten a can of air and blown all the dust out?

    Seriously.

    Try that before you spend money on anything.



  • @blakeyrat said in Help me find my bottleneck (if any):

    @benjamin-hall said in Help me find my bottleneck (if any):

    I've noticed increased stuttering in reasonably recent games played at 1080p and am considering my options.

    Have you gotten a can of air and blown all the dust out?

    Seriously.

    Try that before you spend money on anything.

    I cleaned it reasonably recently (within the last few months at least). A quick visual inspection shows no appreciable dust buildup since--I can read all the labels without issue.

    Thanks though.


  • area_pol

    @Benjamin-Hall

    Is the cpu running at 100% of a given core?
    (The overall % of cpu usage may be misleading. A cpu-bound single core-program will display as 1/n_cores % cpu usage, unless you configure the task manager to display 1 core = 100%, 2 cores = 200% etc)

    Is the RAM nearly full? Is disk swap space used?

    There should be some way to diagnose the GPU memory/processing usage (I only know the one for NVidia).


  • Impossible Mission Players - A

    @adynathos said in Help me find my bottleneck (if any):

    There should be some way to diagnose the GPU memory/processing usage (I only know the one for NVidia).

    Included in Task Manager as of Windows 10 Fall Creator's Update



  • @adynathos said in Help me find my bottleneck (if any):

    @Benjamin-Hall

    Is the cpu running at 100% of a given core?
    (The overall % of cpu usage may be misleading. A cpu-bound single core-program will display as 1/n_cores % cpu usage, unless you configure the task manager to display 1 core = 100%, 2 cores = 200% etc)

    Is the RAM nearly full? Is disk swap space used?

    There should be some way to diagnose the GPU memory/processing usage (I only know the one for NVidia).

    Playing through Dragon's Dogma: Dark Arisen with a Chrome tab playing Youtube videos, I was sitting at ~30% CPU (so probably pegged one core, given that it's a 4 core cpu), RAM ~40% (mostly not from the game, since Chrome was doing more than the game) and GPU at ~50% (using the windows 10 task manager tab). I'm not sure how to determine how much swap is actually being used.

    I have another program (a map-builder) that chews up ungodly amounts of memory (hex maps at three levels, ~500k individual hex grid points) and when it's saving I start hitting ~70% memory usage.

    Does that tell you anything?


  • Impossible Mission Players - A

    @benjamin-hall
    0_1512091693814_d6e550a2-1805-46bb-a020-f3f71f32ca0f-image.png

    Can tell you if you're overloading any one individual core.

    I second the "blow it out" - dust builds up in fans pretty crazy fast (especially if there are any pets or smokers in the house), and the computer should be taken away from its normal environment (preferably outside), opened up, and blown out periodically. I try to do mine about once a year, but I don't have either pets or smokers - computers in higher dirt/dust environments should be a lot more frequently cleaned. Also don't forget the fan in the power supply as well.

    8GB RAM is on the edge of what I would consider low these days, but Task Manager shows you're not maxing it out (and for that matter, I don't routinely exceed 8GB with just normal video game play either, so I guess I'm just an epeen snob...)



  • @izzion I didn't notice anything unusual in that game (all cores were used about equally, so it's probably multi-threaded or not CPU heavy). I did notice that the CPU frequency dipped (compared to unloaded) quite a bit--no load it's at about 2.2 GHz, in the game it was down to about 1.6 GHz. Could it be under thermal throttling? It only has the stock cooler on it.

    I'm going to try another game--this one's an MMO that was giving me very low frame-rates before.



  • @benjamin-hall definitely thermal throttling. If anything, the clock speed should go up when gaming.

    Stock cooler is fine unless you're overclocking or unless it's clogged with dust.


  • Impossible Mission Players - A

    @benjamin-hall
    Yeah, sounds like thermal throttling, I’d bet the stock cooler is caked with dust and not getting the designed airflow through the metal. And/or one of your case fans isn’t working correctly so the hot air isn’t going anywhere.



  • @gąska playing Tera, the clock speed is constant at 2.5, so I think I was misreading something. The GPU is under pretty bad stress, and the CPU is around 50/60, spread between cores

    And noticeable hitches in frame-rate, with texture/character pop in like crazy. Is there a good frame-rate overlay program I can record frame-rate with?



  • Here's the inside of the case--could the GPU be blocking airflow? There's a fan at the bottom facing in, and one at the back facing out. The cables are a mess, but the case doesn't have a good way to route them around the back.


  • And then the murders began.

    @izzion said in Help me find my bottleneck (if any):

    @benjamin-hall
    Yeah, sounds like thermal throttling, I’d bet the stock cooler is caked with dust and not getting the designed airflow through the metal. And/or one of your case fans isn’t working correctly so the hot air isn’t going anywhere.

    And/or the thermal compound is toast.

    Have you tried using CPU-Z or another tool to see what the CPU temps are like when it's (we think) throttling?



  • @benjamin-hall said in Help me find my bottleneck (if any):

    Here's the inside of the case--could the GPU be blocking airflow?

    Not if you're not overclocking. Just take out dust from fans with canned air/paintbrush; if it doesn't help (and you measured the temperature with some software so you know for sure that's the problem), you might try replacing thermal paste between radiator and CPU.


  • Impossible Mission Players - A

    @benjamin-hall
    I wouldn’t expect airflow problems with that setup, unless the power supply is installed upside down so that it’s fan is pointing into the case instead of out the bottom of the case. Can you feel air blowing out the back of the case? Does it feel warmer than ambient?



  • @gąska said in Help me find my bottleneck (if any):

    @benjamin-hall said in Help me find my bottleneck (if any):

    Here's the inside of the case--could the GPU be blocking airflow?

    Not if you're not overclocking. Just take out dust from fans with canned air/paintbrush; if it doesn't help (and you measured the temperature with some software so you know for sure that's the problem), you might try replacing thermal paste between radiator and CPU.

    What temps should I be looking for?


  • Impossible Mission Players - A

    @benjamin-hall
    25-35C idle, 60-70C under load. If it’s going over 80 it will likely begin thermal throttling.



  • @izzion said in Help me find my bottleneck (if any):

    @benjamin-hall
    25-35C idle, 60-70C under load. If it’s going over 80 it will likely begin thermal throttling.

    Either this monitoring software (open hardware monitor) is lying, or...wow. It's showing CPU Core temps of 99C. With basically no load (just youtube, 4 tabs in Chrome). Temperature sensor #6 (wherever that is) is also absurdly hot (85+C)


  • Impossible Mission Players - A

    @benjamin-hall
    Yeah, that's in the "omg you're frying the chip" temp range. Given that the CPU fan doesn't look inordinately caked in your picture, that sounds like the thermal paste has failed or was improperly installed.

    1. Get some Q-Tips, Rubbing Alcohol, and thermal paste
    2. Remove the CPU fan
    3. Clean any remnants of previous thermal paste from both CPU and cooler with Q-Tips & Rubbing Alcohol, allow to air dry thoroughly (shouldn't take more than 5-10 minutes)
    4. Apply thermal paste to CPU per directions
    5. Reinstall CPU fan
    6. ???
    7. Profit!


  • @izzion guess I know what I'm doing after work tomorrow. Sigh. And there's no good parts stores near me either.

    Thanks all.



  • Although, if I were to upgrade, is it worth actually doing it piece by piece with this old of a MB? That's not a burning concern, more looking towards the future and wanting to possibly get into AR.


  • And then the murders began.

    @benjamin-hall IMO, with Intel's "we need a new socket every 2 years" approach, you generally have to replace CPU and MB at the same time, you can't do it piecewise like in the old days.

    If you have the slots to spare, throwing in another 8 GB of RAM would probably be my next recommended upgrade.



  • @unperverted-vixen slots I have. I'm only using 2 of 4.



  • As I removed the cooler, I found that at least one of the clips hadn't engaged right and the cooler was hanging loose. A bottom one to be sure, but I'm guessing that's not a good thing. Now I just need to get more thermal paste tomorrow and hope that fixes things.



  • @benjamin-hall this might very well be the root cause. The radiator has to be tightly pressed to the CPU to work correctly.

    You might have just saved yourself $800 for new PC.


  • area_pol

    I second the idea about changing thermal paste.
    Once broken paste prevented my laptop from booting at all, replacing paste fixed it.



  • @adynathos said in Help me find my bottleneck (if any):

    I second the idea about changing thermal paste.
    Once broken paste prevented my laptop from booting at all, replacing paste fixed it.

    Thirded.

    Maybe two years ago, I had a machine at work always hitting ~90-100C temperatures on the CPU (which caused it to throttle aggressively). Cleaning the old paste and applying new one fixed the problem (I also switched out the fan for a larger/quieter one, but in hindsight, that probably wasn't strictly necessary).

    From what I'm told, the machine worked fine for another 1.5 years until a colleague took it over and irrevocably bricked the motherboard.



  • @benjamin-hall said in Help me find my bottleneck (if any):

    WD Blue 1TB HDD (old and probably on its last legs, but I need the storage)

    I've seen multiple WD Blue drives running 24/7 for 5.5 years in my home server before I said "fuck it, I better change it before it starts failing".

    WD drives (except the green ones) are very reliable



  • @timebandit said in Help me find my bottleneck (if any):

    @benjamin-hall said in Help me find my bottleneck (if any):

    WD Blue 1TB HDD (old and probably on its last legs, but I need the storage)

    I've seen multiple WD Blue drives running 24/7 for 5.5 years in my home server before I said "fuck it, I better change it before it starts failing".

    WD drives (except the green ones) are very reliable

    Nice to know. I'll keep it running till it dies--it only has (part) of my Steam library and my recorded TV media on it, none of which is irreplaceable.



  • I've had RAM failing. I've had CPUs overheating internally so cooling doesn't help. I've had GPUs just giving up on life suddenly. I've had front jacks short-circuit with each other so what was going on line out was also going on mic in, making it effectively unusable for communication. I've had CRT screen going from 4:3 to 9:3 due to mechanical damage. I've had LCD screen blinking in red every so often (loose cable inside laptop). But I've never had a disk failure.


  • Discourse touched me in a no-no place

    @gąska said in Help me find my bottleneck (if any):

    I've had RAM failing. I've had CPUs overheating internally so cooling doesn't help. I've had GPUs just giving up on life suddenly. I've had front jacks short-circuit with each other so what was going on line out was also going on mic in, making it effectively unusable for communication. I've had CRT screen going from 4:3 to 9:3 due to mechanical damage. I've had LCD screen blinking in red every so often (loose cable inside laptop). But I've never had a disk failure.

    Ever had a mobo failure? You'll think you're going crazy when it happens to you as the effects are so utterly weird. In particular, they look like other parts are going wrong, but replacing them doesn't help at all. I've seen service engineers completely stumped by this, and also seen the wonderful situation where the computer would work after yet another module was replaced until the poor guy left the building to go back to base… All definitely the true face of Murphy.

    Disk failures aren't very common (unless you get bargain basement disks) but they're rather annoying to recover from since they're the bit that is supposed to keep your data safe. Most other components are drop in replacements, but disks aren't. Keep backups and save yourself trouble. (I've seen one disk failure. It was in a laptop that was about 8 months old, and thank god I had a backup…)



  • Hm. Reading this thread has reminded me to do my regular preventative maintenance on my PC, and I'm starting to think there's something up. OpenHardwareMonitor is showing two of the temperature sensors sitting at just around 100C constantly, while another one is reading -17C. Not entirely sure which ones these are though, and the CPU core itself shows at around 25-30C across all cores sitting here idling with just Chrome open, but I do see the chip throttling itself randomly. Wish those thermal sensors had some sort of label to tell me where they are...

    Well, this thing has had a good run...perhaps it's time to start looking anew...



  • @e4tmyl33t Hm. Maybe it's not thermal throttling, just idle throttling or something. I fired up Rimworld and it's sat at a pretty constant higher clock speed than the CPU's base, and the odd temps haven't moved at all.


  • Discourse touched me in a no-no place

    @e4tmyl33t said in Help me find my bottleneck (if any):

    the odd temps haven't moved at all

    They might not be connected to anything at all. The temperature monitoring software I've got has GUI slots for several sensors that my system just doesn't have. In my case, it also manages to show them as disabled, but I could well believe that disabling not working right and instead showing random hardware crap.


  • Impossible Mission Players - A

    So, you know how I said "you should blow out your case at least annually" upthread?

    Here's what happens when you forgot to do that for two years (and it's a lot worse in a water cooled rig like I have):

    0_1512238986847_IMG_0033.JPG

    Whoopsie. Guess that's why the cobbler's kids have been complaining about sore feet...


  • Impossible Mission Players - A

    Haven't gone into the major major towns where I was having the most heat problems before (CPU hitting 75C, GPU pegging the 3D processing core and fans going nuts), but so far the cleaner and bigger CPU cooler + the newer GPU with hybrid cooling (basically watercooled with a closed loop cooler already attached, so I can be lazy and not futz with a custom loop) has gotten me down to this:

    0_1512240495742_031ec6d5-c756-4c16-82df-a25f5bcb43b4-image.png


Log in to reply
 

Looks like your connection to What the Daily WTF? was lost, please wait while we try to reconnect.