[PS4/XB1] Due to an error, we have extended current events (such as dungeon highlights) by one day. The new end time is now 10/15 at 0500 UTC. Thanks for your patience!

Low FPS on perfectly capable PC

2»

Comments

  • LapomkoLapomko ✭✭✭
    edited March 11
    Theres no point to discuss about this because I have 8700k overclocked to 5.0 GHZ and my fps is garbage when i'm in highwatch or doing dungeons and everyone is popping their ultimate to 20fps. The fps count is bollocks because I have 240hz monitor with g-sync and it shows much lower fps and different otherwise. Already using all the super minimal ini file with broxy to have more fps but you can't do anything anyway unless you hide everyone.

    You want higher fps? Force the devs to rewrite the game into unreal engine 4 with competent new developers. But suprise they don't want to because their new game is made in the same unreal engine 3!
  • There is no perfectly capacle pc for running tera, or any of bluehole's crap
  • I do understand that AMD chips are generally not accepted as "ideal" for gaming..

    I am using AMD ryzen 7 2700 process along with radeon rx580 8GB but i am pretty okay with FPS in general.

    even in CS with a lot of ppl, i tends to stay above 60FPS ...
  • LapomkoLapomko ✭✭✭
    edited March 14
    Ryzen made a big step in improving their IPC and prior that it was horrible compared to intel and generated bad name for them. Though compared to intel their IPC is still lower and especially for Tera that can only do simple multithreading to 1 or 2. For that you want a CPU with higher IPC which is more beneficial.
    Also don't forget Intel cpu with unlocked multiplier can easily reach 5.0 and lucky 5.3ghz these days while AMD struggle to get 4.0ghz+ so you also have that.
  • I'm pretty sure its not a hardware based problem. I can run anything else perfectly fine. Tried everything suggested here and only improved by about 5fps but now I'm crashing alot more.
  • LiverneckLiverneck ✭✭
    edited March 14
    I have the standard bad fps of 20-25fps in highwatch when its really heavy populated. That's about as bad as it gets though. In corsairs bg, i can get 40-60fps as long as i turn off shadows. And everywhere else in the game fps are 60-80fps.

    This is on a 1080gtx gpu and cpu is a 4770k i7 @4.4ghz windows 7 32gb of ddr3.

    Worst fps i had ever experienced was in the 2012 Nexus days when there was hundreds of people in 1 node. The fps was about 4-10 fps at best.

    The most priceless thing i see all the time is the game giving the message "due to insufficient memory, texture resolution has been reduced" or something around that line. This is the type of optimization we are dealing with here, the stable genius programmers of this game are so stupid and inept, they think an 8GB video card is low on memory, even though i'm looking right at my overlay and its showing ~1-2gb of 8gb being used.

    I wish i could help you, but if you are getting 20 fps everywhere, something is really wrong. Even this badly optimized game will run decently in some areas. 20 fps is the number you see when theres 40+ people all infront of you.

    Even someone above already said they have a 8700k at 5ghz, and still has bad fps in crowds. This is so true. Proof of this was when I upgraded from my 2600k to 4700k, i noticed zero improvement in performance.

    Optimization should be done a long time ago. Just simply finding a way to put every 10 players or so onto a separate cpu thread would help if those threads are available. A 16 thread cpu could then in theory handle 160 people with much less lag than the 1-2 threads this terrible game currently uses. No matter how crazy you can overclock your cpu, it just bottlenecks at a certain point and even having 5 ghz doesn't help you much.

    And the people saying 2080ti RTX helps you are incompetent. Even when i ran this game on a 970gtx at 21:9 ultrawide i didn't get a gpu bottleneck. You can run this game on a high end GPU from 2012 and have room to spare. Even in 2012, I could only max out my 680gtx at 1440p, at 1080p it was always below 100% usage.
  • LapomkoLapomko ✭✭✭
    edited March 14
    This is true about i7 2600k and 4770k. That's the same jump I made but 4770k was very poor overclocker so almost no difference. I made a jump from 4770k to 8700k and that gave huge improvement though because of the generation gap and it hit 5.0ghz.

    They couldn't have anticipated when they developed this game way back that cpu cores would increase and the megahertz race would hit hard silicon limit. Dual cores were most common cpu back in the time too and multithreaded programming was also very difficult and time consuming. This game also use Directx9 rendering API which can only do very simple multithreaded drawcall from cpu to gpu which is why you see the first thread or two being heavily saturated. This is why they developed these API's: Mantle ,Vulkan, DX12 to fix this cpu bottlenecking drawcall issue.

    Also being mentionened several times to fix these issue they would have to port the game to Unreal Engine 4 to benefit from Directx12 ability to fully multithread the amount of drawcalls. And would never happen because the huge expenses and manpower it takes to do for a dying game. Completely not worth it for them.

  • ElinLoveElinLove ✭✭✭✭✭
    Lapomko wrote: »
    This is true about i7 2600k and 4770k. That's the same jump I made but 4770k was very poor overclocker so almost no difference. I made a jump from 4770k to 8700k and that gave huge improvement though because of the generation gap and it hit 5.0ghz.

    They couldn't have anticipated when they developed this game way back that cpu cores would increase and the megahertz race would hit hard silicon limit. Dual cores were most common cpu back in the time too and multithreaded programming was also very difficult and time consuming. This game also use Directx9 rendering API which can only do very simple multithreaded drawcall from cpu to gpu which is why you see the first thread or two being heavily saturated. This is why they developed these API's: Mantle ,Vulkan, DX12 to fix this cpu bottlenecking drawcall issue.

    Also being mentionened several times to fix these issue they would have to port the game to Unreal Engine 4 to benefit from Directx12 ability to fully multithread the amount of drawcalls. And would never happen because the huge expenses and manpower it takes to do for a dying game. Completely not worth it for them.

    I won't agree with the team's future views on computer hardware, as the game is not THAT old. You could trace it's development as the proper game code (kek, "proper game code"... But you get it) around 2009 at it's earliest, before that if I recall it did live as a concept but no coding there.

    2009 was already Phenom x4 days, Pentium 4/D have died already and gave way to Core architecture, and not even going into extreme cases like server processors (Opterons had like ~16 cores or what). Basically, they saw already the change even if duals were the most common, they were just... well... Bluehole Studios. I've read around these forums that this is common with other Korean studios. Thankfully Pangya! for PSP was coded in Japan...

    It's true tho that we probably won't ever see a re-coded TERA with proper game engine. They just ain't taking any bigger bets on this game anymore since they have a new, equally unoptimized buggy cash cow. We can only hope something very similar to TERA but better MORE LOLI RACES comes, so then if this one eventually goes kaput or we just want a change we've got something equally nice or better.
  • LapomkoLapomko ✭✭✭
    edited March 16
    ElinLove wrote: »
    Lapomko wrote: »
    This is true about i7 2600k and 4770k. That's the same jump I made but 4770k was very poor overclocker so almost no difference. I made a jump from 4770k to 8700k and that gave huge improvement though because of the generation gap and it hit 5.0ghz.

    They couldn't have anticipated when they developed this game way back that cpu cores would increase and the megahertz race would hit hard silicon limit. Dual cores were most common cpu back in the time too and multithreaded programming was also very difficult and time consuming. This game also use Directx9 rendering API which can only do very simple multithreaded drawcall from cpu to gpu which is why you see the first thread or two being heavily saturated. This is why they developed these API's: Mantle ,Vulkan, DX12 to fix this cpu bottlenecking drawcall issue.

    Also being mentionened several times to fix these issue they would have to port the game to Unreal Engine 4 to benefit from Directx12 ability to fully multithread the amount of drawcalls. And would never happen because the huge expenses and manpower it takes to do for a dying game. Completely not worth it for them.

    I won't agree with the team's future views on computer hardware, as the game is not THAT old. You could trace it's development as the proper game code (kek, "proper game code"... But you get it) around 2009 at it's earliest, before that if I recall it did live as a concept but no coding there.

    2009 was already Phenom x4 days, Pentium 4/D have died already and gave way to Core architecture, and not even going into extreme cases like server processors (Opterons had like ~16 cores or what). Basically, they saw already the change even if duals were the most common, they were just... well... Bluehole Studios. I've read around these forums that this is common with other Korean studios. Thankfully Pangya! for PSP was coded in Japan...

    It's true tho that we probably won't ever see a re-coded TERA with proper game engine. They just ain't taking any bigger bets on this game anymore since they have a new, equally unoptimized buggy cash cow. We can only hope something very similar to TERA but better MORE LOLI RACES comes, so then if this one eventually goes kaput or we just want a change we've got something equally nice or better.

    They were at the mercy of the API's at that time and the average consumer is not gonna get a 16 core opteron just to play games... Also remember at that time cpu's were around 2.8/3.2ghz. Developers were still hoping at that time the advancement of computer architecture and megahertz would get better. Remember legendary intel sandy bridge that could overclock to 5.0 ghz and the IPC was very high? That's when Cpu started to become stagnant after that and that came out in 2011..
Sign In or Register to comment.