Strange Performance issues?

I recently got a brand new laptop in February, one much more powerful than my last. I downloaded guild wars 2 and for a while, the game ran really slow for some reason and would cause my computer to crash. That no longer happens but the performance is really bad when I have the settings maxed out. I have a gtx 1070 with a 7700HQ processor. I can run more demanding games at much higher fps. Here, I get around 20. I looked into people saying they get performance issues when their system gets hot, but the performance for me is just terrible right as I start the game up. Does anyone know of a solution, or am I just going to have to play on all the lowest settings to barely get 60 fps?

I have the same specs as you do, and many of those who has better specs run at similar frame rate. Turn render sampling to native and lower character model limit, besides that , there's nothing much you can do

I also have a laptop with 7700 and a GTX1070 but I never had any problems with running GW 60+ fps on maxed settings. Are your system options all set right to maximize performance over quality? Do you play with laptop charger on or off? Are you sure your laptop is always using the gtx over the one from i7?

A. Check your temps. Most laptops have massive cooling issue.... especially "gaming laptops" given the higher end components generate massive amounts of heat. And before you pop "I can play blah blah game".... let me finish. You'd be shocked how "much more demanding games" usually aren't. It'll make sense in a second.

B. The game is heavily CPU bound.... or more specifically, its extremely bottle-necked due to how it interacts with GPUs via the DX9 API. Visually this game stacks an insane amount of effects..... a LOT more then you'd think, even compared to the newer games. This on top of the fact DX9 isn't suited for the kind draw call volumes the game puts out. If anything, newer games are a lot more resource budget conscious due to how expensive effects are; and made more complicated by some sketchy practices in GPU optimizations over the last decade or so.

C. The type of bottleneck means it can't be brute forced easily with newer hardware. In fact, newer hardware going parallel exemplifies the problem, as per core performance is limited to manage heat. Depending how the game was designed, and how good it is at leveraging multi-threading, you can get a wide range of performance results across various CPU/GPU setups.

D. if you're letting it run in 4k... just don't.

E. You're never going to get consistent 60 FPS short of standing someplace where nothing happens. I know this is a hard pill for the younger kids... but you can live with 30 FPS. I've lived through a lot worse.

F. You don't need to run "lowest setting". You just have tone down the biggest performance killers. Shadows, Effect LOD (on), LOD distance, Ambient Occlusion (off), and lower the model limit to high or medium.

And monitor the temps. The Heatpipe setup in most laptops have a rapid saturation rate due to the very limited amounts of material used to make it. Less material to dissipate heat over = less able to store heat before having to radiate it out. My Razor (which runs shockingly cool for what it is) gains 15C in the time it takes to load the game, and hits its running temp (75) in less then a minute after that.

@"starlinvf.1358" said:B. The game is heavily CPU bound.... or more specifically, its extremely bottle-necked due to how it interacts with GPUs via the DX9 API.

You were right the first time here: the game is CPU limited -- specifically, if you eliminate every other bottleneck (eg: enough memory, fast enough video card, etc) then a single thread becomes the slowest part of the entire thing. GW2 does use more than one thread -- it'll keep three more cores at 80-ish percent while that main thread is capped at 100 -- but because all CPUs bring enough extra cores to the table, nobody talks about it that much.

The DX9 API, however, is not the root of the bottleneck. Before you go there, even the developers, ANet, have publicly stated that switching to DX11, or DX12, would have zero effect on performance because that is not the limitation. (Not to mention that, honestly, very few games are actually limited by the internal locking in the DX <= 11 API, compared to elsewhere in the processing pipeline.)

So, yeah, you are quite right that it is CPU limited -- but that is predominantly dynamic model placement related. This is why ramping down the model counts in the options is the best way to boost FPS if you need to. (At least, for most people, if you have a 10 year old GPU, it prolly won't help that much, etc.)

D. if you're letting it run in 4k... just don't.

Solid advice, and I completely agree. Either select "subsampling" in the graphics options, or switch to a lower resolution. They both do the same thing, draw the 3D part in lower resolution, the only significant difference will be what resolution text and HUD stuff is drawn at.

E. You're never going to get consistent 60 FPS short of standing someplace where nothing happens. I know this is a hard pill for the younger kids... but you can live with 30 FPS. I've lived through a lot worse.

nod Agreed. Even on a pretty much top end system, the Teq effects explosion, with 150 people all sitting in one exact stop hammering stuff drives a well cooled, desktop 1080Ti and fast CPU down to ~ 20-30 FPS. It'll hit fairly solid 100 most other places, and 144 capped by vsync if you find a nice, quiet instance though.

It'd have zero performance gain, because the entire engine needs to be refactored for stronger multithreading and asynchronous tasking. You're right in that its not "just the API", but their whole engine architecture is built around DX9 era design, and not even fully optimized under that. The engine makes a massive number of calls to the GPU (someone once said 1:1 with effects in many cases, which sounds believable), and the thread is mostly doing this in sequence. Batching is supposedly the solution to this (cutting overhead)- but as I understand it, implementing it changes how you time engine tasks, since it can accidentally fall out of sync due to latency. Small scale is relatively easy (as easy as a several millions lines of code will get), and they undoubtedly due use it in several areas. But its in real time rendering where it needs more parallelization, yet is also the area where its the least tolerant to it.

Its a catch 22, largely uphill battle. But we pretty much have no other direction we can go with, since we've hit the practical limit for clock speed on silicon. Or Intel magically makes a multi-op per clock CPU that doesn't explode from the heat. Or finally comes out with an Alt Material based CPU I've hard rumored for several years now.

Or just Quantum Computers.

Ok guys so I thought that i had my 1070 selected but I had to go into the nvidia settings and select it for the physx settings. I booted the game up and maxed out settings and got a good 45fps. But like what happened when I first got this laptop, the game crashes after a few seconds. I started to walk around LA and it crashed lol. Is there some solution to go into the actual application settings in the nvidia control panel or should I turn the in game settings down?