Developer Spotlight – Meet Gpx1000!

Ghostchild@Seradon wrote:

There will be some changes to War Golem on the design side that should help with performance but will make the fight harder. The new code build should also be helpful so keep an eye on performance once the new code is on live.

Wigin wrote:

Interesting nugget there Wigin. Thanks!

Hi all, sorry for being MIA, little sister got married last friday and was out of office for the festivities.

I'd like to clarify a few things; bugs range to all sorts of things. The wireframe test in the original article was a place where the FPS dropped to 3-4 in a very specific location. This turned out to be an art asset bug; but took a while to figure out if it's art, code, or design that needs to change. Other times floors are transparent; this is usually caused by the camera believing it's outside when your character is inside; so art would need to change the collision such that the camera never leaves the indoor scene. Sometimes design wants to try using features that aren't quite ready for primetime and thus cause performance penalties which constitute other bugs. While performance bugs are just some of the bugs we have worked on, they are probably going to be the most noticable.
Also; it's important to note that I'm not the only programmer on Vanguard. Todd Schmidt and David Airmet are really kicking butt in getting things put together in intelligent ways; as a team we've made a lot of progress in a very short period of time. Among other changes, we moved from an engine that is mostly dynamically linked dlls to statically linked objects. This results in better performance for lower end machine at the cost of some performance to really high end machines (I have theories as to why; but we're probably not going to worry about a machine that does 105 fps vs 110 fps if the trade off is that machines that couldn't play, now can so likely to not spend a lot of time pursuing this).
When/if an area gets bad; please call it out. This is the sure-fire way to get someone to look at it. I can't promise to fix everything, but I sure can try.

Wigin wrote:

It's not just War Golem though. Sparkles, Karax, Slim, Shendu...they're ALL horrible right now but with Karax being UNKILLABLE due to the lag. The others we'll still have wipes due to the lag but will reattempt until we either get through it or just say screw it and have people log off upset. When you have the vast majority of your guild logging in to raid and you are faced with this on a nightly basis for 3 plus months now, you can't even imagine how hard it is to keep people interetsed to even log in let alone not bash SOE and this game to everyone they come in contact with! Patience has been a virtue we follow but your (SOE's not you Wigin directly) treatment of the playerbase is DEPLORABLE! If it was any other game title in the stable, it would have been fixed months ago.

Ashreon@Halgar wrote:

Multithreading is just one of several things we can do. There's even some next generation things that are in the same parallel programming vein that I would love to try. However, please remember, that more often than not; multithreading doesn't buy as much as one might expect. It's not a silver bullet, but is certainly a technique that we can further utilize beyond what we currently do.

Wigin wrote:

lets see, a handful of guilds can presently do war golem (even before the lag fest) and now the performance "may" improve but changes to the encounter will make it more difficult. Not to mention he still sits on top of Shendu and near Sparkles. And this is 'good' news.
how about a more definitive answer like we really did test for lag and can assure you it will not be a problem going forward before changing the encounter given that many raiders wont even log in for a war golem raid any more and karax is off our schedule completely.

Well what we've been trying to say for a while is that there was general lag introduced into the game about 120 days ago, there was a big soe wide 8 hour maintenance and after that every just went pear shaped badly. Im not sure what was done in that maintenance but it introduced all these ping issues we see now, we never had this before.
There were always fps issues with vg in raid settings, but we've always been able to work our way trough it, this is different however, the game just seems to lock up be the fps isnt dropping much. if you keep an eye on your roundtrip meter itll spike up to 3 seconds for one roundtrip. its new and it was introduced recently ( about 120 days)
There is also the huge drop in fps when a raid engages, i own a good gaming rig, i7 2600k 6950 ocz revo3x2 as ssd and 16gb ram. when i stand in some deserted chunk in full settings, i get between 100 and 130 fps. but when my guild engages zodifin, or arachnidon sunchine i have an avarage fps of about 20, with dips down to 1*'s. I have an creative xfi titanium in my system, but if i use it for vanguard i bluescreen within minutes so ive always used my onboard realtek chip for vanguard. By some mistake i ended up without sound and disabled it completely the other day. Even on my system, just doing this, netted me 25% more fps during encounters, i think someone should really look into this.
So be sure to stress test your tweaks and changes in a raid setting too if possible, many guilds including cobra are happy to have some gms or devs tag a long to see whats happening.
targets that generate bad fps:
arachnidon sunshine ( in shimmering shallows)
arch magus zodifin ( ceros isle)
Arch magus shendu, during the 42-25% "raid stage" ( Tomb or lord tsang) also always good for at least 2-3 people actually crashing.

Targets that have bad ping issues: ( all of them, but in particular)
war golem
karax woefether
zodifin ( was very bad today)

I'm not certain I know which maintanence you're talking about. I am going to be playing the game from outside the soe internal network in my downtime and will watch the roundtrip ping times with an eye towards what you're referencing.
Audio is definitely a subject area where I want to make some improvements. There being such a large frame rate jump for turning off audio is not a good thing. However, you shouldn't get a BSOD for a properly working audio driver. Do me a favor and try updating your system's OpenAL. http://climpxwss01.creativelabs.com...s/AllItems.aspx
The way that the audio within works is that we're using the audio stack as provided by the OpenAL API, which in turn sits on top of directX, which in turn sits on top of audio drivers, which in turn sits on top of windows kernel; which naturally sits ontop of actual hardware. At any point in that stack if something fails, then there is an opportunity for a BSOD. Switching to a different card just means changing the audio stack.
OpenAL is a Creative product as such they're usually the best in the industry at making OpenAL compliant devices; however, they've had problems with XFi Titanium on Win7 64bit. However, the fact that switching the audio stack solves the BSOD problem for you is an indicator that the audio stack beneath the game might have an issue.
I'm not entirely certain what the process is for getting a bug report from you through the SOE system to ultimately me looking at it; I am still kinda new to the process. However, if the driver update doesn't work for you, I'll ask how to get your bug filed through the system and we'll take a look at it as this forum is probably not the best place to do the problem solving back and forth.

Thanks for the reply gpx, ill see if i can address my bsod issue rather than avoiding it, and ill keep you updated.

regarding the maintenance that caused our current dispair, ive found the announcement for it. im not sure which one of these 2 it was but one was on tuesday and the other the day after.
http://forums.station.sony.com/vg/p...?topic_id=57039
http://forums.station.sony.com/vg/p...?topic_id=57023
The ping spikes were first complained about on the forums in this thread i believe :
http://forums.station.sony.com/vg/p...?topic_id=57073
Figure out whatever happened in these patches and your likely a lot closer to the root cause of the ping spikes. We have suspected /speculated they switched datacenters or hosting companies or something alike but stuff like that never gets comfirmed.
post from sillius about it, but it never went away:
http://forums.station.sony.com/vg/p...&topic_id=57065

Pleasure to meet you , GPX, never met any of the VG Devs team before when I played back in 2009 hehe. It would be nice if SOE or Forum Administration could setup a thread for you to work on major bugs players are experiencing. The only other feature I know of is the /bug command in game which is dictated by how well game wants to run that day until the latency issues are fixed. If we had something on here we could better get the information to you so you could track down the problem with coding faster than waiting until it reaches the coding teams level.
Perfect example of the system was with EQ2's Shader 3.0 Launch ( Still working out some kinks) the Developer on this project had players submit any bugs they had while on test server testing the 3.0 to include a picture, BI0s, and current drivers installed with description of what they were doing when bug happened, and how it affected them. Bugs were able to be tackled with 3.0 in 24-72 hours unless the dev was on leave. This response is the same with Dev's working with other parts of the game and I feel its what this game needs is an out of game method of assisting developer's with polishing Vanguard into the amazing game it can be. This doesn't mean whine and cry about things, this means as a professional, informing the dev over that part of Vanguard's inner workings of the existant bug.
As said many times this is neither the time nor place to discuss such things. Thank you for tackling the field you are working in and Best of luck to you!