Re: GLX in bad state

I do not have EverQuest and there does not appear to be a demo available 
for this game.  Hence, I cannot check anything regarding this game myself.  
This alone disqualifies it as a performance measurement for me.

I watched a video of EverQuest and it looks like the Half Life graphics.  
EverQuest 2 looks like the Half Life 2 type of graphics.  This is more the 
direction I am investigating.

>From your post I am getting the feeling that you are getting very low 
fps in Windows and this is not the case in Linux.  Can you quantify this 
difference in fps? 

You hint that by limiting the field of view in Windows you manage to lower 
the fps sufficiently to be able to move around.  This sounds to me like 
there is a rasterization bottleneck in Windows.  There are benchmarks 
which specifically target this functionality and can provide useful 
comparisons, e.g. the fill-rate test in 3DMark03.

Try running EverQuest like this:

   wine --debugmsg -all,fixme+ddraw EverQuest.exe

and describe the result you get in your next post.  It is possible that 
your DirectX setup is doing something that WineX is not doing, e.g 
Multisample Antialiasing.  Also look at your graphics and sound options 
for differences.

>From my point of view, if Windows lags behind Linux in performance then 
that would be something for the Windows people to worry about.  If you 
wish to impress them then find a benchmark which validates your 
assertions.

If you do not like benchmarks, I am not forcing you to use them.  People 
who object to them are generally people whose toils get summed up by a few 
unflattering numbers.  For many they represent a very good way to guage 
the performance of a system.

Please understand that I am not organizing a competition between WineX and 
DirectX, so you do not need to be dogmatic about the fairness of the 
rules.  Generally, I use my understanding of the processes involved in my
attempts at finding the problem areas when running a game.

On Tue, 13 Apr 2004, LaPeer, Jerry K. wrote:

> "Firsly, quantify what you mean by "plays better"."
> 
> Two examples,
> 
> The Bazaar.In WinXp I need to either look straight down or use first 
> camera view (top down view) to be able to move.In WineX3.3 (Feb Build) 
> I can move quite freely in the Bazaar in normal view.This is generally 
> associated with a line of sight lag issue.How much of that is 
> WineX/DirectX/Video Driver is way beyond my ability to design any test
> that could quantify "Why can I move in normal View in WineX but not
> DirectX, in the bazaar".The Bazaar is not a graphical nightmare.  It 
> is simply a place with a lot of players, that have a lot of items on
> their character.I am 100% confident the "lot of items" is the cause of 
> the lag.

In a multiplayer game the server would probably do most of the item 
tracking calculations.  I doubt that this is a CPU problem, unless the server 
resides on the same computer as the client.

> Jagged Pine Forest and/or anyLuclin Expansion zones.These zones are 
> much more graphic intense than the preceding zones.In WinXP I need to 
> look down at about a 45 degree angle to the horizon.This reduces the 
> overall view by enough to allow me to move, in an acceptable way, while
> maintaining an acceptable view of my environment.Now I choose Jagged 
> Pine as it is the opposite of the Bazaar.Meaning quite often, the zone 
> is totally empty.Just me, a lot of detail inanimate objects and some 
> animate objects (Non Player Characters).So to repeat I need to reduce 
> my field of view in WinXP, by a significant percentage.In WineX I do 
> not need to adjust my Field of View.I can move in JP as easily/fluidly 
> as I can in any of the original zones (considerably less graphics details).

Sounds like a rasterization problem, i.e. the video card assigns color 
values to pixels in video memory.  Less pixels - less work.

> Now how do you design a series of test to measure the "Exactly why does
> WineX give a significantly better play experience than DirectX".

Why would you want to?

> As far as using performance monitors, your talking to the wrong person.
> In my experiencethey are at best ...... unreliable and way to
> situational to be of any really value.

When you are sitting in front of your monitor to play a game, the 
situation demands the best performance.  Benchmarks test your situational 
performance.

> The only thing that I would accept is a program (in source format that
> can be compiled on both systems.Assembler being the preferred source 
> language as C/C++ is way to unpredictable as a cross platform
> measurement tool) that performs all of the DirectX calls in a plain
> vanilla way.Anything that would cause the OS to become involved in any 
> way, would invalidate any comparison, in my opinion.

Unfortunately, the OS is part of the situation.

> When it comes to WineX the Video Version vs DirectX the Video version,
> it isn't clear to me how one can uncouple the respective package from
> the installed Video Drivers. I maintain my systems with the latest 
> versions of everything I am using.I will use non WHQL versions of the
> NVidia Drivers on XP, if that is what NVidia the Company says is the
> most current.I make this statement to underline that my WinXP is as 
> current as I can keep it.My Linux is as current as I can keep it.  I 
> have zero doubts in my mind that the WinXP Video Driver from NVidia has
> a different performance profile than the Linux Video Driver from NVidia.

Video drivers are a part of the situation also.

Nadav Frum
----------