Pages: 1 [2] :: one page |
|
Author |
Thread Statistics | Show CCP posts - 10 post(s) |
Lord Zuku
The Military
|
Posted - 2007.04.25 11:23:00 -
[31]
Oveur, will it be possible in Trinity 2.0 engine to move UI parts that are now static (ship HUD, Overview, Solarsystem info, targets...) around?
|
TOGAKURE Daisuke
Occam's Razor Combine R0ADKILL
|
Posted - 2007.04.25 14:35:00 -
[32]
Edited by: TOGAKURE Daisuke on 25/04/2007 14:31:59
Originally by: CCP Oveur And how do you assume that will improve our current network layer? This is about the performance on the server and client of our own software, not internet topology and transport layers.
Of course in no way. I was just referring to a very small part on your blog where you mentioned that you've asked people to give you traces from their networks to measure performance and pl (...I think you originated from single AS at that point), and you relaxed your filters on single hosts at that time.
Now that I checked I can actually ping the login server (or whatever host the client initially connects), somebody clearly has more clue now :) Dunno when that happened but it's positive improvement. Seems to still lack a reverse dns though ;)
It's the small things that give me the idea about the general quality of your team there, from my real-life occupation and work history pov only of course.
|
Hunin
Minmatar EVE Empowerment League Navy Interstellar Alcohol Conglomerate
|
Posted - 2007.04.25 16:04:00 -
[33]
Originally by: CCP Oveur
Originally by: Nifel Will the new EVE client (or the current EVE client even) use RAM for cached items instead of the hd? Preferrably shared between multiple clients >_>.
I'm pretty sure it's already using RAM.
Some of the cached data might be in RAM, but definitely not all. I have seen some significant client improvement when moving the /cache folder to a RAMdisk. Most computers have more RAM available nowadays so the client can use more RAM for temp storage. Will you be moving more of the graphics/data located in the /cache folder to RAM?
|
OneSock
PLuSQuAMPERFEkT iNc
|
Posted - 2007.04.25 16:29:00 -
[34]
Are you actually going to introduce UI options to allow a user to turn off in game graphics ? for example, if in a blob fight, you might want to turn off all ship graphics and just leave the tactical overview and ship icons. We don't really need to render any eye candy unless we really want to. yes no ?
|
skydvejam
Minmatar Black-Mesa THE V I G I L
|
Posted - 2007.04.25 16:47:00 -
[35]
Originally by: OneSock Are you actually going to introduce UI options to allow a user to turn off in game graphics ? for example, if in a blob fight, you might want to turn off all ship graphics and just leave the tactical overview and ship icons. We don't really need to render any eye candy unless we really want to. yes no ?
I think that would be a great idea, might even help a lot of the client issues that the devs say the problem is.
|
Nifel
Caldari Reikoku Band of Brothers
|
Posted - 2007.04.25 19:39:00 -
[36]
Originally by: CCP Oveur
Originally by: Nifel Will the new EVE client (or the current EVE client even) use RAM for cached items instead of the hd? Preferrably shared between multiple clients >_>.
I'm pretty sure it's already using RAM.
OK, if so I'd like to change the question to why the RAMDrive thing (basically puts the entire cache folder of the EVE client in RAM) is so noticable in the client's performance under stressful situations o_O. It's really helped with speeding up loading times on jumpins, warpins, and generally has made things so much easier with the old client.
"When I die I want to die peacefully in my sleep like my grandpa. Not yelling and screaming like the passengers in his car." RKK Ranking: Sama |
Max Hardcase
Art of War Cult of War
|
Posted - 2007.04.25 19:45:00 -
[37]
Do drones require alot of network traffic to update positions etc ? eg Does a drone equal a ship for network traffic ?
|
Regat Kozovv
Caldari Deep Core Mining Inc.
|
Posted - 2007.04.25 20:37:00 -
[38]
Thanks for the update!
Regarding the new sound system, can you confirm wheither or not it will be using OpenAL?
|
Goumindong
Amarr Merch Industrial We Are Nice Guys
|
Posted - 2007.04.26 01:32:00 -
[39]
Originally by: CCP Oveur
Originally by: Nifel Will the new EVE client (or the current EVE client even) use RAM for cached items instead of the hd? Preferrably shared between multiple clients >_>.
I'm pretty sure it's already using RAM.
Not for a majority of called items/objects. It stores cached items in the "cache" folder in the eve directory on your harddrive.
This is why there was a large thread about moving your cache folder into ram via a software ram drive[which gives huge loading speed benefits]
---------------------------------------- Thou Shalt "Pew Pew" |
Hunter GlobaGateways
Caldari The Edge Foundation Zenith Affinity
|
Posted - 2007.04.26 01:55:00 -
[40]
Edited by: Hunter GlobaGateways on 26/04/2007 02:01:49 Hi there Oveur, I see this thread got your attention so I am hoping you can answer me in full here or in a series of new dev blogs in a bit more in detail than you currently do here in this blog.
The aim of this post is to ask a question that hopefully leads to a full understanding of how your technical software and hardware integration really are going.
If done correctly the effect for the players should be a super scalable and redundant server system, capable of scaling to xx xxx number of habitants all online one solar system if community of players wanted to achieve this. In effect the available hardware with the correct software has the architectural potential capacity to remove server lag. That is, if the software is written correctly, LAG should be a thing of the past.
Prologue: Its nice to see you are progressing on the client side of things here addressing it all with a major re-write to all code, more or less in all of EVE, whereas code that is labelled and mature gets moved and ported to C if needs be.
Now looking at your existing load balancing SQL servers and blade server farm, Today you have some Xxxx SQL servers and are running the server database on RAM drives, but the solar systems run on low cost blade servers, in a architecture to scale and load balance the star systems all over these blade servers. Knowing that C is fully multi threaded and also scale both sideways and up ways, in terms of multi core server CPU's, where there can be multiple CPU's in one CPU but also multiple CPU' housing units can be linked with Infiniband. An example that exist off the shelf is IBM X3850 that can house today 4 CPU's and each CPU core has in effect 2 CPU's. In a not to far future new CPU’s will enter from Intel that each have 8 cores and 4 of those can be expected to be housed in one server unit. Now today and since 2003 16 such server can be linked IBM hardware that gives massive bandwidth between the servers. That’s architecturally available today, so 16 servers, each of 4 cpu’s clustering sideways and up ways and scaling to be a total of 128 core CPU’s. Also knowing that new server entering the market this year, these new servers do link as well with Inifinand technology.
Now knowing that the scaling 16 * server as one working unit, the link technology have been available to purchase directly now from IBM since 2003, I am starting to wonder if we can expect to see a real architectural as a total re write or just a maturity code optimization to make what is there work better and more optimal but not architecturally re write the founding code for our bellowed solar systems and the software layers that runs this. Source
|
|
Hunter GlobaGateways
Caldari The Edge Foundation Zenith Affinity
|
Posted - 2007.04.26 01:56:00 -
[41]
As a recent dev blog Link hinted on, and talked some of was the issue of multiple Infiniband expansion options if you so choose to do it. Now this is where multi server load balancing comes on a massive scale. If properly written code in C, CPU balancing is not so hard to do these days. Next layer is to make the code work on clusters of servers and finally is the level of making the code act as one until binding all layers and having the needed I/O bandwidth between not only the CPU's inside the servers but also the needed bandwith between the servers to make all of that work together as one unit (pardon the mouth full)... Now Infini band lets you dot his today, technically the bandwidth 30 Gpbs time 30 have been available to do so now for about a year, and the founding technology is mature now for some years. Microsoft demonstrated the technolog in 2001, later launched a SQL server and the pulicly available libraries to do this in 2002. In 2003 a new world record of I/O operations of running load balancing and clustering database of a terabyte was publicly demonstrated. We are now in 2007. What I am trying to say here is, the technology in software to scale up as well as wide have been design proven for some years now. Also knowing that you are either having IBM to help you enter the Infiniband age jointly, or are or are you doing this work to create the working code and rewrite this yourself, or are you in effect bringing in the developer labs of IBM to joint solve this issue together with the developers of Python and porting your code to a re write in C where python don’t work for you.
Knowing you are now addressing this and probably lots of these issues, in this years need for speed initiative going this year, can we please get a full and detailed architectural explanation of what can be done, what will be done and when can we expect to see this or part of this solution affect the general EVE users, while at the same time make sure the star system scaling limits per solar system, where today’s current limitations are detailed out, and also stipulated when the new techno this represents will be technically developed and become available. What the community want to know for them I believe it is most important issue of solar system and node scalability and constellation scalability and reliability. I think CCP are at the same time best served to inform all of EVE, and not limit these details only a certain limited and close group of people who fills certain star system prior to others.
With this publicly available to all Eve players, I am sure all benefit and can expect to better understand how and when to plan for their joint attacks against one enemy and what are the limits of today’s and the future expectations of your architectural re write will represent for the EVE users and I am sure you yourself can understand the 0.0 political landscape and the delicate balance that exists today that may or may not be the result or a factor of the current system limitations.
Cheers then Hunter
|
Solbright
|
Posted - 2007.04.26 11:21:00 -
[42]
NOOO! Bugger off to one of those sleezy server disoriented threads! :P
|
Linda kays
Gallente University of Caille
|
Posted - 2007.04.26 12:42:00 -
[43]
Thourgth I'd hand in a few cents here. In my book client optimazing would involve a "split/port". It all nice with all the pretties ingame, but once in while I'm using a really really old laptop to play on. and with 14.1" screen size the graphics aren't really what I'm looking at. I keep a watch on the IRC chat channels I'm in and do very basic commands. Like mine here, target that from the overwiev. Stuff like that wher I don't really see the ships, station enviroments or pretty much anything that isn't text based. I'd really really love to see a client release that is cut down to the very basics, and demands CPU power over Graphics. I'd even love to se a client where I could log into the game but not leave the station I'm in. Make this for me and I'll promise to use it, I even think other people will.
|
Xenofur
Di-Tron Heavy Industries Freelancer Alliance
|
Posted - 2007.04.26 15:05:00 -
[44]
Originally by: Linda kays I'd really really love to see a client release that is cut down to the very basics, and demands CPU power over Graphics.
actually, the current client does use more cpu than gfx power. it's simply badly written and doesn't let the gpu do nearly enough work, while being also badly written in that way too many things can lock the client instead of running asynch.
|
Nate D
New Atlantis Tek Corporation
|
Posted - 2007.04.27 00:10:00 -
[45]
Whooo! GO OVEUR!
-Nate
A New EVE Voice In Game Channel |
Nyphur
Pillowsoft
|
Posted - 2007.04.28 10:19:00 -
[46]
Any chance you could just change drones so that they look like 5, 10, even 35 (for the moros) normal drones but they act a single entity with a single HP reserve? It'd reduce lag further than a 5-drones maximum, which I always thought was a poor solution.
The moros looks dumb with 5 drones :/.
Eve-Tanking.com - For tanking spreadsheet and resources. |
Jim McGregor
|
Posted - 2007.04.30 17:13:00 -
[47]
Sounds good to me. --- Eve Wiki | Eve Tribune |
j0sephine
Caldari Reikoku Band of Brothers
|
Posted - 2007.05.04 15:03:00 -
[48]
Mhmm it's a bit disappointing to read the UI code optimization is something that's only starting to being looked into now (it was sloppy since its creation a few years ago so it certainly took some sweet time to go about this framework re-design et al) ... but it's good to hear it's at least being worked on ^^
(although knowing it's likely a year --projected implementation date plus obligatory delays-- before we actually see it... well that stings a bit. But guess there's no way around it :/
|
Rektide
|
Posted - 2007.05.10 19:59:00 -
[49]
so we have to upgrade to that gigantic heap of garbage vista? afaik dx10 is vista only. cop out.
|
Solbright
|
Posted - 2007.05.11 13:47:00 -
[50]
No, CCP Oveur - "We will utilize some new spiffy shader stuff, but with a fallback path to DX9"
|
|
Silc'n
|
Posted - 2007.07.08 02:08:00 -
[51]
Greetings pilots, greetings devs,
I have a question, I'd appreciate an answer from the dev team or anyone who knows...
Rewrite the whole thing in c? Wouldn't it be just easier to develop a new python compiler that generates optimized code?
Thanks!
|
|
|
|
Pages: 1 [2] :: one page |
First page | Previous page | Next page | Last page |