I havent checked into those motherboards UnskilledPRO suggested yet (after all I dont have the cash at the moment) but I have been interested in something else lately.
I'd like to focus attention on the fact that Gateway seems to run perfectly smoothly (unlike Deck that is unplayable). Where is that difference coming from?
Trying to reserch this matter, I downloaded a few "smaller" maps. For instance, 1on1-outOfGas, and 1on1-Instinct. I also downloaded Campgrounds. All three of these maps run perfectly. I even went ahead and raised my settings to world/texture detail of 3/3, 1280x1024 resolution. They still run great, over 30fps most of the time. Heck, I even got more than 100fps at some point (even though that is very rare and I have also set 60 as max fps last time I checked).
Now, in campgrounds, I was surprised at how detailed the map was (reflections all over the floor etc.) and that it would run so well, but the match I was playing was me+2 other bots free for all DM. So I played another one and set bot number to max on the sliding bar. I set that to 31 bots, but I think the map caps that to 12 bots max. Anyway, with that many bots, when on the central area where the UDamage spawns, I had noticable framerate drops most of the time, all the way down to 18fps and even less on some occasions.
This is unlike what happens in Deck which is more or less unplayable even with no bots at all.
So, isnt this at least a little suspicious? What seems to be the main bottleneck that is causing this performance drop? Do you think that the video card is insufficient (wait a minute, its not that old a card and it cost a fortune too) or is the computer out of RAM or is it in need of more CPU punch?
Anyone know of any utilities I can run while playing ut3 (they should therefore be light on resources) that keep a log or a chart of cpu and memory usage during my game? Using such a utility I might be able to pinpoint precisely what needs upgrading the most Since the minimum requirements of UT are a 2.0 Ghz single core cpu, my 3.2 Ghz cpu *should* be enough to run it at a playble fps.
By the way. I have found that the difference between "Lowest possible settings at 1024x768 resolution by Brad G's Tweaker utility" and "World/Texture detail at 3/3 at 1280x1024 resolution" is almost unnoticeable! I'd say its at most 5fps or less. So my 2 cents are, this is not about my video card is it! Ok, so is it the cpu or the ram then.
Oh, by the way. What do you recommend I set my AGP RAM aperure size to? I learned by googling that the size is how much system ram should be used to help the video card with its tasks. Maybe I should set it to 0, since my system ram is only 1GB but the video card's 512MB are certainly up to the task?
Edit: Hmmm I run into some interesting findings tonight. Looks like I might make it with my current system after all. I may even be able to get away without even increasing my RAM! The procedure I have followed just might end up being a good help for the community because it can pinpoint your systems bottlenecks effectively as long as they are not VGA-card caused, or so it seems. I will make a detailed post with my findings tommorow
~~ *** Kinnison's Guide to benchmarking your PC's gaming performance with Windows XP Built-in tools *** ~~
Allright, here goes. After googling and googling, I ended up not being able to find any utility that will keep a log of my cpu and ram usage for a relatively large period of time, while I play UT3. But I learned thatthere is a way to do that just with Windows.
Go to Control Panel->Administrative Tools->performance,
a window showing current system info like cpu % usage , disk I/O operations and memory use comes up. On your left hand side there are several options, one of which is for "counters log files" or something similar (sorry I have Greek WinXP version). Choose that and a "sample log file" or some such appears. That thing is actually a "template", a setting that determines a specific combination of system data to keep track of. That settings file also includes the option to create a log file on your disk (at a folder location of your choice), and to have it record the values for the chosen system values at set intervals, default should be 15 seconds (which seems reasonable for unreal....).
Since that is a sample file, right click anywhere in the blank area near it and choose "new log file settings..." or some such. You will have to enter a name to identify your file, choose something obvious like "Ut3 benchmark log" or something.
Now a window will pop up where you'll put in all the info that tells Windows exactly what and how you want it to log in your PC. First, on the top line there will be a file name. That shows you, what the created log file will be called when you next set windows to log your info.
Right underneath that one there is an empty list of counters and under that 2 buttons. Click on "Add counters" or "add meters" or something similar but not the "Add objects" button. When you click add meters you get to choose where to get the info from: your current PC or another one on your local network, if you wish. Then there is a drop down menu with various parts of your computer that windows can "benchmark": processor, memory, physical disk, and others. Choose what you are interested in (*HINT: since it can be tedius loading up UT3 each time and playing through maps just to get measurements, if you are not sure if you need a certain measurement, just add it anyway. You will probably get a 1-10MB log file -worst case scenario- for every 10-15 minutes of UT3 play. The more counters you add the larger the log file will be*). Chooce what you need then click "Add". You wont get any indication that its been added, but you will get a warning if you try to add the same thing twice, so its ok. keep adding until you are satisfied that you'll get a good idea of how your systems various parts will be faring, and then hit close.
Now you are back to the config files' settings window. Under the list of the added counters and measurements there will be a setiing for the time interval in seconds in which to take measurements. I went with the default of 15secs but it might be better to use something like 4 or 5 seconds instead. Or even 3 just to be sure.
Move on to the next tab of this window should be named "Log Files" and you can chooce how windows will vary the filenames of your created system performance log files with each new log file created. I recomend manually moving your log files each in each one folder every 2 benchmarks and adding in there a text file where you mention what you where benchmarking, just so you wont loose track of them in the end.
The final tab "Scheduling" or similar, is super important and you do not want to skip it.
You can set Windows to automatically start and stop the logging sessions at set times of day ,but I recomend you forgo all scheduling and set everything to "not automatic (use shortcut menu)" or some similar option. That means that you will have to manually issue a command to start and to stop each and every one of your log files. You need to do that so that you can then compare the different log files that respond to different tested conditions.
When you're all set click ok and your settings for the log files are saved. Now you can just right click on the settings and choose "Begin". As soon as you do, Windows will create a log file at the set place and will add entries to it as specified by your chosen timeframe. You can now close all windows, including administrative tools and Control Panel. It will continue to log your chosen info until you open again the performance tool (Control Panel->Administrative Tools->performance) and right click your conifguration settings and choose "stop".
This is the time where you launch your UT3 (or other game) and play as normally. Make sure to write down your game settings, like what video options you were using (resolution, detail settings ..) and what you where playing (ie. a team deathmatch on map Biohazard with a total of 11 bots as well as myself). I highly recommend playing only one match. You might want to make it a long one (with high frags and time limits) so that it lasts more than 5 minutes to get, perhaps, more statistically sound measurements. But if you wish you can also do away with a 5 min session. When you are done just exit UT3 (or even Alt+Tab if you like) open Control Panel->Administrative Tools->performance select your log configuration file, right click it and choose "stop". Now you ve got a detailed report on how your machine did while you where playing. You did note exactly what settings, map etc. you were playing with, right?
After that, you can at any time go to "System Overview" or "System watch" or whoever its called, that normally shows the current values of your PC, and just above the graph there is an icon of a cylinder. Click it or alternatively press Ctrl+L.
Yet another window appears with radio buttons: "Source Log :"or some such followed with options "Current Activities", "Log files", "Data Base" or similar. Choose to view data from Log Files and open the log file that was created while you were gaming. The collected data will appear in the form of a chart. Right underneath it are measurements that you have chosen to appear, out of the recorded ones. This means that you need to right click on the chart, choose "add counter" or "add measurement" or whatever, and choose, for each part you were monitoring, "add all". So you will have to do this multiple times, for cpu, ram etc.
This will make all the collected data appear on the chart.
There is a colour code to distinguish the measurements but it will probably be more convenient to click on the light bulb icon on top of the chart and then click the measurement you want to have it appear in a bolder line on the chart. This way they are easier to distinguish.
Now, most of that data fits on the default axis which goes from 0 to 100. So both proccessor usage as a % time percentage and, say, disk %time usage fit on that scale. If some of your measurements are not percentage ones, you will have to change the vertical axis scaling so that you can see it. For instance, using the "remaing memory MB" measurement, you will probably have to change the axis to 1000 or 2000 (1000MB=1GB) so that you can see how much memory you have left. in this case, a "memory left MB" measurement of 400 would mean your PC has 400MB left of ram. The lines move around as time passes, depending on how the values change.
Changing the axis values is no big deal, just right click the chart, choose properties, then on the "graph" tab there are 2 text boxes to type in maximum and minimum values for the vertical axes to use. Change this to be able to better see your chosen measurement. Then if you move to another measurement change it again.
Repeat the whole procedure with different settings and in different in-game situations. Take separate log files for each situation and then compare them. So If in one map you get low fps ingame (you can see that by playing or use the "stat fps" command on the unreal console) log it as mentioned above and then see whats the problem. Is your CPU constantly hitting 100% work load? Are you running out of ram? Is the system writing and reading to the hard drive like there's no tomorrow?
This should be at least indicative of what is your bottleneck. Better than upgrading blindly and hoping it will help.
*with an AGP Aperture Size of 128MB*
with ut3 settings: 1280 x 1024 resolution
Texture/World Detail set to 3/3
i played a duel on custom map 1on1 instinct vs a "skilled" bot.
the graphs seem to show almost stable cpu % usage
that is at average around 68% while playing the match
and Ram available seems to be around 160MB.
while playing the fps where adequate to play, typically over 30fps.
Then with the same settings I took another log file while I wa playing a deathmatch on Deck, with me and 10 other bots.
Cpu is again appearing to be stable around 66% (yeah its less) while playing the match
RAM appears to have 80MB available.
FPS where almost unplayable while in the central bridges area of Deck and they where generally horible most of the time.
This seems to be a lack-of-ram issue.
I will take more benchmarks with different AGP Aperture sizes. I expect that chaning agp aperture size shifts your system ram to be used for the agp card instead of for the cpu. Supposedly, reducing agp aperture size (if we suppose that my vga cards 512MB are enough) will free up system ram allowing UT3 to overcome the insufficient memory bottleneck that kills fps on bigger maps like deck. I will post when I have the next set of measurements.
AGP Aperture should be the same size as the video card's memory.
UT3 is unlike every other game I've played, it hardly uses any ram at all, firefox uses more ram than ut3, it doesn't max out 1gb.
If you don't care how bad it looks ut3 can run on anything with good framerates
AGP Aperture should be the same size as the video card's memory.
UT3 is unlike every other game I've played, it hardly uses any ram at all, firefox uses more ram than ut3, it doesn't max out 1gb.
If you don't care how bad it looks ut3 can run on anything with good framerates
hmmm is that ram as in system ram or ram as in video-ram ? if that is so it would indeed make a huge improvement to put agp aperture higher. BUT! my motherboards BIOS appears to only be able to increase it up to 256MB....
*bump* !!!! I am surprised how quickly this thread dissapeared into the neverending forum pages. I guess there are a lot of people struggling with their pcs lol Anyway, I havent taken more measurements yet, but still feel free to post any ideas, comments suggestions, etc. if you people feel like it
as noted in the other thread I'm interested in results.
Hopefully I can figure out the Performance monitor. So that i can get detailed results.
Thanks for the how-to!
Comment