Multiboxing Hardware Science!

[MiRai]

Instead of continue to derail the other thread where this topic began (http://www.dual-boxing.com/showthread.php?t=42416&page=3), I figured I would just open a new
thread to keep things on track.

A quick recap:
I did some basic initial 10 boxing testing with some trial accounts during non-peak times:


zjEk-e-_zLg
http://www.dual-boxing.com/images/fenril/tenboxtest-full.png



People wanted to know more so I thought I would give them more! I jumped on last night during peak times
and tweaked some video setting macros and this is what I whipped up:



L7Pw96TZz0w


YouTube description:
All Windows @ 1920 x 1080 (Custom Video Settings)
Using ISBoxer (Increased swap times due to my window swapping macro!)

2600K @ 4.6GHz [Venomous-X Cooler]
EVGA GTX 580 3GB
Intel 510 SSD

Finally got a chance to hit Stormwind at peak times with this 10 boxing test setup to see what I should be
expecting performance-wise and to play around with my video settings to get tolerable FPS. I had to use
FRAPS's DWM recording feature to capture the Windows Task Manager, GPU-Z, RealTemp, and ISBoxer's
Video FX. The DWM recording feature adds a large amount of CPU processing overhead while in use so I
decided to cap the recording to 30 FPS -- I felt this gave more accurate readings across the board instead of
using 60FPS and having the FPS erratically jump around.

If you're considering 10 boxing World of Warcraft this was just an initial test to see where my hardware
stood performance-wise. Actual PvP and PvE raiding may require additional tweaking or lower settings than a
simple major city walk through and I obviously cannot test either of those at the moment... maybe one day.

Note: The big words "File Change" (in both videos) are for those who are curious where the FRAPS files split
themselves. Some people experience frame rate dips during the file creation and I was trying to show all
angles of performance here.

Both videos are recorded and rendered in a resolution of 1920x1080 and should be viewed at that
resolution in full screen if you want to see everything that was happening.

This discussion is open to any and all that feel the need to ask any questions about 10 boxing or whatever.
Sam was most interested in trying to decode certain things with VRAM and what-not but it'd be nice to see
some 10 boxers post their specs and add to this.

This thread is not bound to World of Warcraft and anyone who has useful information on other
games (i.e. RIFT, EQ, etc.) should feel free to post as well. I do ask that if you're going to post
about hardware-related information that you at least provide some actual benchmarks
(screenshots and/or video as I have done throughout this thread) pertaining to your system as
opposed to just stating what you think you might be getting.

However, this thread is bound to legitimate game servers (i.e. no private servers). Anyone who
posts any benchmarks or tests in this thread that come from private servers will have their posts
deleted without warning. In addition to private servers being against the rules of discussion here
on this message board, private servers can run modified, altered, or old patches of content or
code which can easily skew any real and up-to-date information.

QUICK LINKS TO MORE TESTS THROUGHOUT THIS THREAD:

World of Warcraft Related:
Orgrimmar Test (http://www.dual-boxing.com/showthread.php?p=336615#post336615)
2600K Hyper-Threading Test (http://www.dual-boxing.com/showthread.php?p=336875#post336875)
World of Warcraft SLI Test (http://www.dual-boxing.com/showpost.php?p=339227&postcount=40)
30 Boxing (Not In A Major City) (http://www.dual-boxing.com/showpost.php?p=341846&postcount=48)
20 Boxing - 2 Laps In Orgrimmar (http://www.dual-boxing.com/showpost.php?p=342006&postcount=51)

[Alge]

Nice work!

Still, it's a pity you didn't buy a GPU with enough RAM :)

[MiRai]

Nice work!

Still, it's a pity you didn't buy a GPU with enough RAM :)
No kidding, I'm a sucker for good looking video settings though. I need one of these (http://www.newegg.com/Product/ProductList.aspx?Submit=ENE&N=100007709+600007787&QksAutoSuggestion=&ShowDeactivatedMark=False&Configurator=&IsNodeId=1&Subcategory=48&description=&Ntk=&CFG=&SpeTabStoreType=&AdvancedSearch=1&srchInDesc=) to perform further tests
for comparison but I don't exactly want to drop $800 on a video card for a "test" and the built-in Crossfire will
most likely hinder test results.

[Sajuuk]

Pheh. Pop another 580 3GB in and rock on.

[ebony]

the 2nd video peektime on your realm?

[Ualaa]

I like the video feed use.

Stormwind doesn't seem that busy, for a peak hours.
I'm used to far more people crowding its streets.

Ten boxing is even neater than five boxing.

[MiRai]

Pheh. Pop another 580 3GB in and rock on.
I've never been able to use the second GPU's VRAM while attempting to split the load between cards. If
someone has been successful in doing so I'd like to know (A GPU-Z screenshot would be great).


the 2nd video peektime on your realm?


I like the video feed use.

Stormwind doesn't seem that busy, for a peak hours.
I'm used to far more people crowding its streets.

Ten boxing is even neater than five boxing.
The time in the video is 23:33 which would have been 21:33 server time so it could have been a bit
"peakier" so to speak for a week day. Orgrimmar's population is much more concentrated compared to
Stormwind's... perhaps I should have created a Horde team instead (I like my Human Females, though).
Kil'jaeden is listed as "High" almost 24 hours a day and we have a very good H:A ratio compared to most of
the High populated realms, although Horde does have more among their ranks.

It is funny you mention it, because when I went over towards The Canals I was going to head over to the
second Auction House and see how many people were over there; but, I decided I didn't want to walk that
far and put a transition in the video if the venture wasn't worth it.

[Sam DeathWalker]

Excellent. I am amazed that you can render 10 screens at the full resolution with a single gpu, although a very good one.

Clearly the 3G option is paying off as well.

Your switching is still a bit (not much though) worse then my alt/tab method but then again you can see all 10 of your characters whereas I can only see 6 of mine at a time (one per computer). But I can see in full screen and don't have to pip in and out to do things with those characters. But my system requires 6 1920X1200 monitors whereas yours only needs one or two.

Ya well I guess there isn't much more to be said, for 10 boxing the 580 3G is the best (and I don't think that duel gpu is going to help you that much in wow, nor will duel cards).

1 Box: 460 2G (might not get all effects though)
5 Box: 6950 2G
10 Box: 580 3G

As you can see with you superior equipment you did much better then me (but there are more people in Org):



Spin / Fly around Test (all about the same) Nvidia440 3GVideoRam 5 Clients (4 in background on lowest options); I play full screen 1920 X 1200.

5 Clients: Low Options: 13fps Ultra: 9fps

1 Client: Low Options: 27fps Ultra: 14fps (63/361 ms ping)



Spin / Fly Around / Stationary Test (all about the same) Nvidia460 1GVideoRam 5 Clients (4 in background on lowest options); I play full screen 1920 X 1200.

5 Clients: Low Options: 22fps Ultra: 14fps

1 Client: Low Options: 35fps Ultra: 19fps



Its fairly clear that going over 10 boxing on one computer isnt going to happen, as you were close to the 3G max as it is.


I notice that the GPU and Memory controller loads are relativly low, meaning it might be possible to get by with a bit weaker gpu (although we need to know if you were runniing full ultra in the main), if you can find one with 3G. And it does seem like the CPU is getting a real workout as well so maybe CPU needs be stonger then what I would have thought.

[MiRai]

Excellent. I am amazed that you can render 10 screens at the full resolution with a single gpu, although a very good one.
Too many people are giving me crap about peak times for the Alliance that night so I've decided that I'll create
10 new Horde tonight and take them into Orgrimmar... this should be exciting. :)


Clearly the 3G option is paying off as well.
I am definitely happy I went with the 3GB route.


Your switching is still a bit (not much though) worse then my alt/tab method
I am going to do my best (depending on Orgrimmar lag) to adjust my settings accordingly to show a nice quick
swap instead of that delayed one. It's mainly the Texture Resolution and Shadows that I will need to adjust.

[Sam DeathWalker]

Ok after spending the last couple of hours looking up video cards it seems there is little choice for 5 or 10 boxing.

http://www.tomshardware.com/reviews/best-gaming-graphics-card-geforce-gtx-590-radeon-hd-6990,2879-7.html



1st; Discrete: GTX 580 Discrete: HD 5970

2nd: Discrete: GTX 295, GTX 480, GTX 570 Discrete: HD 4870 X2, HD 6970

3rd: Discrete: HD 4850 X2, HD 5870, HD 6950


Given that few are 2 or 3G (some are 1.5G but that seems insufficient) you choices are:

5 or 10 Box: 580 3G (5970 is a dual gpu card)

5 Box: 6970 2G or 6950 2G (or the more expensive and harder to find GV-R5876P-2GD may as well get the 580 3G instead of this)

1 Box or Budget: 460 2G


Here's the cheap (off brands) on the 580 3G $539:

http://www.overstock.com/Electronics/Zotac-ZT-50103-10P-GeForce-GTX-580-Graphics-Card-772-MHz-Core-3-G/5968635/product.html?cid=123620


6950 2G $299:

http://cgi.ebay.com/HIS-ATI-Radeon-HD6950-2GB-DDR5-2DVI-HDMI-2x-Mini-/280684416732?pt=PCC_Video_TV_Cards&hash=item415a184edc


:eek: $250

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=230633058098&ssPageName=ADME:B:SS:US:1123

[MiRai]

You want Orgrimmar? You got Orgrimmar @ 7:42PM Friday (http://www.youtube.com/watch?v=CD2LRROpph0) night.


xWDQTbXtVlc



10 Boxing Test in Orgrimmar on Kil'jaeden.

All Windows @ 1920 x 1080 (Good Settings)
Using ISBoxer

2600K @ 4.6GHz [Venomous-X Cooler]
EVGA GTX 580 3GB
Intel 510 SSD

You didn't like Stormwind on KJ, so I give you Orgrimmar. Again, I had to use FRAPS's DWM recording
feature to capture the Windows Task Manager, GPU-Z, RealTemp, and ISBoxer's Video FX. The DWM
recording feature adds a large amount of CPU processing overhead (~30%) while in use so I decided to
cap the recording to 30 FPS -- I felt this gave more accurate readings across the board instead of using
60FPS and having the FPS erratically jump around.

If you're considering 10 boxing World of Warcraft this was just an initial test to see where my hardware
stood performance-wise. Actual PvP and PvE raiding may require additional tweaking or lower settings than a
simple major city walk through and I obviously cannot test either of those at the moment... maybe one day.

The big words "File Change" are for those who are curious where the FRAPS files split themselves. Some
people experience frame rate dips during the file creation and I was trying to show all angles of performance
here.

For the window swapping portion of the video I spun my mouse wheel like a madman to show that window
swapping was super fast and instant as it should be. Then I showed it in a much slower manner. My last test
included a custom window swapping macro that increased swap times, this video did not.

Yes, I realize that some of the Video FX on some of the windows are missing. I had a sync issue that I didn't
feel like manually fixing while FRAPS was eating away at HDD space.

[Sam DeathWalker]

Well you are at good effects, and 30 fps max but still the gpu dosnt seem to be stressed to much.

You do drop to 20ish fps when loading a lot of new characters (like when you first run out of the enterence). And it is far from clear if you would get more then 40fps if you were unlimited.

But no matter there seems no lag or stuttering and the game is very playable at anything over 20fps and even down to 10fps, as long as there is no lag or stuttering.

I didn't take into account that your cpu had to do fraps and stuffs so ya the cpu seems more then strong enough for the job and any fps loss is on the gpu I would think.

Ya the switching is excellent, as good as mine. Mine dosn't mess up the screen though but thats nothing. It is really amazing that you are rendering 10 screens at 1920 X 1080. The 9 alts are also at good efffects?

I really don't think I miss anything by not tileing, maybe you drop aoe targeted spells a lot better though, but outside of that the screens are just to small to be really usefull (even considering you put the 9 on another monitor). I would think for 4 though tiled on a seperate monitor you could see a lot more.

Still overall there is no question you can easy 10 box with your system, and that you have the best video card at under $550 for the job. Any more spent would have just given way to small return for the $.

[MiRai]

Well you are at good effects, and 30 fps max but still the gpu dosnt seem to be stressed to much.

You do drop to 20ish fps when loading a lot of new characters (like when you first run out of the enterence). And it is far from clear if you would get more then 40fps if you were unlimited.

But no matter there seems no lag or stuttering and the game is very playable at anything over 20fps and even down to 10fps, as long as there is no lag or stuttering.

I didn't take into account that your cpu had to do fraps and stuffs so ya the cpu seems more then strong enough for the job and any fps loss is on the gpu I would think.

Ya the switching is excellent, as good as mine. Mine dosn't mess up the screen though but thats nothing. It is really amazing that you are rendering 10 screens at 1920 X 1080. The 9 alts are also at good efffects?

I really don't think I miss anything by not tileing, maybe you drop aoe targeted spells a lot better though, but outside of that the screens are just to small to be really usefull (even considering you put the 9 on another monitor). I would think for 4 though tiled on a seperate monitor you could see a lot more.

Still overall there is no question you can easy 10 box with your system, and that you have the best video card at under $550 for the job. Any more spent would have just given way to small return for the $.
Yeah, instead of tweaking the video settings to get the very most out of them I figured it would be easiest to
just leave them at a set standard. The initial loading of all the people in Orgrimmar and Stormwind are always
big hits but everywhere you see the video at a solid 30 FPS I would normally be getting anywhere between 30
and 60 FPS. The slaves are actually capped at 15 FPS so I could even lower those 9 other screens down to 10
and essentially free up 45 FPS of CPU/GPU for even more main window performance.

There is some stuttering from time to time but I believe it's because the VRAM and CPU are completely maxed
and the GPU is having to swap textures around so frequently. If I wasn't using the DWM recording feature, or
even recording at all, the CPU would have much more headroom.

As for swapping, the screen only really gets messed up when I swap at lightning speed. I have a G700 mouse
and it has the ability to switch whether the mouse wheel is completely free spin or whether it's "clicked." I put
it in free spin mode and just spun it as fast as I could so the windows are swapping before they've been fully
re-sized. When I slowed down the swapping it looked much cleaner. Also, to confirm, the slaves are at good
with shadows turned down.

[Umbaalo]

Thank you for your donation in the name of multiboxing science!

[MiRai]

Thank you for your donation in the name of multiboxing science!
You bet. There has been a lot of talk recently about how powerful the current Sandy Bridge chips could be and
I thought I would just give those theories an initial stress test in World of Warcraft. Perhaps when I get a
break from school here in the next few weeks I can bust out some RIFT trials and see what I can do with that.
Trion might actually have add-on support out by then which could make testing more interesting... we'll see.

[Umbaalo]

Sounds very interesting! I'm really kicking myself over not waiting for the sandy bridge chips and started building my rig last fall. I'm happy with it, but I could have saved quite a bit of cash knowing what I do now. I guess that's the nature of the beast though when it comes to technology.

I've only pushed my i7 980x to 4.1GHz... But I always use a lot of caution with my hardware.

[MiRai]

Given that few are 2 or 3G (some are 1.5G but that seems insufficient) you choices are:

5 or 10 Box: 580 3G (5970 is a dual gpu card)

5 Box: 6970 2G or 6950 2G (or the more expensive and harder to find GV-R5876P-2GD may as well get the 580 3G instead of this)

1 Box or Budget: 460 2G
EVGA actually offers a GTX 560 2GB SSC (http://www.evga.com/products/moreInfo.asp?pn=02G-P3-1469-KR&family=GeForce%20500%20Series%20Family&sw=) and a GTX 560 Ti 2GB (http://www.evga.com/products/moreInfo.asp?pn=02G-P3-1568-KR&family=GeForce%20500%20Series%20Family&sw=). It would seem that although the Ti is
technically better than the non-Ti 560, because it's an SSC card (stock factory OC'd) that it may be on par with
the big brother Ti version. Anandtech does a review (http://www.anandtech.com/show/4344/nvidias-geforce-gtx-560-top-to-bottom-overclock) on Asus's GTX 560 DirectCU II (http://www.newegg.com/Product/Product.aspx?Item=N82E16814121445) and it is also a factory OC'd
(slightly faster than EVGA's OC) card that seems to hold it's own against the Ti cards.

While I was looking at Newegg I realized that Asus's card is cheaper than both of those EVGA cards and would
perform better than them, but... it only has 1GB of VRAM. Guess that's where the price increase on the EVGA
cards comes from.

However, Tom's Hardware (http://www.tomshardware.com/reviews/nvidia-geforce-gtx-560-ti-gf114,2845-12.html) shows that the GTX 560 Ti is slightly better than the 6950 2GB. So, for anyone
looking for the nVidia equivalent of the 6950 2GB, EVGA's GTX 560 Ti 2GB would be it.

[Sam DeathWalker]

Good call, seems $50 cheaper then the 6950 also.

http://cgi.ebay.com/EVGA-GeForce-GTX-560-2GB-GDDR5-SC-Video-Card-/320712570431?pt=PCC_Video_TV_Cards&hash=item4aabf5763f

Well $30 after shipping considered but still it seems a better deal:

http://cgi.ebay.com/NEW-RADEON-HD6950-PCIE-2GB-DDR5-CTLR2PORT-DVI-MINI-H-/380346853762?pt=PCC_Video_TV_Cards&hash=item588e706982

[MiRai]

There was a prior discussion (http://www.dual-boxing.com/showthread.php?t=41394) (starts around page 3) on Intel's Hyper-Threading here on the forums and I just wanted to do my own quick test on a 2600K before my trials ran out on this 10 box setup.

Video settings were identical from my 10 boxing Orgrimmar test run (http://www.dual-boxing.com/showpost.php?p=336615) during all the screenshots taken. FRAPS is not running in this test and WoW is the only application which is using any CPU in all screenshots. Games were launched through Inner Space after a fresh system restart.
Orgrimmar [Tunnel] [HT = Disabled] (http://i.imgur.com/D3kgT.jpg)


45% CPU Usage


60 FPS


Orgrimmar [City] [HT = Disabled] (http://i.imgur.com/hydcN.jpg)


100% CPU Usage (Pegged)



20-27 FPS
Orgrimmar [Tunnel] [HT = Enabled] (http://i.imgur.com/wJWPg.jpg)


15% CPU Usage


60 FPS
Orgrimmar [City] [HT = Enabled] (http://i.imgur.com/tcV05.jpg)


75% CPU Usage



40-45 FPS

I also did a quick check to see how much View Distance on the main client affected FPS and it did not see to be that large of an impact. I was really just double checking to make sure that View Distance did not affect the distance at which characters could be actually seen. I do realize that if the View Distance setting actually affected this it would be a major dis/advantage for people depending on the machine they were using to play.

I always advocate that Hyper-Threading does not have a large impact on increasing gaming performance and that it has, at times, shown to actually decrease performance in certain applications. It does, however, happen to show a nice performance increase while 10 boxing World of Warcraft.

I have yet to do a 5 boxing test.

[honeypot2011]

great to see this and interesting to see the limits..good for anyone in future wishing to try to 10 box..

one thing..have you tried recording with a differant screen recorder and seen the results?say camtasia

the reason i say this..while ago now i rember a certain multiboxer streaming and recording using fraps..but he found that

it actually locked the wow fps to 30 fps when recording...of course this may have been an old version..and if there was an updated version this may not happen now..

was just curious..anyway great job :)

EDIT : ignore the 30 fps thing..probly should have watched your vids first :P

[Bollwerk]

I'm honestly surprised that hyperthreading helped, especially by such a large margin. Maybe hyperthreading on the new procs is MUCH better than the older versions.

[MiRai]

I'm honestly surprised that hyperthreading helped, especially by such a large margin. Maybe hyperthreading on the new procs is MUCH better than the older versions.
I was curious about that as well but I don't have a 1366 chip to test with. =\

Accepting donations!

[wyofiddler]

You're results for Hyper-Threading vs. non-HT were very interesting to me, since I'm researching a new computer build and wondering if the i7-2600k is really worth $115 more than an i5-2500k CPU. You only lose 0.1 GHz on stock CPU speed with the 2500k (3.3GHz vs 3.4GHz for the 2600k) but the 2500k does not provide HT. 5 boxing Wow and Rift is what i use my computer for the most, so if HT really makes that much difference I'll hold out for the 2600k for sure.

Once you do a five boxing test I would love to know if HT has that much impact on performance running 5 accounts rather than all 10. And if you run a Rift 5 boxing test that would be even cooler... let me know if you need Rift trial invites, I'd be glad to put them in for ya :)

[MiRai]

You're results for Hyper-Threading vs. non-HT were very interesting to me, since I'm researching a new computer build and wondering if the i7-2600k is really worth $115 more than an i5-2500k CPU. You only lose 0.1 GHz on stock CPU speed with the 2500k (3.3GHz vs 3.4GHz for the 2600k) but the 2500k does not provide HT. 5 boxing Wow and Rift is what i use my computer for the most, so if HT really makes that much difference I'll hold out for the 2600k for sure.

Once you do a five boxing test I would love to know if HT has that much impact on performance running 5 accounts rather than all 10. And if you run a Rift 5 boxing test that would be even cooler... let me know if you need Rift trial invites, I'd be glad to put them in for ya :)
I am definitely wanting to do the 5 boxing WoW and RIFT tests ASAP as well. My last day of class is tomorrow
so expect something for it very soon (most likely this week.)

[Silence]

Thanks for this...

I'm in the process of putting together a new PC and was contemplating if the 3GB on the 580 was worth it... knowing myself I probably would get it anyway even if it wasn't, but it's a nice incentive to speed things up ;p

As I only 5-box I'm interested in your results... keep it up!

[Lyonheart]

I am definitely wanting to do the 5 boxing WoW and RIFT tests ASAP as well. My last day of class is tomorrow
so expect something for it very soon (most likely this week.)

GIT'R DUN!! Jk!

Thanks for this. i wont be building my new power rig until later this year ( i ended up buying a laptop ) but this research is very helpful 8) Thank you!

[Tonuss]

Sorry I missed this thread until now. Awesome work, Fenril. I doubt I'll be upgrading until later this year, but it's good to have this kind of data.

I'm honestly surprised that hyperthreading helped, especially by such a large margin. Maybe hyperthreading on the new procs is MUCH better than the older versions.
I don't follow the hardware scene closely these days (or at least, as closely as I used to) but I do recall that people were predicting a 30% boost to processing capability from HT in the Sandy Bridge CPUs. Fenril's test seems to bear that out.

[wyofiddler]

Fenril: Any updates to post on 2600K hyperthreading performance boost for 5-boxing groups in wow and especially in Rift? I'm dying to know how it compares to the wow 10-boxing test results you have posted already.

[MiRai]

Fenril: Any updates to post on 2600K hyperthreading performance boost for 5-boxing groups in wow and especially in Rift? I'm dying to know how it compares to the wow 10-boxing test results you have posted already.
Sorry, I've really kinda been dicking around lately. I did do an initial 5-box HT / non-HT test and HT showed
good gains over non-HT. I was able to peg the non-HT 2600K at its stock clock on a few different video settings
but I want to redo the test. I wanted more population in the city during the test so I was waiting until 4.2 hit
since I knew it would bring people back and then the last 2 weeks have been kinda retarded for me.

However...

I do have a second GTX 580 3GB on the way and it will be here this Friday. I want to do HT / non-HT and SLI
/ non-SLI Load Splitting types of testing in World of Warcraft this weekend. So, unless I run into some sort
of driver issue or something with the second card expect some WoW results this weekend. As for RIFT... let me
get back to you on that. :)

[Kekkerer]

Excellent news, would be very interested in your opinion whether it's worth getting a second GTX580 3GB, my initial idea was to go up to 3 monitors with 1 GPU running my main and the second one 4 slaves.

You have the Palit version?

[MiRai]

Excellent news, would be very interested in your opinion whether it's worth getting a second GTX580 3GB, my initial idea was to go up to 3 monitors with 1 GPU running my main and the second one 4 slaves.
I, personally, have mixed feelings about more than 2 monitors per computer for multiboxing. So far, it would
seem that I'm alone when I say I take a performance hit and my FPS becomes "spikey" if more than 2 monitors
are hooked up to my system while multiboxing. This had been true for me for well over a year and it only went
away after I built my second system and used my old system to house my 3rd and 4th monitors. The only
things that were not changed out in an attempt to troubleshoot the issue were the monitors (they're all
identical) and GPU manufacturer (nVidia). If I had different monitors or an urge to buy an ATi card I might try
and troubleshoot the issue again but, I was so tired of dealing with it that when I built my new system I just
kept my old one instead of selling it off.

I had experienced this performance loss using 2x GTX 260's / GTX 260 | GTX 460 / 2x GTX 570's -- All non-SLI
configurations. After that I RMA'd one of the 570's (the other one is wasting away in my old system), picked up
the GTX 580 3GB, and never looked back. If this second GTX 580 proves to be worthless for multiboxing or
other games like Crysis 2 / Civ 5 with DX11 on super ultra settings... then it's going back as well and I'll just
wait for 28nm GPUs. :)

But...

Please don't take what I've said as a deterrence to try out what you're wanting to do. Like I said, it seems I'm
the only person that has ever reported this issue before and I've not heard anyone talk about such an issue in
the 2+ years I've been around (if they did... I don't remember). I mentioned it to Lax awhile ago and we were
never able to sort anything out even though I figured it was probably hardware all along.



You have the Palit version?
I'm a loyal EVGA (http://www.evga.com/products/moreInfo.asp?pn=03G-P3-1584-AR&family=GeForce%20500%20Series%20Family&sw=) fan.

[MiRai]

I've been thinking of the list of "benchmarks" I'm going to go through with for this weekend and here's what
I've come up with:

5-Boxing World of Warcraft

CPU Usage Tests [2600K]


Hyper-Threading VS No Hyper-Threading
3.4GHz Stock VS 4.4GHz OC

Since WoW is CPU intensive I will cycle through all the available video setting presets (Low - Ultra) on all five
clients and find where the average CPU usage lies for each one in each of those categories. I am going to throw
my custom window swapping macro into the mix as well so we can see how much of a benefit something like
this gives versus having all clients on the same settings. I believe the GTX 580 3GB GPU will not be a
bottleneck until around High/Ultra so it should initially put a lot of pressure on the CPU.

FPS/GPU Usage Tests [EVGA GTX 580 3GB]


Total VRAM Usage With Each Built-In Preset / My Custom Preset(s)
Main Window FPS / Alt Window FPS
SLI VS No SLI [Windowed]

An Attempt To Split Loads Between GPUs

I plan to head to either Stormwind or Orgrimmar during peak play times this weekend to conduct testing. Since
I've re-RaF'd I'm dealing with lower level characters and can't reach anywhere else that might be useful at the
moment. I don't feel that testing in BGs or Raids will give a good average amount of FPS results, although, it
may give results for minimum FPS in heavy situations. I would say it's a trade off between View
Distance/Environmental versus lots of Particle Density. With 50 - 100 players and NPCs in view in either major
city at peak times on Kil'jaeden, I feel that it should give enough stress on Particle Density as well as View
Distance and the environmental settings. Raids have "spikey" Particle Density hits depending on who is casting
what spell and all of that, when a city will have more of a constant flow of Particle Density.

The above tests might sound a little confusing to some of you at the moment but they make sense to me and
are a baseline for what I'll be touching on. If anyone has anything they'd like to see tested just post in this
thread and let me know (you've got 48 hours). Also, if you believe I'm completely retarded with anything I've
mentioned here, again, let me know and perhaps we can come to an agreement on what situation is best for
testing (although, I cannot guarantee that I will be able to test it out). :)

[Tonuss]

So far, it would seem that I'm alone when I say I take a performance hit and my FPS becomes "spikey" if more than 2 monitors are hooked up to my system while multiboxing.
Not sure if this is related, but I have had a performance issue with dual video card/monitor setups. The setups I have tried were:

Dual Radeon 5850 cards, each connected to one monitor.
Single Radeon 5870 card connected to two monitors.

I would use either keyclone (early on) or pwnboxer (later on) to put the main WOW on one display and the rest on the second display. Whenever I was focused on the primary display, it would slow to 30 fps max, but it could run at 60 fps if I was interacting with one of the characters on the second display. The cards are powerful enough that running 30 fps was generally smooth and playable, but it was annoying to know that the minor stuttering that I would see was not due to the card being underpowered.

My current system still uses the 5870 with a single monitor, so it won't be an issue. I plan to try IS Boxer whenever I get back to multiboxing and perhaps I'll have an opportunity to test a multi-monitor setup in the future. But by then I hope to have one or two of those 3GB 580s, so that might skew the totals. :)

[MiRai]

Not sure if this is related, but I have had a performance issue with dual video card/monitor setups. The setups I have tried were:

Dual Radeon 5850 cards, each connected to one monitor.
Single Radeon 5870 card connected to two monitors.

I would use either keyclone (early on) or pwnboxer (later on) to put the main WOW on one display and the rest on the second display. Whenever I was focused on the primary display, it would slow to 30 fps max, but it could run at 60 fps if I was interacting with one of the characters on the second display. The cards are powerful enough that running 30 fps was generally smooth and playable, but it was annoying to know that the minor stuttering that I would see was not due to the card being underpowered.
My FPS spikes were pretty severe. All 4 or 5 windows would get locked at about 10-15FPS and monitoring the
GPU's usage via GPU-Z would show that the GPU load would just shoot up to 100% at random times and it
wouldn't matter whether I was in a main city or in lonely Redridge. There was no method to the madness
what-so-ever which is why I could never pinpoint the issue and I assumed it was hardware, although, I do
remember that decreasing View Distance to a very low value seemed to remedy it for the most part but not
entirely (it would just happen less frequently). At this point though, I do like having 2 computers... one for
gaming/video editing that I can keep clean while I clutter up the second one with browsing/IRC/whatever else.

My current system still uses the 5870 with a single monitor, so it won't be an issue. I plan to try IS Boxer whenever I get back to multiboxing and perhaps I'll have an opportunity to test a multi-monitor setup in the future. But by then I hope to have one or two of those 3GB 580s, so that might skew the totals. :)You will definitely not be disappointed with ISBoxer one bit... or a few 3GB 580's for that matter (assuming
your CPU can handle it). :)

[Kekkerer]

I don't think that I've seen you mention what your FPS limits are for main/slave windows, may just have missed it though. Might be worth playing around with that as well, I run my slaves at 15 fps and main 60.

[MiRai]

I don't think that I've seen you mention what your FPS limits are for main/slave windows, may just have missed it though. Might be worth playing around with that as well, I run my slaves at 15 fps and main 60.
Were. This is all in the past since I've switched to 2 computers and split my monitors between them The FPS
problem went on for over a year and of course I tried everything from swapping hardware to limiting FPS to
adjusting video settings to whatever else I tried. I just ended up having to push through it like a boss.

[remanz]

sorry for hijacking the thread a bit.

Did they add new graphics changes with recent patches ?

My machine used to handle Org valley of strength ok upon cata lanuch (40-50 on main , capped at 30 fps on slaves). I was ok all the way until 4.1. Then valley of strength got pretty bad. My framerate dropped to 15-20 on all windows of wow. My hardware config did not change, nor the graphic settings.

Another obvious example is at the gundrak's snake boss chamber. With all those little snakes crawling on the ground, my framerate dips dramatically. It wasn't the case when I did it in 4.0.x

I am thinking about upgrading to GTX580, currently using the mega fail GTX 295. But I was monitoring the GPU usage with afterburner. It never exceeds 60% on both GPUs, even when my framerate is struggling. Does it mean the card never operates at full capacities ? If that's the case, would upgrading the card help at all ? Or this is more CPU bound.

[MiRai]

sorry for hijacking the thread a bit.

Did they add new graphics changes with recent patches ?

My machine used to handle Org valley of strength ok upon cata lanuch (40-50 on main , capped at 30 fps on slaves). I was ok all the way until 4.1. Then valley of strength got pretty bad. My framerate dropped to 15-20 on all windows of wow. My hardware config did not change, nor the graphic settings.

Another obvious example is at the gundrak's snake boss chamber. With all those little snakes crawling on the ground, my framerate dips dramatically. It wasn't the case when I did it in 4.0.x
My first guess was going to be that more people came back to the game on your realm and because 10 more
people are in the city on one screen... it's really 50 more people because it's rendering on five screens. But
then when you mentioned the FPS drop in the instance so, my first guess was quickly nullified. :)

I have not heard of any graphical updates since Cataclysm has been released but, I have been reading about
many, many people complaining about nVidia's 275.33 drivers. People are seeing FPS drops and overall
performance loss (on all games) with them. They're reverting back to 270.61 or even 267.91 and waiting for
the new 280.xx drivers before upgrading again. Might want to give it a try if you're using the 275.33 drivers.
Or if you're not using 275.33... then try upgrading. :)

I am thinking about upgrading to GTX580, currently using the mega fail GTX 295. But I was monitoring the GPU usage with afterburner. It never exceeds 60% on both GPUs, even when my framerate is struggling. Does it mean the card never operates at full capacities ? If that's the case, would upgrading the card help at all ? Or this is more CPU bound.
First and foremost (which I'm sure you already know) the GTX 295 is a dual GPU card sitting in SLI mode
(unless you've manually disabled it). I might try disabling SLI mode (if you've still got it enabled) just to see
what that does for you. Also, the nVidia 200 series GPUs are only DX10 compatible. While they're able to
emulate some DX11 stuff it's not a true DX11 card, so if you're using DX9 at the moment I would like to believe
that DX11 would help improve overall performance.

You're right in thinking that because the GPU never exceeds 60%, a video card upgrade probably won't give
you any real performance boost besides any gains that DX11 might give. Check those usage numbers again
after you've disabled SLI (again, if its enabled) and see where they fall. Other than that, WoW is definitely CPU
bound so when you're boxing and you see that performance loss make sure you've got your Task Manager open
to the Performance tab so you can see what CPU usage looks like. You can overlay the Task Manager on top of
your game windows if your layout covers all of your monitor(s).

[remanz]

Thanks Fenril.

I will give disabling SLI and reverting drivers another try before I bust out the cash for GTX580. I tried disabling SLI before. Didn't really improve anything, nor degrade anything. So I figure I just left it on. It was just my wishful thinking that one day, all of sudden, WOW would support SLI fully. and with both card running at 90-100%, the game would perform well. and it really bothers me to have my GTX 295 is running as a GTX 260, lol.

[MiRai]

Sooo... about those tests...

The World of Warcraft SLI tests were very disappointing (surprise, surprise) and there is no need to even
post the results. World of Warcraft is just too CPU bound for SLI to really give it anything more than your
processor can already provide it.

SLI acts funny with WoW in the sense of splitting the GPU load amongst the two cards. In all of the tests I
had run, a single GTX 580 would produce less heat, consume less power, and have lower overall load
compared to its SLI counterpart configuration yet, give the same FPS results. Look below...


GTX 580 3GB No SLI ("Good" Video Preset @ 60 Main/30 Alts (~28 In-Game FPS))

http://multiboxology.com/bin/images/wow/benchmark/580Solo.png

2x GTX 580 3GB SLI ("Good" Video Preset @ 60 Main/30 Alts (~28 In-Game FPS))

http://multiboxology.com/bin/images/wow/benchmark/580SLI.png


In the above screenshots we're focusing on GPU Temp, GPU Load, and VDDC Current (if you'd like). This is
how all of the tests looked. The SLI configuration is working harder, consuming more power, and producing
more heat which only helps rise the ambient temperature of my computer room.

Also, the newest nVidia 257.33 drivers were giving me a slight stutter when playing solo with SLI enabled.
Downgrading to the 270.61 drivers fixed that so I conducted these tests with that driver.

Final word on SLI in World of Warcraft (for now) -- avoid it and just get a great single GPU card. This has
been the same advice most people have been giving out for a few years now and nothing seems to have
really changed (blame the game engine). I am going to keep playing around with it from time to time to
see if I stumble upon the holy grail of video settings that change our lives as multiboxers (don't hold your
breath). As for ATi's Crossfire setups, I have no idea and you're all on your own with that one. :)

[MiRai]

In addition to the SLI testing, I did attempt to utilize the total VRAM of both cards and "split" the games
between them essentially giving me 6GB of VRAM.

I was successful and it does indeed work.

However, due to what seems to be a bug on Blizzard's end this can only currently be accomplished using DX9.
Using DX11, I was unable to successfully set which monitor I wanted certain windows to launch on w/o getting
an error during the selection in the drop down menu. If I tried to force this through the config.wtf the windows
would revert back to their default 800x600 resolution which is not what I wanted. There are a few reports of
this on Blizzard's Technical Support forum but, changing from DX11 to DX9 alleviated the problem.

Doing this while using DX9 and five clients did not seem to be worth anyone's time. DX9 was really showing its
age because I was easily able to set all 5 clients to Ultra and they would barely take up over 1GB of combined
VRAM between the cards. On the other hand, using High settings and a single GTX 580 I can take up over 2GB
of VRAM in DX11.

Even in it's current bugged state, I can definitely see this feature being useful for people who 10+ box. I'm
going to guess that I could throw at least 10 characters on each video card and still be under the 3GB limit in
DX9 and without using horseshit video settings. If someone really wanted to crank down the settings I'd bet
you could jam 20 characters per 3GB card on there. I'm not saying a single CPU could handle this by any
means... but, perhaps I just sparked another idea for the next test. :)

If you are wanting to attempt to use this, you'll need to create a special window layout with whatever software
you're using because you need to keep the game windows running on the monitors they launch on. If you start
swapping game windows between monitors you do take a performance hit.

I was going to provide a screenshot for this but, it accidentally got overwritten in my late night stupor.



As for the Hyper-Threading VS No-Hyper-Threading testing. Soon™

[Knytestorme]

Thanks for the work Fen, seems like grabbing a 3Gb 580 is the go for now and hold off on SLI until we see how a single handles RAGE, BF3, Skyrim etc later in the year

[EDIT]

To add to what you said about DX9 and 11 I just noticed something interesting.

4 DK's in starting area under DX11 = 994Mb usage with some artifacting starting on a 1Gb card
Same 4 in same spot with same settings under DX9 = 667 Mb usage.

[MiRai]

Thanks for the work Fen, seems like grabbing a 3Gb 580 is the go for now and hold off on SLI until we see how a single handles RAGE, BF3, Skyrim etc later in the year.
SLI did immediately give a noticeable boost in both TERA Online and Civilization V completely maxed out so it
isn't a complete loss but, I'm still not entirely sure I can justify the $600 price tag of this second card. I keep
telling myself just give it a few weeks but... I have a feeling this card is going back and I'll just eat the
restocking fee.

[remanz]

SLI did help with boxing rift right ? A lot of new games had problems supporting it at launch (just look at Crysis 2, the once benchmark hardcore PC game doesn't support SLI at a launch, GG).


but I would need a new PSU first.


Another mistake that I made with my current built, (other than purchasing a GTX 295 lol) is to go with a mediocre power supply. The PSU does degrade over time. It is slowly dying. and I had this constant problem of rebooting my PC automatically when I try to push the graphics. Damn you "witcher 2"~. as soon as I do anything in that game, it reboots.

longer than 2 hrs of rift, reboots. I first thought it was software crash or my card was overheating. But more and more, I think it was due to the PSU as how can the card overheat when I am just 5 seconds into the game. It was at 50 degree to start with. Insufficient power/unstable current seems to be the cause.


on a side note,
Speaking of upgrading hardware, have you guys noticed those "rent a gaming PC/or graphic cards services"? With the frequency that we are upgrading, maybe it will be worth while just to rent cards/PCs than buy them. 2k for 2 years, that's like 84 dollar every month.

[Knytestorme]

SLI did immediately give a noticeable boost in both TERA Online and Civilization V completely maxed out so it
isn't a complete loss but, I'm still not entirely sure I can justify the $600 price tag of this second card. I keep
telling myself just give it a few weeks but... I have a feeling this card is going back and I'll just eat the
restocking fee.

Yeah, looking at costs now that my tax has been done.

Looking at benches and your results here I'm wondering if a 2Gb 560Ti would be just as effective for 5-boxing as a 3Gb 580. Looking at $319 vs $609 I'm not sure if it's worth it atm or in the long-run if I look to move to Ivy Bridge next year when I'm assuming the 6xx's will be out.

[MiRai]

Yeah, looking at costs now that my tax has been done.

Looking at benches and your results here I'm wondering if a 2Gb 560Ti would be just as effective for 5-boxing as a 3Gb 580. Looking at $319 vs $609 I'm not sure if it's worth it atm or in the long-run if I look to move to Ivy Bridge next year when I'm assuming the 6xx's will be out.
There's also something like the 2.5GB 570 (http://www.evga.com/products/moreInfo.asp?pn=025-P3-1579-AR&family=GeForce%20500%20Series%20Family&sw=) that falls in between (even though I personally don't like those fan designs).

[Sam DeathWalker]

4 DK's in starting area under DX11 = 994Mb usage with some artifacting starting on a 1Gb card
Same 4 in same spot with same settings under DX9 = 667 Mb usage


I guess there is no way to know if wow or win7 uses the extra video ram for cacheing. Probably not ....

Looks like then for up to 7 boxing 2G ram will do the job for sure, even with DX11.

[MiRai]

I got bored last night so I thought I'd see how far I could push my hardware.



http://www.multiboxology.com/bin/images/wow/30boxtest.png



CidQTmuJ8QE



All windows are rendering at 1920x1080 with instant swapping enabled. Even with 30 screens there is zero
lag while swapping between any of them. Main @ 60FPS / Slaves @ 10 FPS. I would advise against taking
these 30 characters into a major city anywhere near peak times as you can see my system RAM is almost
maxed out.

I split the GPU load between my video cards and they are rendering 15 clients each. As you can see I could
have put all 30 on a single card and I would have still been way under the 3GB limit. This is of course using
DX9 since DX11 would have most likely maxed out the VRAM. If your interest is to box as many characters
as you can on a single machine it's unlikely that you care how high your video settings are anyway.

I would say running 20 on a single machine would allow you the system RAM headroom you would need to
visit major cities and PvP with; although, it would still be cutting it close during intense situations.

[Tonuss]

That's hysterical. :)

I upgraded to a 3GB 580 a week ago. I had a problem with one of the RAID 1 drives where I keep my data and figured that if I was going to pop open the case to install a new hard drive, I may as well fulfill a few dreams. :p

[Sam DeathWalker]

Humm so 15 clients don't even take up 1G of video card ram ...


How do you say you render 15 clients per card, there seems only one screen, are they tiled on the 2nd screen?

If they are not tiled they are not being rendered (although there are 15 clients per card) to my way of thinking. Or if they are all in the background not visible they are also not being rendered. If that is the case your fast swaping times is just because the 580 can redraw a new screen faster then your fps. The fact you have 10fps in background would tend to indicate that you are not gaining anything by tiling in full resolution. Your card is fast enough to render a full screen in less then 1/60th of a second. Which I think should give you instant swaping no matter what.

I wonder what would be the result if you were in different zones, does the card use so little Video Ram because its able to share data between the 15 clients? How does the ram go up from when you load 1 to many is the first load a lot lot bigger then 950/15? If it is then the video ram is being shared between clients. Or does each client just take 60MB of ram?

So it takes 30 clients to use up 16G of system ram. More good infos.

Good testing for sure though.

Do you think you can 15 box on your system (with just one 3G card) with no trouble in populated areas?

[MiRai]

Humm so 15 clients don't even take up 1G of video card ram ...


How do you say you render 15 clients per card, there seems only one screen, are they tiled on the 2nd screen?

If they are not tiled they are not being rendered (although there are 15 clients per card) to my way of thinking. Or if they are all in the background not visible they are also not being rendered. If that is the case your fast swaping times is just because the 580 can redraw a new screen faster then your fps. The fact you have 10fps in background would tend to indicate that you are not gaining anything by tiling in full resolution. Your card is fast enough to render a full screen in less then 1/60th of a second. Which I think should give you instant swaping no matter what.
I don't know where this is coming from. Whether one screen is on top of the other 14 or not, they're all
being rendered all the time as you can see in my screen shots below. Don't mind the increase in VRAM, I
accidentally moved the camera with my mouse repeater on and I gained 300MB, at least that's what I'm
blaming it on. The main screen is in focus so that it's at 60FPS and all windows are 1920x1080.

http://multiboxology.com/bin/images/wow/tiled_allshowing_GPU_load.png


http://multiboxology.com/bin/images/wow/tiled_ontop_GPU_load.png


I wonder what would be the result if you were in different zones, does the card use so little Video Ram because its able to share data between the 15 clients?
No idea and that isn't on my list to test at the moment; although, such a test might give a clue as to
whether or not textures are shared.

How does the ram go up from when you load 1 to many is the first load a lot lot bigger then 950/15?
I don't know but after going through Orgrimmar I think my ceiling is ~14GB since I can't seem to go any
higher than that. I didn't look at my Resource Monitor see if Windows was holding any memory hostage in
Standby or not. Here's a screen shot after I did a single lap in Orgrimmar before I recorded the video below.

http://multiboxology.com/bin/images/wow/20box_inside_org.png


Or does each client just take 60MB of ram?
From my Orgrimmar screen shot...

1593MB / 20 = ~80MB

However, the main is obviously taking up more than any slave so I would say it's closer to 65MB per slave
and like 300MB for the main client.

Do you think you can 15 box on your system (with just one 3G card) with no trouble in populated areas?
15?! I'll give you 2 laps in Orgrimmar with 20! This was at about noon (12PM) on a Saturday -- Obviously
not peak time.


tGwx2jHFzPI

[Sam DeathWalker]

I don't belive that the gpu actually renders anything that is not visible on the display.

Thats not to say it dosn't load the textures or set aside the memory for all clients, just that the gpu dosnt do the calculations, even if the data is available if its not going to the display. I am sure it loads the textures for all clients all the time though.

In the first screen yes the GPU is rendering all 15 clients at the full resolution, but at 10fps.

In the 2nd screen the GPU is only rendering 1 client but becuase your fps and effects are higher (are they?) more GPU load is being used. And in the 2nd screen more of the world is showing. Although the 2nd screen is also shown in the first I am fairly sure its at 10fps in the first screen and 60fps in the 2nd.

My point is that in the 2nd screen with only one client showing the other 14 clients put ZERO load on the GPU.



Well that video card is something for sure. Excellent information keep up the good work.

300Mb for first client with most effects and 65M for addional lower alts is excellent infos.

Maybe 2G cards are not needed, it seems that there is a lot less video information needed by the card then I had thought although the fact that there are lots of 4K read and writes would tend to support that very little video information is being used. They have made wow to be played on some really poor computers.


Sure wish we could find out if it uses the vram that not "in use" in the gpuz display as cache.

These people clearly are not stupid obviously if something can be done to speed up game play it probably has.

http://developer.amd.com/documentation/presentations/legacy/Chapter02-Mittring-Advanced_Virtual_Texture_Topics.pdf



That is very bad for stable performance and MultiGPU (SLI/Crossfire) scaling but it was a manageable
solution at the time. Streaming allowed us to stay within the 32 bit limits with run‐time
data requirement sometimes exceeding the limits. On 64 bit and enough main memory
or when using half resolution textures the texture streaming is not necessary and performance is more stable.


Thus the conclusion is that the more Vram (and system ram) you have the better even if gpuz shows only some being used I am sure the rest is being used for cache purposes. Might have something to do with "lod" from the above artical.


The LOD Bias affects MIP-Mapping in determinating the MIP-Level.

MIP-Maps are collections of one and the same textures in different resolutions. To allow perspective effect and avoid underfiltering, objects in the back of a scene need to be displayed smaller and the textures in use need to have a lower resolution. The MIP-Map Detail Level globally determines if MIP-Mapping happens with the best possible resolution (and therefore best imagequality (http://forums.guru3d.com/showthread.php?t=201527#)) or with best performance in mind through all MIP-Levels.

Now LOD Bias means global displacement of the MIP-Level determination. With this value set below zero, textures of higher resolutions will more likely be applied, while with a LOD Bias greater than Zero MIP-Mapping will "prefer" textures of lower resolutions.



gxTextureCacheSize is a CVar that affects the D3D9 API; if you're already using DirectX 10/11 mode, this CVar will not have any affect. This variable decides how much memory - in megabytes - to use for caching textures. The default value is '0', meaning the game will handle this on its own. Some players have found better performance - particularly when turning - by manually selecting a value for this. As a general rule of thumb, ~70-75% of your VRAM is best. For example, if you have a 1.5 GB GTX 480, a good value to try would be 1152 MB. The syntax would look like this:

SET gxTextureCacheSize "1152"

For quick reference, these are 75% values. Feel free to use more or less as depending on specific configuration settings.

128 MB: 96
256 MB: 192
384 MB: 288
512 MB: 384
768 MB: 576
896 MB: 672
1 GB: 768
1.125 GB: 864
1.25 GB: 960
1.5 GB: 1152
2 GB: 1536


Ok that answers the shared question also. 75 percent of the video ram is for texture cache, which no doubt is available to all clients.

They want 25 percent free assuming only one client so with 10 or 20 box maybe less texture cache should be set aside according to a 80M per client rough estimate.

[MiRai]

I don't belive that the gpu actually renders anything that is not visible on the display.

In the first screen yes the GPU is rendering all 15 clients at the full resolution, but at 10fps.

In the 2nd screen the GPU is only rendering 1 client but becuase your fps and effects are higher (are they?) more GPU load is being used. And in the 2nd screen more of the world is showing. Although the 2nd screen is also shown in the first I am fairly sure its at 10fps in the first screen and 60fps in the 2nd.

My point is that in the 2nd screen with only one client showing the other 14 clients put ZERO load on the GPU.
Before I take a screen shot, I click on region 1 to make sure it's in focus and getting full attention. I do this
every time (at least I try to remember to do this every time) because it gives a more accurate load reading as
if someone was actually on the first client. I mentioned this before the screen shots above:

The main screen is in focus so that it's at 60FPS and all windows are 1920x1080.
GPU-Z has a handy feature called "Always on top" so that it can stay on top of windows even when it's out of
focus (like the Windows Task Manager).

But for you Sam... I'm going to go the extra mile and do another test. I should have the results later today.


300Mb for first client with most effects and 65M for addional lower alts is excellent infos.

Maybe 2G cards are not needed, it seems that there is a lot less video information needed by the card then I had thought although the fact that there are lots of 4K read and writes would tend to support that very little video information is being used. They have made wow to be played on some really poor computers.
Remember, this is all in DX9. If someone wants all the eye candy of DX11 with 5 or 10 clients they'll need over
1.5GB and probably over 2GB if they really had some settings turned up.

[MiRai]

Here's a big-mother-of-a-picture to help show you what I'm talking about and here's what is going on in it...

Top Monitor:
15 tiled game windows with Client #15 (upper left window) at foreground/background FPS of 60/60. This
ensures that even when it's in the background it's acting like it's in focus. Client #15's video settings are also
identical to Client #1's video settings. The GPU-Z that is overlaying the 15 tiled windows is reading the GPU
that is rendering all of those 15 clients.

Bottom Monitor:
14 tiled game windows that are covered up by the main window itself (Client #1). Client #1 in this particular
screen shot is in focus so it is also getting 60 FPS just as Client #15 is. The GPU-Z that is overlaying this
main window is reading the GPU that is rendering the main window and the 14 tiled windows behind it.

Miscellaneous Information


All windows, except for 1 and 15, are capped at 10 FPS
Each monitor is being rendered by its own GPU
Each GPU is set to render 15 clients each (if this wasn't obvious from the above text)

Rendering in Direct X 9
All windows at 1920x1080
Windows Aero is disabled





http://www.dual-boxing.com/images/fenril/tiled_vs_onescreen.png




Now, you can clearly see that both GPU-Z's are practically identical regardless of windows being in
view or not. You cannot honestly tell me that Client #1 just so happens to magically match the 15 tiled
game windows above it. I believe this proves my point that even when a game window is not in sight, it is
still being rendered.

Note To Everyone And Anyone: If you happen to see a flaw in what I'm doing here please let me know or
point it out. I am not an engineer or a someone who codes how games work so I only believe that what I'm
showing here is indeed correct.

[Sam DeathWalker]

I am coming to the conclusion that more Vram is way way better then a SSD, or even more system ram.

Wonder if there is a 3G card thats like a 460 or 560 or something less then the 580.


Ok as to the above. 1 and 15 are 60fps others are 10fps

In first shot you have 2X60 and 13X10 or 120 and 130, meaning that it takes about as much gpu power to render 1 and 15 as it does to render 2-14.

Still the total is 270 fps for all 15 clients. (can isboxer tile at different fps, I find that a bit odd).

In the 2nd you have 1X60 for a total of 60fps

So the 1st requires 4.5 times the gpu power (time) of the 2nd.

We see a 1percent difference in gpu power. (18 vs. 19).


1X = .18Y
4.5X = .19Y

Solve for X and Y

3.5X = .01Y

X = .01/3.5 Y

X = .0028 Y

Y = 357ish times X.

So if your gpu is 360 or so times what you need to render 60fps (i.e. it can render 21,000 or total fps on all clients /running at 100percent gpu load/, which might be possible with that simple screen) then the 1 percent difference is explained.

Also the gpuz thing being in only 1 percent could cut that in 1/2 depending on where exactly in the percent reading it is (i.e. could be 1.01 to 1.99 and still show 1 percent, big difference).


The point of all this is that your 580 is so powerfull as compared to the test you are using that even the smallest difference is a lot.



Anyway the bottem line for the reader, based on your excellent tests, and the infos I found is that money should be spent in this order:


VRAM
System Ram
GPU (this depends on how high resolution and how high effects you want mostly, if you dont care then spend more on cpu)
CPU / MB (MB is basically a reliability issue)
SSD

[MiRai]

Ok as to the above. 1 and 15 are 60fps others are 10fps

In first shot you have 2X60 and 13X10 or 120 and 130, meaning that it takes about as much gpu power to render 1 and 15 as it does to render 2-14.

Still the total is 270 fps for all 15 clients. (can isboxer tile at different fps, I find that a bit odd).

In the 2nd you have 1X60 for a total of 60fps

So the 1st requires 4.5 times the gpu power (time) of the 2nd.

We see a 1percent difference in gpu power. (18 vs. 19).


1X = .18Y
4.5X = .19Y

Solve for X and Y

3.5X = .01Y

X = .01/3.5 Y

X = .0028 Y

Y = 357ish times X.

So if your gpu is 400 or so times what you need to render 60fps (i.e. it can render 24,000 total fps on all clients /running at 100percent gpu load/, which might be possible with that simple screen) then the 1 percent difference is explained.
Dammit Sam, I have no idea what all of this means. You mention 2 screen shots but, there is only 1. It's a
double monitor screen shot and all I did was press the Print Screen button and then paste it into MSPaint --
No hacking, chopping, or trickery was done to the above screen shot. I don't know what all the math is
about but, my above post was to show you why I believe that clients which are not being seen by the human
eye are still being rendered in the background by the GPU... that is all.

As for ISBoxer tiling at different FPS, all I did was change the background FPS for that particular slot in the
Character Set itself. You could make all 30 windows have unique foreground and background FPS if you'd
like, there is no funny stuff going on here.

[Sam DeathWalker]

No I'm not say there is any tricks at all, I'm just saying that the 1 percent difference that you show: 19 percent of full gpu load with 15 clients showing on one monitor vs. 18 percent with 1 client showing on the other monitor proves my point, even though you feel that because the are about the same it proves your point.

If your gpu can put out 500fps total for all clients then that 1 percent is 5fps which would prove your point, as they are "about the same".

If your gpu can put out 19,000 fps total for all clients then that 1 percent is 190fps which would prove my point. As the game is putting out 250 total fps with one gpu (60 plus 60 plus 13 X 10) showing 15 tiled clients vs. 60fps with the one client showing. Which is a difference of 190fps


I think that on the test you used the fairly static and repeating and common to all clients log in screen that 20,000 fps over all clients for a 580 running at full load is not a stretch.


I don't belive that they would have the gpu render screens that are not seen, considering they dont waste time rendering objects that can't be seen behind other objects.

[MiRai]

No I'm not say there is any tricks at all, I'm just saying that the 1 percent difference that you show: 19 percent of full gpu load with 15 clients showing on one monitor vs. 18 percent with 1 client showing on the other monitor proves my point, even though you feel that because the are about the same it proves your point.
Sam, it proves my point. Client #1 and Client #15 are identical clients therefore they put out identical loads
and take up identical amounts of VRAM. If that wasn't the case, then the bottom monitor should be reading a
much lower number than it is, because if what you say is true, the other 14 clients aren't being rendered.
You're telling me that 1 single client @ 60 FPS is taking up the same VRAM and putting out the same GPU load
as 15 clients using up a total of 200 FPS at the same resolution? That's retarded...

If your gpu can put out 19,000 fps total for all clients then that 1 percent is 190fps which would prove my point. As the game is putting out 250 total fps with one gpu (60 plus 60 plus 13 X 10) showing 15 tiled clients vs. 60fps with the one client showing. Which is a difference of 190fps
I don't know where you're getting 60 + 60 + 130 from. Out of clients 15 through 30 only #15 is rendering at
60 FPS. If you want to add up the FPS, the top 15 clients it would be 60 + 140 (14 clients x 10 FPS) for a total
of 200 FPS. In your world, the bottom monitor is only rendering 60 FPS yet the VRAM and GPU load match the
top monitor.

I don't belive that they would have the gpu render screens that are not seen, considering they dont waste time rendering objects that can't be seen behind other objects.
This is so ridiculous. You're basically saying that if a tree falls in the forest and no one is around... it doesn't
make a sound. Just because you don't see the windows doesn't mean that the characters don't see the world
around them. If what you are saying is true... then the slave characters would never be able to assist the main
character if their screens weren't showing because they wouldn't be able to see what they should be assisting
and would receive the error "You have no target."

EDIT: I made another video for you.

Important Information:


Client #1 has higher video settings than Client #2
Client #1 is located in the tunnel to Orgrimmar looking at the ground

Client #2 is located in the Night Elf starting area

Client #1 is rendering at 60FPS foreground and background
Client #2 is rendering at 10FPS foreground and background
The inactive window sits behind the active window



OqdAGf7NEwA


Now, because both clients have different video settings each window should take up different amounts of
VRAM, however, as you can clearly see in the video... the VRAM doesn't change when I swap windows. Also,
both clients are rendering at different framerates which affects the load on the GPU, however, as you can
clearly see in the video... the GPU load does not deviate more than 2% from the reading at the beginning of
the video.

All game windows that aren't minimized, whether in the foreground or background, are rendered
at all times.

Is there anything else I can do for you Sam?

[Sam DeathWalker]

The amount of of ram isnt going to change it allocates the space.

Maybe we are hung up a bit on the word "render". To me render means that the gpu does all the calculations needed to display a full frame. It has to determine which objects are in front of each other, the lighting, the character and world textures etc.

Everything except rendering is always done for all clients, postion of characters, loading the textures of nearby characters, movement exptrapolatioin, etc.

Lets say a weightlifter(gpu) can lift 1000lbs. At 1000lbs he is at 100percent(zgpu shows 100), at 500lbs he is at 50percent, at 100lbs his zgpu shows 10percent.

You design a test that has him lift 2lbs vs. 10lbs and say that "10 lbs is FIVE times 2lbs". Surely his zgpu will show a difference of FIVE times because when he lifts 10lbs he is doing FIVE times the work of lifting 2lbs.

Lifting 2 lbs shows his zgpu at 1 percent because the zgpu only measures in single percentages 2/1000 is rounded up to 1percent for display purposes.
Lifting 10 lbs shows his zgpu also at 1 percent because 10/1000 is in fact 1 percent.

The zgpu reading is the same for both (1 percent) yet one test was 5 times the other test. Thus you conclude that because the zgpu reading is 1 percent for both that he must be doing the same work (10lbs and 2lbs) 12lbs both times.

Notice that when you ran two clients in the 2nd test you are at 12 or so percent and with 15 clients tiled you are at 19 or so percent.

Why isn't running 15 clients using seven times as much gpu load as two clients?




Let me ask you this, after the gpu renders the display that you see its sent to the display monitor. Where does the GPU put the rendered frame from the background when its done rendering the full screen? How many frames does it save? If you were writing the program would you render background data?




If what you are saying is true... then the slave characters would never be able to assist the main
character if their screens weren't showing because they wouldn't be able to see what they should be assisting
and would receive the error "You have no target."


You're characters arn't real people ... Position data is calculated (not rendered) for all clients at all times. Knowing that the leader is at coodenate 1021, 938 and is moving say in the X direction at 30 per second then your computer uses that informaiton to calculate his postion untill the server sends new information, thats all that is needed for position data.

You don't have to "see" his textures to follow him, if you know his location in the world. Thats why you get character running forward when ping lag dosnt allow your computer to get server data to update the characters position. Then when the information fromt the server comes in the character snaps back to the server position.

All the visual informationo is client side the server just sends poitional data, what the characters are and what they are wearing etc. how much damage they took and just minimal information. Your computer then takes taht minimal information and builds it into a world. The server probably dosnt even render any of the world even server side, but it does all the positional calculations and damage and all taht.

[MiRai]

Notice that when you ran two clients in the 2nd test you are at 12 or so percent and with 15 clients tiled you are at 19 or so percent.

Why isn't running 15 clients using seven times as much gpu load as two clients?
Because it was the login screen Sam. I could load up 150 login screens but I wouldn't be able to load 150
characters into the game world because there's much, much more going on in the game world.

Sam, I do not understand any of your examples or math but, how about a simple question...

How come I can successfully FRAPS a game window that is behind another game window?

I await your answer...

[Sam DeathWalker]

Ok that might be a good test. Make 10 clients on one gpu.

Set all to 60fps forground and background (all 10 at 60fps all the time) and full ultra including max multisampling (make sure you only using one gpu, if you have a 2nd card in disable its driver in device manager). Have 9 tiles in background (3X3) and one tile full screen forground. Have all look at something in the distance not at ground, a crowed place is better but not necessary. Do not send the background to a second monitor (of course if you send outputs to a 2nd monitor it will be rendered). Have only one active gpu and one active monitor.

Bring the one tile to the full screen forground hiding all others in the background. Make sure fraps is off completly (check process in the task manager). Do a SS and see what gpu load is. Swap the windows and bring the 9 tiles to the forground (is that possible?) and the one client to the background take SS and see what the gpu load is (memory means nothing).

Then do the exact same test except with fraps recording the background only (not the forground). SS with the one client in the forground and fraps recording in background, then 2nd SS (showing gpu load) with 9 clients in forground and 1 client in background with fraps again only recording the background.

Those 4 gpu load figues should give us the answer.

If they are the same in both sets compared to the other in the same set and the first set is a bit lower then the 2nd set (cause fraps gonna take up a bit) then you are right. I.E.

GPU load:

1 (no fraps 1 client forground) 38percent
2 (no fraps 9 client in forground) 38percnet
3 (fraps 1 client forground) 42percent
4 (fraps 9 client forground) 42percnet

If we see something like that you are correct.

Of course wow will render both the forground and the background if the forground goes to one display and the background goes to another display.

[MiRai]

Ok that might be a good test. Make 10 clients on one gpu.

Set all to 60fps forground and background and full ultra including max multisampling (make sure you only using one gpu,
I think you're completely mistaken as to how powerful you believe a single GTX 580 is or how much CPU &
GPU Ultra w/ MSAA requires. I can't even run 5 clients @ 60/30 FPS using the Ultra preset and no MSAA
(x1). Why your test requires super maximum video settings is beyond me. I mean... why don't I just go into
the nVidia control panel and increase all those settings to max as well so I can have Ultra++++ settings
across 10 clients? What more is that going to prove rather than keeping the clients at Fair or Low settings?

I'm done with all of this and I will not be running anymore of your tests in an attempt to help you believe
what I say is true. It's unfortunate that you can't see where I'm coming from in all of these different tests
and that isn't my problem. Why don't you run some tests of your own?

The last test that I FRAPS'd which was the Tedrassil/Orgrimmar test with two characters in two completely
different zones. The character in Orgrimmar had 60 FPS at all times and the character in Teldrassil had 10
FPS at all times. FPS affects the GPU load as anyone can see by using GPU-Z and changing their FPS in real
time.

You Believe: If a window cannot be seen because it is hidden behind another window then it is not being
rendered and that only the top window (out of how many ever are stacked on top of each other) affects the
total FPS that is being rendered/used by the GPU.

I logged in the same characters from that same test but only one at a time and both characters' single
window GPU-Z readings were... wait for it ...much, much lower than the 234MB VRAM / 13% GPU load
readings from the video. Why? Because the GPU renders all game windows that aren't minimized.

Why can I successfully FRAPS a game window that sits behind another game window? Because the GPU
renders all the game windows that aren't minimized.

I keep saying "that aren't minimized" because here's a little test you can run on your own. Try to FRAPS a
single game window and minimize it, give it a few seconds, and then unminimize it. Play that video back and
you can see that the video freezes in place during the time it was minimized. Why? Because the GPU isn't
rendering that window anymore because it's minimized. You can even watch the GPU load drop when you
minimize a window because it isn't being rendered anymore in its minimized state.

When on the flip side, in my last video, jumping between game windows did not affect the GPU load. The
2% that the GPU load did fluctuate was most likely because of people passing through the Orgrimmar tunnel
and the GPU had to... wait for it again ...render those people even though they were in the background
window!

To all of the readers of this thread: I hope these last few page have been fun and exciting for all of you who
do read and keep up with this thread. Hopefully, your tubs of popcorn are not near the bottom because I'm
sure this thread will continue to experience more excitement in the future.

[Sam DeathWalker]

I think you're completely mistaken as to how powerful you believe a single GTX 580 is or how much CPU &
GPU Ultra w/ MSAA requires. I can't even run 5 clients @ 60/30 FPS using the Ultra preset and no MSAA
(x1). Why your test requires super maximum video settings is beyond me. I mean... why don't I just go into
the nVidia control panel and increase all those settings to max as well so I can have Ultra++++ settings
across 10 clients? What more is that going to prove rather than keeping the clients at Fair or Low settings?



Prior test were usless because the gpu was not being stressed at all. I just picked some high stuff at random as it seems like the 580 was just to strong, any setting that use like 80 or so percent of the gpu load is fine.



Why can I successfully FRAPS a game window that sits behind another game window? Because the GPU
renders all the game windows that aren't minimized


FRAPS itself might require the window to be rendered thats why my test is run with fraps on and with fraps off.

Just as if you have two monitors and send background to one and forground to the other even if one is turned off your gpu will still render whats being sent to it..

Ill have to run some tests myself. Obvisouly if what I say is true then gpu load with one client should be the same as gpu load with 10 clients (if nine are all in the background). Should not be that difficult to test properly.



Ok I did a test. And anyone can do this with their computer its easy enough.

Two Clients. Set the FPS back and forground to 100 on both. Set DX9 or 11 the same on both.

1st Client. Full Ultra with 8 X multisampling stand in the center of the area and look far.

2nd Client. Low with 1 X multisampling stand against the wall looking into it.

When I show the 1st client I am at like 80ish gpu load, when I show the 2nd client I am at 50ish gpu load.

BUT THIS ONLY WORKS IF BOTH HAVE "FULLSCREEN" checked.

If you check "windowed" or "windowed fullscreen" it will probably render the background.

Keep in mind your computer will change the fps to try and compensate and of course as fps goes down so does gpu load. Probably it might be best to set both at 20 or 30 whatever causes the 1st client to get like 90 percent or so.


Thats not to say that Fenril was in any way wrong, probably he never runs with "fullscreen" checked in all clients (in fact IsBoxer might not allow that when tiled).

[Sajuuk]

Why would you run your game in fullscreen?

Seriously, WHY?

And not just for one of your stupid tests Sam. To boost performance? Sorry, but if I'm multiboxing I want to have a quick glance at my other clients to make sure they're where they should be, not humping a tree or some nonsense, so I'll have them run in non-fullscreen mode (again, WHY?! I switch between other applications all the time!) and tile them as I see fit.

[MiRai]

Ok I did a test. And anyone can do this with their computer its easy enough.

Two Clients. Set the FPS back and forground to 100 on both. Set DX9 or 11 the same on both.

1st Client. Full Ultra with 8 X multisampling stand in the center of the area and look far.

2nd Client. Low with 1 X multisampling stand against the wall looking into it.

When I show the 1st client I am at like 80ish gpu load, when I show the 2nd client I am at 50ish gpu load.

BUT THIS ONLY WORKS IF BOTH HAVE "FULLSCREEN" checked.

If you check "windowed" or "windowed fullscreen" it will probably render the background.
Why are you testing in Fullscreen mode? Is that is how me multibox? Is that how anyone around here plays
multiple clients on one machine? In Fullscreen mode? No, they don't. When one fullscreen window is in
focus... everything else can be considered to be in a minimized state hence the reason why that one game
window hijacks the entire computer.

Thats not to say that Fenril was in any way wrong, probably he never runs with "fullscreen" checked in all clients (in fact IsBoxer might not allow that when tiled).
No, Sam I don't. No one plays in Fullscreen because you can't tile windows in Fullscreen. It's been like that
since the beginning of multiboxing time. I have no idea why you don't know this since you do, in fact, tile game
windows yourself.

[Sam DeathWalker]

No I don't tile (check my SS on my web site) I play fullscreen and alt/tab between clients (cause I have 6 monitors I see enough of the world anyways).

Lax should look into that. If you could have all your tiled character in the background (say 4) and the main in the forground full screen that would save some gpu power for sure. But you will lose a LOT of switching time. And coodenated mouse movement might not work (although you dont need to render the world to know where your mouse is). But you might not lose to much switching time if you have a really powerfull gpu like the 580 cause it can render the screen in less the 1/60th of a secound.

Does everyone use coodenated mouse movement (mostly for ground click attacks if you run mellee you don't need it)?

Well, actually the reason to tile is to get common mouse movement on all screens at the same time, and to see more of the world I guess.

If you have a 2 monitor set up, it might not be that bad to put 2 guys on your main screen and 3 guys on the alt screen and do full screen on all, (or 1 and 4) you wont get coodenated mouse movement (although lax might be able to fix it so that you can on background sceens also I am not sure, if that dons't occurs as it is). That would save some gpu power for sure. And switching times are very very fast. And you never have to squint looking at a tiny screen.

Also I am not sure that the rendering thing dosn't work with "windowed (fullscreen)" checked or not Ill do more tests on that later.

Now that I think about it I can set my background fps to 60 like my forground (or at least 30) given taht there is no rendering .... make my auto follow tight.

[MiRai]

No I don't tile (check my SS on my web site) I play fullscreen and alt/tab between clients (cause I have 6 monitors I see enough of the world anyways).
I stand corrected on how your Window Layout works, however, you are the first and only person I know of that
alt-tabs between fullscreen game windows.


Also I am not sure that the rendering thing dosn't work with "windowed (fullscreen)" checked or not Ill do more tests on that later.Windowed (Fullscreen) and Windowed are both "windowed" in these tests. One has a border and the other
doesn't. All of my tests that I've shown here in this thread are in Windowed (Fullscreen) mode. I've conducted
other, smaller tests on my own that were just Windowed and the results are all the same. As to what the
technical differences are between the two modes, I don't know (or if there are any differences besides one
window being 'maximized').

[jiagu-wimem]

what here talk about ? not hardware tools ??