1.03.15527 system performance reports
 

In short, I get a 10% framerate reduction with the beta patch. If you are going to post your feedback here please:

1. Add your system specs to your signature.
2. Make sure you have cleared the cache before testing (see here).

There have been a lot of comments about framerate drops with 1.03.15527 beta. Smooth performance is one of my biggest concerns (I actually thought the cockpit sounds were fine as they were :)) so I compare results after each patch.

I like the new colours and appreciate the patch but it is important to note that this patch HAS resulted in a significant performance hit as you can see below.

The below are in triplicate (yes, I'm a scientist!) although this is not necessary given the tight results. Results are before and after patch, vsync on and off with the following video settings. I used FRAPS to measure the first 60 seconds of a low-level Dover dogfight test track attached to this post:

http://farm7.static.flickr.com/6187/...aca1b95d_z.jpg

http://farm7.static.flickr.com/6210/...5c66d4a8_b.jpg

Comments:
- I think the most important column is the minimums given that a reasonable average with low minimums can result in stuttering and lags.
- Gameplay is smooth without stutters on my system UNTIL a lot of planes appear. Then I get significant input lag (reported here and here) which is annoying.
- Given that I only play with vsync due to atrocious tearing this performance drop is fine for me.

Jimbo, did you defrag the game files from the Steam menu?
Didn't help me but might help you.

Yes, I did but forgot to add that above. I'm not complaining since it doesn't affect my gameplay with vsync, just putting a figure on the reduction.

I agree. While I haven't really experienced any stutters or really serious performance issues, I'd estimate I've lost around 10% as well. I'm sure some optimisation work for the proper patch will get performance back again. :)

I hope your sureness is sure! :-P

This are my video settings you can see that only the ground shadows are on medium in combat in free flight I set it to high;
http://s1.postimage.org/m5p1ujcq1/video_settings.jpg
Now this the picture is the result;
http://s1.postimage.org/4fpmatzdf/sh...907_125242.png
Not too bad…
My system is at the bottom of this reply.

Have a nice game.

My PC; ;)
Win7 x64
Asus p6x58d
I7 975 3.75Ghz
System & game HDD 256Gb 10.000rpm.
12 Gb (6*2Gb) ddr3 1333Mhz memory
Geforce gtx580 1.5Gb
Trackir 5
Logitech G940.

Well, I also got a small performance drop - not that it has any effect as the frame rate is so high that I really don't care.... But for the sake of it here are my Fraps benchmarks for the Black Death track with 1680x1050 resolution @ 120 hz and settings on default high + 2x AA

Pre patch:

Frames, Time (ms), Min, Max, Avg
14571, 216436, 20, 153, 67.322

After the patch:

Frames, Time (ms), Min, Max, Avg
13133, 217544, 10, 136, 60.369

So there is a drop but most of the time I average 70-80 fps in normal gaming so there is really no need for more.

And by the way, the Black Death track really benefits from the new lighting and sounds. A lot... Was the blast wave from explosions there earlier or did I just miss it? And I don't remember the dust kicking up when the delayed fuse bombs landed?


EDIT - here are my patched results for jimbop's track (no vsync only). Really thought my rig should be a bit faster than yours. They almost identical but the 580 should give an edge over the 480 one might think... I guess it's the fact that you run a tiny bit lower res that gives you the edge, and obviously it's the CPU that limits in this setup more than the GPU and there we are identical.

Frames, Time (ms), Min, Max, Avg
4963, 60606, 61, 157, 81.890

You missed it. :-D

Interesting, thanks for posting.

Hi Mazex, I see you have AA set to x2, I have tried all settings of AA from x2 to x8 and never noticed any change, are you seeing a change with AA on?

According to the patch thread FSAA is not enabled till the next patch along with optimization improvements. This info was from the sukhoi forums.

Then the next patch will be the big one. Optimization and AA are game breakers, landscape colour and sounds are not (although I do like the new improved ones very much).

Someone said that 2x works on some configs and I try to convince myself it does that some way, even though I deep in my heart know it doesn't. As I use x16 in all other games I don't really know how x2 looks ;)

You will be happy to hear the next patch will have new aa then mazex :D

same performance with a little drop here and there with less stutters almost visible ;-)

Well, seems my heart was right after all... Was going to check out the new Thames beach - and found both the beach and a horizontal grass of rather horrifying proportions. When I paused to take a screen shot I did try one with FSAA set to "OFF" and one with x2. It definately works at x2 at least for me.

FSAA off:
http://img508.imageshack.us/img508/9538/fsaa0.jpg

FSAA x2:
http://img13.imageshack.us/img13/232/fsaa2.jpg

And no, I did not do any jpeg optimization or such...

But look at the antenna wire...:(

Cheers
Uffe

Game is running fine...maybe better but not worse from last patch.

AA has always worked in this game, the difference in the cockpit is incredibly obvious. There are limits to what AA can do in certain situations with modern shaders. However the shortcomings really only ever becomes obvious from external views.

Then I think you are the only one that has it working with any reasonable amount of usefulness.

JD
AKA_MattE

regarding performance,....with the same settings, i have the same fps like before the beta patch.
but i have set textures from medium to high and landshading from low to high, and the fps loss is way smaller than it was before the patch.at least thats my first impression.i think they did some optimizations regarding the textures.

What a pile of apologetic whatever. Its not working and you can't even force it on the card! Even the devs admit this FACT. LOL:rolleyes:

But still, many here say that it does not work at all which is obviously not true. Look at my example shots. Sure it's not x16 but look at the lower "edge" of the plane for example. The coolers etc look way better on the x2 shot than on the shot with FSAA set to "off"...

If I run black Death track with or without 2xAA set in the game menu I see a big different on, for example, the airscrews on parked planes.
Forcing AA on the card (CCC) doesn't do nothing.

OK, lets magnify the images I posted on the previous page...

Not working? Not working good I can sign but...

http://img853.imageshack.us/img853/4...comparison.jpg

that's just an opinion and far from fact

Yes Mazex, it is working at the lowest setting, but even you point out the poor rendering of the antenna wire . That being said in terms of minimizing the vast majority of Jaggies CoD is not working yet, so you could say in a very general term it has been working all along (albeit very poorly) but that statement is not really indicative of the true performance of FSAA in CoD.

But this is just semantics, it certainly is working on your rig.


JD
AKA_MattE

Not fact AA is slated for release in next version :D

thanks JG52Krupi for the heads up
love to see the new animations and the AA :-D

Well I'll post my stats when I get back home but currently mine is a game of 'stop-start' graphics. I can be playing smoothly then I get a freeze of 1/2 a second, flies okay again, then stutter,stutter, smooth, stutter.

This was while playing Hawkinge and I was using 1600x1200, everything on Max except Trees on medium and Textures set to High not Original.

I did turn Grass on but SSAO off as-well as Epilepsy Filter.

Graphics look fine and the draw distance pop-up is now much better, but the freezes are a pain. I'll go through the defrag process later, but my system is;


Windows 64-Bit Ultimate Edition SP1

Asus P6X58D-E Motherboard

Intel® Core™ i7-950 Processor (8M Cache, 3.07 GHz, 4.80 GT/s Intel® QPI) stock speed.

XMS3 — 6GB Triple Channel DDR3 Memory Kit (TR3X6G1600C8), XMS Memory for triple channel systems, 1600MHz, 8-8-8-24, 1.65V

MSI N560GTX-Ti Twin Frozr II OC Edition

1000W Thermaltake Toughpower PSU

Dell 24" Monitor 1920x1200 @ 60Hz

4x 1TB Samsung Spinpoint 3 HD @ 7200rpm + 6 External 1TB HD

Track IR 4 with Track Clip Pro

Simpeds Rudder Pedals

Thrustmaster Cougar HOTAS with Modded Hall Sensors and U2Nxt Gimbals on Stick

Hall Sensors on Throttle

It does not say the next patch IIRC. It says it will require more time which could mean 3, 4 or 10 patches. Be careful with assumptions please to not cause frustration in future.

In several modern games old FSAA technology is useless but new technology is time consuming and expensive.

MLAA and FXAA can be used meanwhile on modern cards I think. There was a thread on the forum about it some time ago.

i meant landscape colour and sounds are real deal for my taste. now we have transparent shores:mrgreen: the rework on the graphics engine is not perfect (somewhat excessive blue saturation) but is a huge step in the right direction (transparent shoreline, better water, darker trees, more natural colour for landscape, better shadows, less flickering, better effects, clean canopy glass,...)

i hope this settles the debate on whether if graphics improvements are important, of course they are important;-)

Lets not forget this is a BETA patch...it's not final yet. So maybe when it does go final the code will be tightened up and it won't be a hit to fps.

I'm going to keep playing my older version until this goes final.

I realize that I am running Cliffs of Dover on a non supported system but I wanted to report how it was working prior to and after the Beta patch.

The largest signifigant change was the sound lockups I was having when firing the guns. If I held the trigger down too long (couple of three seconds) sometimes the gun fire sounds would get stuck in a loop and lockup the Laptop. At first the only way to stop it was to do a hard shut down on the system. Later I found that I could stop it by muting the speakers. This would clear out the issue and I could continue on.

With this patch that condition is completely gone.

Other than that there have been no noticeable performance increases or decreases which is good.

It runs at a solid 20 - 24 fps second which seems low but it is still rather impressively smooth none the less.

MysticPuma,

Same here. The stutters are back and fps have dropped quite dramatically. Something's wrong with this beta patch although some report no fps loss at all which I find very strange.

Regards,
FP

Ive had a Huge increase in performance I was getting low 20's prior now 50+ average, seems whatever they did made my crossfire start working better 11.8 drivers. I also see long blades of grass at times.

Seconded. FPS seem to have dropped by 1/3. Zooming in decimates FPS, and so does flying through clouds.

System specs:

i5 2500k
P8P67 EVO
8 GB RAM
GTX 580 Lightning (driver v. 275.33)
Asus Xonar STX

Yeah, thats whats happend to me too. Have reverted back to official version and performance is back to reasonable levels again.
My system specs;
E8400@3.6ghz
GTX460 756mb vram
4g DDR3 ram

performance is very smooth,lovely flying over kent in the tiger moth,trees look better in fact the whole land area seems much more natural and its a pleasure to fly over it now.everything on max apart from buildings

No problems here, looks pretty good and flys pretty good. TC

I don't know if it's lower FPS but anything higher than 30 is good enough for me and i do use Vsync, so i don't have any use for more than 60FPS because i have a 60Hz IPS monitor.

In that sense i prefer having 40 FPS with more consistency than having 60 FPS that dips into the low 20s. Think about it, if the FPS are limited somewhat on purpose the PC has more time to render what it needs to display and this helps maintain a more constant frame rate.
This has been a very often used tweak for FSX as well.

Well, in the current beta patch that's exactly what seems to be happening, most people report lower FPS but more smoothness.

In other words, losing 5-10 FPS is only a big deal when that drop results in your minimums getting below the human eye's comfort zone. If you have 50 FPS pre-patch and 40 post-patch it's no big deal, especially if post-patch is steadier.

And if that drop gets you below 25-30 FPS there is a very easy fix with minimal loss of visual quality: drop your texture resolution one notch, it's the graphics setting with the most impact on FPS. ;-)

Yes, blackdog's post highlights what I said in OP about the minimum figure being the most important. The drop from 90 to 80 in my case is immaterial but if you were borderline before (say 45) then this patch could hurt a bit. It will be interesting to see what happens with the release version.

Personally I can't understand how people can play without vsync anyway. I get really bad tearing when it is disabled but maybe that is a gtx 480 problem?

And Land Detail! These two settings are the "Holly Grail" of tweaks to run CloD fluid without losing much visual quality! :grin:

I believe what blackdog said is happening to me as well and I agree with him absolutely.

Looking at the earlier shots with 2xAA, it looks like that for me too, however if I use only 1xAA the game looks a lot better (going to post shots if I can), in external view I pretty much cannot see any jaggies when looking at the side of the plane (same position as the earlier screenshots).

I am under the impression that forcing vsync on in the nvidia cp gives better performance than the ingame vsync.

Running 270.xx whql driver, after using the latest whql driver, I like this one better, my card does seem to have problems with the latest driver even in win7 desktop environment. I read in another game dev's forum that the whql driver after the 270 is nvidia's new "safe standard" (so 270.xx whql is their last "risky" perfromance driver?) and people in that forum were advised to use 270.xx for their game due to issues with the newer drivers. The other game I am referring to is DragonAge 2 btw.

Oh, and one other thought.

I did play around with overclocking my graphics card and processor, just won't keep up the processor overclocking until I get a decent after market cooler.
Please correct me if I am wrong, like to hear your thoughts on this. It seemed graphics card overclocking didn't really do anything much for me in terms of fps, however with an overclocked processor the game seemed to run much, much better.
I am happy with the perfromance on stock speeds, so like I said no need to push it without proper cooling.
Could it be that the game reacts very positively to processor speeds, you know how some games are more gpu sensitive and some cpu sensitive, which one cannot necessarily tell from the graphical appearance of the game?

Turn AA off, then fly say.. a Hurricane, and look at, for instance, the metal frame around the gunsight. Now turn it back on and look again. There is a clear and obvious difference in pixelation.

The difference has also been shown externally in screenshots in this very thread. Yes there are areas where it is not working as well as it could, but it is certainly working to some extent and always has.

Here are a couple of pics with only 1xAA (for me it works better than 2xAA or up). As I said the 2xAA looks for me the same as the pics earlier in the post and I cannot see a difference between 2xAA and 16xAA.
So yes it still needs some fixing but it is just plain wrong to say it doesn't work at all.

Land shading and clouds are another 2 you can add to the list. Big hit from both of these.

This latest update cost me about 10% in terms of performance, which seems to be in line with what many have reported. Out of interest, I increased my base clock rate by the same amount, and gained the loss straight back again. This is somewhat unusual, as rarely does an overclock produce an equivalent gain in performance – it’s usually less. I tested this a couple of times using the Black Death track as a benchmark, with the same result.

I tried raising the multiplier to 20 on my previous mobo with a much lower increase in frame rates (my previous DDR-II 1066 RAM didn’t have much headroom), but my current DDR-III 1333 RAM runs happily at 1500MHz with a base clock of 224MHz (cpu at 3.80GHz) with no voltage adjustments on either the cpu or memory. For cooling I am using a CoolerMaster Hyper 212, which keeps the cpu temp down to around 52˚C under load (ambient around 26 ˚C when I tested).

It seems that this latest update has placed more load on the system rather than the graphics card, as increasing my card’s clocks produces a negligible increase in frame rates. Roll on Bulldozer and DDR-III 1866 memory :)

I've been running the Black Death-track and notice a 10% performance hit as well. I have to mention though: disabling the logo.wmv really does make a difference on my machine. I ran the Black Death-track with and without it. By disabling it I gained no less than 10 FPS.

FPS in free flight for me is same but have problem when in 6 to any planes i have big fall of FPS from 40 to 5 FPS, like stuttering and i cant shoot on him because is slide show :( when plane get out from my sight FPS back to normal.
sound is great now :)

Very interesting, it seemed to me as well that raising the FSB has the best effect, then raising the mutiplier. Nice clockincrease without raising the voltage. Going to order a cooler tomorrow and then it's overclocking time :grin:.
If you don't mind, are you raising (if at all) the multiplier in bios or with a program in windows. I was planning on FSB overclock in bios and then maybe a little multiplier increase in windows.
Ughhh, and I need a new mobo :rolleyes:.

It would be helpful if you could post your system specs.

Are there any planes in particular that cause the framerate drop?

Yup, this one too. I did a small experiment very early on (i had the game for less than a week), i wanted to see which one can give me more FPS with less loss of visual quality: land detail or land shading?

In the end, i found out that lowering land detail is fine and gives me some FPS to spend on land shading. I think land detail has something to do with draw distance too (how far away from you the LOD transitions happen for the landscape) which means that only the far away landscape is degraded. Shading on the other hand seems to be a blanket effect and it affects all visible landscape.

So, i turned down land detail and in return for some blurriness across the horizon i could turn up land shading, shading is more important for navigating and depth perception anyway (especially when flying low).

I use the AOD utility in Windows as my clever mobo resets the RAM timings if I raise the bclk in the BIOS. For eg, I can set the RAM to run at DDR 1600, but then the mobo resets the timings to 11, which results in a big drop in performance. Using the same settings mentioned above in the BIOS causes the RAM to run at cas 10, instead of the default 9. I can set the individual timings in the BIOS, but experimenting to find the limits is very time consuming. If I use AOD, the timings remain unchanged, although they can also be set using this utility.

I don’t like changing the multiplier as this disables the PowerNow feature and leaves the cpu screaming along at maximum all the time, irrespective of whether it’s doing anything or not. This is probably not really something to be concerned about, but being a pedantic Virgo, it annoys me. And previous testing has shown that there is little to be gained in terms of gaming performance by running the cpu faster on its own.

I am also not much of an overclocker, preferring to spend time playing games on my rig rather than trying to extract the last drop of performance out of it. But it is nice when you can achieve a worthwhile gain without too much hassle.

Lets not forget this is a BETA patch...it's not final yet. So maybe when it does go final the code will be tightened up and it won't be a hit to fps.

I'm going to keep playing my older version until this goes final.