AMD 7970HD 3Gb and CoD
 

S!

So today The Card arrived. Just got it installed with latest drivers and had time to quickly run a test with CoD. For AMD users oh the joy, the 3 lines of death are GONE!! YESSS!! Spit vs 110 track..maximum 68fps minimum 2fps(when it loads into cockpit), average 60FPS! :) I have VSync on.

Will do some more testing and report in this thread, but from what I tested the card is DAMN fast! :D Could not be happier!!!!!!

Congratulations! :grin: And great news on 3 lines of death :)

Could you run the test with SSAO on and off, and with vsynk off and on also just to compare to gtx580 performance tests from here http://translate.google.com/translat...hp%3Ft%3D68723

What is your AA setting and resolution? With vsync ON triple buffering must be forced ON I think to avoid short drops to 30 fps in complex scenes.

gtx 580 results vary from 51 average to 65 with vsynk and SSAO off. Clouds ON in realism settings.

43 avg with SSAO and vsynk ON AA 4x 1920x1080

You can pause the track after loading cockpit and only then start the counter to avoid 2 min fps. You may also first run it once without a counter to load all textures into memory and record results only in the 2nd run. This is how they tested 580 and other cards there.

Do you experience the stripes bug on the border of land and water?

PS. E.g. sample results for 580: What are your CPU clocks?

Excellent news! It's just so damn expensive :evil:

S!

I run the card on the stock settings 925/1372MHz, 1920x1080 32-bit resolution. SSAO is Off as well as AA. Will post my CCC and game settings, need to plug in the joystick etc. again..had CoD kind of shelved before the patch :oops:

Thank you very much! Could you please run "The black death" track. The results for gtx580 are for this track. I did not notice you were running a different one, sorry.

S!

Ok, made some quick runs with Black Death at 1920x1080 resolution and I think you find out something interesting in results. Mind you I did not clear cache before or between the tests, it MIGHT have an effect but anyway..here settings and numbers!

CCC Settings, Performance driver 11.12 from AMD

Antialiasing Mode : Use application settings, Morphological filtering OFF, Filter Standard.
Anisotropic Filtering : Use application settings
Tessellation: AMD Optimized (I think has no effect in CoD whatsoever)
Catalyst AI : Performance, Surface optimization Off
Vertical refresh : On, unless application specifies
Triple Buffer : Off (OpenGL setting only)

First runs I ran at what I had from my 6970HD. Cache was NOT cleared nor game restarted between runs in any phaze.

In-game settings...

Full Screen : On
Antialiasing : Off(in all tests)
Epilepsy : Off (in ALL tests)
SSAO : Off
Vsync : On

Model detail : High
Building detail : High
Land Detail : High
Forest : Low
Visual effects : Medium
Texture Quality : High

Damage decals : High
Buidlings amount : Unlimited
Land shading : High

Grass : Off
Shadows : On
Roads : On

Black Death ran once to load textures. Then the benchmarking run with FPS counter on. Counter was reset on each run to get consistent results.

Minimum: 10fps
Maximum: 68fps
Average: 45fps

Next run with VSync OFF, otherwise all remained same.

Minimum: 10fps
Maximum: 112fps
Average 48fps

Next run with SSAO ON, VSync Off. Otherwise similar settings. This was interesting!

Minimum: 10fps
Maximum: 111fps
Average: 47fps

Next run SSAO ON, VSync ON. Otherwise similar settings. Again interesting!

Minimum: 10fps
Maximum: 67fps
Average: 44fps

Then I changed graphics settings to VERY HIGH, VSync ON. Interesting average, should run another run?

Minimum: 5fps
Maximum: 67fps
Average: 34fps

Last run VERY HIGH, VSync OFF. Interesting results again..

Minimum: 5fps
Maximum: 112fps
Average: 45fps

So here a quick and dirty testing. Will do some cahce clearing and other tweaks. My system runs ALL at STOCK speeds, no overclocking. Signature for more info :)

Thanks for sharing so detailed report!

Your registry may also have data remaining from 6970. It may help to make a clean reinstall of the drivers with DriverSweeper or DriverCleaner in safe mode.

The stock speed of your CPU is only 3.3GHz that can limit such a fast card like this one similar to gtx580 sample in my 1st post above. Do you have any utility to log your CPU load during the test? "Black Death" is very processor dependant I guess.

Highest texture setting is "Original". It is higher than "High".

S!

Did clear the game cache to see if any difference. Settings in game are VERY HIGH. In CCC same as previous post. Results with and without VSync below.

Vsync ON

Minimum: 5fps
Maximum: 67fps
Average: 43fps


VSync OFF

Minimum: 5fps
Maximum: 112fps
Average: 45fps


So as an conclusion so far can say SSAO has almost no effect on any area of performance as it had before. Propably for best game play I would use VERY HIGH as base but change the following:

Building Detail: High, from Very High
Forest: Low or Medium, from maximum
Texture Quality: High, from Original
Visual Effects: Medium, from High

Will run a test with those changes and report back shortly.

This would be very strange unless processor is the bottle-neck.

Could you log your processor and or GPU load percentage?

S!

Ran Vsync on/off tests with settings above. Results were OK, not much difference but again the track does not represent actual gameplay IMO. But the results :)

VSync ON

Minimum: 6fps
Maximum: 68fps
Average: 44fps

VSync OFF

Minimum: 11fps
Maximum: 112fps
Average: 46fps

Do you use a logging program of some sort Ataros, will gladly provide CPU/GPU usage stats for you :)

Thank you! This one has a "Log to file" checkbox on the 2nd tab for GPU. http://www.techpowerup.com/downloads...-Z_v0.5.8.html

For CPU logging I have not used any recently, maybe someone else can help.

My concern is that with SSAO ON your tests deliver 3-4 fps difference with gtx580 (running 4x AA). But switching SSAO off gives gtx580 substantial fps increase.

Maybe a clean reinstall of drivers with DriverSweeper may help. I would also reinstall directX and .Net after the clean driver installation just in case :)

S!

Ok, will get that proggie :) I did clean up old drivers and all. But never hurts to check again. Still the results are very promising as there are no graphical anomalies at all and the shimmering is less too. Back in a moment..testing again :)

I am sure that the drivers are not fully optimised yet and there will be 10-15% performance increase in the future at least.

PS. Do you have the UBI logo deleted? :)

Very good news. So you do not have any stripes on the border of water and land? They bother me a lot.

S!

Yes, got UBI logo disabled. Strangely enough the log shows that my GPU never went above 150/300MHz?!! Maybe the program does not fully support this card yet as it showed no GPU usage either..

Also waiting for new drivers to improve the performance and add to that the patch that should fix a lot of things for us performance wise :D So good times ahead...

Will continue testing and writing back..Need also to test online as well.

You can try ATI Tray Tools to make sure. It does not log to file but shows on screen display. http://downloads.guru3d.com/download.php?det=733
Go to Tools & Options >> OnScreenDisplay >> Enable OSD on 1st tab and then >> Flash OSD to configure which info to show (CPU, GPU, etc.)

S!

The GPU kicked in as should, runs at full 3D speeds. Did some testing. Even SSAO did have almost no effect on tracks it has in game when flying low. Increases ocasional stutters, without it smooth. So left it off as I did not notice any IQ difference either.

As this gave you only 1 FPS I make a conclusion that your processor is the bottleneck. You can get up to 10 more fps if you overclock as shown in my 1st post for gtx580.

Do you have any stripes bug on the border of water and land? I mean not 3 lines of death on the horizon but water stripes going inside land and land stripes going into water.

S!

Have not noticed any stripes so far and also the strange anomaly at the edge of draw distance with better textures is smoother. Clearly visible now over France but less so over England.

As of overclocking..even the 2500K does it nicely I am not very keen overclocker anymore. Too much fiddling around. Sure maybe a few MHz could help, but I want to OC only the CPU in that case as 1600MHz DDR3 should be fast enough.

As of testing other games...World of Tanks ran great. GTA IV had a strange anomaly, it let me play only in 800x600 because I was at "edge of running out of resources"..WTF?! 200mb of 3Gb used and game says no resources?! Now how I can reset the game settings to get it to work again..

Anyways..back to testing. Last night took a quick swing at a Repka dogfight server. Ran smoothly without stuttering, so very nice :)

Very interesting to hear about your FPS with the new card. I am getting I believe the same or close to frame rates with my 3 gig nvidia card. I am
also not a big fan of overclocking, I have tried it a few times, however I don't see the point in possibly frying your cpu/motherboard for a few more FPS, it's
to me just not worth the risk. I think your stock CPU with your graphics card should not be perceived as a bottleneck. I can't imagine you even noticing
a 10 FPS increase from say 60-70. There are many great articles on what exactly the eye sees, how it perceives it, and I'd be happy with 60 FPS.

For comparison performance in ArmA2 http://www.bit-tech.net/hardware/gra...0-3gb-review/4

(visibility 3600 m. only, not maximum 10 000)

Flanker, did you try pseudo-fullscreen mode? Luthier mentioned it may help ATI cards.

Also if you can quickly look through this list maybe you notice something that can hold the card back if it is not the CPU or drivers (like antivirus real-time shields, epilepsy, etc.) http://forum.1cpublishing.eu/showpos...4&postcount=13

Hello Flanker, do you think you could test something for me? My frame rates are roughly those of yours. When I am landing, and press for the 3rd person view of my plane, as it touches down, and the dust starts, my frames drop under 10 FPS. Can you tell me what your FPS are with your card with this scenario? Thank you!

S!

Will do :) Reporting back in a moment.

S!

FPS dropped from 60fps to 40fps at worst dusting while landing. So not a big hit compared to 6970HD where it went very low. Still the particles, smoke etc. could use an optimization :)

Ataros, my bet is on the drivers as the card is so new. A few driver revisions might give us a nice boost etc. But checking those links you put up :)

Thanks Flanker, that is pretty amazing. Were you inside the cockpit when you noticed the frame drop from 60-40, or were you on the f6 or f8 3rd person chase scene when the dust kicked up? If so, I cant understand why the geforce 580 with 3 gigs would take such a huge hit, my cpu runs at 3.4 and its the fastest AMD multicore out there. I may have missed it, but what screen resolution are you running?
Thanks again for your help!

S!

I was in F2 view looking back at the dust. Running 1920x1080 resolution. Just feels strange that only game with problems is IL-2, rest run without a glitch now. Well, I remain confident about the performance patch Luthier & Co are working on :)

Could you please make a screenshot when you see it next time over France? BTW make sure Anisotropic Filtering is set to Application Controlled in drivers. Otherwise this could be the reason.

S!

Allright, will get you the screenie :D I use Application Controlled in CCC as most games work better with their own settings than via forcing them. Less tinker, less hassle most of the time.

EDIT: Here is the pic..you see the yellowish is the low rez and other is the normal.

S!

Here more easy to compare. In Textures-picture you see the French coast where high resolution meets the low resolution aka view range. In the GB Textures you see the transition is less pronounced, look at 1oc, the lower resolution is lighter green but blends better. I hope this clears it :)

Yes, I can see, thank you. Does the normal green-textured area make a square field around the player if viewed from a higher altitude?

S!

Yes, the "view range" of high resolution is a big square around you. It could be bypassed maybe with a bit heavier fog/haze as you do not spot a plane over coast of France from Dover anyway, at least in the game.

Thanks for testing! I hoped I get this square only because I run the game on low or sometimes med landscape details settings. I hoped it should disappear on high settings.

S!

Let's hope the new patch will address these issues as well. But at least no 3 stripes of death anymore! :D

You are killing me Flanker, I am just waiting for the non-reference versions to start showing up before I purchase one.

JD
AKA_MattE

I'm just waiting for Nvidia to come out with their cards before picking between that or the AMD card.

S!

I wanted a non-reference one too but small supply of cards only here so had to take what I could. I am happy with it and a lot. As of the nVidia..I am interested in it too, but I am sure it will be a lot more expensive than AMD. At least there was over 150€ difference here between 6970HD and 580GTX so was not worth it...But anyway..great to see new tech coming :D

Everything I read indicates the new Nvidia Keplars are not going to be out for some time. This means, for now at least, your limited to the gtx580 3 gig or Flankers new AMD card if you want to get these high frame rates. I still do not understand while Im having such an issue with my "dust". What clock speed are you running your cpu Flanker? I believe you said 3.3?

Thanks for your help with my questions.

I was contemplating getting a 7970 myself but have decided to wait a bit. If the games graphics engine has been optimized I hope SLI/X-Fire is fixed. I might get another GTX570 if that's the case. If Kepler is out by that time I 'll weigh that one in as well.

S!

Priller, I run the 2500K at stock speeds, so it might be this 3.3-3.4GHz. Kepler seems to be a nice thingy, but remains to be seen. And AMD is for sure planning for a refresh before that :) Interesting times!

BigC..I am sure that when Luthier's team get the GFX engine working with proper SLI/CrossFire an another 570GTX will do just fine.

Yes, I too read that AMD will have newer line/updates out by the time kepler is even out.

Even if it turns out that the AMD 7XXX card is the better one, Nvidia releasing their new flagship product can only help pricing. AMD usually drops their prices a little if a new Nvidia card has comparable specs and performance, at a similar price.

S!

Which I see is a good thing. The flagship cards are after all not the main market but the mid range of 200-300Euro/Dollars. But for the companies the most important is of course right to brag with the fastest card on market :D

For now I am more than happy with the performance of this AMD 7970HD. It screams thru all games I've thrown it and by a margin over my previous 6970HD. The card is quiet, uses less power and temps are in check too. So far seen 30'C on idle and some 63-70'C while gaming..my case and cpu fan make more noise than this card :D Now it is just to wait for the drivers to mature a bit more, AMD has done great work with the card and their long term plan seems to pay off now :)

Did you manage to see typical GPU load percentage with the ATI Tray Tools?

For my old card it is 100% in most situations but IIRC for gtx580 it can be as low as 60-70%.

S!

Managed to get the usage with GPU-Z. Clocks ticked in normally 925/1375MHz, memory was used up to 921Mb average being 800Mb or so. GPU usage varied between 20-91%, average around 40-50%.

Only 40-50%? Is that a cpu bottleneck or is the games code at fault?

S!

Editor, my bet is on the game code. Eagerly waiting for the patch to see what it brings to the table :)

On the other hand this shows that the game scales well with CPU speed:

S!

Seems the 4GHz is the sweet spot as above it the gain is less. I rather run stock as other games cause run extremely well anyway.

Does anyone, AMD or Intel even make a cpu that runs stock at 4.0 or above?

The Intel Core i7 3960X comes close in turbo mode (3.9 GHz), but you really would just get the 3930X (3.8GHz) and overclock it for the $500+ difference ;)

Hi.

Thank you very much Flanker35M and Ataros. Very intresting results. I also wait the non-reference models like Asus DCII 7970. Have to wait middle of february.

http://www.tomshardware.com/reviews/...rk,3123-3.html

Here is the 7950

S!

7950HD seems to be a strong performer in it's price class really :) Times will get really interesting with nVidia's offering coming out at some point.

Thank you.

Hm... Anyone knows if CloD is more DX9 or DX10 application (or both depending on conf.ini setting)?

Does changing Render=D3D10_0 to D3D10_1 or D3D11 or D3D11-0 in conf.ini change performance?

dx 10 for CoD
 

Hi there Ataros, this is from a post Luthier did on performance.



E. NOTE: DirectX 9 offers slower performance and lower graphic fidelity. If your system meets Recommended system requirements, upgrading to a DX10-capable operating system will offer better performance boost in Cliffs of Dover than perhaps any hardware upgrade.

http://forum.1cpublishing.eu/showthread.php?t=19819

Hi, MajorBoris. Thanks for info.

Then we are lucky we have a DX10 engine at least unlike some other games.

Hi!

What does D3D10_0 and D3D10_1 means? Off and on? Thanks. I did try D3D11_0 but it had a performance hit for me. Now I use D3D10_0.

I think these are different versions of DX10 or different ways the game uses DX10.

S!

I use the setting DirectX 10_0 meaning it uses DirectX 10.0. Not sure if changing to 10_1 does anything. But that we know that 11_0 does not fully work yet. And as of AMD not working in DX 9..I doubt it. Played older games and they run just fine.

I'm trying to figure out why the 3gig card Flanker is using does not take the FPS hit on particles, ie, dust, gun firing that the 580GTX 3 gig experiences. Any ideas?

It has different architecture and is generally faster according to many other tests. Thus it may be faster in CloD in complex scenes where FPS is minimal.

But I am not sure it does not take the FPS hit on particles. To prove it you have to run the same track under the same conditions (CPU speed & cache size, memory, game and driver settings, software running, etc., etc.)

Welcome to the land of Big VCards Flanker.

I ran the Spit vs 110 track. so you have a compare. Vsync on i get .
maximum 68fps (spikes) minimum 60fps, average 60FPS!

Vsync off avg FPS 89 Max 118

My nvidia settings all standard except I set me pre rendered frames to 8

No real point in comparing FPS. The game is so unoptimized over land it's just a mess. It should easily pull 100fps on full detail over land without any stuttering.

Hi Priller,
Why do you say it does not take an FPS hit?

From what I saw in the posts I read, Flanker said it does.


~S~

By the way, I do not hear good news about the next NV generation :(
I hear about 2Gb VRAM (only) and special code delivered by NV in order to be added to the game code in order to achieve vastly superior performance = re-translated: No special graphics code written, graphics performance will be unimpressive :(

S!

Indeed not good news for nVidia :( I hoped there would be a nice competetive card from them, but this "add a code only for NV card to make an application to work fast" sounds VERY bad. Adds more discrenpancy to gaming etc. when special codes are needed. On the other hands shows that AMD's long term planning is paying off now, they have very good lineup and a solid product to upgrade later on.

And yes I take a hit in FPS with dust but not so much as with the 6970HD. Actually now my gaming is VERY smooth, only occasional hiccups when something is loaded. But otherwise very nice :) Drivers to mature and new patch = good times!

source? Nvidia isn't releasing much data on their new cards at all! We just have to sit back and wait for real news/release time.

Source: some people who are running the benchmarks ;)

I think all the dust particles , bullets , explosions ect are mainly to do with the CPU. I upgraded a few days ago to a 2500k from a E7400 and now all the micro stutters when in combat have just about all gone. The rest comes down to bad coding in the game itself.

If the engineers at Nvidia see any advantage to adding a larger frame buffer, Kepler well have a larger frame buffer. Ram is pretty cheap.
Some of the same sites that have the leaks specs also say its at least 45% faster, we won't no the truth until actual cards get out in the wild.

S!

As r0bc said, when we get the actual tests we will know. Until then it is only wild speculation and marketing hype :) But nevertheless interesting to see how Kepler will do. One thing is sure though, it will be premium priced some 100-200€ over 7970HD as always with nVidia.

More Kepler info... time will tell.

http://www.3dcenter.org/news/die-akt...rformance-chip

http://www.brightsideofnews.com/news...-leak-out.aspx

Not sure I believe the prices ;)

thanks Kruppi, interesting posts!

@Editor: Unfortunately my first "prediction" is confirmed, only 2Gb VRAM seems to have become a widespread rumour :(

Time to start praying that Luthier's new graphic patch will handle VRAM more efficiently... 2,4Gb VRAM is what we need with max settings :(


~S~

Hmm well my 5970 which only has 2GB and no crossfire seems to work okay :D... okay not with max settings but hopefully with the next patch we won't need 3Gigs of vram to run the game well lol

S!

Is that CUDA for OpenCL and similar things and has nothing to do with gaming? If so then what use it is to stuff a lot of them if no game uses them? The article was a bit green tinted so I will wait for actual figures when Kepler is released. But it can be a strong contender to AMD Tahiti series which is good as then the price wars begin :D

Not necessarilty. As NV cards, or more accurately the driver allows to outsource some calculations onto the CPU, this increase in fps may not be mainly due to ocing the cpu but a result of nv drivers. so, in case of nv-gpus, it would be nice to know, if that particular setting (iirc "Multithreading" or something alike) is activated or not, and best, such tests should be run with it on and off to get a conclusive result.

This is of much concern for me, as i´m planning on replacing my Q6700 (as it is already 4 years old, hence i don´t call it an upgrade :D ) and would like to know, what cpu will i really need. not just for clod, but other games as well.

S!

How many games really use several cores efficiently, as what I've read it is not "just like that" to program real multicore support or such. Will be interesting to see how Kepler will be and the upcoming AMD updates. Seems the GPU contest is getting very interesting :D

Well, my cpu...AMD Phenon 1100T x6 at 3.8 with win 7 64, 16 gigs of ram, and the 3 gig nvidia 580GTX SHOULD be able to easily handle particle effects, it does with no issue whatsoever in DSC A10, and shouldn't with this game as well. When frame rates come down from 60 flying to 8 upon landing kicking up dust, that tells me its not my cpu or gpu.

Thanks all for your help!

As far as CPU's, the 2500k is a great pick. It doesn't run much above 50% in CLoD but atm that's due to my out of date HD 5770 card bottlenecking it.

All these GPU stats don't mean a thing to me until I start seeing the benchmarks for games. I'll then buy the best card between AMD/Nvidia base off of price/performance.

S!

It will be interesting to see the game benches coming out. But again in most cases the FPS games are used to measure as they are the majority. Very few reviewers use flight sims so I guess that task is up to us to do :D

Sounds like you just volunteered! :)

I think I have one:
http://www.tomshardware.com/reviews/...-gcn,3104.html

Look at the difference between the two GPUs in terms of
Stream processors: 2048 vs 512
Texture Units: 128 vs 64
Texture Fillrate: 118.4 Gtex/s vs 49.4 Gtex/s
Memory Bandwidth: 264 GB/s vs 192.4 GB/s

Except for the Memory Bandwith, the 7970is more than 100% percent powerfull than the 580GTX.



I admit, I started showing signs of weakness....I was holding one 7970 in my hands today... :(


~S~

Hold on brother, the wait is coming to an end....Hurry up Nvidia!

ROFL Editor, nice one! :D :D :D

I am afraid the NV cards we need are the models planned for August (the GK110 core), not the GK104 core NV will introduce now.


@Flanker: How is the noice of the 7970? I see that this was a point of misery in the tests...

~S~

S!

I find the 7970D quite quiet. It idles at around 30'C and under load seen figures like 60-75'C. The fan has not gone near 50% ever. I play with headphones so can not hear the computer next to me anyway.

Is the 680 released this month a Gk104 or Gk110 and if not which ones are earmarked to be Gk110

Apparently, this month is a GK104 and the GK110 is expected after August.

~S~

Ok I know it is all speculation but I have read in a few places that the 680 is a gk110 and it might be out in February

Where can you find this ?

Pure Speculation (unfortunately :( )

There will be the GK104 series now. Apparently two versions of it just as they did with the GF114 core:
The GTX560 is marked GF114-325.
The GTX560ti is marked GF114-400.

And the GK110 series which will be the high-end core will come in August.


~S~

Right ok thx for explanation

How much power does a 7950 or 7970 take? I have a 650w Corsair psu. Is that good enough?

I'd be up for upgrading later this year, along with a 27" monitor.

I`m running a 7970 with a 600w PSU, all ok

S!

I use a Corsair TX650 and is more than enough. I do NOT overclock which would require a bigger PSU. The 7970HD draws LESS power than 6970HD. Also getting rid of traditional HDDs in favor of SSDs is good, quite a bit less power draw and noise..even sometimes you miss the rattle of HDD :D Oldskool!

Flanker,

This maybe of interest to you... http://forum.1cpublishing.eu/showthr...d=1#post392655

If it does, WE WANT PICTURES! (and fps) :D

~S~

S!

Will test when home from work and post here. Just as sidenote..was wondering why game all of sudden ran slow. Cleared cache, checked GPU setting etc. At last found it..due Steam cloud the conf.ini files differed for some reason AGAIN! So checked and saved = smooth again. So worth verfying these files now and then..might help.

Thanks for this! Never use cloud myself but good to know. If true, it probably merits a separate thread on it own, it may get lost in here...

Flanker, when you post pictures, is it possible to also have the fps frame rate shown? (fps SHOW START)

~S~

S!

Maybe Ataros could add the conf.ini check to his tips & tricks thread? Will turn on FPS when testing!

HD7970 vs GTX580
 

First of all I can re-confirm, as it has been mentioned before in this thread, that the HD7970 3GB GPU
works perfectly with a 600W PSU.
So no additional cost for the upgrade.

I have 2 basic systems running CloD based in P67, 2500K and 8GB RAM with the only significant difference
of the GPU a HD7970 3G vs GTX580 1,5GB.
They both run in 1920x1080, they both share, as much as possible, graphic configuration in the driver
control panels, the both share graphic options, as much as possible, in the game.

In game Graphic Options values in high, but Low in Forest, Low in building detail, Medium Effects,
Medium Decals.
Unlimited buildings.
Grass, Shadows & Roads on.
VSYNC in game On for ATI, VSYNC forced On in driver for Nvidia.
I do not want to enter much into detail because at this moment this is not the objective of this post.
Even though I am open to any specific request.
All the comparison has been done with the Black Death Track, no flying low over London for me…

With both systems I have very similar results.
Average always over 50FPS with consistent values in the 51-53 FPS.
Maximum limited to 60FPS (VSYNC ON).
Minimum 3-6FPS for Nvidia 12-15FPS for ATI.

Here, in the minimum, is where the most significant difference appears. ATI is able to handle better,
twice as better, the effects related to dust, explosions, fire and so on. Still the frame hit is excessively
stressful for the GPU.

When the hurri is blasted by explosions while is trying to take off is the moment of maximum stress
for the GPU. There are others, for example the ball of fire after the vertical crash of the wingless hurri,
or the right turn of the bf110 around the huge smoke column.

No difference in AA, that is bad. External tool FXAA works perfect in both configurations.
Shadow “flickering” same for both, neither able to correctly manage the shadows of smoke,
bad for both.

Others consideration, after a 4 times cycles of track playing the ATI reaches a stable temp. of 70ºC
with the only standard available fan/cooler. GTX580 seems to be cooler but the model I have do not
uses the standard cooler. Noise level is always at acceptable values.
Nvidia seems to be more comfortable with the original Il-2 at max. settings (not shown in FPS counters),
but ATI has solved all issues with the OGL drivers.

Beware I also belong to the "no manned cars, no manned tanks, no manned flack, just planes" lobby group.
So this could be all wrong. ;)

Salutes, Majo.

Thanks for all the comparisons Majo, much appreciated! :)

It confirms that ATI7970 is the strongest card today and at the same time not enough (we need to wait for a year or more until GPUs are able to give us those 60+ fps we have with IL2 - or a miracle with Luthier's new graphic model).

Maybe it is the 1,5Gb VRAM of the GTX580 that is the limiting factor causing this lower min fps?
Is it possible to look how much % percent of the VRAM is utilized? I know for NV you can do this with NVInspector but I have no experience with ATI.

I rememeber well, the max VRAM measured (flying over London) was 2,4Gb VRAM. So maybe the GTX580 1,5Gb has its hands (ehem..chips) full very quickly thus struggles with the memory I/O much longer than the 3Gb 7970

~S~

S!

Need to run Black Death and this weather test to see how 7970HD fares on my system compared to Majo's. Not much difference between except I use Z68 chipset.

Which Drivers are you guys using? My 7970 shows up today and I see they have a 12.2 driver out.

JD
AKA_MattE