Harley

Welp, It's somewhat bittersweet to report that after installing that new 7800xt, that Iraq now flies remarkably well. A smooth and steady 60FPS, likely more if I turn off the cap. Apparently, it was the junction temp limit throttling the ol radeon vii, causing the driver timeout. I bombed a couple things in Baghdad last night, because why not? Still need to tweak and optimize the driver because surely I can turn some settings up a little, and then I can also likely turn some DCS settings up higher, too. My guess with anybody experiencing CTD in Baghdad: enable SAM in the Adrenalin driver, or whatever the NVidia equivalent is. Mine enabled by default. I can only assume that it will need this enabled as it seems far more detailed than other maps, and the autogen is crazy. Also, apparently GDDR6 and whatever else these newer GPUs have can handle whatever is being asked of them more than the previous generations can. This was the answer that worked for me. I will need to do more optimizing and testing, but it works now. Oh Radeon vii we hardly knew ye.

If the legacy is such a.good seller, it seems that it would be counter productive to offer the superior version, expecting sales to continue for the C model. Unless it were an expansion that required the legacy hornet. *BAM* Make it an expansion to the C model. I'd buy it. I already have the C. I want the extra pylons.

Got lucky on Monday night. I've not run anything that isn't Asus for some time, but I managed to scoop up a fresh sapphire 7800xt on the ol newegg. They are now gone. Will be delivered Friday night. Then the testing will begin. I imagine Baghdad will pose fewer problems then.

Man, the scalpers getting ahold of all the newer generation GPUs should be a dang crime. Can't get ahold of anything short of +50% or more of MSRP, if at all. I'm ready for a new GPU, like a 7900xt, or even a 7800xt, but nothing available anywhere. Frustrating. 4 figures for a GPU is intolerable.

That's what I would figure, honestly. Seems like that wouldn't be difficult to change.

Yes, if I am understanding what you're asking. Any position of the speedbrake, partial or full, I think should result in a countering FCLS retrim, based on the panel position. Full deploy should full retrim, and a portion should retrim proportionally. The switch position is what does it now, where holding the switch in the deploy keeps the flight controls in the retrimmed position. But that switch position is a momentary function, and will spring back to the center "hold" position, and will improperly retrim the elevator function back to the same position as if the speedbrake were stowed upon release of the switch. The airplanes I work on do this with the flaps as well, and retrim the horizontal stabilizer position based on flap position. I know the super hornet does a net zero function when deploying speedbrakes to keep the nose at the same position, and it works really smooth, although that really is a very different bird. But I still think that the retrim function should be encoded against speedbrake panel position, not the switch position. Just to note, it's not a deal breaker, and I love the legacy hornet enough to be building a dedicated sim pit with it, but that part I think is not correct, and changes the nose angle during a strafing run. Hard to compensate for.

I will try this next. I suppose it's just a matter of throttling settings for this particular map if I want to bomb the garbage out of Baghdad.

Ok. Did some work this morning. Set settings to the default "low" preset, and had the AMD adrenaline monitor open under the performance>tuning tab. Started DCS, set at the runway at the airport in Baghdad (don't remember the name) and saw in the temp monitoring that the GPU Hotspot temperature was at 95° while sitting on the runway. Took off and flew around Baghdad for the first time, but it was very bare. But that was after taking the ol radeon vii apart this morning and cleaning/reapplying new compound to the contact areas. That's likely the issue. Throttling while it's running the sim, and timing out as reported by the driver. It's definitely hot, even though the design specs do say that this temp can get to over 105°, but there's surely not much headroom before it hits that, and changing the fan settings definitely rival the sound of the jet when set to the outside view. It's working really hard trying to do something. I will focus my attention there, although it may only end up with me taking out a 2nd mortgage to buy a new GPU. The price point is definitely out of control. So, who's trying to watercool a radeon vii in 2025? This guy.

Looks great! Just for clarification, that wasn't my video. It was one I found that helped me understand that there's something going on. I get very smooth performance everywhere, even in in the areas around Baghdad, but I can Not takeoff from there, nor can I fly over the city. I will throttle settings and see if that affects anything. Maybe I am just pushing the detail too hard over Baghdad.

I don't mind upgrades, it just seems that either 1 or 2 conditions exist: I don't understand the game engine or their more recent developments, or that their more recent developments require something that my components just can't do, or aren't being allowed to do. I am only lacking the South America and KOLA maps. Otherwise, only Baghdad and sometimes Cairo give me problems. It handles all other areas I have without issue. Steady and smooth 60fps all the time. Scenery up at decent levels, also. It just doesn't seem logical, even if it is the case that upgrades are in order. Why only those 2 cities, and only one consistently? The whole rest of the Iraq map seems to play nice, and not draining resources. But, Baghdad reliably crashes as soon as I go wheels up, or fly over it. No drops in fps, nothing. Not the slightest indicator that a problem is coming, just a crash.

Perhaps someone else can chime in if they are seeing the same or very similar issues. I'm not sure if it's because Iraq and Sinai are supposedly still in an "early release" phase of development, and if optimizations will be made, but many videos over Baghdad show heavy demand on resources and low frame rates. It seems there may be some more work done, but no word yet.

Well, after some homework and testing was done, and even posted the crash log into the help room and the crash log analyzer in the DCS discord server, some direction was made, although ineffective overall. If I fly over Baghdad, especially leaving from the airport in Baghdad, DCS crashes. Sometimes also over Cairo on the Sinai map, but that's not even consistent enough to test for yet. I think that the solution lies with making Baghdad a success, the rest of these issues go away also. So, I manually throttled back on the memory and core overclock that the AMD driver automatically adjusts within the driver, disabled/Un-installed ReShade, ran the slow repair on DCS, deleted unnecessary files with it, and added the tweak to the windows registry that another forum post said may work, and all net no change. So, I flew over Afghanistan, released at a similar time, for testing, and I can fly over Khandahar all day. Difference is all the buildings, developed areas, and perhaps some other properties and differences that I don't know about. The F-18 flew over the air base in a circle until it ran out of fuel and augered in. So, it's not the developer of the map, but maybe some of the populated areas that are using some textures or something else that DX11 handles poorly with this GPU. Interesting bit: the discord help tells me that the critical error causing the crash is DX11 related, that it's constantly removing the GPU. So, I don't know what to make of that. It seems that there is no real consistency as of yet. Users from either GPU camp seem to be having these issues, and apparently, it's popular enough that a quick Google search has many posts about this "dreaded" issue, and there are no answers as of yet. I'll post the part showing what someone pointed out is the critical issue. ``` 2025-02-11 20:39:34.800 ERROR DX11BACKEND (14108): DX device removed. Reason: 0x887A0006 2025-02-11 20:39:34.815 ERROR DX11BACKEND (14108): Can't map structured buffer. 2025-02-11 20:39:34.815 ERROR DX11BACKEND (14108): Failed assert `0` ```

Ok. That confirms for me that I will wait to download and install the newest update. Nobody wants these issues.

I tried that the other day. I'm not familiar enough with the discord interface to figure everything out. Working on that. I was having issues logging in and what not. My password didn't work, and I was frustrated with that. Seems like I found the right area to do that, and something else didn't work right. Thanks again.

Thank you for looking into that. The top gun mod is a loading screen image set, and music. Likely not the cause, but I will remove it for testing. Also, this was happening with a different GPU and a newer one I just installed. They are both similar in age, but it almost seems to be a driver issue. I will remove all AMD drivers and retest. Thank you for your help. It's definitely a challenge to figure out what is happening here.

That is a great way to say that. I concur. I think the way it should work is that the pitch should not try to retrim until the board begins to travel during the stow cycle. The net effect should be as close to zero as possible with this function. It likely would be if the trim were programmed to function against the board position instead of the switch position.

typical video of someone departing Baghdad. It looks like it really pushes the hardware pretty hard with the framerate seen here. Maybe it is just something not optimized yet, and we're all working on something that isn't going to work yet.

Welp, I've tried several things, all with no real improvement, although I saw some differences. Un-installed AMD adrenaline software. For you Nvidia users, likely no surprise that it didn't help. Installed instead, the last pro version, first driver only, and then the whole feature set. With the driver only, I did get further into Baghdad than before, but crashed. Almost certainly a driver issue. Still using the pro version, because the blue accents are a welcome difference, although the driver itself is supposed to be more stable/reliable, it also makes me feel I can trust it more. I think I will try just the standard driver only option next. Un-installed ReShade, just incase the "structured buffer" issue and ReShade created some sort of conflict. Nothing. Low altitude, high altitude, still a CTD over Baghdad. If anyone is having any success navigating through Baghdad, I would love to know their hardware specs. I did see the GPU finally load up past 10GB of used memory, which was pretty neat for me. I've not had a GPU with more than 8GB in many years. This map will use some hefty amounts of VRAM, so expect stutters if your particular GPU has less than 12. Surely this map is going to use plenty of resources. I wonder if it is working for anyone. I've looked some, but still not seen much for videos of Baghdad city being explored in detail, which begs the question: I wonder if many folks are able to yet. There could be missing files/textures/something that the GPU is being called to do and just isn't available or finished yet. I don't know a lot about programming language or what else to conclude at this point. It seems that this region is in need of optimizations or something else that just hasn't been addressed yet, although the errors some of us get are fairly common, historically. Interesting if nothing else. Any input? Ideas?

Ok. There are crash log entries, and this thread seems to be alive with the sound of music. Maybe someone can sift through these and tell me what area I need to look at. It appears that in one of those files, it shows an entry wawnting to reprocess the f-18 texture or something like that? I am not sure that is it, but it appears to be the last thing it tried to do before a stack of the very nearly same error message flooded the file. Can someone make sense of these? It always opens an AMD error reporting window about a driver timeout, and some folks have also showed a work-around to increase the timeout duration, and if someone sees that this is possibly where I should look, I'll try that. I get a little nervous about using regedit, because it's not really intuitive, and I'm not sure if I'll be able to undo some of the things that an impatient person can do there. I wanna drop some bombs, man! dcs.20250201-040316.crash dcs.20250201-040316.dmp dcs.20250201-040316.log dcs.20250201-040316.trk dxdiag.txt hash

I'm going to hunt down the crash log and post it in here to see if someone can decipher this Rosetta Stone. I see so many entries on it that looks like some driver timeout, but not an answer for why. The last thing of any importance is something about reworking the F-18 texture or something of that nature. dcs.20250201-040316.crash dcs.20250201-040316.dmp dcs.20250201-040316.log dcs.20250201-040316.trk dxdiag.txt hash

There already multiple threads about this topic. I'm looking for more information from people that also have this issue.

Having similar CTD, and reviving this old thread to use these tweaks. It's still an issue, or has resurfaced recently.

I think perhaps setting the clock manually within the driver, or if there's a preset, set it to the power saving option. It's going to be a try and test hassle to me at this point, unless someone can decode exactly what some of the error codes mean, but it does look like we're already off to a good start. Nothing like going into debt to upgrade hardware that otherwise seems to work just fine, eh?

I downloaded the latest chipset drivers and all AMD utilities available for my setup. I will test again later. Seems to be a driver issue for sure. It's always the same errors. I don't think it's hardware anymore.

I tried this on the ground to correlate what I was feeling with something I could see. From outside, the speedbrake deploy causes the pitch trim to nose down, perhaps to cancel the nose up that should result from the speedbrake. Then, as soon as I release the speedbrake button, the pitch trim is what changes the pitch, even while the speedbrake is still deployed. I think it's that causing the issue, not the speedbrake. Pitch trim on the elevons increases as soon as I release the button. Maybe this is proper, maybe not. I can't tell now. There's so much going on.