What is my problem in P3D v5?

[Prpn 366]

Manufacturers have taken different approaches in hardware and software for ages. This can lead to different results using the same API (DirectX) no matter how platform and hardware agnostic you try to make it. One piece of hardware/software can simply be more fault tolerant than another. My main point here has been that LM failed to even test it on the hardware and software of one manufacturer and basically left that testing as something for the customers to do.

Trying to absolve LM from that by pointing to AMD and Windows is easy, but LM should have done their testing more thoroughly before shipping a half broken product. 

[Chock 17,769]

Sorry to disappoint those pointing the blame at AMD cards and suggesting they are not a good choice for flight sims, but I'm running P3D V5 on a Radeon card with no problems. I have preferred AMD cards for years. I've run all versions of P3D I've ever had on them and FSX and FS9 on AMD stuff before that.

[SteveW 1,792]

37 minutes ago, Prpn said:

Manufacturers have taken different approaches in hardware and software for ages. This can lead to different results using the same API (DirectX) no matter how platform and hardware agnostic you try to make it. One piece of hardware/software can simply be more fault tolerant than another. My main point here has been that LM failed to even test it on the hardware and software of one manufacturer and basically left that testing as something for the customers to do.

Trying to absolve LM from that by pointing to AMD and Windows is easy, but LM should have done their testing more thoroughly before shipping a half broken product. 

The differences you refer to are not the fault of Windows or LM, but the fault of the hardware and driver resulting in problems specific to that item. When looking at particular addons that exhibit problems with any hardware might well be a problem of LM or that addon. That's what I referred to by being sure to use proper techniques of elimination to determine where the problems lie.

From Chocks post it can be seen that if some don't get a problem and others do, then those should be looking at their setup in particular and not blaming AMD or LM or Windows.

Edited June 15, 2020 by SteveW

[SteveW 1,792]

54 minutes ago, Prpn said:

platform and hardware agnostic you try to make it.

You've mis-interpreted my posts completely:

I've not said "hardware agnostic" at all, which is entirely different. I clearly stated hardware abstracted.

In other forum posts I have mentioned Hardware abstracted with device capabilities or 'DEVCAPs' in Windows programming parlance. SO the obvious one regarding GPUs at present is the availability of RTX or not.

In the situation with printers it may be easier to understand that one printer might not be able to accomplish borders as small as another.

[cwburnett 1,809]

59 minutes ago, Chock said:

Sorry to disappoint those pointing the blame at AMD cards and suggesting they are not a good choice for flight sims, but I'm running P3D V5 on a Radeon card with no problems. I have preferred AMD cards for years. I've run all versions of P3D I've ever had on them and FSX and FS9 on AMD stuff before that.

You're using a pre-NAVI card, right? RX580/RX590 or something like that? This problem is specific to the NAVI cards...the Radeon VII, RX5500, RX5600 and RX5700 cards.  And it is actually a very specific problem. View change TO the cockpit only in aircraft with PBR textures.  It has been nailed down pretty well by the LM team as to what specific conditions cause it and as far as I know, there aren't any exceptions, or at least none have surfaced on the LM forums.  If you do have a NAVI GPU and aren't having this problem, LM wants to know about it 😉

Edited June 15, 2020 by cwburnett

[cwburnett 1,809]

1 hour ago, SteveW said:

GPUs are abstracted hardware, so it doesn't matter what the GPU card is, so long as it complies with the system. DX requires that all GPUs are capable of the common set of instructions or they are not compatible. Something is wrong with your AMD GPU or the drivers or something else such that it does not perform as expected. That's between AMD and Microsoft, LM could do nothing about it, they won't be releasing an AMD specific codebase and an Nvidia specific codebase, just one DX specific codebase is all they can do.

Steve, with all due respect, and I have a lot for you given all I've learned from you over the years, if this were true, then Beau Hollis and Rob McCarthy from LM would not have both acknowledged the problem and stated that they're working with AMD to solve the problem (as well as acknowledging they didn't even have one of these GPUs to test on). To be clear, this is not one person's card. This is not one model of card. This is the entire line of NAVI GPUs from AMD. Their current generation. We have not heard a single report of a NAVI GPU user without this problem in v5. I'm confident it is driver related; AMD is notorious for driver problems, but that doesn't stop application developers from working with AMD to sort it out so that the millions of users with AMD GPUs can run the same software. There are countless cases of applications crashing due to graphics drivers, Nvidia and AMD alike, and in each case the application developer works with the GPU manufacturer and out comes a patch for the application or an updated driver and the problem is solved. This is no different. Of course LM won't release a codebase specific to each GPU, but there's clearly a problem that is very specific to the NAVI GPUs, and to suggest that the current gen AMD NAVI GPUs are not DX12 compliant is illogical, as I don't see how they'd have a WHQL certified, yet non-compliant driver unless a whole lot of people are asleep at the switch.

I think it is safe to say that there is a problem, that LM is working on solving it with AMD and that we simply have to be patient. Of course, I'm not patient, so I swapped GPUs with another machine I have that had an RTX2070 and all is well for me, but not everyone is lucky enough to have a spare RTX card around...

[GSalden 4,700]

3 hours ago, Prpn said:

Stating something like this with such certainty is fun but meaningless. Meanwhile both Nvidia and AMD users see the same kind of DXGI hang ups and I get the distinct impression that this is a LM code issue, with Nvidia drivers simply being more forgiving. Example, the PMDG 747 view switching issue. Outright crash on many AMD cards, and I've seen a lot of users reporting it can take a long time on Nvidia cards as well. My guess is, error in code being handled more gracefully by one card than the other. And if you only ever develop for one specific manufacturer and not even testing it a single time on any other card, you get this kind of half-*cheeked* public beta job.

My V5 is superstable as was V4.5.

I did however experience the DXGI Hangup in both my pc’s after updating to the latest NV drivers.

Went back to 442.92 and both running again like a charm...

See my vids in my signature.

If the coding really was that bad then everyone would have CTD’s and that is by far not the case

Like I have written it several times :

The stability of your hardware icw the stability  of your software is the stability of your system.

And even if you can run benchmark tests for hours that is no guarantee that your Fs is running without issues as in total it is more taxing ..

[xxd09 68]

I may have a problem in that my Win 10 version is 1709 ie behind the curve!

Some work to do-have automatic updates on but not doing the business!

xxd09

[CaptainNick 575]

34 minutes ago, xxd09 said:

I may have a problem in that my Win 10 version is 1709 ie behind the curve!

Some work to do-have automatic updates on but not doing the business!

xxd09

Download the Windows Update Assistant from Microsoft...it should help catch you up.

[CaptainNick 575]

I have always thought, based upon the few times people have benched AMD cards vs the similar level Nvidia cards for the past two or three generations in regards to popular flight sims....that for flight sim applications Nvidia always seemed to be superior. 
 

Now if flight sim wasn’t ones main use...the AMD cards have given amazing value for performance depending on the game. Especially the 5600/5700xt models. It was one of the two reasons I went with a 2070 Super vs the 5700xt. I only really play P3D, DCS and a little Xplane, along with the nvenc encoder on the RTX cards sealed the deal on the price difference. IF I played other stuff and only sometimes flight sim I could have easily gone the other route.

[SteveW 1,792]

13 hours ago, cwburnett said:

Steve, with all due respect, and I have a lot for you given all I've learned from you over the years, if this were true, then Beau Hollis and Rob McCarthy from LM would not have both acknowledged the problem and stated that they're working with AMD to solve the problem (as well as acknowledging they didn't even have one of these GPUs to test on). To be clear, this is not one person's card. This is not one model of card. This is the entire line of NAVI GPUs from AMD. Their current generation. We have not heard a single report of a NAVI GPU user without this problem in v5. I'm confident it is driver related; AMD is notorious for driver problems, but that doesn't stop application developers from working with AMD to sort it out so that the millions of users with AMD GPUs can run the same software. There are countless cases of applications crashing due to graphics drivers, Nvidia and AMD alike, and in each case the application developer works with the GPU manufacturer and out comes a patch for the application or an updated driver and the problem is solved. This is no different. Of course LM won't release a codebase specific to each GPU, but there's clearly a problem that is very specific to the NAVI GPUs, and to suggest that the current gen AMD NAVI GPUs are not DX12 compliant is illogical, as I don't see how they'd have a WHQL certified, yet non-compliant driver unless a whole lot of people are asleep at the switch.

I think it is safe to say that there is a problem, that LM is working on solving it with AMD and that we simply have to be patient. Of course, I'm not patient, so I swapped GPUs with another machine I have that had an RTX2070 and all is well for me, but not everyone is lucky enough to have a spare RTX card around...

OhOh - I was explaining how stuff works - that's all - not what some kind of problem was.

[SteveW 1,792]

So I say again - with goofy use of the printer example I gave very clearly - the AMD printer might not be able to do the smaller border like the Nvidia printer (lol) and that has been missed by LM, AMD or Microsoft. My posts are correct.

[xxd09 68]

Windows Update Tool and Creative Media Update Tool both failed to do the business 

To the computer shop!

xxd09

[Chock 17,769]

14 hours ago, cwburnett said:

You're using a pre-NAVI card, right? RX580/RX590 or something like that? This problem is specific to the NAVI cards...the Radeon VII, RX5500, RX5600 and RX5700 cards.  And it is actually a very specific problem. View change TO the cockpit only in aircraft with PBR textures.  It has been nailed down pretty well by the LM team as to what specific conditions cause it and as far as I know, there aren't any exceptions, or at least none have surfaced on the LM forums.  If you do have a NAVI GPU and aren't having this problem, LM wants to know about it 😉

Yes, that's true, but I just wanted to point out that there are quite a few people on this thread who are, quite frankly, talking b*llocks in suggesting that AMD GPUs are somehow best avoided for flight sims. These people are talking nonsense. That there is a problem related to V5 and some AMD cards is a specific issue. It's not as if this kind of thing has never occurred with any NVidia card ever.

Edited June 16, 2020 by Chock

[cwburnett 1,809]

6 hours ago, Chock said:

Yes, that's true, but I just wanted to point out that there are quite a few people on this thread who are, quite frankly, talking b*llocks in suggesting that AMD GPUs are somehow best avoided for flight sims. These people are talking nonsense. That there is a problem related to V5 and some AMD cards is a specific issue. It's not as if this kind of thing has never occurred with any NVidia card ever.

I couldn't agree more. 👍