Aapje

To start off with, I'm going to ignore your irrational constraints and give the actual best advice. Then later on I'll give the fanboy advice for the CPU. For the GPU, a 5070 Ti should be able to be about to hit your performance targets. A 5080 is going to have about 15% more headroom, but it probably going to cost a lot more than 15% extra. Check your local markets for the price difference and think about whether you want to spend the extra amount. A 9800X3D or 7800X3D is going to be the best CPU. The 9800X3D seems to be much more overclockable, so it is probably worth the price difference. The Intel option is a 14900, which is slower and runs way hotter & is thus much more problematic with cooling. You'll also be buying a motherboard with zero upgrade potential for the CPU, while an AM5 motherboard will get at least one more generation. Of course you are free to make bad choices in life, but after already making the bad choice of getting the 10600 (a stopgap solution released because Intel couldn't get their 10 nm process working), do you really want to make another bad choice just because Intel used to do well for you? PS. Get 64 GB of DDR5-6000

Yes, but you need an adapter: https://docs.google.com/forms/d/e/1FAIpQLSflUrHfM9AB0f-0WGyhOLxvXQNMr7E8TBDSYYJUBQ8kqaDC7g/viewform

You should give your rough location so people will know if shipping is going to be an issue. And your picture is broken.

Just keep in mind that Virpil has long delivery times and that you should take the displayed delivery times with a huge grain of salt. If you get their CDT stick, I would suggest getting at least one Throttle Module with it. You can never have enough axes, IMO.

There is a Pimax subforum.

List of known issues/workarounds: https://github.com/mbucchia/Oasis-Driver-for-Windows-Mixed-Reality/wiki/Known-Issues

I think that your body & mind gets mixed signals if you use a motion rig with a monitor.

@wct24 You need to wait 1 day and 14 hours for the Oasis driver to be released, before you can use it. I suggest listening to some music while you wait: Let's see if the WMR headsets will live forever.

Like I said, I'm not criticizing your choice. I'm just talking about the reality that quite a few people are looking for a bargain, and those people are susceptible to overly rosy descriptions of products, which can result in them making poor buying decisions. So we should be careful not to overhype a product that doesn't deserve it.

I don't mind it when someone makes an informed decision to get something that I wouldn't suggest, but I don't like it when claims get made that are not true. In particular because it can cause other people to make decisions based on incorrect info.

The VKB STECS also has (quick switch) detents, dual throttles, configurable throttle stiffness and indeed a ton more controls. Having so many is not just a benefit in being able to map more stuff, but also that you have more choices in what kind of control you map to what function, and what location you pick. For example, I mapped 'eject' to a recessed button that is right under my finger, but also very hard to accidentally press. I use the other recessed button for 'jettison', which similarly is not something I want to do by accident, but which you may want to do quickly. Having many buttons of different types, gives you that sort of options. The STECS not designed to mimic a specific airplane. If you take the spring out of the white up/down lever on the button box (STEM) of the STECS, then you can configure it in the software to be an axis, that you can use for zoom. The Virpil CDT VMax has axis dials that should work for that purpose too. Although I'm always a bit wary about recommending products that are only available for pre-order, as we can only judge their promises, not what they actually deliver. Then again, it is Virpil, so the quality tends to be fine, but not so much the delivery times. Given the very limited number of controls, it seems rather overpriced for what you get. And I doubt that it is really higher quality than the top options of VKB and Virpil. At most there is less to break because it is missing so many features. I just watched a teardown, and it has pots, which tend to have jitter, which wear out and suffer from dirt. It also has plastic sliding on plastic, while the high end throttles have proper mechanical levers. With the plastic sliders, they had to choose between minimizing the stiction or having the throttle actually have a decent amount of friction. They chose the former, which is the least bad option, but not being able to have both is due to the subpar technology being used. My experience as an engineer is that a lot of people are very bad at judging actual quality, and they tend to judge things as quality just because some part of the product appears impressive to them, even when the quality is actually quite subpar. A lot of people also have experience with few products and often not the best quality gear, so they don't know how good it can be. For example, I think that if you actually move a STECS or Virpil throttle, you will notice that the feel is so much better than the CH Pro. The CH Pro surely was a good product for the 90's, but it's decades later now and better stuff is available. The Moza throttle doesn't have FFB, just vibration. Only their joystick base is anything special, IMO. Their grips are not very good. So I would definitely mix-and-match if you go for the Moza base.

You always have a risk, including of messing up the install. If you spend less in the first place, then you have money left over for a possible replacement. I think that the idea that if you spend a lot, you will avoid risk, is a bad take. IMO, the best way is to keep a financial reserve (in general, not just for PC stuff), so you can take a hit. This goes for everyone except the most poor or the most rich.

That's a very strange conclusion. B650 is still easy to get and otherwise you can get the B850, which is essentially the same chip but with a higher minimum spec for the boards. It seems more like you are rationalizing going to the more expensive boards (which is fine). Keep in mind that the X670E mainly offers more USB and SATA ports. Do you actually need that, because... You absolutely don't need to get a X670E or even a X670 for that.

There is no better CPU. Server-type tasks tend to consist of lots of work that needs to be done in parallel, but not necessarily superfast. That is a very different use case from games, and games are not going to run optimally on such a chip. But I agree with the other people that you have very weird load that doesn't seem consistent with just the game putting load on the CPU. It's not normal for games to fully saturate all cores. Can you post a picture of the performance view when DCS is not running?

The best place to find stuff like this is typically Aliexpress. This looks like a JST PH connector. Search for that. Do you have male connectors on both ends?

Pot kettle. Only if you are extremely pedantic and insist that established ways of speaking are incorrect. Except that it shares the most expensive, most crucial and most central component. But I guess in your world, If I get a hair cut or change clothing, I become a different person.

I don't really understand what your objection is to what @AngleOff66 said, since you seem to be agreeing with him or her that it is not that complicated (for Chinese experts) to create a new PCB for a 48 GB card. I assume that everyone understands that you can't just turn a 24 GB card into a 48 GB version by watering it regularly and giving it some fertilizer. Of course, some extra memory modules need to be added and you need a PCB that can connect those to the GPU chip. This seems to be missing the point, since @AngleOff66 is not arguing that someone should set up a conversion factory, but that the ability to turn a 4090 24 GB into a 48 GB version, proves that Nvidia could manufacture those or let AIBs make them. Now, the video is not actually needed to prove this for those in the know, since Nvidia has already been selling pro cards with the same chip as the 4090, with 48 GB. So of course they could also sell a 4090 with 48 GB as part of the consumer line. First of all, this comment is directly countered by the video, where Chinese vendors consider it a sufficiently sustainable market to develop a custom PCB for and offer it as a product. There is also enough of a market for them to show up on eBay: https://www.ebay.com/itm/376271640799 Secondly, you are ignoring market segmentation, monopoly practices and market manipulation. So it's not necessarily the case that there is no market for a 48 GB 4090, but rather, that Nvidia doesn't want to offer people this option, but instead wants them to buy the pro cards for double the price or more. Nvidia segments their lineup by memory and is known to retaliate against vendors and AIBs that don't do what Nvidia wants. So from the perspective of a US/EU company, it is probably more sensible to buy a pro card with 48 GB, like the RTX 5880 or 6000, than a modified card that may result in Nvidia putting the company on a support and sales blacklist. However, this is not a free choice based purely on the merits of the product, but a choice driven by market manipulation. You can't simply conclude that these companies would make the same choices if Nvidia would allow AIBs to make 4090's with 48 GB, and if Nvidia would properly support those products with drivers and...support. The math for Chinese companies can easily be different, due to them (as a country) being on a sales blacklist already anyway, they may benefit less from Nvidia support due to language issues and thus may need to be more self-sufficient due to it, their shenanigans may be more opaque to Nvidia by being much more isolated from Nvidia in various ways (for example, a large Silicon Valley company is likely to have workers move to or come from Nvidia, so their internal secrets may become known to Nvidia that way), they may feel protected by the Chinese government, etc, etc. Thirdly, you ignore that the BIOS/drivers probably don't work well for gaming on these modified cards, so that means that they are only viable for AI/business use. Of course, Nvidia could easily release a 48 GB card with proper BIOS/driver support. Surely interest by gamers would increase a lot if such a card would actually work for gaming. Ultimately, the only way to truly prove or disprove the viability of such a product, is for Nvidia to allow such a product to exist, which they don't. You can't just draw conclusions based on a manipulated market. For example, when the Iron Curtain was up, Eastern European countries claimed that their people were very happy with the Trabant/Yugo/etc, but once customers were allowed a free choice, they showed that the sales on the manipulated market didn't actually match their real preferences.

The 5800X3D is no longer sold new, but you may be able to get a decent deal for it second hand. A 5700X3D should perform similarly in most games. Having very many cores is not so beneficial for games, while the X3D-cache has a big benefit. An alternative is to upgrade to AM5, getting a cheapish AM5-motherboard, 64 GB of DDR5-6000 and a 7600X. For example, like this: https://pcpartpicker.com/list/TykKdH DCS can really benefit from 64 GB of RAM, and by going to AM5, you are not spending more money on an obsolete platform. A big benefit of that route is that you later on do another big upgrade to a much faster X3D-CPU and get a big boost. The 7600X is about as fast as a 5700X3D/5800X3D, but where the latter is the end of the line, you can later on replace that 7600X with a 9800X3D or a future generation X3D-chip. That new motherboard also has PCIe 4, so that means that it can use faster NVMe's and is more futureproof with regard to the PCIe speed for the GPU. PS. It is absolutely not necessary to get an AIO cooler. Not sure what cooler you have on there (the stock one??), but there are very good air coolers for amazingly low prices, like the one that I specced in my build. These are more reliable and longer lasting than AIOs. Of course, regardless of air cooling or AIO, you need decent airflow into the case.

@Nedum I suspect that the smooth autorotation might be the issue, since XRNecksafer would then need to continuously change the view position. You might want to try with Snap rotation.

According to the Steam page, it works on both 10 and 11.

Why would it? It is a replacement for WMR, so it is irrelevant whether or not you are running a Windows version that still has WMR in it.

It's the second best choice purely for gaming (the 9800X3D being the best). It's weak relative to the price for multithreaded productivity software, if you use that.

For DCS, I would pay that difference, since the 9800X3D seems to be far less stuttery:

Like any tool, it needs to be used to its strengths and not its weaknesses. You also don't use a drill as a hammer and then complain that it is a useless tool. There are specialized AIs to generate 3D models from 2D images, like TRELLIS, which seem to do a decent job. I don't know whether they are good enough for modelling planes with. But just like human modellers, they base their model on images of the real thing. So it's not just a generic component or cockpit model, but it would be specific to a certain plane. In theory it would even be possible to make an even more tailored AI trained for modelling real life object in sims, training it on already created 3D models and the 2D images that those are based on. I once worked for a non-profit company that made the software for an entire industry, where everyone who used the software was required to invest in it, and then they were charged a fee for the software that covered the cost of maintaining/improving it (so no profit margin). In principle, the companies/individuals who make airplane models for flight sims, car models for driving sims, etc could adopt the same model, where instead of merely investing, they all hand over their 3D models and the 2D images that those are based on, to the non-profit. Then the non-profit could train an AI to be optimal for this kind of work. For example, the hallucination level could be set very low, and because it would just be trained on relevant images, it wouldn't even know how to generate irrelevant imagery.

From his comment, I took that he also uses productivity software that benefits from lots of cores. Intel can be a good choice for a mixed use machine, since AMD strictly segments its lineup by the number of cores, so there are no cheap many-core options, while that is the case for Intel. I don't know the relative performance of these chips in DCS specifically, so I don't know what advice to give though.