Jump to content


  • Posts

  • Joined

  • Last visited

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

  1. I mean, I do use my nose as a stylus at times, but not generally in this context...
  2. From a practical standpoint, even fixed foveated rendering is a big performance improvement. What I've found is I'm only generally using the periphery for object tracking, not fine detail, so it does not need to be at near the same resolution as the center 30 degrees or so does. The things that really require high resolution, like visual ID, VFR navigation, and reading labels, I'm generally looking directly at it, as close to center as I can be.
  3. So finally got it installed, and used the fpsVR logging to capture the CPU frame times of the F-14B cold start with the 5800X and 5800X3D. With the 5800X and DDR4-3600 CAS 16 memory the median CPU frametime was about 9.2ms. The 5800X3D with DDR4-2666 it was 7.8ms, and with the 5800X3D with DDR4-3600 CAS 16, it was 7.2ms. Frame rates were still bottlenecked by the GPU, but it looks to be providing a lot of additional CPU performance in VR. I also saw similar performance improvements in the CPU limited Il-2 VR benchmark. For standard 2D gaming, it does seem to be providing around a 7-12% lift over the 5800X. I think what we're seeing is the impact of the dual rendering of VR. It seems like having two almost identical render scenes benefit greatly from very large cache sizes, to a degree that other types of games do not. Overall, I think us VR users can more than double the GPU power on one of these without becoming CPU limited.
  4. @Nahen It's been a while since I followed SATAL, but I gather it was the Radar Warning Receiver that was considered excessively omnicient. But again, that was a while ago.
  5. I think it's mostly because the FC3 F-15C is getting dropped from things like SATAL because it's electronics package is pretty magical. Likely the interest in a A2A mode F-15E will drop once a full fidelity F-15C is in the works.
  6. I've been seeing pretty solid stock of them. From the benchmarks it's comparable to an I9-12900K on average, and everyone is waiting on Zen4 later this year, so I'm thinking we'll have solid supply for a couple of months before they wind down production. I was able to pick one up Wednesday. Haven't had time to set everything up and run all of my benchmarks yet, but will try over the weekend. Based on the Il-2 and MSFS numbers, I'm expecting a 10-15% reduction in CPU frame times from the 5800X, myself. But, we shall see.
  7. If you're happy with your framerates, stick with the 1080 Ti. For VR, I would not expecting any of those cards to do much better than it, so probably wait until the next generation stuff comes out later this year. Between AMD, nVidia, Intel and the consoles all being on different high yield nodes, and the mining crash, this next generation should be less bonkers than the last one.
  8. From some of the MSFS benchmarks I've seen, the 3800X3D is slightly faster than a 12900K with DDR4, but slightly slower than a 12900K with DDR5-6400. So, it's not an absolute win, but if you've already got an AM4 with high 3800hz DDR4, it appears to be a solid last upgrade before you replace the entire platform. I'm probably going to grab one because I've got a 5800X platform and a goodly amount of CAS16 DDR4-3600, so it's just a bios and CPU swap, for a likely decent step up in VR performance. However, if I wasn't so heavily invested in the AM4 platform right now, and mostly GPU limited, I'd expect Raptor Lake and Zen 4 with DDR5-6400+ to roflstop it later this year. As it is, what I *really* need is either a 7900 XT or whatever the Lovelace *090 gets called to feed the frames. Even with the fholger's FSR, and a 3080 Ti, my GPU frame times are still nearly twice as long as what my current CPU can do.
  9. If you haven't, go check out Moore's Law is Dead on YouTube. He has some deep breakdowns on what is coming. Basically, AMD figured out how to do GPU chiplets and vertical stacking, so they can dice up the top end GPUs into small, high yield parts, and use multiple nodes to build the core GPU. They're going to be using both the 5nm and 6nm nodes for the RDNA3 parts. Further, TSMC's 5nm node is having exceptional yields, so we're going to get a double whammy of a higher yield design being built on higher yield nodes. And, because these are chiplettes, they're able to both shift production ratios after fabrication, and build out much larger GPUs than they could with monolithic designs. nVidia is about a generation out from solving the chiplette problem, so they are responding by pricing very agressively, and by pushing their GPU performance to the absolute limit. It sounds like they're also paying extra to get space on TSMCs 4nm production line, so they won't be pulling from the same wafers as AMD. Even if our niche doesn't play nice with RDNA3, nVidia will be bringing some heavy hitting hardware in. Finally, with Intel getting into the game, they are going to be buying market share with Alchemist. Basically pricing at cost so they can get their foot in the door. The drivers are a big area of concern, but their cards should help the current low end demand. The numbers aren't huge, but it should help alleviate the demand for RX 480 replacements. Q4 this year, provided TSMC is still standing, is going to be wild in the GPU market.
  10. The HP Reverb is 20ppd and known because the screen has a relatively fixed distance from your face. A 4K display, at a distance equal to its diagonal, has a 60ppd resolution. That was their design viewing distance, as the human eye peaks around 60ppd. Basically that means a 27" 4k monitor will have a 60ppd at the typical 27" ( little bit longer than arms length) placement. Closer it will have less effective pixel angle density, further, more. Also, individual peak resolvable resolution changes from person to person, depending on your eyes and vision correction. Using that conversion, a G2 is more the equivalent of a 1280x720p monitor. Certainly not as good as a 4k monitor, but definitely better than a 640x480. That said, once headsets hit 30ppd, we'll be at around 1080p performance, which should be about the point where they become equally capable. Given 4k can already meet human vision limits, I don't expect 8k to provide any meaningful advantage. Sim developers may even have to start implementing stuff like airy disk emulation and atmospheric attenuation at that point.
  11. At least according to Fedex, looks like mine has been sitting at Charles deGaul for the last week or so. Not sure how long that usually takes.
  12. So what happens if you try to gift a module to someone who already has it? The sale bounces? Or disappears?
  13. They're going to come down this year. Mining is cratering, and Intel and AMD are adding new process nodes to the pile. I think by the end of the year we should see significant too end performance improvements, and enough availability of 30, 40, Arc, RDNA2 and RDNA3 cards that the 10&20 series parts can be replaced by considerably better cards for relatively not insane prices. Straight up inflation is going to bite, but we should see a lot of GPU power hitting the table pretty soon.
  14. He mentioned something about the computer as well and some system that should be calibrated before or during flight. I also recall at one of the AIM-9/AIM-4 fly-offs, the Hughes team had brought a metric ton of support equipment for the falcon, that the Sidewinder team didn't need or care about at all. I think it was more than just the coolant, but have never really dug into the century series fighters to know what. I've just seen enough odd things in the whole Sidewinder vs Falcon stuff to suspect there is something very wonky that was going on in the Falcon that later IR missiles just didn't bother with. If you look at most very 1st generation systems, they pretty often do things in really weird ways compared to what is found to work later. Look at the differences between the Mannlicher M1894, the Luger and the M1911, and realize they competed against each other. I think if we really want to see the AIM-4 Falcon, it's probably worth digging into how it actually functioned, and was intended to function. The reality may be stranger than we can guess.
  15. So, I see how you can buy a module as a gift for someone who you already know who it's going to, but I'm wondering, is there a way to buy modules for delivery at a later date? Thank you, Harry Voyager
  • Create New...