Jump to content

Kathrine Jansma

Resident
  • Posts

    189
  • Joined

  • Last visited

Everything posted by Kathrine Jansma

  1. The requirements probably confuse hyper threads vs. cores. A typical CPU core these days allows Hyperthreading, so for Intel & AMD you quite often get 2x the hyper threads vs. cores. So a 4 core CPU with 8 hyperthreads should be fine. Recent work on the performance viewer allows the viewer to distribute work to a thread pool, this improves a few things like faster rezzing speeds for texture heavy environments and others. So having extra cores helps to get a better experience. Why 8? That is probably a rule of thumb thing. To get an optimal experience a lot of variables need to be balanced. If you have a really fast network, you want many threads to decode the textures fast as well. If you have a slow harddrive you want more threads too, as the worker threads for disk access block longer. And if you have a good graphics card that offers nice multi threaded support (like NVidia cards usually do), you might also want more threads. Add one or two threads to make your virus scanner behave and any other background things, and you reach some number in the 4-8 threads range. With just four hyper threads, you might not get any real benefits from the new thread pools. Basically 1 for AV, one for the mainloop, one for the disk writer/read thread and one for texture decode and a bit for extra GL threads. Of course, the operating system will allow more threads to run, but the best performance shows when you have about one application thread per real hyperthread of the hardware.
  2. No, going to higher temperatures for longer does not damage your CPU. In ancient past some CPUs had no throttling when overheating and burned out (e.g. Pentium II era), but that is no longer the case. Modern CPUs have sensors and automatically slow down when getting too hot. Being hot is not a problem. Modern CPUs have a base clock and some boost clock speed. If the CPU stays cool, it can trigger the boost and run a bit faster. You can look at the specifications at arc.intel.com. Intel Core i51135G7 Processor 8M Cache up to 4.20 GHz Product Specifications It tells you the processor may become up to 100°C hot without problems. And that it can boost up to 4.2 Ghz when a single core is working and it is cool enough. It is sometimes worth it to pay someone to clean your fans for a laptop, remove dust etc. Especially if you smoke or have a lot of dust around your place. Some places recommend to clean fans/vents every 6 months to two years. That usually takes half an hour to an hour for an experienced person with the right tools and can speed up your computer a little bit. Can be 20%. Last time i had someone do it, it was like 30€ for the work and it helped a bit. You could use compressed air, but thats less efficient for cleaning the fans. There are some guides that recommend it as an easy way to go. So basically do not worry. You may be able to squeeze out a few more percent of performance or get the machine cooler if you have the fans cleaned.
  3. TOTP based MFA works in a lot of desktop options. I tend to use KeePass with a TOTP plugin or KeePass XC which can do TOTP out of the box. Works fine.
  4. It also dies when your provider has carrier grade NAT to isolate you from the lack of IPv4 addresses. But the simple solution would be to hand every viewer a "websocket" endpoint that the viewer connects to to get notifications for IMs. The viewer would then use this endpoint to fetch IMs, not much different to a long poll. The LL just needs a websocket capable endpoint in their AWS deployment and push messages there, instead of routing it via the region.
  5. Sounds a bit like some kind of BOM (Bill of materials) format for assets (e.g. something like CycloneDX (https://cyclonedx.org/use-cases/) for software).
  6. Lets see how long that EU data protection actually lasts. There are plans to undermine it. See for example https://netzpolitik.org/2021/eu-commission-why-chat-control-is-so-dangerous/ But curious if the XMPP interface is really still there, as Jabber/XMPP isn't a totally bad protocol. (even if the Signal protocol as used by other messengers is clearly superior for privacy).
  7. Stuff like that can work, if your customers/users accept the state of things. For example look at Apple Inc, when they upgrade their OS X offerings. They are quite liberal in throwing out the old junk and forcing customers to upgrade, developers to do new versions of applications etc. Even hardware in RL sometimes gets washed away by progress. For example germany forced everyone to move to higher fidelity TV sets some years ago. The old distribution channel was simply cut off and you needed new hardware. Similar to how phone providers force people to move away from older mobile phone standards. Or it can fail miserably. Have a look at Modern Windows Apps, when Microsoft tried to also do an Appstore. The important business questions are: Would enough designers and developers create better replacements for the old stuff? Would people buy into the new stuff and make the creators happy? Does it help LLs cash flow and bottom line earnings. LL is probably in a bit of a torn state, as they get cashflow not just from Marketplace and L$ conversion fees. Those are mostly driven by sales, so new stuff and people consuming is great. But Land and Premium membership fees might be linked to people that get angry about old cherished stuff breaking and leaving over that.
  8. Totally. They changed it to the DejaVu font. (with some Helvetica in between, that got exchanged for DejaVu) https://bitbucket.org/lindenlab/viewer/commits/d6101558a171dbd2390792ac1e78d09fc2c27711#chg-indra/newview/skins/default/xui/en/fonts.xml
  9. At least not XSLT and FOP. Sanity points saved. Or Docbook SGML & DSSSL... or TEI. Or groff/nroff?? Hmm, LaTeX for really nice typesetting? Maybe just add a new inventory type "Rich Text Notecard", that can show pictures inline and has a few rich text features. Markdown or whatever is the current trend.
  10. The newer AMD driver is a little faster then the old one. But NVIDIA is still significantly faster. In addition there are some reports of instabilities or excessive memory usage with the 22.7.1 driver. Like AMD removed a bunch of OpenGL extensions in the update, which may cause issues. )
  11. Sounds like a misguided idea. It would be more useful to have a setting to switch notecards to use a fixed width font for that. Trying to do ASCII art with proportional fonts just doesn't fly. The font is part of the skin support, so anyone might tweak that to his or her liking. The style for the text editor just says "SansSerifSmall". This gets mapped to the actual font later, probably ending up with the DejaVu fonts in a default install. See https://bitbucket.org/lindenlab/viewer/src/master/indra/newview/skins/default/xui/en/fonts.xml this already lists different fallbacks for the default font. So if the original Linden viewer already takes precautions to handle missing default font files, why should you expect to get the pixel exact font rendered as-is? If you want that, use a web page or screenshot. Even if you use the same DejaVu font, there are version differences, e.g. (https://dejavu-fonts.github.io/NEWS.html) so any TPV that depends on an OS package for the font might get different versions. Not to mention font rendering differences between different versions of Freetype, Windows, OS X, etc. Various TPVs just use Lindens OLD default fonts: https://bitbucket.org/lindenlab/viewer/commits/9ec432034dc3c45d7ce763eb02dae4cc7f6b8da8#chg-indra/newview/skins/default/xui/en/fonts.xml So LL itself broke this "feature". Not to mention the fact that it breaks user customization. IF i override a font explitly, for whatever reason, i want MY choice to show. Especially for something mundane like a text file. If layout is really important, plain text is often the wrong medium. PDF or other layout formats might be more appropriate, as can be seen by some product descriptions that simply link to a PDF outside for more fancy layouted instructions.
  12. Thats like checking any malloc() return carefully and then still getting killed by the OOM killer randomly. Usually not worth the trouble. You technically CAN handle low memory situations and the usual app compatibility guidelines recommend testing and handling it (Windows App Verifier tests low memory situations too, e.g https://docs.microsoft.com/en-us/windows-hardware/drivers/devtest/application-verifier-tests-within-application-verifier#low-resource-simulation). But most applications do not care and just crash or shutdown on their first malloc error. Handling all malloc failures is doable sure, especially if you override the platform allocator and have some memory pool you can throw away in emergencies. But thats not always done and usually needs some dedicated effort to make it happen.
  13. Seems AMD finally released the performance optimizations to Open GL for their Windows driver. I updated to the 22.7.1 driver today with a Vega 56 card and the FPS seems to have increased massively. Highly recommended.
  14. Isn't the usual personal space comfort zone usually "an arms length"? FB violates that by quite a margin with those "Personal Boundary" tubes.
  15. Hopefully its not the "Second Sight" thing. That had some quite dystopian/cyberpunkish aftermath already: https://spectrum.ieee.org/bionic-eye-obsolete The company went down and the patients are up for nasty surprises.
  16. Technically MS already dropped support for 32-Bit versions with Windows 11, the servers already dropped it with 2008. But thanks to the WoW64 subsystem you can still run 32 Bit processes on nearly all Windows machines (Server Core being the exception, its optional there).
  17. Actually, with the investments of Steam and others into DXVK and Proton (https://www.protondb.com/) and stuff like Lutris (https://lutris.net/), thats not necessarily true anymore these days. Gaming on Linux mostly works, but you might get troubles with "Anti-Cheat" crap in some multi-player games, that try to install Windows kernel drivers or similar deeply scary stuff.
  18. That depends on the Windows edition. Windows Enterprise licenses and various others offer downgrade rights that do not loose any features. Just one random example, Dell lists that option under "Business Laptops": "Operating Systems" : "Windows 10 pro with Windows 11 pro License". https://www.dell.com/en-us/work/shop/dell-laptops-and-notebooks/sr/laptops/windows-10-pro-with-windows-11-pro-license?appliedRefinements=37829 Other vendors offer similar things for the "Windows 11 Pro" stuff.
  19. According to some account on the CoolVLViewer forum for the ARM port, it ran on a chromebook style Mali T860 GPU with 5-10fps, so yes, probably not worth it. http://sldev.free.fr/forum/viewtopic.php?f=10&t=2212&start=30#p10896
  20. Indeed. Someone managed to port the necessary parts to Linux ARM64, so it should run on a modern Raspi or similar hardware. http://sldev.free.fr/forum/viewtopic.php?f=10&t=2212&start=30
  21. Sounds a bit like the optimizing JIT compilers. Observing the scene and then specializing. But you do it the other way round, starting with static assumption and moving to dynamic once it happens.
  22. Windows Server makes quite a decent workstation (unless you need Bluetooth support or other semi fancy stuff like that). Had one running as a workstation for some years.
  23. Should not be a checklist. It should be automated away...
  24. Case in point, look at the abysmally bad OpenGL performance of AMD cards, even on PCs. AMD stopped supporting OpenGL in their performance tools, the debugging tools, basically everything and moved the bet to Vulcan. Apple just says "its deprecated, we officially do not care for OpenGL", AMD just lets it wither and die silently.
  25. Technically, using the old date is correct in most cases. It should list the year of first publication... https://www.copyrightlaws.com/copyright-symbol-notice-year/
×
×
  • Create New...