Jump to content

DuckDodgers

Member
 View Profile  See their activity

  • Posts

    287

  • Joined

  • Last visited

 Content Type 

Everything posted by DuckDodgers

  1. DuckDodgers
    Hardware vendors can't catch a break. The US government is turning up the heat on export controls for advanced AI tech. Quotes Dunno what to say more about this farce, but that the looming black market for high-end GPUs amid tightening trade restrictions will push up the prices even more for all of us. Source: https://videocardz.com/newz/u-s-revises-chip-export-rules-to-china-geforce-rtx-4090d-likely-to-be-affected
  2. DuckDodgers
    In a new report, the White House Office of the National Cyber Director (ONCD) has called on developers to use "memory-safe programming languages," a category which excludes the popular languages. Quotes The market has been flooded with developers who have degrees or certifications with only a single semester of Java programming. Doesn't matter how "memory safe" the language is. With the incredulously low standard for programmers these days, they are all a threat. Stop blaming the language! Sources: https://www.tomshardware.com/software/security-software/white-house-urges-developers-to-avoid-c-and-c-use-memory-safe-programming-languages
  3. DuckDodgers
    Quotes How the tables have turned, indeed. A decade ago such proposition would have been unthinkable, but neither the competitive environment was that aggressive on all fronts. Also, Nvidia's market evaluation alone is several times that of AMD, Intel and Qualcomm combined. Source: https://www.tomshardware.com/pc-components/cpus/intel-ceo-pat-gelsinger-i-hope-to-build-chips-for-lisa-su-and-amd
  4. DuckDodgers
    Well, there's the future of computer graphics Nvidia was probably dreaming all the time. Low resolution input model -> nextgen DLSS inference -> 4K+ 120Hz output.
  5. DuckDodgers
    Microsoft continues its battel to get rig of old unfashioned hardware for its latest Windows 11 release, this time with deeper changes. Quotes Well, you either have a PC with "new" CPU after 2008 so it doesn't matter or you don't care for Windows 11 anyway -- in both cases nothing changes for the adoption rates of this Windows generation. Source: https://arstechnica.com/gadgets/2024/02/windows-11-24h2-goes-from-unsupported-to-unbootable-on-some-older-pcs
  6. DuckDodgers
    CUDA and OpenCL were a small revolution for GPGPU Compute back then. I could run an early sample code for path-tracing on a Radeon 5870 under OCL in 2009. The BrookGPU library was the real garbage before that, where compute code calls were mapped directly to the graphics pipeline with all the weird consequences, incl. system lockup due to unstable kernel-mode drivers not meant for this type of workload.
  7. DuckDodgers
    A new rumor mill indicates Intel is hard pitching its offerings to Microsoft for the next Xbox. If the rumor has some leg, this will be the first time in 20 years the chip veteran could find its way in a gaming console. Quotes Considering the mail source of the rumor (see below), don't put too much hope in it -- it's still hearsay with generous amount of assumptions. On the other hand, a potential deal for Intel would reinvigorate their further GPU R&D that would also be a positive contribution to the PC market competitive landscape. Sources: https://www.techpowerup.com/318820/intel-looking-to-grab-microsoft-xbox-semi-custom-soc-business-from-amd https://www.tweaktown.com/news/95974/intel-wants-to-be-inside-microsofts-next-gen-xbox-console-would-built-in-the-usa/index.html
  8. DuckDodgers
    China's economy is much too dependent on both exports and imports than US is. In fact, the US of all developed countries is by far the least dependent on the global trade and could wield a trade war with ease against China, taking much less damage. It also helps to have the world's reserve currency, the most powerful navy and hundreds of military bases and outposts around the world.
  9. DuckDodgers
    China was barred from accessing NVIDIA's H800s and A800s, which were already cut-down variants previously developed by the firm, but now is being reported that the US has blocked off access to an older AI product, the A100. Quotes Looks like Mr. Huang would need more "dance" classes with all the tightening export restrictions around Nvidia. Source: https://wccftech.com/us-government-blocks-nvidia-a100-ai-gpus-chinese-firm-technology-transfer/
  10. DuckDodgers
    Remember all those 4090 cards being frantically separated from their GPU chips in China? Well, someone found a business opportunity for all the leftover PCBs and coolers. Quotes Like usual, just beware of too good offers from shady dealers. Source: https://videocardz.com/newz/scammers-are-now-selling-rtx-4090-graphics-cards-without-gpus
  11. DuckDodgers
    AFAIK, after the source video file is uploaded to YouTube, the service encodes separate copies for all available resolutions with the appropriate bitrate. Say you have 2160p source file, so YT will encode individual copies at the original resolution and then 1440p, 1080p, 720p and so on down to 144p is the lowest I think. And this is how the "bitrate" is adjusted during playback, by switching between the different video file copies on the server. The only way to get higher bitrate, since YT doesn't expose such option, is to simply pre-encode your source material at the highest option (2160p 60FPS).
  12. DuckDodgers
    The China-specific variation of RTX4090 will reportedly only have reduced Tensor core functionality, but the rest of the GPU capabilities will remain the same. Quotes It would be rather curious to benchmark the impact of the reduced Tensor cores on the DLSS/FG/RR performance against the normal RTX 4090 and how much the AI hype is holding water. Source: https://videocardz.com/newz/chinese-nvidia-rtx-4090d-to-launch-today-with-reduced-tensor-core-specs
  13. DuckDodgers
    The 1.0 and 1.1 versions are affected and will be recalled. Users must stop using them immediately. Quotes Can the industry just update the ATX standard to 24V and move beyond this connector melting lunacy? Sources: https://videocardz.com/newz/cablemod-recalls-12vhpwr-1-0-1-1-angled-adapters-users-advised-to-stop-using-them-immediately
  14. DuckDodgers
    Intel is tightening the certification requirements around their new mobile platform to keep the integrated GFX performance leveled. Quotes The OEMs still haven't learnt from the Vista disaster with underpowered hardware, while Apple is charging $200 for 8GB of RAM upgrade. The PC ultra-portables need to level up. Source: https://www.pcgamer.com/intels-slapped-a-16gb-dual-channel-minimum-entry-requirement-on-core-ultra-laptops-if-you-want-top-gaming-performance/
  15. DuckDodgers
    Uncle Sam fears too much compute power could end in the wrong hands. Fiction now has become reality -- legislating export controls by the FLOPs and data types! Remember this PowerMac G4 ad? All the hype just because of a bunch of SIMD instructions added to PowerPC.
  16. DuckDodgers
    NVIDIA is set to introduce the GeForce RTX 4090 D graphics card in China, aligning with US export regulations. This move comes in response to the suspension of sales for the NVIDIA GeForce RTX 4090 in China and certain markets since November 17. Quotes $1599 for the new SKU and the 4090 keeps at $1999 -- Nvidia (stocks) wins again while we simply get another overpriced D. Sources https://wccftech.com/nvidia-geforce-rtx-4090-d-china-exclusive-flagship-gaming-gpu-us-compliance/
  17. DuckDodgers
    NVIDIA Reveals Alan Wake 2 Game Performance
  18. DuckDodgers
    Back at GDC 2016 MS made it clear that D3D11 will continue to live alongside D3D12 for the foreseeable future. Apparently, even they were aware of the difficult path the new API have to go through before gaining momentum.
  19. DuckDodgers
    The upcoming sequel to Alan Wake is already famous for its rather outlandish PC system requirements, but that's not all. A large swath of PC systems will simply not be able to run the game at all. Quotes Here's some supplementary material on the expected PC graphics card compatibility for the game: Source: https://wccftech.com/alan-wake-2-dev-says-gtx-10-and-5000-gpus-arent-supported-due-to-lack-of-mesh-shaders/
  20. DuckDodgers
    NVIDIA's pending release, DLSS 3.5, builds on the previous DLSS 3 tech, further advancing its Deep Learning Super Sampling (DLSS) technology. While AMD FSR3 and XeSS have shifted to an open-source approach, NVIDIA continues to refine its proprietary DLSS technology. Quotes Finally some ML love for RT rendering. There's so much regular image denoising can do with so few ray samples per pixel, particularly when using upscaling where there's even less than one RT sample per pixel. Source: https://videocardz.com/newz/nvidia-announces-dlss-3-5-with-ray-reconstruction-launches-this-fall
  21. DuckDodgers
    Here's extensive listing (raw dumps) of instruction throughput and latencies rates for a wide variety of x86 CPUs over the last 30 years: http://instlatx64.atw.hu/ This is more structured database on the same topic, with extra info on issue ports and micro-ops: https://uops.info/table.html
  22. DuckDodgers
    FADDs are more common in compiled code and the cost of implementation in hardware is lower than another FMA unit, just to use the accumulator for additions. Way back with Skylake, Intel doubled the ADD rate for the SSE/AVX op's to match the rate of MULs at marginal cost.
  23. DuckDodgers
    AMD's implementation of AVX512 in Zen4 is probably the most balanced and efficient approach to date. Zen4 can sustain 1x512-bit FADD and 1x512-bit FMA every cycle with minimum transistor and power cost, while to achieve this Intel had to slap a second 512-bit FMA unit, with all the overhead resulting of that for their server parts. Intel turned around with Golden Cove, but they fusied off AVX512 in ADL, so ordinary customers were denied the performance benefits anyway. There are very few practical use-cases for 2x512-bit FMA to be worth the extra cost and FADD circuits are cheaper to implement. By the way, mask operations in AVX-512 are still much faster in Zen4 than any of Intel's designs. Kind of a sheme here, for one of the premiere features in the instruction set. On the other hand, Intel still beats AMD in load/store bandwidth.
  24. DuckDodgers
    The x86 decoding and branch prediction stage in any modern implementation takes significant area and power budget, because an x86 instruction length cannot be determined before full decode. This makes superscalar front-end even more expensive, compared to any typical RISC architecture.
  25. DuckDodgers
    It's ironic that Intel's first attempt to kill x86 with even more advanced CISC architecture called iAPX in the early 80s. This new extension to the ISA (for the first time since AMD64) actually diminishes a lot of the advantages of ARM, although the decoding overhead will still be consuming more power I guess. The only question now is the rate of adoption of the market and the x86 ISA will indeed be proven as immortal.
×