8G5 will certainly have Pegasus. If X2 is releasing in July 2025- as the leaked Dell roadmap suggests, I could still see the possibilt that X2 may have Pegasus. I don't see what's the point of releasing X2 without a major uplift in CPU performance.The document, on the other hand, puts "Oryon V2" in the timeframe of X2/X Gen 2/etc, which should be based on Phoenix-L, not Pegasus.
You aren't helping me to explain by mixing the wrong things together 😅What I meant by CPU is the block itself, i.e., is Oryon "V1 to V2" like Cezanne to Rembrandt or like Phoenix to Strix?
SoC Brand : Snapdragon X = Ryzen AIYou aren't helping me to explain by mixing the wrong things together 😅
Cezanne/Rembrandt etc are all SoC codenames (not actually used in branding), Zen3/4/5 are the CPU cores for those SoC's.
SoC (system on chip) is the whole silicon microchip with several separately designed islands, all connected by a common data bus and power network.
CPU is the main processor of the chip, and GPU runs gfx + massively parallel compute, and now the XPU/NPU or whatever AMD calls it for running AI inference tasks.
The lawsuit looms above all, like a dark stormcloud.As to how oRYON will be branded in the future I have no idea, at the moment they are probably just praying that the ARM lawsuit doesn't go nuclear and spent minimal time worrying about it.
I suggest they should brand the big core and little core as Oryon V1P and Oryon V1E.Bear in mind that they still haven't introduced their efficiency core yet, so their will probably be something like Oryon Gold 1xx and Oryon Silver 1xx for Snapdragon 8 Gen4, while a future SD8 gen with a new CPU µArch would be branded something like Oryon Gold 2xx and Oryon Silver 2xx.
But that is all very speculatory on my part.
No that suits fine.SoC Brand : Snapdragon X = Ryzen AI
SoC Tier : Elite = HX (??)
SoC codename : Hamoa = Strix Point
CPU Brand : Oryon = Zen
CPU generation : Oryon V1 = Zen5
CPU core codename : Phoenix = Nirvana
This is a very crude analogy.
Too Intelish.I suggest they should brand the big core and little core as Oryon V1P and Oryon V1E.
All? Naaaah.The lawsuit looms above all, like a dark stormcloud.
I know what they are, and I'm sure you know too, that's why I'm not calling Ryzen 5000/6000 or whatever. You're probably confused because you come from the idea that I don't know what I'm talking about.You aren't helping me to explain by mixing the wrong things together 😅
Cezanne/Rembrandt etc are all SoC codenames (not actually used in branding), Zen3/4/5 are the CPU cores for those SoC's.
Will it? I precisely gave a counterexample. We have Kryo Prime in the same year (and the same "Gen X" for that matter) based on both X2 and A715, which are (very) different microarchitectures. Same thing happened with Kryo 670 and 680 (equal generation/series, very different cores).Bear in mind that they still haven't introduced their efficiency core yet, so their will probably be something like Oryon Gold 1xx and Oryon Silver 1xx for Snapdragon 8 Gen4, while a future SD8 gen with a new CPU µArch would be branded something like Oryon Gold 2xx and Oryon Silver 2xx.
I'm not putting any stock. I just pointed out inconsistencies in the naming convention.8G5 will certainly have Pegasus. If X2 is releasing in July 2025- as the leaked Dell roadmap suggests, I could still see the possibilt that X2 may have Pegasus. I don't see what's the point of releasing X2 without a major uplift in CPU performance.
Which is why i would also advise not to put too much stock into that Dell slide. For one, it was devised last year- and plans/timings may have changed since then, and some details are dubious (more than 2 year gap between Oryon V2 and Oryon V3!?).
Samsung Electronics has chosen MediaTek’s Dimensity 9300+ application processor to power its latest premium tablet, the Galaxy Tab S10, instead of Qualcomm Snapdragon chips, media report, noting the move could be aimed at getting the upper hand in negotiations with Qualcomm. It’s the 1st time MediaTek chips have been chosen for the Galaxy Tab series.
Apple is accelerating its move away from reliance on Qualcomm. In 2025, two new iPhone models will ditch Qualcomm's 5G chips and adopt Apple's in-house 5G chips: the iPhone SE4 (1Q25) and the ultra-slim iPhone 17 (3Q25).
Would make sense to use those two models as first-gen and then iterate enough to do a full release with the 18s and ditch Qualcomm modems for good.
Apple is accelerating its move away from reliance on Qualcomm. In 2025, two new iPhone models will ditch Qualcomm's 5G chips and adopt Apple's in-house 5G chips: the iPhone SE4 (1Q25) and the ultra-slim iPhone 17 (3Q25).
The M3 Air has a smaller battery than the Dell XPS and a higher resolution screen than the tested Dell XPS.Dell XPS 13 2024 Snapdragon version review:
Dell XPS 13 (2024) review: The new battery life champ
Even MacBooks can’t keep up with this laptopwww.tomsguide.com
The rumours are very strange indeed:I'm however extremely skeptical of the idea that would use it in the fabled iPhone 17 "slim". There are conflicting rumors but that's supposedly going to be the most expensive iPhone 17, and point to the future of iPhone design the way iPhone X did. I have a really hard time believing they would feel their modem is ready for what would be their most important iPhone 17 model, but not for the rest of the iPhone 17 lineup. Unless they have a more integrated solution (including all the various filters, antennas, etc.) than Qualcomm's it doesn't seem like you'd get much benefit internal size wise from it to where it would be necessary for the "slim" model.
Could Oryon V2 have 256 bit SVE2 SIMD pipes?
I guess it would greatly help with x86 AVX2 (256 bit) emulation, which a lot of Windows software uses.
Of course, AVX2 can still be emulated using 128 bit SIMD pipes, but it would be slower.
I would disagree. That argument is more fitting for AVX-512, which right now is still niche. AVX2 on the other hand is has become quite pervasive by now, with hardware from both Intel and AMD supporting it for many years. Many AAA games use AVX2, and reviewers noted that applications that use AVX2 won't simply run on Snapdragon X laptops, because the Prism emulation layer doesn't support it256 bit AVX2 is pretty useless for non-hpc use. Not many programs can use such a wide vectors.
Idk. Intel not supporting wider vectors on their E-cores seem to be causing problems on their side...Intel E-cores use a similar approach to ARM, more 128-bit units which actually improves performance unlike widening vectors.
There are uses outside of HPC for wider than 128-bit vectors: for instance game engines relying on bitboards (Chess, Othello, etc.) and encoding/decoding. That's the reason why beyond floating-point instructions, the SIMD extensions have some, at first, seemingly odd instructions that operate on integers.256 bit AVX2 is pretty useless for non-hpc use. Not many programs can use such a wide vectors. Intel E-cores use a similar approach to ARM, more 128-bit units which actually improves performance unlike widening vectors.
There are uses outside of HPC for wider than 128-bit vectors: for instance game engines relying on bitboards (Chess, Othello, etc.) and encoding/decoding. That's the reason why beyond floating-point instructions, the SIMD extensions have some, at first, seemingly odd instructions that operate on integers.