I think snapdragon is comparing their chips to M3 with 8GB of ram in ram heavy benchmarks, while using 16+GB. Or, they’re using their most powerful chip with full 80W+ comparing it to M3 which maxes out at like 20W. Something doesn’t seem right with how they present their numbers and I want to see real benchmarks
Semi-Accurate is truly holding up to it's name 😅 cuz the xElite is kicking ass 🎉😂 against m3 benchmarks and apple had to announce an early M4 release.😂
I was with you until the part where you said that measured performance was 50% of what was claimed. That's more than a rounding error. No way Qualcomm would have juiced their numbers that much.
Yhe US id noy s lion but a bald headed vulture that roam the sky scavangingy the sky for decay meat.. Now cam. Any dhoe potrayyhe ud ss a loon is beyond sny sane thinking human
It will become obvious if the benchmarks are wrong once real laptops with this processor hit the market. If performance is not what was claimed then everyone will know and Qualcomm's reputation will suffer.
Ok I have read some of the comments. As of date Taiwan has been helping China behind the scenes. And I think Japan is going to do the same. Ukraine Saga has exposed the real devil. Huawei has been helping the US despite of the sanctions with keeping up their sponsorship. Good versus evil good always wins in the end
You are comparing 4nm Qualcomm Chip using EUV litography and Armv9 architecture with a 7nm Kirin 9010 using DUV litography and ArmV8 architecture. If you want a fair comparison you should compare Kirin9010 with SD 870. Both is 7nm and ArmV8 architecture. 😁 😁😁😁
It is sick that the US has the power to tell other countries to prevent China's growth because the US cannot compete with them. China must invest more into research and development of their own machine. Become totally self sufficient.
Comparable to Intel's 10th generation of which intel chip, i3, i5 & i7 ? If it's just i3 then Loongson still need to buck up at least to the level of i7 10th generation.
By the time all these fabs in the U.S. get built and become market viable, China's focus will pay off. Not only will China not be needing TSMC, NVidia, Intel etc., they will be producing for their own massive market as well as competing against grossly over priced "Made in 'Merica" offerings and we all know who will win that competition. This chip war nonsense is doomed.
My God, I hate this new Intel naming scheme so much that I think I'll probably switch to AMD just because it's so bad and there's no reason for the change. Plus with how unstable their newer high end chips are, just one more reason to stay away. I also hate E cores, I'd rather have half as many P cores to replace the E cores- get rid of the extra complexity, I don't want any E cores, I want more P. AMD it is for me for the foreseeable future. I've had Intel chips ever since the 4 series but I don't like where they're going, especially with this brute force approach that just keeps sucking more and more power each generation as AMD gets more and more efficient.
the LGA 18xx Socket will see 2 generations of CPU Core Ultra 200 and then Core Ultra 300. that's it. Intel is very stingy with future support, they rather sell new boards and chipsets every 2-3 years to ppl just to make like harder
I do kinda hate this mix of P/E cores myself- take 4 of those E cores and turn them into 2 more P cores, I'd rather have less complexity and more Performance Cores.
28 дней назад
Canon had already said last year that it will not sell its NIL lithography machine to China.
@@dbdesignlabs i think even if they are not affected by USA, they are continuously targeted by the Chinese government as the bad neighbours. Why would Japan help an unfriendly neighbour?
it not even beat the Snapdragon gen 1, just barely surpass the snapdragon 888 in the multi core benchmarks due to multi threading. let alone M3. the Architecture of Taishan v130 was suppose for server processor. since it has multi thread capabilities, kirin 9010 also inherit this features, but the downside it, Mobile platform are not suitable for multi-threading , u gain nothing in the real world usage, but it indeed provide extra "marks" when running benchmark. Chip designer can always put extra physical cores for more demanding task, the reason huawei not able to do it was processing node limitation by SMIC, since 7nm with multi patterning technology, combine with Huawei redesign of core architecture still not sufficient to compensate the lower density of processing nodes. ( SMIC fabricate their chip with ASML 14nm DUV lithography system. with 4x multi patterning technology to achieve similar density of 7-6nm ) if u want to know the real "performances" of kirin 9010, look at the single core performances, it barely made any differences since kirin 9000. , the spike of multi-core benchmarks was mainly due to multi-threading and has nothing to do with real world performances.
Your ANAL-ysis has nothing to do with the upcoming next-gen Kirin chip above, Taishan architecture is modified here, not 100% same as server versions. It's 9000s from 2023 not "9000" from 2020, 9010 is a mid refresher from 9000s, upgraded generation annually, if you thought you can shift to new generation in 8 months without a full 12+ month period is unrealistic. The new chip was successfully debugged and had 1.1 million benchmark scores on Geekbench and it is a new node, 5nm without EUV machine and the yield rate was passed 72% as of end of Q1 2024. H2 2023 50% Yield: Report - July 2023 breakthrough = 2mths later, Aug breakout. Improved yields = results, 10m units March, 20m units April '24 sold with activation of 17.33 million units. 72% Yield - April 2024 for H2 2024. Latest flagship April 2024 launch quantities three times previous generation August 2023 launch. Real performance is raw performance not paper specs. TSMC is not doing great in it's 3nm generations, lots of revision for the best batch with EUVs with the stochastic effects. ASML is not the only machine, Canon DUV and NIL, with Chinese chipmaking components and EDA softwares, customised engineered.