true iBegginer

5.6K posts

true iBegginer banner
true iBegginer

true iBegginer

@true_ibeginner

Cutting-edge technology addicted. Apple не best, но лучшее в B2C. Получил ретвит от Benedict Evans и лайк от Max Weinbach. Работаю ¾-stack+mobile разработчиком

Россия Katılım Ağustos 2017
182 Takip Edilen76 Takipçiler
true iBegginer retweetledi
dosdude1
dosdude1@dosdude1·
The original Android phone (HTC Dream/G1) and the original iPhone... Both running the same OS.
dosdude1 tweet media
English
25
103
1.5K
46.7K
true iBegginer
true iBegginer@true_ibeginner·
В презентации SGS24 недоумеваю с момента, когда через AI переводчик спрашивают как пройти к дворцу. Такое ощущение, что люди не умеют пользоваться картами во время поездок.
Русский
0
0
0
41
true iBegginer retweetledi
Austin Byrd
Austin Byrd@AustinTByrd·
Can't believe it's 2023 and we're still getting scammed by Steve Jobs impersonators. Your instincts are telling you something was off. Your instincts are right. $700 +$24/month for ChatGPT inside an Alibaba projector.
Austin Byrd tweet media
English
381
323
6.5K
1.8M
true iBegginer retweetledi
Vladi Miro 👁️
Vladi Miro 👁️@mirvla·
Просто Яндекс работает по International Fixed Calendar (IFC) 🤣 По сравнению с 12-месячным григорианским календарем IFC намного проще: – Год разделен на 13 месяцев (каждый по 28 дней) – Календарь каждый год один и тот же. – В каждом месяце ровно 4 недели. – Каждое число каждого месяца приходится на один и тот же день («16-е число» — понедельник). – 13-й месяц назывался «Солнце», так как дополнительный месяц приходился на период с июня по июль, в период летнего солнцестояния. Основатель Kodak Джордж Истман был поклонником IFC и использовал календарь для отчетности Kodak с 1928 по 1989 год. Почему? В каждом месяце было одинаковое количество дней (и выходных), что облегчало сравнение отчетов о прибылях и убытках за разные месяцы и годы
Vladi Miro 👁️ tweet media
Чё там у смешариков?@aintmybike

Если вы, как и я, задавались вопросом, почему Яндекс списывает за подписку с каждым месяцем все раньше и раньше, то я спросил за вас:

Русский
8
23
308
86.9K
true iBegginer retweetledi
RGcloudS
RGcloudS@RGcloudS·
Truth time 👁️ Deep stalking kernel level Does SMIC really made kirin ? Does 7nm actually possible with legacy DUV machine ? Who made Huawei Kirin 9000s ? Why there's a lot of weird information, like in Image #1 , Kirin 9000s classified as 5nm 😶 Image #2, known information of Hisilicon cpus. kirin 9000 truly listed as 5nm, both 9000 & 9000s recognized as the same chip Back to Image #1, there's 2 unidentified codes "72_3330" "72_3394" The hexadecimal should be implementer 0x48 part 0xD02 implementer 0x48 part 0xD42 What does 0xD02 0xD42, etc means? It's a code set by Arm to name their cores ¹developer.arm.com/documentation/… other cpu parts from different manufacturers also use these kind of code { apple qualcomm broadcom, literally almost everyone } you can check other names here [ not updated for 2022+ ] ²#L61" target="_blank" rel="nofollow noopener">gitee.com/openeuler/kern… So, 0xD42 is Arm A78AE which is used for automotive & industrial solutions that makes no sense if kirin use that cpu 💀 Image #3 By the time A78 & A510, huawei already got sanctioned So they're just guessing next arm core value, A510 was correct prediction 0xD46 But A78 is wrong. arm name it 0xD41 Huawei guessed D42 { A78AE } How about 0xD02 ? Well, another suspicious thing There are 2 cores with same name As you can check on Link² { Number 113 & 114 } 0xD02 is Hisilicon Taishan CPU TSV200 which made almost 3 years ago, first submitted 2 years ago here gitee.com/openeuler/kern… 0xD02 renamed to Linix core 9100 months ago before mate 60 launch approximately 4 months ago here ↓ gitee.com/openeuler/kern… First of all, you can rename your cpu into anything you want. But you can't rename 0xD codes, because it's a MIDR_EL1 ID value set by ARM { refer to Link¹ } MIDR_EL1 register is read-only. cannot be modified through software means. The value of MIDR_EL1 is stored on the chip eFuse, and eFuse (one-time programmable memory) generally can only refresh data once and cannot be modified twice. even if the code { 0x } is found to be repeated, the early chips cannot be changed to other values That's why huawei can only change cpu name but not the value pretending it was brand new core but not at all. Apparently huawei did think to name Kirin 9000 to 9100, but the end up with 9100s Image #4 In Mate60 teardown, 9000s there's " 2035 " Means the chip was produced since / in 35th week of 2020 which is 24-30 August 2020 kirin 9000 already in production before arm officially introduced 710 & 510, and huawei guessed it's core value wrong Summary Kirin 9000s is actually kirin 9000 made by tsmc 3 years ago not SMIC That's why the score of both chips are too similar. obvious Wait how? some of you know that chinese has spy everywhere, high patriotism there's a lot of IP theft news already { i don't mean to be racist at all } Huawei already anticipated this before US issued sanction So much overstocking is it really possible to stock that much? Yes. in 2020, TSMC capacity of combined 6 & 7nm was 140000 wafers per month. Let's say huawei booked 50% 70000 wafers 1 wafer = around 800 Hisili chips Tsmc yield is super high that time Maybe take 85%? 1 wafer = 680 chips 680 x 70000 = 47.600.000 chips per month In just 3 months, Huawei can overstock 142 million chips of combined Kirin 9000E & 9000 SMIC 7 N+2 yield was too low { <20% } Mate 60 was too good & smooth Even samsung at 35% yield on gen 1 & exynos 2200 was so trash It doesn't matter how talented you are, you can't produce 7nm with legacy 1980 DUV machine, multiple steps of stacking to create compared to EUV Especially huawei targets 40 million of mate60 sales with such yield? nonsense I'm just wondering, why did huawei so pessimistic losing the battle on court and decide to overstock tons same chip? are they really spying ? well. who knows Everybody spies anyway, doesn't matter who you were Just like Ireland decide to keep huawei 5G, because both sides are spying Oh well, forget politics Huawei & SMIC successfully pranked the whole world 🫠
RGcloudS tweet mediaRGcloudS tweet mediaRGcloudS tweet mediaRGcloudS tweet media
English
31
31
191
50.8K
true iBegginer retweetledi
gaut
gaut@0xgaut·
it’s over for engineers
English
285
648
6.2K
2M
true iBegginer
true iBegginer@true_ibeginner·
Календарь подсказал, что в 2021 великий Куо ванговал в этом году складной айфон и айфон без вырезов 9to5mac.com/2021/03/01/kuo…
Русский
0
0
0
46
true iBegginer
true iBegginer@true_ibeginner·
Сон был сегодня в руку, что сижу на ДР @eldarmurtazin и говорю тост :) Календарь напомнил про видео 2019 m.youtube.com/watch?v=XLYyg2… в Apple системный кризис, клопы в магазинах, клавиатуры-бабочки, еще пару лет и Кук похоронит компанию :) Видео теперь почему-то доступно по ссылке
Русский
1
0
0
60
true iBegginer retweetledi
Mishaal Rahman
Mishaal Rahman@MishaalRahman·
Google is working on a new developer option in Android that will swap out your device's Linux kernel with one that uses a 16K page size. Compiling the kernel to use 16K pages could provide a significant performance improvement but also break many apps. A page is a fixed-length contiguous block of virtual memory. Android currently uses a 4K page size, while iOS and macOS use a 16K page size. Some workloads benefit significantly by using a larger page size, like kernel compilation, as shown below: But although ARM64 Linux binaries are compatible with 4K, 16K, and 64K pages by default, many native apps and Android OS components that assume a 4K page size might just break. (These comments by the lead Asahi Linux developer go into a lot more detail about the potential challenges: #issuecomment-1518600436" target="_blank" rel="nofollow noopener">github.com/k3s-io/k3s/iss… reddit.com/r/AsahiLinux/c…) This is why Google is incrementally experimenting with 16K page size support in Android. They've been working on compiling the kernel with a 16K page size and fixing various issues since last October, but late last week, a Googler submitted a series of patches to AOSP (that have not yet been merged) that add a "developer option for booting with 16K pages." 16k_dev_ui: android-review.googlesource.com/q/topic:%2216k… If/when these patches are merged, a new "enable 16K pages" toggle will be added under Settings > System > Developer Options. When you toggle this option, Android will warn you that "some applications may not be compatible with this mode." Android will then call update_engine - the system OTA service - to swap the kernel to a 16K compatible one. I have no idea if/when this developer option will land, nor whether Google will actually go through with changing to a 16K page size compatible kernel. If they do, it won't be a while and the transition will be slow given how much might break.
Mishaal Rahman tweet mediaMishaal Rahman tweet mediaMishaal Rahman tweet mediaMishaal Rahman tweet media
English
18
86
638
102K
true iBegginer retweetledi
Justine Moore
Justine Moore@venturetwins·
Self-driving cars are fixing the birth rate
Justine Moore tweet media
English
331
412
7.4K
2.1M
true iBegginer
true iBegginer@true_ibeginner·
Это ВАЗ Нива? Ну, и commies цвет и стайл на слайде
true iBegginer tweet media
Русский
0
0
0
53
true iBegginer retweetledi
Techmeme
Techmeme@Techmeme·
Humane co-founder Imran Chaudhri demoed an AI-powered wearable with a projected display that does not need a nearby phone and uses voice and gestures for input (@inafried / Axios) axios.com/2023/04/21/hum… #a230420p30" target="_blank" rel="nofollow noopener">techmeme.com/230420/p30#a23
English
0
1
11
7.9K
true iBegginer retweetledi
Maxim Shulga
Maxim Shulga@maxbeard12·
Maxim Shulga tweet media
ZXX
3
14
90
0