林彥宇

[林彥宇 @Liopleus]

@koidetakushi Thank you for the wonderful work. Although I think the execution of the ending could be better, I like the overall direction and think you guys did a great job with the screentime you have.

[林彥宇 @Liopleus]

@MOFA_Taiwan Please take more action to remove stray cats and dogs if you want Taiwan to remain this way.

[林彥宇 @Liopleus]

@lauriewired 1. Where's the latency data for local 3080? Feel like this graph is hiding the disadvantage of cloud gaming.
2. While it's more efficient, the surplus from the extra efficiency mostly go to the profit of Nvidia instead of lowering the price of consumers.

[林彥宇 @Liopleus]

@AMD Annnnd you said FSR 4 won't be available for this APU right after Nvidia put DLSS on their RTX Spark marketing materials.

[林彥宇 @Liopleus]

@KuwaLee2000 因為這跟 GB10 幾乎是同一顆晶片，所以開發 DGX Spark 的時候 Linux 驅動大部分都搞定了。

[林彥宇 @Liopleus]

@harry03994688 很多人也低估了 AMD 在 SoIC 封裝上搞了好幾代產品累積的技術資產，將 SoIC 應用在量產產品上需要的 physical design 經驗跟製造公差處理能力是很難追的。

[林彥宇 @Liopleus]

@ShanghaoJin INTC 毛利從來就沒有 90%，因為要充分利用自家晶圓廠產能，做了很多低毛利的中低階處理器。
他們毛利最近 20 年都沒有超過 70%。

[林彥宇 @Liopleus]

This is hilarious lol

[Framework @FrameworkPuter]

Every time we engage with an influencer on X dot com, Dell sends them an XPS. Anyone want a free Dell XPS?

[林彥宇 @Liopleus]

@FrameworkPuter @yacineMTB I hope the battery life catches up with the recent Windows competitions. Framework has fallen behind on this regard in recent generations due to other brands improving a lot.

[林彥宇 @Liopleus]

@pavandavuluri 2026 will be the year Windows Update finally become exciting.
Although I feel like this is partially a response to MacBook Neo.

[林彥宇 @Liopleus]

@Mojo_flyin @JumboShrimp787 AMD licenses x86 from Intel and Intel licenses the x86-64 instruction set from AMD. If Intel want to license x86 to Nvidia without AMD consent, it would need another 64-bit instruction set, otherwise it would be limited to 32-bit.

[林彥宇 @Liopleus]

超���仁菜

[まさここ超百合爆彈@百合ONLY E01 @Kokokoro__]

仁菜的反���
#girlsbandcry
#ガルクラ

[林彥宇 @Liopleus]

@SemiAnalysis_ I think that's because there are plenty of systems using Instinct with ConnectX, more than system using Pensando currently, so they need to prioritize the hardware combination more customers are using?

[林彥宇 @Liopleus]

@kaorukakoii 真心推薦 Qudelix 5K，可以藍牙或 USB 轉有線耳機還能做 EQ。

[林彥宇 @Liopleus]

@freeshiuan 台灣的虹韻也有做客製化耳塞，我自己目前用了三個禮拜覺得非常滿意，是選零壓睡眠專用款。
ACS 的製作精度可能比較高，但虹韻的價格不到一半，等待時間也比較短，預算有限的話可以參考：
https://t.co/c9OlrkA98Y

[林彥宇 @Liopleus]

@MH_Canta You should use GeForce Now to play when you're overseas if the place have decent internet.

[林彥宇 @Liopleus]

@AnushElangovan Might be off topic but is it possible to have 2 Strix Halo IOD connected together and act as a dGPU with 256GB LPDDR5X?
Put it into a PCIe card and it won’t have any competition at $3000.

[林彥宇 @Liopleus]

@SapphireTech Can you go into detail about how you achieve even distribution of power?
Buildzoid can't find the mechanism by looking at the photo of PCB. He didn't test the actual card so there might be something he missed:
https://t.co/VzD9rjujay

[林彥宇 @Liopleus]

@jimkxa @elonmusk Resetting and starting from zero isn't always a good idea. If you stop taking antibiotics in the middle of treatment you'll get drug-resistant bacteria/virus, even if the pause is short.

[林彥宇 @Liopleus]

@gindi4711 @aschilling If that's the case Nvidia would've used standard GDDR6 on the 4080, since it would cost less and not lose any market share.