#hardware acceleration settings

[squarshedbrains]

How To Enable Or Disable Hardware Acceleration In The 360 Secure Browser

How To Enable Or Disable Hardware Acceleration In The 360 Secure Browser | PC Tutorial | *2024

In this tutorial, we'll guide you through the process of enabling or disabling hardware acceleration in the 360 Secure Browser on your PC. Hardware acceleration can improve browser performance or help resolve issues with rendering. Follow this step-by-step guide to optimize your browsing experience. Don’t forget to like, subscribe, and hit the notification bell for more 360 Secure Browser tips and tricks!

Simple Steps:

Open the 360 Secure web browser.

Click on the 3 bar hamburger menu in the upper right corner and choose "Settings".

In the left pane, click on "Advanced" to expand it the section, then choose "System".

In the center pane, toggle on or off "Use Hardware Acceleration When Available".

[merrilark]

Actually if you watch on your computer, you can take screenshots using the Firefox browser!

streaming companies will say um we're increasing your subscription fee. no password sharing. no screenshots allowed. please subscribe to a separate channel for this movie and another for this tv show. free trial but put in your card details so we can charge you if you forget to cancel. this title is a rental only that's 4.99 please. this title is not available in your region. you are begging people to torrent at this point Like ye are off the edge of the map matey here there be pirates argh argh argh 🦜☠️

[agnoll]

firefox just started doing this too so remember kids if you want to stream things like netflix or hulu over discord without the video being blacked out you just have to disable hardware acceleration in your browser settings!

[kaisaniku]

I finally finished the second chapter of this doodle comic!

Basically from a translation machine:

It's called notes, but it's really bullshit time:

(I didn't put any asterisks in the "text" in order to make the picture cleaner and not make this little comic seem too serious)

①WX says that their body "doesn't have any of those really advanced things" and that "the concept is just bionic", which might make you wonder how WX can agree with their own shortcomings when they are always been so confident in their own machine body. This is kind of a guess: WX's comments in the wiki about Wagstaff being "afraid of progress" and "short-sighted", combined with WX's usual machine-worshiping and violent tendencies, such as their comments about Maxwell "He has that kind of power but he doesn't go out and destroy human". So it's an offense to them, but they actually agree that even though they claim to be superior, I guess WX would like to see their bodies more advanced and even more radical.

② Yes, this WX has acceleration circuits installed.

③ Gnome get✓ Do you guys remember the line where WX checks the Gnome

④ "You're not as kind as you look", how should I put this one, because the Wilson I understand is more or less with a little bit of darkness in his mind, he's someone who is full of emotions but often restrains them with reason, he fears and hates the negative impulses and instincts within himself. And this trait I think probably comes from his good nature, his education and survival experiences and the influence of the Shadow Throne. (But honestly after suffering in CONSTANT for so long, it's normal for whoever it is to be a little psychologically unhinged, not to mention the SANITY setting.) And after WX's soulwalking, they easily have some "psychic empathy" with Wilson... ...... Well, how did that happen Wilson?

⑤ Well I know the reasoning of the empathy module episode is weird ...... In fact, it's mainly because when I drew this plot I didn't have a good understanding of the empathy module, and simply thought of it as something like "emotional deficiency", so this episode was supposed to be Wilson saying "why do I still feel emotional ups and downs blabla" and then WX explaining that it's because of "memories of emotional experiences", which would have made a lot more sense.

⑥I guess it's my own personal setting: although the game doesn't make a distinction, I don't think WX as a robot would have a "headache and blurred vision due to lack of sanity". The system will be affected, sure, but the physiology won't necessarily feel it. It's hard to go from luxury to frugality, and since there is no experience in the eternal realm in the human era, WX can't adapt to the negative impacts of sanity reduction at all.

⑦Wilson's curiosity and desire to explore and then equipped with WX's hardware strength is simply ON FIRE. and "adapting to the human body so quickly" this conclusion mainly comes from the last chapter when the two people just transformed the body, Wilson's side is very difficult, while the WX on the contrary, it seems to be very easy. Even when they suddenly possessed internal organs, blood, light weight and so on, there was no adverse reaction. Wilson, who loves to observe, has always had suspicions (sorry however I didn't draw this clue out)

⑧ on the one hand, just learned a shocking secret, excited and energetic Wilson, on the other hand is the history of the exposure, and is also experiencing unprecedented headache WX. so the two temperament is not quite the same as usual.

⑨ "Wiped of most of their human memories" from the game's credits: "Suddenly recalling the memories of his past life, WX-78 soon decides to change his fate on his own."

⑩ Those of you who have fought Shadow creatures online might know that the only way a teammate's Shadow creature will have hatred for you is if you've forcibly attacked them. What happened here is that WX forced an attack on Wilson's Shadow Creature (except that Wilson was still relying on his headache of empirical judgment and didn't realize that his SANITY was too low), and then WX's own shadow creatures that were looming all showed up as well, which is why it became so much more. I don't have a very comprehensive understanding of this mechanic online though, and it doesn't seem to be very rigorous, so that's probably what it is anyway.

I accidentally added a lot more, mainly because of the limited ability to express the drawing ... Hope you enjoy!

[noneatnonedotcom]

obligitory batman like character. the gentleman badass.

this character doesn't have physical superpowers instead he's mastered two distinct mental techniques that better allow him to learn. mainly in an attempt to impress a girl. now the only question is just how far will he be able to push himself?

build under the cut

Man of the New World: (200cp)

"As the world moves on from hitting each other with crude sticks and fists, power begins to change hands.

The strongest man in the room goes from an undoubted leader to mere muscle for the true power, the man with all the knowledge.

A combination of natural prodigy and years of effort gave you a position among Britain’s most vaunted minds. While your most famous talents are focused in a single academic discipline, such as Moriarty and his world-renowned skill with mathematics or an unsurpassed brilliance with biology, you are broadly skilled with many other fields.

From several respected academic paths to the different styles of the arts and even a deep understanding of philosophy, you match the learning in each of what most men try a decade or more to come close to.

Though a touch barbaric, you even indulged in learning a martial style of the empire, having enough experience in boxing or baritsu to knock back-alley heads together without trouble.

It’s made possible with a particular mental technique that allows you to absorb and retain information easily, the creation of a continually expanding ‘memory palace’ style territory within your mind.

Information is easily memorised and efficiently stored, granting the ability to rapidly make connections between even seemingly disparate pieces of data that you have gathered. Knowledge of disciplines you once studied thirty years ago remains as fresh as the new fad you spent an afternoon exploring."

Thought Acceleration: (200)

"Even a human brain operating at peak capacity has its limits - it is, after all, what amounts to biological hardware with set limitations.But what if those limitations could be surpassed?

You are able to accelerate your thought processes to twice their normal speed, letting you quickly multitask and react faster than body and nervous system would normally allow, running calculations and strategies in a blink rather than a minute.

In time, you could even accelerate this to three, or four or more times in speed. However, much like overclocking a computer, this is a stressful technique that is exhausting and, if overused, can burn your brain out like an overheated engine. Practice and careful moderation may let you overcome this, but for now, be cautious in how you use this."

In for a Penny, in for a Pound of Flesh: (200) "Training? Guns? Numbers?

All is for naut against true battle experience.

You have no issue taking on entire armed groups on your lonesome.

The more you kill, the larger groups you can take on, and the stronger individuals you can challenge. With a few wars under your belt, taking on a small army or a stone-skinned monster is child’s play."

Stealth Efficiency: (100)

"The strong rules the weak, so how can the weak defeat the strong? Like how two brilliant NEETs from another universe always say, the weak’s weapons against the strong are their intelligence and strategy, not their power.

Apparently, you learned a different lesson from this ideology: no one, regardless of their power, can prepare themselves from an attack they don’t know is coming, including those stronger than you. Your attacks are twice as effective and powerful when used against an unadvised opponent, but this power boost loses its effect once they perceive your presence.

This is perfect for a first surprise attack to end a fight before it starts, letting even a weaker servant defeat a stronger master."

Stealth expert: (200) "You’re a Master of stealth. You could infiltrate highly armed military bases and no one would ever even know you were there."

Stuntmaster: (100)

"The absolute best part of being a skilled warrior as a Solar Exalt? It’s how your every movement follows with both grace and grandeur, how you become a stuntmaster that makes combat look almost like dancing.

Your attacks become eye-catching and graceful, and while no more effective, your skills in battle will look visually impressive in addition to your already-existing effectiveness. This can also be applied to any sort of skill, power or trade you put effort and enthusiasm into, letting you perform all of your abilities while looking absolutely fantastic doing so."

Stunt Master: (free)

"You are a master of stunting. Any time you try to do something amazing the world will conspire to make your attempted stunt just that little bit more dramatic and incredible. Maybe there just happens to be a rope to let you swing across the ballroom dramatically as you make off with the nobles treasure, or maybe there's just the right amount of handholds and footholds on two opposing walls, at just the right distance from each other, that would allow you to dramatically wall jump your way up.

So long as you strive to do things in an amazing and/or over the top fashion you will be slightly rewarded for doing so, Fate conspiring to make you that little bit better at whatever you're doing because apparently Fate is really fond of over-the-top action sequences. You can still fail when you're doing over the top stuff though, so don't get too out of hand, this just represents a slight-to-moderate increase when you're purposely trying to be amazing. "

Beautiful: (free) "You are very beautiful, guaranteed to be beautiful enough make the majority of the opposite sex think you are hot. This won’t shift your appearance to something you find ugly. "

so the basic idea behind this character is a sherlock holms/ batman like character who's main abilities are mental. his mind palace plus thought acceleration puts him well above street level and his skill with baritsu means he's not a slouch in combat anyways. but his skills in stealth and using surprise to take down his enemies really pushes him to the level of someone like batman, plus like batman he's got plot armor in the form of his stunt master abilities. Everything he does is over the top and stunning to behold. and as he keeps fighting he keeps getting better his skills somehow never hitting a peak letting him fight even superhumans.

his one downside is that he starts off as a kid trying to impress a girl and he never really out grows that. no matter how many years and adventures he has under his belt this character will always find himself in trouble by chasing a pretty face.

He'll also always start young, having to grow into his own body to truly be the force on the battlefield he's destined to be. but because of his more esoteric mental abilities, he's far more than just a brute able to plan and strategize around enemies and react to or predict things, moving faster than human thought. a genius.

This character doesn't need to be the hero. if you think they'd do better as a phantom thief or big bad evil guy, or even just a merc doing work for the highest bidder or prettiest face he'll fit into and thrive in any situation.

@heliosthegriffin i'm picturing this as an alternative Shadow Knight, but this jaune learnt mental abilities rather than trying to out fight the Grimm learned how to go all in on the advantages he has. Still having neo as his assistant and having a less heroic and more mercurial bent to him, he's just trying to keep his girl entertained after all.

@weatherman667 @howlingday how would you two use a character like this?

i like the idea of a guy who gets mental powers but thinks he's stuck being a street level hero for the rest of his life, un aware that he's constantly taking on threats that the heavy hitters just can't handle because they're too big to find the root cause of the villainry of the city.

[nostalgebraist]

In explainers about AI, every neuron of the first layer is always connected to every neuron of the second, etc. but is that really necessary? If you have 100 neurons per layer and every neuron connected to 10 neurons in the next layer then every neuron in the first layer can still affect every neuron in the third, and this would have about 1/10 the computational costs, so you could compensate by having bigger or more layers. Would a neural network with sparser connections between layers have a commensurate drop in capability to make this not worthwhile? Are neural networks like this already used? Obviously there would be less parameters to work with, but it seems like a lot of parameters are redundant anyway- being set to basically 0, so it's not a priori obvious that sparser connection would reduce performance significantly, within limits.

This is definitely a thing that people sometimes do.

It's referred to as "weight sparsity," and often it's something that people do to a network after training ("sparsification" – taking those "basically 0" parameters you mentioned and making them actually zero). But sometimes people train them from scratch this way too.

The tricky part is getting it to be significantly faster in practice. Yes, in principle you have to do fewer floating point operations because you can skip a bunch of terms with zeros in them. But to reap these gains in practice, you need to write specialized kernels, and the actual gains may depend on the amount and "structure" of the sparsity as well as the nature of the hardware you're using.

(I realize that's sort of vague – I'm not familiar in detail with sparse NN acceleration so I don't know what the biggest challenges are, just that they exist)

Anyway, some links that may interest you: here's an OpenAI blog post from way back in 2017, and here's something much more recent about sparse ViTs

[abby118]

Since this came up earlier today - if you want to take Netflix screenshots without the screen turning black:

Chrome -> Menu -> Settings -> System -> toggle "use hardware acceleration when available" -> restart your browser

[noxiatoxia]

I've been really, really wanting to make the UTDP mistranslation post I teased a week back but my brain issues have been so against me as of late. I know everything I wanna point out, but my fatigue and brain fog has made it impossible to sit down and put the words into English as of late :( mostly just sleeping.

Hmm, I can share something else with you guys, though.

So, remember this post? I've been doing a lot of research since...I just sit for 10 minutes reloading the same save file over and over and I've gotten interesting results.

Let me break it down.

NATIVE SYSTEM: The system the game is for.

MEDIA NO.: Each game for the PSP/PSVita gets assigned its own unique Media Identification Number...this number can be different from the physical box number it comes in. It's basically the ROM data number.

GAME TITLE: Self explanatory.

YEAR/PUB.COMPANY/PORT CREDIT: Year is self explanatory. The Publishing Company is who is in charge of, well, publishing the game. This is usually the company responsible for the Translations also. The Port Credit is for who was credited for making the ported versions. More on this in a second.

REGION: Region code for where the game was released. JP is Japan, US is America, EU is England, Germany, and Australia, and ASIA is Taiwan and Hong Kong.

TESTED ON: Either the physical console + model number OR emulator + update version.

DOES GLITCH?: Again pretty self explanatory. If the game experiences RAM overflow issues.

RELOADS: How many reloads of a save file it took to see results.

STAND-ALONE?: As 1・2 Reload is a compilation game, this category exists to say if the game is by itself or part of a compilation game.

So...SDR2 was released firstly on the PSP. This is the first ever release, so there's no port credit to be had. It can't be a port if it's the original.

Then, Spike releases an updated version for the Vita for both DR1 and SDR2 that fix some stuff from the PSP versions. In Japan only, this was released on the Vita as 1・2 Reload. This version only has both the Vita remakes in one game. The port was handled by Shade Inc., so it was outsourced.

Now, when it was translated for English and Traditional Chinese, the US, EU, and ASIA got both games sold separately. We do not know who the porting credit is given to, as none can be found, but it can be assumed it was Spike themselves who gave these versions of the game out, while Shade Inc. simply put them together for 1・2 Reload.

Anyways, so far...I've found something interesting. It seems this RAM overflow glitch did not exist until the Vita version was made, so it's a porting bug. However, it does NOT occur in 1・2 Reload, meaning Shade Inc. must have fixed it themselves.

All standalone versions of SDR2, however, suffer from the glitch. The US and EU versions suffer at the same rate, and for some reason the ASIA version glitches at an accelerated pace.

For all of these I did the same set up: I saved in the middle of dialogue with Komaeda in the prologue, right after you get the e-Handbook. I saved, closed the game, then would start reloading the save files.

The reason there is two numbers is because the first one (60 or 55) is when the system menu started to bug out, and then the second (62 or 57) is when the overworld started to lose textures. As you can see, they are both the same rate (took 2 more reloads between both the two) but the ASIA version just glitches faster. The results are also seemingly replicable, as the same type of glitching happens everytime without fail.

Screenshots of what I mean:

Now, you may wonder what the point of testing it on both hardware and emulator is. Well, it's to determine the cause. See, if it didn't happen on an emulator but did on console, that would point to it being a physical RAM issue, and not vRAM. However, it is happening on emulators as well to the exact same rate and in the exact same way. It's for sure a vRAM issue.

Emulators however seem to be more stable with error checks. For example, in all of these versions, once you get to this point, opening the student profiles crashes the game. However, on emulator, I could actually open it and look at them...not that there's much to look at.

What's most interesting perhaps is this is a SDR2-only issue so far. I tried this on the US Vita release of DR:THH, and did not encounter anything wrong after reloading up to 80 times. I will check the EU and ASIA versions of DR:THH another day, as well as the Steam and Switch versions of both.

I don't own the Android or iOS ports, so maybe I'll buy them for this sake.

I plan to make a video and include all of the recordings I've taken once I finish all my research. As you can see in the chart, I need to get around to certain emulator and physical tests that I just haven't yet.

Oh, and in case you want to try this yourself...it's completely safe, won't hurt your game or save data. I've saved while everything was messed up like this, and as soon as you force close the game and reopen, it's all back to normal.

[wolfpup026]

Catch up/get to know you

Thanks @cha-melodius for the tag! 💕

💫 last song:

💫 favorite color: green

💫 last film: The Big Bounce (working on some art for this now 😊)

💫 last show: The Righteous Gemstones

💫 sweet/spicy/savory: sweet!

💫 relationship: in a relationship

💫 last google search: chrome hardware acceleration (was trying to find the setting to turn that off)

~~~~~

LAST SONG I SANG OUT LOUD: I never sing out loud 🙈

FAVOURITE CRISP FLAVOUR: salt and vinegar

LAST BOOK I OPENED: do comic books count? then the TVA comic

EARBUDS, HEADPHONES, OR NOTHING: nothing if possible

LAST PLACE I WENT TO OTHER THAN HOME: the mall (to get donuts lol)

A COLOUR THAT LOOKS GOOD ON ME: black

LAST TRAILER I SAW: Thunderbolts

No pressure tags: @in-my-loki-feels, @distracteddream, @mirilyawrites, @mobius-m-mobius, @thosegayoldmen, @andthekitchensinkao3, @lgwilt, @insert-witty-user-name-here, @devilbearingtrouuble

[danrifics]

how the heck did you screen record the stream? kiswe won’t let me. i have vod access but wanted to keep a copy for myself. i’m almost ready to just ask for your link because i have no idea how you did it.

I used google chrome, so you need to go into your settings and turn off hardware acceleration and this for some reason turns off the block that stops screen recording! I then downloaded OBS and used that to do the screen recording!

[aturnoftheearth]

okay hello I keep watching your beautiful AMVs and I'm trying to make one right now and I was wondering if you'd be so kind as to offer tips on where you get your clips from? The video quality is INCREDIBLE but there are certain scenes (especially from The Man Who Would Be King) that I can only find on YouTube very grainily. Love your edits so much, thank you for making them!!!!

AKFNFKSHFKFK HIIIIII 💞 okay so! to get by the black screen on netflix, i go to my browser settings (i use firefox but i used to use chrome and it worked the same way) and search for “hardware acceleration” and make sure it’s disabled. once that’s done, it should screen record no problem!

[stevebattle]

Mars 1 (1990) by Andrew Cattell. Mars 1 competed in the 1990 UK IEE Micromouse competition in London, and again in 1992 where "Derek Hall’s Motor Mouse 2 managed a good best run but picked up some penalty points as did Andrew Cattell’s Mars 1." – UKMARS

"Mars 1 is a wheelchair design with DC (PortEscap ironless rotor) motors. Home made opto discs on the motor output shafts give 1.4mm resolution plus some backlash. Eight downlooking opto sensors are used on each side, set to allow peeking over ythe walls into adjacent cells. The inner set are used to check heading error before a turn by measuring distance travelled between 1st on sensing wall ahead and 2nd. A crude compensation is added to turn. The processor is an 8031 micro @ 6Mhz with 8K EPROM, 8k RAM, battery backed for post-run diagnostics using a BBC micro. The motor current control loop sets basic acceleration uses pwm for steering corrections. This was unstable till a disc of foam rubber, with grease on, was squeezed between gears on inner ends of the axle producing a sort of LSD (limited slip differential). All moving parts run in ball bearings, including the castors and their pivots. Power is provided by 6 Nicads of 600mAh capacity and the all-up weight is more than 2 pounds (1kg). Mars 1 took 3rd place in the London competition in 1991, but it was obvious that the resolution of travel distance and control of motors were too crude to be really competitive. A bug somewhere in the hardware was the final straw." – Andrew Cattell (via Pete Harrison)

[the-art-of-sanshoku]

Since there is more Kamen Rider on youtube now, if you stream it to your pals (especially to your non US and Canada friends who are sadly region blocked) over discord to group watch and find that they only see a black screen you can fix it by turning off "use hardware acceleration when available" in your browser settings (at least for Mozilla and Chrome)

[deramin2]

FYI, If you're watching things on streaming services (like The Legend of Vox Machina) and you can't screenshot or screenshare because it goes black, that's caused by hardware acceleration in your web browser and can be turned off.

This article details how to disable hardware acceleration in Discord and multiple web browsers. It's a very simple setting to just toggle.

Once you disable this, screenshotting will work just fine and you can have a screensharing watch party with your friends over Discord.

Also please switch to Firefox because chromium-based browsers including Chrome, Opera, Vivaldi, and Edge are getting increasingly hostile to user rights (in everything, but especially media playback). Firefox is the only browser out there focused on users over corporations.

Amazon will artificially lower the video quality to 1920 × 1080 on Firefox specifically because it blocks Amazon’s invasive video DRM. But that resolution is just fine, and fuck Amazon.

[karnpuffs]

Hellooo I love your blog! How do you get clips for your videos and such? I thought Disney+ blocked all of that kinda stuff so I've been having to just stalk YouTube and hope the clips I need have been uploaded on there. (I need to make Syril AMVs - there are not enough of them)

Aaah thank you! And you're right, there's no legal way to create clips from Disney+ content 😉

If you only want to take screenshots: Disable hardware acceleration in your browser settings (works for me, at least in Firefox)

But for entire video clips, you'll probably have to sail the high seas 🏴‍☠️

[mariacallous]

On April 15, U.S. chipmaker Nvidia published a filing to the U.S. Securities and Exchange Commission indicating that the government has restricted the company from selling its less advanced graphics processing unit (GPU)—the H20—to China. The company is now required to obtain a license from the U.S. Commerce Department’s Bureau of Industry and Security to sell the H20 and any other chips “achieving the H20’s memory bandwidth, interconnect bandwidth, or combination thereof” to China, according to the filing.

Similarly, a filing from AMD stated that the firm is now restricted from selling its MI308 GPU to China—and likely any chips that have equal or higher performance in the future. Intel’s artificial intelligence accelerator Gaudi will also be restricted under the new control threshold, which reportedly appears to limit chips with total DRAM bandwidth of 1,400 gigabytes per second or more, input/output bandwidth of 1,100 GB per second or more, or a total of both of 1,700 GB per second or more.

The possible new threshold not only restricts the advanced chips that were already controlled but also the less advanced chips from Nvidia, AMD, and other chipmakers, including Nvidia’s H20, AMD’s MI308X, and Intel’s Gaudi, which were used to comply with the export control threshold and intended primarily for sale in the Chinese market.

The new restriction came roughly a week after NPR reported that the Trump administration had decided to back off on regulating the H20. Prior to that report, curbs on the H20 and chips with comparable performance had been widely anticipated by analysts on Wall Street, industry experts in Silicon Valley, and policy circles in Washington.

The latest set of chip controls could be seen as following on from export restrictions during the Biden administration and as continuation of the Trump administration’s efforts to limit China’s access to advanced AI hardware. But the new measure carries far-reaching industry implications that could fundamentally reshape the landscape of China’s AI chip market.

The impact of the new rule on the industry is profound. With the new controls, Nvidia is estimated to immediately lose about $15 billion to $16 billion, according to a J.P. Morgan analysis. AMD, on the other hand, faces $1.5 billion to 1.8 billion in lost revenue, accounting for roughly 10 percent of its estimated data center revenue this year.

Yet the implications go beyond immediate financial damage. If the restriction persists, it will fundamentally reshape the Chinese AI chip market landscape and mark the start of a broader retreat for U.S. AI accelerators from China. That includes not only GPU manufacturers such as Nvidia, AMD, and Intel but also firms providing application-specific integrated circuits—another type of chips targeting specific AI workloads, such as Google’s TPU and Amazon Web Servies’ Trainium.

The new rule will make it nearly impossible for U.S. firms such as Nvidia and AMD to design and sell chips that are export-compliant and competitive in the Chinese market. That means these firms’ market share in the Chinese AI chip market will decline over time, as they are forced to withdraw almost all of their offerings of both advanced and less advanced chips while Chinese firms gradually capture the remaining market.

The H20 and the upgraded H20E are already only marginally ahead of their Chinese competitors. Huawei’s latest AI chip Ascend 910C delivers 2.6 times the computational performance of the H20, although it offers 20 percent less memory bandwidth, which is vital for the inference training and reasoning models that are a key part of modern AI.

The H20’s memory bandwidth, along with Nvidia’s widely adopted software stack, a parallel computing platform and programming model that enables efficient GPU utilization for AI, high-performance computing, and scientific workloads, have been key differentiators driving demand from Chinese AI firms and keeping them competitive in the Chinese market. China acquired more than 1 million units of the H20 in 2024 and has been stockpiling the chip in response to looming concerns about controls since early 2025.

The narrowing gap between the H20 and Huawei’s 910C highlights the growing ability of Chinese AI chipmakers to meet domestic compute demand without foreign GPUs. As of today, Huawei’s 910C is in mass production, with units already delivered to customers and broader mass shipments reportedly starting in May. Most recently, Huawei is reportedly approaching customers about testing its enhanced version of the 910-series GPU—the 910D. Its next-generation chip—the Ascend 920—is expected to enter mass production in the second half of 2025.

Notably, Huawei is just one of many Chinese firms poised to fill the gap left by U.S. suppliers. Chinese AI chip companies such as Cambricon, Hygon, Enflame, Iluvatar CoreX, Biren, and Moore Threads are actively developing more competitive domestic AI chips to capture this expanding market.

Over the next few years, Chinese firms such as Alibaba, ByteDance, Baidu, and Tencent will likely continue to rely on existing inventories of Nvidia and AMD chips—such as the H100, H200, H800, and H20—acquired prior to the implementation of export controls. For example, ByteDance’s current GPU inventory in China is rumored to include 16,000-17,000 units of the A100, 60,000 units of the A800, and 24,000-25,000 units of the H800. Its overseas businesses likely have more than 20,000 units of the H100, 270,000 of the H20, and tens of thousands of cards such as the L20 and L40.

Advanced chips, including the limited amount of Nvidia’s Blackwell-series GPUs, may also continue entering the Chinese market via illicit or gray-market channels, given the enduring performance advantage and wide adoption of these chips over most Chinese domestic alternatives. The Blackwell GPUs and other cutting-edge chips could still be sold legally to the oversea data centers of leading Chinese AI companies to potentially train their AI models.

Similarly, other leading Chinese AI firms still possess significant chip stockpiles. Assuming export controls continue to restrict Chinese AI companies’ access to advanced computing resources, existing GPU inventories should still enable model development over the next several years. Typically, GPUs have a four- to five-year depreciation lifecycle, providing a window during which Chinese domestic GPU manufacturers can advance their capabilities and begin supplying more competitive chips to support domestic AI development.

Ultimately, time is now on the Chinese firms’ side. As inventories of foreign GPUs gradually depreciate and become obsolete, Chinese firms are expected to shift toward and adopt more domestically produced AI chips to meet ongoing compute needs at a time when local chipmakers offer more powerful alternatives. China’s overall computing demand will steadily rise, given the continued advancement of the AI industry, and such incremental growth in demand will likely be met by Chinese AI chipmakers.

As a result, the tens of billions of dollars in revenue that would have gone to Nvidia and AMD will be gradually captured by Chinese AI firms in the coming years. In a rough assessment, the latest ban causes Nvidia and AMD instant losses of about $16.5 billion to $17.8 billion—about 70 percent of what Huawei spent on research and development in 2024.

This new market paradigm will not only strengthen the market position and financial sustainability of domestic Chinese AI chipmakers but also enhance their capacity to reinvest in R&D. In turn, this will accelerate innovation, improve competitiveness, and fortify China’s broader AI hardware supply chain—ultimately contributing to the long-term resilience and advancement of Chinese AI capabilities.

More importantly, the growing domestic adoption of Chinese GPUs enables local firms to refine their products more efficiently through accelerated and larger feedback loops from local enterprises. As the Nvidia-led GPU ecosystem stalls and gradually retreats from the Chinese market, this shift creates space for local players to build a domestic GPU ecosystem—one that may increasingly lock out foreign competitors and raise re-entry barriers over time.

A total ban on the H20 would likely slow China’s short-term growth in AI compute capacity by removing a key source of advanced chips. But the medium- to longer-term impact is less clear. Chinese AI companies, as previously noted, remain very capable of developing their AI by using a large number of existing Nvidia and AMD GPUs for the next few years, alongside a growing supply of improving domestic alternatives. The U.S. leadership’s ultimate goal of using export controls to constrain China’s AI development remains uncertain, as the gap between the two countries’ AI model capabilities appears to be narrowing rather than widening.

What is clear, however, is the broader industry impact of the new controls. If sustained, they will mark the beginning of a major withdrawal of U.S. AI chipmakers from the Chinese market—paving the way for a significant boost to domestic Chinese AI chipmakers. In trying to isolate China, the United States may end up giving Chinese firms a leg up.