Nvidia Spark

[KompuKare]

Overpriced and running about a year late, should sell well then...

/s

Thing is MediaTek usually execute quite well, so the delay wasn't at there!

Grace seems to be a bit brute force compared to Apple and Qualcomm so despite the launch, this doesn't seem to laptop-first thing after all?

 

[Grim5]

Expensive, minimum price of $2900usd for a N1X laptop

https://twitter.com/x/status/2061865456365940841

 

[AthlonXP1800]

Very interesting, a Nvidia rep confirmed Alan Wake 2 is now running native ARM version. Very impressive how well the game running on RTX Spark, it will be interesting to see how x86 and ARM versions of Alan Wake 2 compare with both x86/ARM CPU, GPU performance and benchmarks when it ship to tech reviewers and see how well it will run on Windows 11/Linux on other ARM CPUs Snapdragon X1, Snapdragon X2, Snapdragon C, Apple Mac M1 to M5/Neo. It will be interesting to see how well it can run on WINE/Proton emulation apps like GameHub and GameNative on android phones without needed to enabled FEX x86 to ARM emulation translator.

 

[AthlonXP1800]

ARM gaming isn't mature enough anyway so benefit of waiting is cheaper hardware prices (hopefully) and better software, which Steam are driving.

ARM gaming are now mature enough after years of beta tests from Linux, Apple, Android users and developers, WINE team, Valve, Nvidia and others.

WINE team announced back in 2025, WINE 10 now supported ARM64 architecture and added FEX emulation implements. In April 2026, Valve announced latest Proton 11 now official supported ARM64 architecture and Proton 11 now feature built in FEX x86 to ARM emulation layer allowing Windows games to run smoothly on ARM64 based Windows, Linux, Apple, Android or other OS and hardware.

 

[Mesai]

ARM gaming are now mature enough after years of beta tests from Linux, Apple, Android users and developers, WINE team, Valve, Nvidia and others.

WINE team announced back in 2025, WINE 10 now supported ARM64 architecture and added FEX emulation implements. In April 2026, Valve announced latest Proton 11 now official supported ARM64 architecture and Proton 11 now feature built in FEX x86 to ARM emulation layer allowing Windows games to run smoothly on ARM64 based Windows, Linux, Apple, Android or other OS and hardware.

Having the tools and it actually being usable and consistent are 2 very different things.

The app ecosystem has improved a lot especially now that Windows are pushing it but definitely not at the same level as what MacOS offers. For gaming, and a few major apps, translation is still hit or miss, which is fine if you aren't expecting a stable and consistent platform. ARM is also highly fragmented, so you're even more limited when it comes to different hardware.

 

[AthlonXP1800]

Having the tools and it actually being usable and consistent are 2 very different things.

The app ecosystem has improved a lot especially now that Windows are pushing it but definitely not at the same level as what MacOS offers. For gaming, and a few major apps, translation is still hit or miss, which is fine if you aren't expecting a stable and consistent platform. ARM is also highly fragmented, so you're even more limited when it comes to different hardware.

Well I can remembered x86 was also highly fragmented back in 1990s, my first PC CPU was Cyrix 486DX2 running at 66MHz. There was so many different hardware running Windows 95 with x86 CPUs from Intel, AMD, VIA, Cyrix, NexGen, IBM, Texas Instruments, STMicroelectronics, UMC, Centaur Technology, Rise Technology, Chips and Technologies Inc, DM&P Electronics, SiS, Transmeta and Zhaoxin. There were about 15 x86 CPUs back since 1990s was capable to running Windows 3.1/3.11/95//98/NT/Me/XP/Vista/7/8/8.1/10 so sadly most are all gone and now there are only just 4 x86 CPUs manufacturers left: Intel, AMD, VIA and Zhaoxin.

 

[AthlonXP1800]

You’re totally missing the point. The X86 AMD chip will run Windows and Linux really well, but this spark laptop will probably be a nightmare with both, as you say. So who is the market because the OEMs will need to shift these ant volume and they’re already getting stuck with dgx spark systems even after offering big discounts. Not to mention just about all the OEMs already sell X86 windows laptops with much faster graphics for a lot less.

Android and Chrome OS run on both x86 and ARM CPUs really well. Nvidia RTX Spark will run well on both Windows and Linux with 30 years worth of experience.

PC Sales Set to Tank as Apple Doubles MacBook Neo Production

New data predicts a sharp drop in PC sales in 2026, while separate reports claim Apple is doubling MacBook Neo production to meet surging demand.

games.gg

Here’s how MacBook Neo sales compare to M5 MacBook Air and Pro - 9to5Mac

A new report explores MacBook Neo sales in its first few weeks compared to the also-new M5 MacBook Air and MacBook Pro models.

9to5mac.com

There are MacBooks Apple shipped in last quarter ended March 2026 accorded to IDC:

• MacBook Neo: 1.1 million units (£599-£699)
• MacBook Air with M5: 900,000 units (£1099-£3457)
• MacBook Pro with M5 Pro/Max: 550,000 units (£1699-£7948)

A MacBook Pro with M5 Max CPU, 128GB RAM and 4TB SSD cost £6648.98 is more expensive than a DGX Spark with same configuration with 128GB RAM and 4TB SSD can be find cost less at £4299. I dont know how many people actually bought MacBook Pro M5 Max with 128GB RAM and 4TB SSD, possible around 1000 or over.

You saying OEMs already sell X86 windows laptops with much faster Nvidia graphics for a lot less?

NVIDIA N1X 48SM Specs

NVIDIA GB20B, 2346 MHz, 6144 Cores, 384 TMUs, 48 ROPs, 131072 MB LPDDR5X, 1067 MHz, 256 bit

www.techpowerup.com

NVIDIA N1X 40SM Specs

NVIDIA GB20B, 2346 MHz, 5120 Cores, 320 TMUs, 40 ROPs, 131072 MB LPDDR5X, 1067 MHz, 256 bit

www.techpowerup.com

NVIDIA N1 20SM Specs

NVIDIA GB20B, 2346 MHz, 2560 Cores, 160 TMUs, 24 ROPs, 131072 MB LPDDR5X, 1067 MHz, 256 bit

www.techpowerup.com

NVIDIA N1 16SM Specs

NVIDIA GB20B, 2346 MHz, 2048 Cores, 128 TMUs, 24 ROPs, 131072 MB LPDDR5X, 1067 MHz, 256 bit

www.techpowerup.com

N1X with 48SM feature 6144 CUDA cores and 29.71 TFLOPs FP32 is barely slower than both desktop RTX 5070's 30.87 TFLOPs FP32 and RTX 5090 Mobile's 31.80 TFLOPs FP32.
N1X with 40SM feature 5120 CUDA cores and 24.76 TFLOPs FP32 is barely faster than both desktop RTX 5060 Ti's 23.70 TFLOPs FP32 and RTX 5080 Mobile's 23.04 TFLOPs FP32.
N1 with 20SM feature 2560 CUDA cores and 12.38 TFLOPs FP32 is barely slower than both desktop RTX 5050's 13.17 TFLOPs FP32 and RTX 5070 Mobile's 13.13 TFLOPs FP32.
N1 with 16SM feature 2048 CUDA cores and 9.904 TFLOPs FP32 is barely faster than RTX 5060 Mobile's 9.684 TFLOPs FP32.

I think Nvidia will likely sell Nvidia ARM laptops and desktop at same price range as x86 like below after saw prices lists from retailers websites but guess prices are all over the places:

N1 16SM ARM laptop = x86 laptop with RTX 5060 Mobile from £899
N1 20SM ARM laptop = x86 laptop with RTX 5070 Mobile from £1099
N1X 40SM ARM laptop = x86 laptop with RTX 5080 Mobile from £2199
N1X 48SM ARM laptop = x86 laptop with RTX 5090 Mobile from £2899

N1 16SM ARM Mini PC = x86 Mini PC with RTX 5060 Mobile from £1399
N1 20SM ARM Mini PC = x86 Mini PC with RTX 5070 Mobile from £1599
N1X 40SM ARM Mini PC = x86 Mini PC with RTX 5080 Mobile from £2899
N1X 48SM ARM Mini PC = x86 Mini PC with RTX 5090 Mobile from £3899

 

[AthlonXP1800]

https://www.tomsguide.com/computing/i-spoke-to-nvidia-ceo-jensen-huang-about-rtx-spark-he-is-willing-to-work-on-an-rtx-gaming-handheld-n2x-and-n3x-are-already-planned-and-the-chip-is-more-like-r2d2-than-a-laptop-cpu

 

[jigger]

Android and Chrome OS run on both x86 and ARM CPUs really well. Nvidia RTX Spark will run well on both Windows and Linux with 30 years worth of experience.

PC Sales Set to Tank as Apple Doubles MacBook Neo Production

New data predicts a sharp drop in PC sales in 2026, while separate reports claim Apple is doubling MacBook Neo production to meet surging demand.

games.gg

Here’s how MacBook Neo sales compare to M5 MacBook Air and Pro - 9to5Mac

A new report explores MacBook Neo sales in its first few weeks compared to the also-new M5 MacBook Air and MacBook Pro models.

9to5mac.com

There are MacBooks Apple shipped in last quarter ended March 2026 accorded to IDC:

• MacBook Neo: 1.1 million units (£599-£699)
• MacBook Air with M5: 900,000 units (£1099-£3457)
• MacBook Pro with M5 Pro/Max: 550,000 units (£1699-£7948)

A MacBook Pro with M5 Max CPU, 128GB RAM and 4TB SSD cost £6648.98 is more expensive than a DGX Spark with same configuration with 128GB RAM and 4TB SSD can be find cost less at £4299. I dont know how many people actually bought MacBook Pro M5 Max with 128GB RAM and 4TB SSD, possible around 1000 or over.

You saying OEMs already sell X86 windows laptops with much faster Nvidia graphics for a lot less?

NVIDIA N1X 48SM Specs

NVIDIA GB20B, 2346 MHz, 6144 Cores, 384 TMUs, 48 ROPs, 131072 MB LPDDR5X, 1067 MHz, 256 bit

www.techpowerup.com

NVIDIA N1X 40SM Specs

NVIDIA GB20B, 2346 MHz, 5120 Cores, 320 TMUs, 40 ROPs, 131072 MB LPDDR5X, 1067 MHz, 256 bit

www.techpowerup.com

NVIDIA N1 20SM Specs

NVIDIA GB20B, 2346 MHz, 2560 Cores, 160 TMUs, 24 ROPs, 131072 MB LPDDR5X, 1067 MHz, 256 bit

www.techpowerup.com

NVIDIA N1 16SM Specs

NVIDIA GB20B, 2346 MHz, 2048 Cores, 128 TMUs, 24 ROPs, 131072 MB LPDDR5X, 1067 MHz, 256 bit

www.techpowerup.com

N1X with 48SM feature 6144 CUDA cores and 29.71 TFLOPs FP32 is barely slower than both desktop RTX 5070's 30.87 TFLOPs FP32 and RTX 5090 Mobile's 31.80 TFLOPs FP32.
N1X with 40SM feature 5120 CUDA cores and 24.76 TFLOPs FP32 is barely faster than both desktop RTX 5060 Ti's 23.70 TFLOPs FP32 and RTX 5080 Mobile's 23.04 TFLOPs FP32.
N1 with 20SM feature 2560 CUDA cores and 12.38 TFLOPs FP32 is barely slower than both desktop RTX 5050's 13.17 TFLOPs FP32 and RTX 5070 Mobile's 13.13 TFLOPs FP32.
N1 with 16SM feature 2048 CUDA cores and 9.904 TFLOPs FP32 is barely faster than RTX 5060 Mobile's 9.684 TFLOPs FP32.

I think Nvidia will likely sell Nvidia ARM laptops and desktop at same price range as x86 like below after saw prices lists from retailers websites but guess prices are all over the places:

N1 16SM ARM laptop = x86 laptop with RTX 5060 Mobile from £899
N1 20SM ARM laptop = x86 laptop with RTX 5070 Mobile from £1099
N1X 40SM ARM laptop = x86 laptop with RTX 5080 Mobile from £2199
N1X 48SM ARM laptop = x86 laptop with RTX 5090 Mobile from £2899

N1 16SM ARM Mini PC = x86 Mini PC with RTX 5060 Mobile from £1399
N1 20SM ARM Mini PC = x86 Mini PC with RTX 5070 Mobile from £1599
N1X 40SM ARM Mini PC = x86 Mini PC with RTX 5080 Mobile from £2899
N1X 48SM ARM Mini PC = x86 Mini PC with RTX 5090 Mobile from £3899

But what is the point for this kind of money? Both are fringe markets already packed with vendors.

 

[TrojanWhore]

But what is the point for this kind of money? Both are fringe markets already packed with vendors.

Yeah, I'd be somewhat interested in a high performance laptop or handheld that could manage on battery for a decent amount of time.
I'd also be interested in a laptop or handheld that would serve (on a dock) as a more portable gaming setup.
Strix Halo is impressive, but it's more than I'd want to spend as I could build a faster ITX setup for the same money.
I have a handheld which can play older games well, especially when combined with a monitor, KB and mouse, (Antec Core HS - Legion Z1E equivalent) and was £550, it's a big gap from that to £2-3k

 

[Blackjack Davy]

But what is the point for this kind of money? Both are fringe markets already packed with vendors.

People havn't been paying attention to what Jensen has been gushing lately he believes the end of personal computing is nigh, everyone will be using "agents" whatever that means, AI based cloud computing where your laptop is merely an access point via whatever AI blob they want to run locally on your hardware. Either he's increasingly detatched from reality and the lives of ordinary people like the tech bros and other Trump allied oligarchs appear to increasingly be or he's a visionary genius, you decide.

 

[smilingcrow]

People havn't been paying attention to what Jensen has been gushing lately he believes the end of personal computing is nigh, everyone will be using "agents" whatever that means, AI based cloud computing where your laptop is merely an access point via whatever AI blob they want to run locally on your hardware. Either he's increasingly detatched from reality and the lives of ordinary people like the tech bros and other Trump allied oligarchs appear to increasingly be or he's a visionary genius, you decide.

Exactly, it was clear what the pitch is here.
The only differentiator is the CUDA cores alongside the shared memory with relatively high bandwidth and the AI software stack.
Unless people actually want to run local agentic AI, this appears to be a niche platform.

It feels like they are building expensive shovels now in case everyday people might decide to prospect for gold in their gardens in the future.
I'd wait to find out if there is any gold before buying a shovel.
A cheap survey of the scene makes sense before buying expensive hardware.

The difference between Jensen and most other people is that he has a better idea in advance on how the software is developing in this area.
He's also invested in local agentic AI actually being useful whereas for most consumers it's an irrelevance.

This may well be his Apple Vision Pro equivalent.
An expensive and cumbersome technology that most people aren't interested in.
If you assign an image to your local AI agentic Avatar, maybe Tim Cook wearing a Vision Pro headset is the perfect fit.
Ideally wearing a MAGA hat and an orange prison issue jumpsuit emblazoned on the chest with the image of a unicorn in flames.
Clippy 2.0 for this dystopia.

 

[AthlonXP1800]

Hands On With NVIDIA's First RTX Spark Laptops & PCs Ft. ASUS, Dell, HP, Microsoft, MSI, Lenovo

Here's our hands-on with several RTX Spark systems, including Laptops & Mini PC, giving us a first taste of NVIDIA's first SoC for Windows.

wccftech.com

ASUS ProArt RTX Spark 16 inch laptop has really very impressive 12.9mm height, it is much thinner than 16 inch Apple M5 MacBook Pro's 16.8mm height.

MSI Prestige N16 Flip AI+ 16 inch laptop featured N1X 48SM probably has around 13.9mm height accorded to MSI product page if it use same chassis used for Panther Lake. The laptop dual fans operating under 18dB idle and under 29dB at full load! That is amazing! When Nvidia launched Blackwell GPUs back in Jan 2025, I wanted a 17 inch laptop with Blackwell RTX 5070 GPU but was very disappointed read many reviews confirmed the laptops fans was running 50-52dB at full load in tests and battery only lasted 5 hours browsed internet. That was too loud, far too louder than my MSI RTX 5080 SUPRIM SOC triple fans so now it look like RTX Spark laptop is finally the perfect one I am look for over a year now.

 

[AthlonXP1800]

People havn't been paying attention to what Jensen has been gushing lately he believes the end of personal computing is nigh, everyone will be using "agents" whatever that means, AI based cloud computing where your laptop is merely an access point via whatever AI blob they want to run locally on your hardware. Either he's increasingly detatched from reality and the lives of ordinary people like the tech bros and other Trump allied oligarchs appear to increasingly be or he's a visionary genius, you decide.

I think Jensen is a visionary genius.

https://venturebeat.com/ai/nvidia-became-a-1-trillion-company-thanks-to-ai-look-inside-their-lavish-star-trek-inspired-hq-the-ai-beat

I think this article gave us a clue that Jensen Haung is massive fan of Star Trek and Star Wars, he named his Nvidia headquarters after USS Voyager and USS Endeavor and also he named rooms after Star Trek places, alien races, planets and starships. He is now closer to achieve his vision to recreate Star Trek computer with very powerful local AI agents in real life and think he is now closer to achieve his vision to recreate Star Wars AI robots R2-D2 and C-3PO but no laser gun, shield, forcefield, lightsaber and break speed of light yet.

 

[jigger]

People havn't been paying attention to what Jensen has been gushing lately he believes the end of personal computing is nigh, everyone will be using "agents" whatever that means, AI based cloud computing where your laptop is merely an access point via whatever AI blob they want to run locally on your hardware. Either he's increasingly detatched from reality and the lives of ordinary people like the tech bros and other Trump allied oligarchs appear to increasingly be or he's a visionary genius, you decide.

Sounds little like communism to me, but I suppose if these “agents” can afford to pay for my hardware I’ll give it a whirl.

I’m not sure what I just watched TBH, but it seem Nvidia is running out of customers and Jenson is attempting to invent some more…

 

[AthlonXP1800]

Really very interesting thing is Nvidia told Leo of KitGuru that the PRISM emulation on Windows 11 26H1 will have no performance hit on N1/N1X ARM CPUs. That is incredible.

Nvidia N1/N1X CPUs will put Qualcomm ARM CPUs to shame. Their old Snapdragon X1 CPUs has 10 to 20% performance hit on Windows 11 25H2 PRISM emulation and new X2 CPUs now has 5% to 10% performance hit on Windows 11 26H1 PRISM emulation.

I can see Snapdragon X1 and X2 owners will switch to superior Nvidia N1 and N1X and Nvidia will command 95%+ ARM CPU marketshare.

 

[AthlonXP1800]

RTX Spark Is Apple’s And Qualcomm’s Kryptonite In One Crucial Area And These Two Only Have A Year To Get Things Right

Apple and Qualcomm may tower over the RTX Spark in more ways than one, but that difference could be nullified if they don’t act quick

wccftech.com

 

[philo-sofa]

Take my desktop, or make it 'non-user serviceable'?

From my cold. Dead. Nerdy. Hands.

 

[img]

RTX Spark Is Apple’s And Qualcomm’s Kryptonite In One Crucial Area And These Two Only Have A Year To Get Things Right

Apple and Qualcomm may tower over the RTX Spark in more ways than one, but that difference could be nullified if they don’t act quick

wccftech.com

WHen one of your points is now you can get 6x the fake frames so performance is good i will wait and see what the real figures are.