• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

[SemiAccurate] upcoming Nvidia chip has a problem that doesn't have a fix as of now

John Carmack says NVIDIA DGX Spark runs at half of the rated power and delivers half the quoted performance​

could be doing things that get memory bandwith starved or a bit underpowered.

That 240 watt is the USB adapter from the wall limit and the networking would be a bit power hungry if used, 200w is probably more the limit if not used and some people do reach it, if you only do inference with it, would often be under 100watt (not a good inference machine per say)
 
could be doing things that get memory bandwith starved or a bit underpowered.

That 240 watt is the USB adapter from the wall limit and the networking would be a bit power hungry if used, 200w is probably more the limit if not used and some people do reach it, if you only do inference with it, would often be under 100watt (not a good inference machine per say)
Wasn't just power limited but according to Carmack, "It gets quite hot even at this level, and I saw a report of spontaneous rebooting on a long run, so was it de-rated before launch?" It's obviously not just a power delivery problem. This would also explain why this was delayed, and obviously not a Windows problem.
nvidia arm hot.jpg
 
Or just a very aggressive form factor here (made to run at 90c), if you use it under heavy load that can destroy the whole point of it (on the desk usage) as it get noisy.

Not sure if they didn't went both too mainstream with it and not enough at the same time (price point not a cheaper without super fast networking option, not good for inference what all the influencer do has virtually none will use it for what it is for,....), if it is a raw dev kit a more like Jetson Launch that do not reach mainstream media... case better at cooling that look worse and so on.

Trying to create a new product class optimised for no one, not sure if it will work
 
Benchmark vs real life performance

https://x.com/ID_AA_Carmack/status/1982831774850748825

John Carmack says NVIDIA DGX Spark runs at half of the rated power and delivers half the quoted performance​


https://videocardz.com/newz/john-ca...ower-and-delivers-half-the-quoted-performance

“AMD swoops in to help as John Carmack slams Nvidia's $4,000 DGX Spark, says it doesn't hit performance claims, overheats, and maxes out at 100W power draw — developer forums inundated with crashing and shutdown reports​

News
By Luke James published 2 hours ago
The $4,000 Grace Blackwell dev kit is rated for 240W and 1 PF of sparse FP4 compute, but early users report 100W ceilings and reboot issues under sustained load.”

https://www.tomshardware.com/tech-industry/semiconductors/users-question-dgx-spark-performance
 
So the spec sheet says it’s a 140w SoC, 100w to the chip and 40 for the rest of the package seems OK.
 
Microsoft looks to be preparing a new version of Windows 11 for release in the first half of 2026, based on a newer version of the Windows platform codenamed Bromine.

The upcoming Qualcomm Snapdragon X2-based SoC and NVIDIA N1X platform requires platform changes that aren't currently available in version 25H2, and the work required to light up Snapdragon X2 and N1X devices appears to be part of the next Windows platform release, codenamed Bromine.


Unlike 25H2, which is based on the same platform release as version 24H2 (codenamed Germanium,) the next version of Windows 11 looks to be based on the newer Bromine platform release.

Microsoft has confirmed that version 26H1 won't ship as a feature update for version 25H2 users, meaning it won't reach general availability. It's possible, and likely, that version 26H1 will only be available on PCs powered by this next-gen Arm silicon from Qualcomm and NVIDIA.

This isn't the first time Qualcomm's next gen chips have required Microsoft to push forward a new version of the Windows platform. In 2024, Microsoft launched version 24H2 on the Germanium platform release, months before it was ready to roll out to all Windows 11 PCs. This is because Snapdragon X was launching in June, and required Germanium to function.

It seems like the same thing may be happening here. Bromine is launching early for Snapdragon X2, with general availability for everyone else happening sometime after.

https://www.windowscentral.com/microsoft/windows-11/windows-11-version-26h1-faq
 
Microsoft has confirmed that version 26H1 won't ship as a feature update for version 25H2 users, meaning it won't reach general availability. It's possible, and likely, that version 26H1 will only be available on PCs powered by this next-gen Arm silicon from Qualcomm and NVIDIA.
So if I get this right then that means the next update is not only not available for previous generation Snapdragon chips but also not available for x86 machines? H1 for ARM and H2 for x86? Microsoft is very much showing their favoritism here. The best thing this could do is make older printer drivers actually work on ARM.
 
So if I get this right then that means the next update is not only not available for previous generation Snapdragon chips but also not available for x86 machines? H1 for ARM and H2 for x86? Microsoft is very much showing their favoritism here. The best thing this could do is make older printer drivers actually work on ARM.
Fat chance…. Bet they use it as a pitch to double down on Microsoft Universal Print Service.

The current Win11 ARM build was so tuned for the X1 that elsewhere in the forum I’ve joked that only a full rewrite would make it remotely usable for another platform.
I guess the Microsoft engineers agreed.
 
Fat chance…. Bet they use it as a pitch to double down on Microsoft Universal Print Service.
That's a thing? Holy shit they did. They created a problem and then solved it.
The current Win11 ARM build was so tuned for the X1 that elsewhere in the forum I’ve joked that only a full rewrite would make it remotely usable for another platform.
I guess the Microsoft engineers agreed.
Does this mean that Windows needs a complete rewrite for each new ARM CPU ever made? That can't be right.
 
Does this mean that Windows needs a complete rewrite for each new ARM CPU ever made? That can't be right.
Well lets be honest it NEEDS one for every new x86 chip as well... they just don't bother. I mean core parking? Intel having to include a "thread director" on their chip packages.

Microsoft just sucks. Yes windows essentially needs a re write to function properly with every new CPU.
 
That's a thing? Holy shit they did. They created a problem and then solved it.
The funny thing is, it’s actually a good print server, kind of like Papercut lite. But it ties in with O365 really well and since the account itself has the printer if they log in to any authorized device their authorized printers follow them, kind of nice actually.
Does this mean that Windows needs a complete rewrite for each new ARM CPU ever made? That can't be right.
Not exactly, but every aspect of the current iteration of the ARM build is tuned exclusively for the X1 platform. Its audio backend supports it and only it, as does the video components, networking, storage interfaces. Etc.
But Qualcomms drivers and documentation is utter crap, and the OS has had everything stripped out of it that isn’t exclusively for the Qualcomm chip and even then it barely supports that.
Every aspect of the OS shows they gave it a lot of thought and really knew what needed to be optimized to make the most of the X1 platform. Then had it developed by the back benchers and their most disposable interns. Because ???? Secrecy??
 
Well lets be honest it NEEDS one for every new x86 chip as well... they just don't bother. I mean core parking? Intel having to include a "thread director" on their chip packages.

Microsoft just sucks. Yes windows essentially needs a re write to function properly with every new CPU.
It does need a rewrite for x86 because look at the performance differences against Linux. This might turn Windows into a very Android like OS in that each version needs to be built specifically for each new hardware.
The funny thing is, it’s actually a good print server, kind of like Papercut lite. But it ties in with O365 really well and since the account itself has the printer if they log in to any authorized device their authorized printers follow them, kind of nice actually.
It's just a DNS server for network printers. Would mean the printer server is still x86.
it's only a printer model.gif

Not exactly, but every aspect of the current iteration of the ARM build is tuned exclusively for the X1 platform. Its audio backend supports it and only it, as does the video components, networking, storage interfaces. Etc.
But Qualcomms drivers and documentation is utter crap, and the OS has had everything stripped out of it that isn’t exclusively for the Qualcomm chip and even then it barely supports that.
Every aspect of the OS shows they gave it a lot of thought and really knew what needed to be optimized to make the most of the X1 platform. Then had it developed by the back benchers and their most disposable interns. Because ???? Secrecy??
So this means Nvidia's chip would need their own optimized version then? It also sounds like X1 early adopters will eventually get screwed out of updates in the future.
 
It does need a rewrite for x86 because look at the performance differences against Linux. This might turn Windows into a very Android like OS in that each version needs to be built specifically for each new hardware.
Its far worse then android. It's windows.
Barely working, cobbled together, good enough junk for the masses.

The last time windows worked as intended was July 1993.
Since then every version of windows has had at least some bit that hasn't really be built to run on the hardware its on.
From that point they added 64 bit, in a hacky way.
Later they added threading. Also in a hacky way.
They have left the file system almost untouched since 1993.

The initial launch of NT I would say that was the last time windows was a cohesive operating system. It has been hacked and poked by multiple MS divisions ever since. Having bits added but in ways to not touch other systems form other divisions. It has had things bolted on. A few times MS has tried to clean up X or Y but generally always broke Z and back tracked or Z got replaced by yet another new bit. All without removing the old bits. How many control systems does windows have these days? (I honestly don't know haven't spent more then 20 min on a windows machine in a few years)

The biggest problem for MS. Is my pithy quip. IS actually the 100% truth. Windows has never been properly built to properly work to the fullest on ANY system since 32 bit hardware 32 years ago. Intel stopped even trying and took most of MS agency away away from controlling their consumer hardware. I mean Intel chips have hardware thread directors on CONSUMER products only. Intel server chips do not have any such thing... cause Intel server chips are designed for and intended to be used on Linux at this point. MS has NEVER had any version of windows properly use a AMD Ryzen of any flavor... and anytime they have tried to push an update to address one issue, NEW hardware is on the market that has new issues. Core parking is a stupid idea, that only exists because MS NEEDS to re built major core parts of windows to properly schedule modern CPUs. Windows can't really properly deal with multi core CPUs. Intel removed the ability of windows to even try... and AMD just couldn't be bothered to do some sort of consumer only version of Ryzen that does the same.

Anyway its the same across every core windows system. Non of it is unified, nor properly designed to work with modern hardware.

As stupid as it is. The hacked together clipped and apparently according to Lakados very X1 only Windows ARM. Is probably the most Unified version of windows since 1993. Which is hilarious as MS needs it to work with new/other companies hardware. lol
 
So this means Nvidia's chip would need their own optimized version then? It also sounds like X1 early adopters will eventually get screwed out of updates in the future.
No "will" about it, already happening.
It launched in a usable state with promises of improvements to come. The improvements happened, then, as new features were added, everything went to shit because of how "optimized" the OS is.
So the new stuff works like garbage because of either Qualcomm drivers or some subsystem being so optimized for options A, B, and C that the addition of D and E causes everything to melt down.

Accounting wouldn't let me replace our X1-based Surface Pros with other units because they were "too new", so we've been downcycling them anywhere we can; the Win10 EoL is helping greatly with that.
They can go into the Janetorial closets to be used for checking email, Time Sheets, or supply orders to replace the Win10 machines that were there before.

However, as it currently stands, Nvidia would need its own "optimized" version, because, unlike the x86 version of Windows, which is generic, the ARM version is as specific as possible.

The x86 equivalent would be Microsoft releasing a version that only worked on Intel, then further making it only work on Intel chips of a specific generation, then limiting it to only 1 of the chipsets for that generation of CPU, and only 1 specific network card and audio chipset, and you can only use the iGPU because they pulled all support for dGPU's and PCIe cards because they didn't feel like dealing with a UEFI that supported them.
 
Last edited:
So if I get this right then that means the next update is not only not available for previous generation Snapdragon chips but also not available for x86 machines? H1 for ARM and H2 for x86? Microsoft is very much showing their favoritism here. The best thing this could do is make older printer drivers actually work on ARM.

More that an out of band update for a few arm chips is happening half a year before the next mainline update because they need more than just a driver package to work (or the drivers were targeted against something that was supposed to ship in 25H2 but ran into schedule problems and had to be pulled for next year).
 
good news. steam FEX tested on DGX Spark

Canonical Builds Steam Snap For Ubuntu ARM64 Leveraging FEX​

Written by Michael Larabel in Ubuntu on 8 January

details on the Steam Snap for ARM64 via the Ubuntu Discourse.

https://www.phoronix.com/news/Steam-Snap-ARM64-FEX

Game Performance​

My team and I have tested the following games on our DGX Sparks. (Note that this is not intended to be comprehensive; rather, it is just a quick a look into the types of games we’ve had success with so far)

GameNative PlatformPerformanceDevice
Cyberpunk 2077Windows x86 (via Proton)200+ FPS w/ DLSSNVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
Counter-Strike 2Linux x86smooth, no notable performance issues (2+ hour session, mostly arms race)NVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
Marvel Cosmic InvasionLinux x86no notable performance issuesNVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
Lonely Mountains: Snow RidersWindows x86 (via Proton)smooth, no notable performance issues. (2+ hour session)NVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
DOTA 2Linux x86smooth, no notable performance issuesNVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
Outer WildsWindows x86 (via ProtonGE-10-17)smooth, no notable performance issues (~20 minute test)NVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
Half-Life 2Linux x86smooth, no notable performance issuesNVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
Portal 2Linux x86smooth, no notable performance issuesNVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
Garry’s ModLinux x86smooth, no notable performance issuesNVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
Hollow Knight SilksongWindows x86 (via Proton)Very smooth (~10 minute test)NVIDIA DGX Spark w/ Ubuntu Questing
Clair Obscur Expedition 33Windows x86 (via Proton)Playable (~10 minute test)NVIDIA DGX Spark w/ Ubuntu Questing
Golf with your FriendsLinux x86100+ FPS, smooth, no notable performance issuesNVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
PEAKWindows x86 (via ProtonGE-10-17)Was previously unstable, but is now stable across 3x ~10 minute tests after a recent game update when testing on 1/7/26. (but have not yet tested with multiple players).NVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
Disclaimer: This snap is experimental, comes without warranty or official support, and is provided on a best-effort basis
 
good news. steam FEX tested on DGX Spark

Canonical Builds Steam Snap For Ubuntu ARM64 Leveraging FEX​

Written by Michael Larabel in Ubuntu on 8 January

details on the Steam Snap for ARM64 via the Ubuntu Discourse.

https://www.phoronix.com/news/Steam-Snap-ARM64-FEX

Game Performance​

My team and I have tested the following games on our DGX Sparks. (Note that this is not intended to be comprehensive; rather, it is just a quick a look into the types of games we’ve had success with so far)

GameNative PlatformPerformanceDevice
Cyberpunk 2077Windows x86 (via Proton)200+ FPS w/ DLSSNVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
Counter-Strike 2Linux x86smooth, no notable performance issues (2+ hour session, mostly arms race)NVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
Marvel Cosmic InvasionLinux x86no notable performance issuesNVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
Lonely Mountains: Snow RidersWindows x86 (via Proton)smooth, no notable performance issues. (2+ hour session)NVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
DOTA 2Linux x86smooth, no notable performance issuesNVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
Outer WildsWindows x86 (via ProtonGE-10-17)smooth, no notable performance issues (~20 minute test)NVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
Half-Life 2Linux x86smooth, no notable performance issuesNVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
Portal 2Linux x86smooth, no notable performance issuesNVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
Garry’s ModLinux x86smooth, no notable performance issuesNVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
Hollow Knight SilksongWindows x86 (via Proton)Very smooth (~10 minute test)NVIDIA DGX Spark w/ Ubuntu Questing
Clair Obscur Expedition 33Windows x86 (via Proton)Playable (~10 minute test)NVIDIA DGX Spark w/ Ubuntu Questing
Golf with your FriendsLinux x86100+ FPS, smooth, no notable performance issuesNVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
PEAKWindows x86 (via ProtonGE-10-17)Was previously unstable, but is now stable across 3x ~10 minute tests after a recent game update when testing on 1/7/26. (but have not yet tested with multiple players).NVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)
Disclaimer: This snap is experimental, comes without warranty or official support, and is provided on a best-effort basis

Well one + for a lot of people.
AI has at least saved us from Nvidia-Nvidia laptops.... for now.
 
GameNative PlatformPerformanceDevice
Cyberpunk 2077Windows x86 (via Proton)200+ FPS w/ DLSSNVIDIA DGX Spark w/ DGX OS (Ubuntu Noble)

Let me guess. 4X frame generation? There's no way that pile of crap could run an x86 game at 200fps without some trickery. Especially when every other game listed is from ancient history, with the exception of Clair Obscur Expedition 33 which is listed as "Playable". Brings up a lot of questions.
 
Let me guess. 4X frame generation? There's no way that pile of crap could run an x86 game at 200fps without some trickery. Especially when every other game listed is from ancient history, with the exception of Clair Obscur Expedition 33 which is listed as "Playable". Brings up a lot of questions.

The CPU is a minor part of what makes games go anymore.

I mean we are at a point where your standard budget 6 core CPU, games exactly as well as the GPU its attached to can manage. (insane data center sized x90 class cards withstanding) I mean how much difference is their really between something like a 6 core Ryzen and a 16 core Ryzen for gaming? X3D aside sure games like cache. The gaming use case for high end CPUs has been dead for a few years now.

ARM systems are going to game decently well, assuming the translation software isn't having to do a lot of extra steps.
Also lets be honest, average gamer is now EXPECTING and basing buying decisions on software trickery.

AI sales have clearly put a pause on Nvidia going after consumer anything. Really though had their N1X stuff still been on the table. They would show it off using the new DLSS 4.5 with native FP4... and you know as well as I do it would sell like hot cakes.
 
The CPU is a minor part of what makes games go anymore.
No, it really isn't. Games need a good bit of CPU power, depending on the game and newer ones are often hitting it harder. Starfield would be an example of one known to hit the CPU particularly hard. In particular where you see the CPU being a real bottleneck is high frame rates. Hardware Unboxed has done some good tests running games on a 4090 or 5090 in 1080p to CPU bottleneck the game. You find that there's only a few rare simplistic games, like Counterstrike 2, that can really push the FPS up to 500+. There's some games that push up into the low 200s on the fastest CPUs, and plenty that cap out in the 100s.

Cyberpunk, for example, hits 219fps in 1080p on High, meaning no RT, but only on a 9800X3D. Go down to a 12900k and it is capped at 153fps. Or in Starfield they had the 9800X3D leading the chard, way out in front, at 140fps, on the other end a 5800X3D was only 93fps. You aren't getting those high FPS numbers, even at a low resolution, without a good CPU or framegen.

Heck it can matter to non-intense games too. I'm playing Pillars of Eternity 2 right now and it is heavily bottlenecked by single core CPU performance. In the big city I'm seeing framerates of 70s or even lower sometimes, yet my 4090 is running at about 40%, despite maxed graphics settings. Reason is that there are basically 2 CPU cores getting hit hard, and the rest loaded fairly lightly (Afterburner calls it a 25% load overall). Now you can whine about optimization and lazy devs all you like, point is that it is this way and if you have a CPU with less single thread performance, you'll get even lower FPS no matter how beefy the GPU is.


Basically what Dukenukem is saying is right that this is a very suspect claim. "Oh Cyberpunk runs at 200fps!" Really? What settings? To me that sounds like a fanboy trying to make this out to be more of a powerhouse than it really is.
 
No, it really isn't. Games need a good bit of CPU power, depending on the game and newer ones are often hitting it harder. Starfield would be an example of one known to hit the CPU particularly hard. In particular where you see the CPU being a real bottleneck is high frame rates. Hardware Unboxed has done some good tests running games on a 4090 or 5090 in 1080p to CPU bottleneck the game. You find that there's only a few rare simplistic games, like Counterstrike 2, that can really push the FPS up to 500+. There's some games that push up into the low 200s on the fastest CPUs, and plenty that cap out in the 100s.

Cyberpunk, for example, hits 219fps in 1080p on High, meaning no RT, but only on a 9800X3D. Go down to a 12900k and it is capped at 153fps. Or in Starfield they had the 9800X3D leading the chard, way out in front, at 140fps, on the other end a 5800X3D was only 93fps. You aren't getting those high FPS numbers, even at a low resolution, without a good CPU or framegen.

Heck it can matter to non-intense games too. I'm playing Pillars of Eternity 2 right now and it is heavily bottlenecked by single core CPU performance. In the big city I'm seeing framerates of 70s or even lower sometimes, yet my 4090 is running at about 40%, despite maxed graphics settings. Reason is that there are basically 2 CPU cores getting hit hard, and the rest loaded fairly lightly (Afterburner calls it a 25% load overall). Now you can whine about optimization and lazy devs all you like, point is that it is this way and if you have a CPU with less single thread performance, you'll get even lower FPS no matter how beefy the GPU is.


Basically what Dukenukem is saying is right that this is a very suspect claim. "Oh Cyberpunk runs at 200fps!" Really? What settings? To me that sounds like a fanboy trying to make this out to be more of a powerhouse than it really is.

You are sort of proving my point though.
Your examples of "Bad" CPU performance is 150+FPS. :) I mean I know Intels latest CPUs are not high end gaming chips... but really how many people if you sat them down in front of unlabeled machines to AB as 12900k vs a 9800X3D is really going to see the difference? Better yet forget the X3D outlier parts... and compare a 12900K vs a 9700X. The 9700x is probably going to to split the difference at like 175fps or something. The number of people that would correctly pick out the extra 20-30fps isn't going to beat chance.

Chips like Nvidias digits/DX1 would compare quite well vs CPUs like the 5600x... sure there are faster x86 chips. But faster isn't that important. Fast enough is all that matters. No one is going to pair an ARM CPU with a 5090 anytime soon. However if we are talking about ARM hardware to compete with things like the Steam Machine? I mean come on current ARM can handle that level of performance just fine.

As I said x3d withstanding... cause games love cache. If we are comparing normally cached CPUs... you would have fun trying to buy a CPU that can't game just fine, even if you have a budget of a hundred bucks.

Sure a couple poorly optimized or CPU heavy games exist. They are the outliers though not the majority or the norm. The average PC game right now runs just fine on a 5 year old 6 core CPU. The only CPUs that stand out as really delivering extra frames are the X3D chips. Remove those from the math and CPUs are just not that important. We tell people to buy their place holder 6 core 7600 type CPUs all the time cause really 150FPS or 190FPS? I'm sorry most people just don't care. Of course WE Care we are on [H] We care. Average gamer doesn't care, average gamer doesn't even have a monitor that is higher then 165hz anyway. More like 120fps if they bought it a few years ago.
 
You are sort of proving my point though.
Your examples of "Bad" CPU performance is 150+FPS. :) I mean I know Intels latest CPUs are not high end gaming chips... but really how many people if you sat them down in front of unlabeled machines to AB as 12900k vs a 9800X3D is really going to see the difference? Better yet forget the X3D outlier parts... and compare a 12900K vs a 9700X. The 9700x is probably going to to split the difference at like 175fps or something. The number of people that would correctly pick out the extra 20-30fps isn't going to beat chance.

As I said x3d withstanding... cause games love cache. If we are comparing normally cached CPUs... you would have fun trying to buy a CPU that can't game just fine, even if you have a budget of a hundred bucks.

Sure a couple poorly optimized or CPU heavy games exist. They are the outliers though not the majority or the norm. The average PC game right now runs just fine on a 5 year old 6 core CPU. The only CPUs that stand out as really delivering extra frames are the X3D chips. Remove those from the math and CPUs are just not that important. We tell people to buy their place holder 6 core 7600 type CPUs all the time cause really 150FPS or 190FPS? I'm sorry most people just don't care. Of course WE Care we are on [H] We care. Average gamer doesn't care, average gamer doesn't even have a monitor that is higher then 165hz anyway. More like 120fps if they bought it a few years ago.
I think someone who drops $4000 on something to be used as a game machine might care. I realize that's not what it was made for, but that seems to be what we are talking about here and the results are very hand wavy and don't seem to really say how it performs. How does the CPU do on games that use it heavily? I do not believe that it gets 200+fps in Cyberpunk, something that only the highest of the high-end gaming CPUs manage.

Also this all gets WAY WORSE when you start talking raytracing, which hits the CPU hard. I'm not saying that you need a bigass CPU for good gaming, I'm saying that the CPU is not irrelevant. Goes double if you go look at their chart and look at 1% lows. If you are running an older 120Hz non-gsync monitor then those particularly matter as any time you drop below 120 there's going to be a hitch. So if you take something like Hogwarts Legacy, one of the worst offenders, a 9800X3D would just about get you a stable 120fps, it gets 111 for it's 1% lows so you are going to see some drops below 120 but not many. However a 7700X, which can do over 120fps, has 74 1% lows, it is going to hitch like a fucker.

Hence the CPU performance of something like this IS relevant, for this use case at least. You can't just throw a beefy GPU at games and say "they'll be fine." No, they won't. Suppose that it's performance running x86 code is not in line with any of these CPUs that HUB is testing and is more in line with something like an 8700k. Still respectable but you'd better believe that'll be limiting your performance in a non-trivial number of cases.
 
I think someone who drops $4000 on something to be used as a game machine might care. I realize that's not what it was made for, but that seems to be what we are talking about here and the results are very hand wavy and don't seem to really say how it performs. How does the CPU do on games that use it heavily? I do not believe that it gets 200+fps in Cyberpunk, something that only the highest of the high-end gaming CPUs manage.

Also this all gets WAY WORSE when you start talking raytracing, which hits the CPU hard. I'm not saying that you need a bigass CPU for good gaming, I'm saying that the CPU is not irrelevant. Goes double if you go look at their chart and look at 1% lows. If you are running an older 120Hz non-gsync monitor then those particularly matter as any time you drop below 120 there's going to be a hitch. So if you take something like Hogwarts Legacy, one of the worst offenders, a 9800X3D would just about get you a stable 120fps, it gets 111 for it's 1% lows so you are going to see some drops below 120 but not many. However a 7700X, which can do over 120fps, has 74 1% lows, it is going to hitch like a fucker.

Hence the CPU performance of something like this IS relevant, for this use case at least. You can't just throw a beefy GPU at games and say "they'll be fine." No, they won't. Suppose that it's performance running x86 code is not in line with any of these CPUs that HUB is testing and is more in line with something like an 8700k. Still respectable but you'd better believe that'll be limiting your performance in a non-trivial number of cases.

Sure no one is buying a digits machine for gaming. :) That whole thing is just an aside on that. Just like no one is really buying the highest end Ryzen Max APUs to game either.

Nvidia has been working with OEMS on consumer N1X systems. It's the same silicon but NOT a $3000 AI setup. Depending what rumors you want to believe... either Nvidia has had issues with the hardware, Microsoft sucks balls so hard they can't make Windows ARM run properly on them.... or the most likely actual reason Nvidia doesn't want to launch anything that eats into precious RAM supply. (or cut into Spark sales)

https://videocardz.com/newz/dell-tested-nvidia-n1x-laptop-just-two-months-ago
By all accounts major OEMs have working hardware??? Who knows if it ever gets released.
 
Nvidia has been working with OEMS on consumer N1X systems. It's the same silicon but NOT a $3000 AI setup. Depending what rumors you want to believe... either Nvidia has had issues with the hardware, Microsoft sucks balls so hard they can't make Windows ARM run properly on them.... or the most likely actual reason Nvidia doesn't want to launch anything that eats into precious RAM supply. (or cut into Spark sales)
Or alternatively: The hardware is fine, but it turns out that there's just only so fast you can emulate x86 on ARM. That combined with cheap x86 chips, and expensive memory, makes it just not all that worth launching.

I don't know what the x86 to ARM emulators are like on Windows but on Mac, where they put a lot of effort into it, the answer is "fine but you do lose a fair bit of performance." Most of the Mac users at work didn't care since they don't do anything but browse the web and use Office, and that all got native fast enough. However we had some whiny people when they first got their shiny new ARM toys because they tried to run some simulation software on them and the performance wasn't what they hoped.

If nVidia is hoping to target these at gamers, I could see that maybe performance is fine, but just not good enough to make them worth it vs a cheap Ryzen chip and then add the fact that RAM is making everything expensive and they decide "let's not for now". It may not be a single factor.
 
Or alternatively: The hardware is fine, but it turns out that there's just only so fast you can emulate x86 on ARM. That combined with cheap x86 chips, and expensive memory, makes it just not all that worth launching.

I don't know what the x86 to ARM emulators are like on Windows but on Mac, where they put a lot of effort into it, the answer is "fine but you do lose a fair bit of performance." Most of the Mac users at work didn't care since they don't do anything but browse the web and use Office, and that all got native fast enough. However we had some whiny people when they first got their shiny new ARM toys because they tried to run some simulation software on them and the performance wasn't what they hoped.

If nVidia is hoping to target these at gamers, I could see that maybe performance is fine, but just not good enough to make them worth it vs a cheap Ryzen chip and then add the fact that RAM is making everything expensive and they decide "let's not for now". It may not be a single factor.

Really with Nvidia having 98% of the GPU market... and at this point lets all be honest the vast majority of the mind share. Nvidia could absolutely sell these assuming the rumor of hardware/MS issues aren't true anyway. How many gamers are gaming at "4k" with performance mode DLSS on (so 1080p really)... and boosting about how great the experience is. I'm pretty sure if you stick an Nvidia inside label on just about anything at this point it will sell for 10-20% more then AMD or Intel anything. I imagine if qualcomm can manage to make a mostly operable windows arm chip the NV chips do actually work. I don't really buy the don't work right rumors. As much as I want to laugh at MS I just don't think that is the reason NV hasn't dropped N1 consumer stuff.
Jensen has been creaming himself for decades trying to do all he could to replace Intel. The ARM purchase fell through, but they have still been full ahead on data center ARM.

My take is the AI explosion went even crazier then expected. So NV Is just too busy to worry about consumer anything right now. They cancelled the 5000 supers. I'm sure they just don't care. Also Jensen might have just got another IN. Nvidia now owns a good chunk of Intel. Intel GPUs are dead in the water, and Intel iGPUs are a joke for the next 2 years with everyone knowing full well Intel plans to release CPUs with Nvidia iGPUs again. I think Jensen might just be sitting back thinking if the AI thing keeps going the way its going, in a few years he'll just buy up the rest of Intel. (and I don't know grabbing 1/4 of the laptop market over the next few years might cause regulation issues he doesn't want) At the point of a Intel purchase or even stock % expansion pushing ARM for consumer isn't required anymore, and having done it only really helps Qualcomm and why do that.

Also to be clear I know that N1 isn't going to hang with a real x86 laptop. I just don't think it needs to and I know how bad Jensen wants to be the king of all compute. I think the situation has just changed a bit for him. He is still planning to fuck us. He might just be biding his time. If Intel bobbles either of their next 2 CPU launches... or more importantly screws up their FAB process launches. Jensen might just make hay while Trump is still around. Pretty sure Trump might just hand him the Intel keys if he wants them.
 
If nVidia is hoping to target these at gamers, I could see that maybe performance is fine, but just not good enough to make them worth it vs a cheap Ryzen chip and then add the fact that RAM is making everything expensive and they decide "let's not for now". It may not be a single factor.
Specially if an intel x86 soc is on its way and that they are busy enough doing others things, will see with the next snapdragon x2 where things are,

and Intel iGPUs are a joke for the next 2 years with everyone knowing full well Intel plans to release CPUs with Nvidia iGPUs again.
RYSY3WoWkw8axwiWyPZ5LF.png
VUf5Ae3vJtm4H7o7bCaVLF.png


would not surprise me if they keep them and nvidia-intel are more about high performance soc than the smaller igpu market.

With how great and historically intel iGPU has been, panther lake apparent nice jump, can of worm to change their formula... more like a DGX/N1 affair but the cpu become an intel one is what I would expect things to go for intel-nvidia, Intel will probably want to protect itself and keep its iGPU/quicksync side of things alive and well (for when any deal made expire) and intel cowos/packaging could make that a very natural way to go.

Digital foundry were able to play with it:

View: https://youtu.be/xPkofuH_bak

seem to match Intel numbers, seem to match discrete rx 6600/3050
 
Last edited:
  • Like
Reactions: ChadD
like this
Really with Nvidia having 98% of the GPU market... and at this point lets all be honest the vast majority of the mind share. Nvidia could absolutely sell these assuming the rumor of hardware/MS issues aren't true anyway.
That doesn't mean Nvidia can just take over the x86 market with overpriced MediaTek ARM chips.
How many gamers are gaming at "4k" with performance mode DLSS on (so 1080p really)... and boosting about how great the experience is.
This can quickly be XeSS and FSR4 in a very easy switch.
I'm pretty sure if you stick an Nvidia inside label on just about anything at this point it will sell for 10-20% more then AMD or Intel anything.
Not during a recession, which we are in.
Intel GPUs are dead in the water, and Intel iGPUs are a joke for the next 2 years with everyone knowing full well Intel plans to release CPUs with Nvidia iGPUs again.
Didn't Intel announce plans for their 770? Doesn't seem like it's dead.
Also to be clear I know that N1 isn't going to hang with a real x86 laptop. I just don't think it needs to and I know how bad Jensen wants to be the king of all compute.
It not only needs to but it needs to exceed x86 in performance and price.
 
Not during a recession, which we are in.
According to who/waht ?

https://www.bea.gov/data/gdp/gross-domestic-product
Real gross domestic product (GDP) increased at an annual rate of 4.3 percent in the third quarter of 2025 (July, August, and September), according to the initial estimate released by the U.S. Bureau of Economic Analysis. In the second quarter, real GDP increased 3.8 percent.

4th quarter estimate are even more spectacular.
https://www.atlantafed.org/cqer/research/gdpnow
Latest estimate: 5.1 percent — January 09, 2026

5.1% real gdp growth not in a reboud quarter, that would be the highest since q2 2000

https://www.reuters.com/business/re...-despite-slower-growth-adobe-says-2026-01-07/
Online spending from November 1 through December 31 rose 6.8% to $257.8 billion, Adobe said. That compared to an 8.7% rise in online spending during the same period last year.
https://nrf.com/media-center/press-...ber-data-shows-strong-holiday-season-spending

CNBC/NRF Retail Monitor’s December Data Shows Strong Holiday Season Spending​

Total retail sales, excluding automobile dealers and gasoline stations, were up 1.26% seasonally adjusted month over month and up 3.54% unadjusted year over year in December,
 
According to who/waht ?
They won't say, but there will be signs. Despite Trump being against regulation, he seems to be proposing ideas even Democrats would be against. Wonder why the Democrats lost? Banks and industry insiders say a proposed 10% cap on credit card interest rates would result in fewer credit card accounts for Americans and a dip in spending for the U.S. economy. Trump says U.S. to ban large investors from buying homes. The thing about recessions is that nobody wants to say the hard R. Nobody will ever say we're in a recession until we're knee keep in it.


View: https://youtu.be/pAJd5w1h4eo?si=oqkwBqqLrenm0ZiD
 
They won't say, but there will be signs. Despite Trump being against regulation, he seems to be proposing ideas even Democrats would be against. Wonder why the Democrats lost? Banks and industry insiders say a proposed 10% cap on credit card interest rates would result in fewer credit card accounts for Americans and a dip in spending for the U.S. economy. Trump says U.S. to ban large investors from buying homes. The thing about recessions is that nobody wants to say the hard R. Nobody will ever say we're in a recession until we're knee keep in it.
so the anti-regulation guy want to make massive regulations now ;) ? of course he is not against regulations, like any politicians; against regulations he does not like and for regulations he like (actual libertarian/anarchist are extremely rare and write for Reason or try to do some communist commune, they do not go into organized politics and if they do, do not win). What does those 2 propositions has to do with the idea that we are January 13 in a recession (despite record consumer spending and this century record real GDP growth) ?

People tend to say we are in a recession after 2 consecutive negative real gdp growth quarter and a giant amount of people want to say it and do say it when it happen, high 3.x, followed by 4.x, followed by 5.x is not nearly close to a recession it is one of the most impressive boom of all time.
 
Last edited:
so the anti-regulation guy want to make massive regulations now ;) ? of course he is not against regulations, like any politicians; against regulations he does not like and for regulations he like (actual libertarian/anarchist are extremely rare and write for Reason or try to do some communist commune, they do not go into organized politics and if they do, do not win). What does those 2 propositions has to do with the idea that we are January 13 in a recession (despite record consumer spending and this century record real GDP growth) ?

People tend to say we are in a recession after 2 consecutive negative real gdp growth quarter and a giant amount of people want to say it and do say it when it happen, high 3.x, followed by 4.x, followed by 5.x is not nearly close to a recession it is one of the most impressive boom of all time.
if you remove the AI related spending (such as power generation) then the economy is indeed in a recession. the inflation of all prices (due to AI) will worsen this effect
 
xbox & PlayStation could soon have a new competitor

Nvidia getting more serious about gaming on Linux and Arm: hiring engineer to work on 'native-speed x86-64 gaming on Linux/ARM64 platforms'​

News
By Jacob Ridley published 17 hours ago
Along with a few more job roles with similar remits.


Nvidia is currently hiring for a Linux Graphics Senior Software Engineer. The role is teed up as such:

"We are growing Nvidia and the Linux graphics driver team with the smartest people in the world. We're looking for outstanding software engineers to help us develop driver solutions for new GPUs on desktop, server and gaming Linux platforms; including our newest addition to the line-up: the ground-breaking DGX Spark. We collaborate with open-source frameworks like Vulkan and OpenGL, and we enable Linux games and apps to take advantage of Nvidia GPUs for both x86 and ARM architecture."


There are a few further points of interest in the application pack, spotted over on Reddit, to do with gaming specifically, which we like to see from the GPU company turned AI poster child.

Firstly, the job is described as working on "high-performance Dynamic Binary Translation (DBT) solutions to bridge the architecture gap, enabling native-speed x86-64 gaming on Linux/ARM64 platforms."


Nvidia is also hiring for a Senior Software Engineer, Graphics Performance to work on future Linux graphics drivers.


Further to both posts, Nvidia is seeking a Senior System Software Engineer, Vulkan Performance, to diagnose "GPU and CPU performance bottlenecks in Vulkan and Proton titles". Nvidia is already a big contributor for Vulkan, which is made by a cross-industry consortium called Khronos. Its members include Nvidia, Arm, AMD, Epic, Google, Intel, Qualcomm, Huawei, Valve, Sony, and on and on.


Across the three job postings, you get a pretty good idea of intent for the future of Nvidia's graphics drivers (traditionally, not Linux-friendly) and Arm-based systems.


https://www.pcgamer.com/hardware/gr...speed-x86-64-gaming-on-linux-arm64-platforms/
 
if you remove the AI related spending (such as power generation) then the economy is indeed in a recession. the inflation of all prices (due to AI) will worsen this effect
Imagine trying to start a war with Iran while the US economy is in the trash can. There are probably people who look at the DOW to determine the state of the US economy.

xbox & PlayStation could soon have a new competitor

Nvidia getting more serious about gaming on Linux and Arm: hiring engineer to work on 'native-speed x86-64 gaming on Linux/ARM64 platforms'​

News
By Jacob Ridley published 17 hours ago
Along with a few more job roles with similar remits.


Nvidia is currently hiring for a Linux Graphics Senior Software Engineer. The role is teed up as such:

"We are growing Nvidia and the Linux graphics driver team with the smartest people in the world. We're looking for outstanding software engineers to help us develop driver solutions for new GPUs on desktop, server and gaming Linux platforms; including our newest addition to the line-up: the ground-breaking DGX Spark. We collaborate with open-source frameworks like Vulkan and OpenGL, and we enable Linux games and apps to take advantage of Nvidia GPUs for both x86 and ARM architecture."


There are a few further points of interest in the application pack, spotted over on Reddit, to do with gaming specifically, which we like to see from the GPU company turned AI poster child.

Firstly, the job is described as working on "high-performance Dynamic Binary Translation (DBT) solutions to bridge the architecture gap, enabling native-speed x86-64 gaming on Linux/ARM64 platforms."


Nvidia is also hiring for a Senior Software Engineer, Graphics Performance to work on future Linux graphics drivers.


Further to both posts, Nvidia is seeking a Senior System Software Engineer, Vulkan Performance, to diagnose "GPU and CPU performance bottlenecks in Vulkan and Proton titles". Nvidia is already a big contributor for Vulkan, which is made by a cross-industry consortium called Khronos. Its members include Nvidia, Arm, AMD, Epic, Google, Intel, Qualcomm, Huawei, Valve, Sony, and on and on.


Across the three job postings, you get a pretty good idea of intent for the future of Nvidia's graphics drivers (traditionally, not Linux-friendly) and Arm-based systems.


https://www.pcgamer.com/hardware/gr...speed-x86-64-gaming-on-linux-arm64-platforms/
I imagine this is Nvidia trying to squeeze out that 15% performance loss in Linux with their GPU drivers. They're also trying to improve CPU emulator but there's only so much you can do with (JIT) compilation. This also goes to show that Nvidia is aware how bad their gaming performance is compared to AMD and Intel, because of Nvidia's use of ARM. Maybe this is the reason why we don't have a GB10 in gaming laptops?
 

NVIDIA teases “new era of PC” ahead of N1 and N1X laptop chip announcement​


The company has confirmed that Jensen Huang will speak at the Taipei Music Center on June 1 at 11 AM local time. The timing lines up with earlier reports that NVIDIA may use Computex to reveal its first Windows on Arm PC platform.

https://videocardz.com/newz/nvidia-teases-new-era-of-pc-ahead-of-n1-and-n1x-laptop-chip-announcement
 

NVIDIA teases “new era of PC” ahead of N1 and N1X laptop chip announcement​


The company has confirmed that Jensen Huang will speak at the Taipei Music Center on June 1 at 11 AM local time. The timing lines up with earlier reports that NVIDIA may use Computex to reveal its first Windows on Arm PC platform.

I'm in the market for a new laptop (current one is too godawful slow with Win 11) - I'm looking forward to this (it will be more than I want to pay so I won't buy it - for now/this generation) - I only need it to run DisplayCal/Calman
 
Last edited:
Back
Top