• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Discussion NV Re-Enter ARM PC market in Q2-2026!

Page 15 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
It apparently taped out last December

But Charlie's leak published only this April

What were nv, mediatek, & Microsoft doing in those 90 days ?
 
marees said:
It apparently taped out last December

But Charlie's leak published only this April

What were nv, mediatek, & Microsoft doing in those 90 days ?
Click to expand...
I mean in 3 years replace Mediatek with Intel too.

This is why you need to own your hardware stack like AMD and Apple do. Intel owns it now too.

Intel replacing its iGPU IP with Nvidia will for sure see delayed products.
 
Jan Olšan said:
What's the problem with the display block? Does it suck compared to MediaTek IP?
Click to expand...
If I'd have to guess, they have issues with integrating that NVIDIA ip and they have bugs that make the thing unstable.

Or everything is just a guesswork and there are no hardware issues at all. I wouldn't be surprised if the real issue is Microsoft or that the drivers just aren't ready yet.
 
So the controlled benchmarks from selected reviewers bits was true. I guess everybody got the same reviewer guide, unless I have missed it nobody ever talks about the CPU part of this thing? [how it does compare in cpu only inference, how fast it can compile new libs, what is the power draw, does it work outside the two linux distros they officially mention].
 
First power measurements I have seen and also some interesting bits https://www.servethehome.com/nvidia-dgx-spark-review-the-gb10-machine-is-so-freaking-cool/

There are a few clear challenges working with the GB10. Somewhat surprisingly, video output is one of those areas that you think any NVIDIA product would nail. The Spark has been challenging to say the least. The LG OLEDs we have that are 1440p, display a garbled mess out of the HDMI port if set to 1440p output in the OS. Likewise, ultra widescreen monitors were a no-go.
Click to expand...

At idle, when we did our power measurements last week, this system was idling in the 40-45W range. Just loading the CPU, we could get 120-130W. Adding the GPU, and other components we could get to just under 200W, but we did not get to 240W. Something to keep in mind is that QSFP56 optics can use a decent amount of power.


Also, in many of the AI inference workloads with LLMs we were using 60-90W and the system was very quiet. There is a fan running, but if you are 1-1.5m away it is very difficult to hear and it never hit 40dba when we were not stress testing the system.
Click to expand...
 
Thibsie said:
Really Why ?
Click to expand...
It won’t be competitive. Both AMD and Nvidia missed the target with 256 bit memory. For LLMs, bandwidth is as important as compute and the Mac Studios have 512 bits up to 128 GB RAM capacity and 1024 bits up to 512 GB of RAM.

The huge knock on Macs is the compute is way too low, but the Apple10 GPU architecture resolves that with the neural accelerators (tensor cores). Comparable compute with double to quadruple the bandwidth.

Apple’s MLX framework now has a functioning CUDA backend, model algorithms run as is on Nvidia hardware. Most people, if not everyone wanting DGX Spark for locally prototyping and fine tuning a planned datacenter deployment can do better with the upcoming Mac Studios as well.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top