First time I am hearing of this Spectrum Brand; also "Dough" is an interesting name for a monitor product series.

Seems to be squarely aimed at GPUs.

AMD's Zen+ 1700X (with no iGPU) has 4.8 billion transistors. 1080 Ti has 11.8 billion.

Even the latest Apple CPU doesn't hit 30 billion:

This will happen shortly, as Apple's M4 system-on-chip for mainstream PCs is already at 28 billion transistors.

Sounds like a solid option for entry-level performance use cases. The price/performance ratio versus the i5-13400F will be key.

If you haven't seen it yet, check out this investigation on Honey (20 minutes, Part 1):

https://m.youtube.com/watch?v=vc4yL3YTwWk

It's fascinating stuff. Open fraud.

I can't speak for formal legal matters (I am assuming such scams are nominally legal in the US), but it goes to show that senior PayPal executives are basically criminals. There is no way they didn't know about this.

Excuse me? I don't do imperial, I am from continental Europe. :)

The plans Arm executives discussed included potentially inching closer to making a complete chip design of Arm's own, according to testimony and documents at trial. Arm sells chip-design blueprints, but most of its customers still spend months completing the chip design.

"It was news to me that Arm is even thinking about (making its own chip)," said Tantra Analyst founder Prakash Sangam, who attended the trial. "It should send a chill down the spine of their customers."

I wonder how competitive a fully ARM developed chip would be compared to say Qualcomm, Mediatek or Nvidia.

I think they will be fine in the iGPU space.

Agreed regarding dropping down from Ultra. But I do think if you're going for a high fidelity Cyberpunk experience (4K), it's worth enabling ray tracing.

Doesn't seem like the 9070 XT is all that good of a solution for native 4K.

Not in a $600 laptop it seems.

RX 8000

It supposed to be released as a RDNA 3.5 product for laptops (iGPUs?).

Sound like a novel and genuinely useful approach to reducing data transfer sizes.

From my understanding, this is an ML powered "on the fly" compression scheme which optimizes based on your particular workload type.

Even conceptually, this makes a lot of sense.

I am assuming this would only work with certain use cases. E.g. I can't imagine this would work well with streaming video (e.g. AV1).

Value is somewhat subjective for non-commodity products.

I could see a lot of LOTRO fans (even ones not into mechanical keyboards) buying such a product.

Lenovo claims the keyboard will last up to 45 days in a completely dark environment before needing to be recharged. When the power is depleted, all it needs is 2 minutes of charging under a light source of 300 Lux intensity to get it back on for use.

This implies that you won't really need much ambient light. That being said, I wouldn't trust the marketing copytext.

Currently, EUV lithography systems consume a lot of power—up to 1,400 kilowatts for high-NA EUV tools. Most of this energy is used to heat tin droplets to a blazing 500,000°C to create plasma that emits the needed 13.5-nanometer light. Using diode-pumped solid-state technology, the BAT laser might improve energy efficiency and heat management compared to today’s CO2 lasers.

I did not know a high-NA EUV tool uses 1.4 megawatt of power. Although the 500,000°C temperature requirement does explain why.

Qualcomm makes AI appliances? The press release isn't very clear, but this seems to be a completely custom architecture (and software tooling suite?).

I wonder what the benefit of such a solution is over Nvidia/AMD? Better suited for on-prem edge uses cases?

Still no confirmation whether FSR 4 will work on older Radeon cards.

Otherwise, nothing too surprising. AMD have given up on the enthusiast/high end segment of the market. Ray tracing improvements (would be interesting to see benchmarks) and more AI compute features/hardware functionality.

AMD has confirmed that the flagship Ryzen AI MAX+ (PRO) 395 will outperform the Apple M4 Pro 12C by 12–86%, while delivering an average of 1.4x the performance of Intel’s Core Ultra 9 288V (Lunar Lake) in 3DMark tests and 2.6x better rendering performance.

That's quiet the claim especially in contrast to the Apple M4 Pro. I am looking forward to independent benchmarks, especially power draw normalized ones.

And this is only GPU performance, I can't imagine this particular Zen 5 implementation will beat M4 Pro in ST CPU performance.

Not a big fan of RGB (especially when it is overused), but this seems like a decent use of RGB with a nice design package overall.

All AI/ML compute.

I believe the last time we saw HBM memory in consumer hardware was one of the Radeon series by AMD. Don't think they did much with it though.