The Hot Take: Seems it was only time with these chips.
AMD's 3D V-Cache CPUs deliver a huge boost versus the Non-X3D part in AI benchmarks, showcasing why they are best suited for RAG pipelines. AMD 3D V-Cache Vs Non 3D V-Cache CPU Benchmarks in AI Showcase a Massive Uplift For RAG Pipelines We know that there are two ways to do AI: the first is LLM, which is currently the most popular model. LLMs are AI models that have been pre-trained on a large set of data and feature various parameter sizes. But LLMs' limitations can be seen when it needs to generate responses on data it wasn't trained on. […]Read full article at https://wccftech.com/amd-3d-v-cache-turns-ryzen-into-surprise-rag-ai-weapon-88-percent-boost-vs-non-x3d/
Read the full article
By ckasprzak | TkOut | April 21, 2026 |
CPU,
Intel,
x86
The Hot Take: I hope this helps them catch up and compete well with AMD as they just keep it coming in the performance department.
New leak information is shedding light on Intel’s upcoming Nova Lake-S desktop processors, with a strong focus on cache architecture. According to recent disclosures, the next-generation chips will adopt a broadcast last-level cache (bLLC) design and significantly increase L3 cache capacity across both single- and ...
Read the full article
The Hot Take: This reminds me of the days with L3 Cache for CPU's. Just the next iteration.
Intel Nova Lake-S "bLLC" cache configurations have been detailed, revealing a maximum of 288 MB of cache for next-gen desktop CPUs. Intel Nova Lake-S Desktop CPUs Will Feature Up To 288 MB of bLLC Cache, 80 MB More Than 9950X3D2 Once again, more Nova Lake-S CPU details have been revealed by Jaykihn, this time focusing on the bLLC parts. While we know that the bLLC or Big Last Level Cache die variants will feature up to 144 MB in single and 288 MB in dual tile configurations, the insider has spilled the beans on the max caches of each bLLC […]Read full article at https://wccftech.com/intel-answer-to-amd-x3d-nova-lake-bllc-cpus-pack-38-percent-more-cache-vs-9950x3d2/
Read the full article
The Hot Take: New out there about the AMD 10k's makes sense why this is coming out now. I wouldn't put it paste intel to leak it to say "Hey folks we're coming back!" Soften the AMD 10k impact on their bottom line. I for one am excited about this news and it means healthy competition is hopefully coming back to the x86 environment at least.
VideoCardz shares the alleged preliminary SKU list for Intel's upcoming Core Ultra 400S (codenamed Nova Lake) processors.
Read the full article
The Hot Take: It's going to be interesting if Intel can make a sudden comeback with actual functional management. I have a feeling this is going to be my next system.
Some high-end motherboards for Intel's upcoming Nova Lake processors have been tipped to include a dual-lever retention mechanism called "2L-ILM." We've seen something similar before with LGA 2011, but that a server platform. Boards with 2L-ILM will live alongside standard ILM (1L-ILM?) variants that are cheaper to produce.
Read the full article
The Hot Take: They did state prices were going up. Now is it inflation or greed?
Following earlier listings, AMD has now confirmed the US $899 price tag for its Ryzen 9 9950X3D2 flagship. The new flagship CPU with dual 3D V-Cache comes with a hefty premium compared to its single 3D V-Cache counterpart, the Ryzen 9 9950X3D.
Confirmed by AMD’s VP and GM of Ryzen CPU and Radeon graphics, David McAfee, the Ryzen 9 9950X3D2, which will be officially available on April 22nd, will be selling for $899, depending on the supply and demand. In case you missed it, the Ryzen 9 9950X3D2 is a 16-core/32-thread SKU that features dual 3D V-Cache, which means that each of the 8-core CCDs has its own cache, leaving the CPU with a total of 208MB of L2 and L3 cache. It works at 4.3GHz base and 5.6GHz Boost clocks and has a 200W TDP.
Although it should be best suited for gaming, AMD is also targeting content creators and developers, saying that it should handle complex workloads and datasets. Unfortunately, AMD has so far shared only select benchmarks, and we’ll have to wait for some official reviews to check out its gaming performance and improvements in latency due to the fact that each CCD has its own cache.
We’ll be keeping an eye on when official reviews go live, and it actually starts shipping from retailers/e-tailers.
Read the full article
The Hot Take: It appears intel is back on the offensive and actually innovating again. This is what competition is supposed to bring us.
Intel's next-gen Nova Lake-S CPUs will feature a 44-core "Dual Compute Tile" CPU config instead of a 42-core configuration. Intel Nova Lake-S 42-Core Desktop CPU Spec Upgraded To 44 Cores, Retains Dual Compute Tile Config With bLLC A few months back, we reported that Intel was working on a range of Nova Lake-S Desktop CPUs that will come in 52, 42, 28, and 24 core configurations, all featuring bLLC cache. It looks like one of these SKUs will be getting a spec upgrade. As per Jaykihn, the 42-core spec has now been upgraded to 44 cores, featuring 16 P-Cores (2x8) and […]Read full article at https://wccftech.com/intel-updates-nova-lake-s-desktop-44-core-dual-compute-tile-cpu/
Read the full article
The Hot Take: If this is true, I feel ARM may have serious competition on their hands.
Many believe that Apple makes the most efficient laptop chips and that MacBooks have the best battery life because the Arm ISA supposedly offers superior performance and efficiency over the crufty x86 ISA. But that is not the case. Apple's products are relatively strong because Apple's engineers do an excellent job at designing them for a
Read the full article
The Hot Take: I'm sorry optimization is optimization. Stop crying.
Intel’s Binary Optimization Tool (BOT) has come under scrutiny following a technical analysis conducted by Geekbench, which examined how the software impacts benchmark performance.
Read the full article
The Hot Take: Reading through them using SIMD, just sounds like they're optimizing the thread pipeline. How can anyone think it's "Cheating"? It's just optimization all hardware vendors do with their silicon.
Geekbench has taken a closer look at Intel's Binary Optimization Tool and found that it can automatically vectorize a large number of instructions.
Read the full article