Intel and AMD Create Alliance To Defend x86 Architecture

At Lenovo Tech World, Intel and AMD announced a new x86 ecosystem advisory group. This group aims to unify the x86 architecture to enhance compatibility and innovation across products.

Both Intel’s CEO Pat Gelsinger and AMD’s CEO Lisa Su emphasized the critical role of x86 in driving technological advancements, especially in AI. Nvidia’s CEO Jensen Huang supported the initiative, noting that fragmentation in x86 was harmful for the industry.

Linus Torvalds and Tim Sweeney are the Luminaries along with founding members Broadcom, Dell, Google, Hewlett Packard Enterprise, HP Inc., Lenovo, Meta, Microsoft, Oracle, and Red Hat.

🔗 Intel | CRN
🧑‍💻 @agamtechtricks

Intel Explores Foundry Alliance with Samsung
Source

Intel has reached out to Samsung Electronics to explore the possibility of forming a foundry alliance. Intel’s CEO, Pat Gelsinger, is reportedly seeking a direct meeting with Samsung Electronics Chairman Lee Jae-yong to discuss comprehensive cooperation plans for their foundry divisions.

If an Intel-Samsung foundry alliance materializes, the two companies could collaborate on various fronts, including process technology exchanges, shared production equipment & joint R&D efforts.

Its' actually a good move because TSMC has established a monopoly and it is crazily increasing prices for its wafers, leading to increase in pricing for SoC's used in smartphones, data centers, gaming handlehelds etc. The price of TSMC's 2nm wafers is expected to double compared to 4/5nm, which may exceed USD 30,000 per wafer.

Follow @TechLeaksZone

BREAKING: Google Tensor G5 Specification Leaked

Google’s Tensor G5 to power the Pixel 10 series next year will be designed internally by Google themselves without relying on Samsung for the bulk of the work. It will be manufactured on TSMC 3nm node process.

Tensor G5 CPU: codenamed 'Laguna'
SLC: 8MB
RAM: LPDDR5X (8533 Mbps)

• 1x Cortex-X4
• 5x Cortex-A725
• 2x Cortex-A520
Sadly Google is NOT using ARM's latest X925 cores which debuted with MediaTek Dimensity 9400 recently.

TPU:
The TOPS value is 18 / 9 TOPS (INT8/FP16), making it 40% larger compared to Tensor G4 but it will be around 15% faster only. The improved TPU also comes with a few new features for Google’s developers, such as small embedded RISC-V cores to allow running operations that are not implemented in hardware, as well as support for on-device training.

GPU:
This is the MOST disappointing part of the whole G5 chip. That's NOT because Google is ditching Mali GPU, but because Tensor G5 will feature PowerVR GPU from Imagination technologies — IMG DXT-48-1536 clocked at 1100MHz.

These GPU are mostly used by UNISOC in their chips for smartphones. The only reason why Google switched to PowerVR GPU is due to virtualization support. It also supports Ray tracing, however, the GPU will just have 2 CU (compute units) only in 2025 for 1000$ flagship phone

Now, to get your an idea of how fast would be the GPU, some crazy people calculated the expected benchmarking value of OpenCL compute & it will score around ...... well ........ Let it be a surprise for Pixel fanbois.

Follow @TechLeaksZone

BREAKING: Google is developing its own custom architecture for future Tensor chips

Google already has its own custom architecture called 'Cypress' but it is basically for Axion processors for datacenters.

But now, Google is also working on a custom CPU core called 'Orion' that will debut with future Tensor chips, atleast 'after' the Tensor G6. Interestingly, Google is even testing some tensor chipsets clocked at 5.1GHz & manufactured on a 3nm node, likely TSMC.

Coming to the Tensor G6, it will power the Pixel 11 series & is codenamed as 'Malibu'. Although other SoCs will already be built on 2nm chip design in 2026, Tensor G6 will still be manufactured on 3nm node (N3P) by TSMC .

CPU:
• 1x Cortex- X930
• 6x Cortex- A730
RAM: LPDDR5X
So, Tensor G6 will be a '7-core' configuration SoC

GPU:
Don't expect Google to ditch PowerVR GPU quite soon. Tensor G6 will have an IMG EXT GPU, having 3 CU (compute units) running at 1100Mhz. It will support virtualization & ray tracing, both.

Follow @TechLeaksZone

Samsung Exynos 1580 Announced

Samsung has announced Exynos 1580 SoC manufactured by Samsung foundry on its latest 4nm node process which will power the Samsung Galaxy A56 smartphone.

CPU:
• 1x Cortex-X720 @ 2.9GHz
• 3x Cortex-A720 @ 2.6GHz
• 4x Cortex-A520 @ 1.95GHz

GPU: Xclipse 540 (4 CU)
• 2x WGPs (up from 1 in Exynos 1480)
• Based on AMD RDNA 3 architecture
• Performance improved by 37%
• Increased GL2 cache capacity
• Twice the amount of FMA/texture unit processing

The Exynos 1580 also has an NPU that supports up to 14.7 trillion operations per second (TOPS) with a memory capacity increased to 2 MB, offering a high-quality AI experience.

Display: FHD+ @ 144Hz refresh rate
Camera: upto 200MP; 4k60fps video recording
RAM: LPDDR5
Storage: UFS 3.1
Connectivity: WiFi- 6E, BT 5.4v

Follow @TechLeaksZone

Here's a first look at Android's upcoming all-in-one Terminal app, which downloads, configures, runs, and interfaces with an instance of Debian running a virtual machine!

Currently, the Debian images are hosted on a Googler's GitHub, but Google plans to host these images themselves soon. For more info on this Terminal app, check out this article.

(Thanks to an anonymous dev for sharing this!)

isnt this lineage updater? its used by the MOONDROP MIAD01 phone