Tag:

AMD

An AMD Ryzen 5 5000 Series X3D chip in retail packaging.

AMD stealth launches the Ryzen 5 5500X3D, a cheaper downclocked version of its oldest 3D V-Cache CPU architecture

by admin June 13, 2025

Without the slightest fanfare, AMD has added a new “X3D” model with 3D V-Cache to its CPU lineup. Give it up, ladies and germs, for the new Ryzen 5 5500X3D (thanks to X user MEGAsizeGPU for spotting it).

It’s a six-core, 12-thread chip with 96 MB of L3 cache memory, which makes it a hardware doppelgänger for the existing Ryzen 5 5600X3D, which launched back in 2023. The difference with the new chip is all about clockspeeds.

The old 5600X3D runs 3.3 GHz base and 4.4 GHz Boost clocks. The “new” 5500X3D makes do with rounding those numbers down to 3 GHz and 4 GHz respectively.

Both chips share AMD’s first-gen 3D V-Cache tech, which stacks a slab of cache memory atop the CPU die. The idea is that the added on-package cache means the CPU cores have less need to pull data from main system memory. Accessing data from on-package cache is much faster than going out to RAM.

The real-world performance benefits of the added cache vary according to application type. Tasks like video encoding don’t see much of an uptick from adding cache. However, games can run quite a bit faster, sometimes as much as 30 to 40% faster than a comparable CPU without the 3D V-Cache, albeit the upside is usually more in the 10% range. Which is why AMD’s Ryzen 7 9800X3D is currently our gaming CPU of choice.

Apart from the 3D V-Cache, the main appeal here is compatibility with legacy rigs running ye olde AM4 socket. Arguably, if you’re building a new rig from scratch to today, it would make more sense to go with a newer AM5 non-X3D CPU, for instance a Ryzen 5 7600X or 9600X. That way you should have more drop-in upgrade options down the road.

But if you’ve already got an AM4 rig and the time isn’t quite right to go for a full system upgrade, the new 5500X3D could be interesting. That will, of course, hinge on price to a great extent.

For now, pricing and availability isn’t clear. The Ryzen 5 5600X3D was a Microcenter exclusive and is no longer available. Right now, the closest alternative is the Ryzen 7 5700X3D, which goes for around $260.

That’s an eight-core chip with higher clocks, so the 5500X3D would probably come in around $200. For context, a Ryzen 5 9600X is currently around $180 and hits 5.4 GHz, though obviously isn’t a drop-in option for AM4 motherboards. So, we’ll have to wait and see how the 5500X3D is priced and how widely available it is.

Source link

June 13, 2025 0 comments

Gaming Gear

AMD says Instinct MI400X GPU is 10X faster than MI300X, will power Helios rack-scale system with EPYC ‘Venice’ CPUs

by admin June 13, 2025

AMD gave a preview of its first in-house designed rack-scale solution called Helios at its Advancing AI event on Thursday. The system is set to be based on the company’s next-generation EPYC ‘Venice’ processors, will use its Instinct MI400-series accelerator, and will rely on network connections featuring the upcoming Pensando network cards. Overall, the company says that the flagship MI400X is 10 times more powerful than the MI300X, which is a remarkable progress given that the MI400X will be released about three years after the MI300X.

When it comes to rack-scale solutions for AI, AMD clearly trails behind Nvidia. This is going to change a bit this year as cloud service providers (such as Oracle OCI), OEMs, and ODMs will build and deploy rack-scale solutions based on the Instinct MI350X-series GPUs, but those systems will not be designed by AMD, and they will have to interconnect each 8-way system using Ethernet, not low-latency high-bandwidth interconnects like NVLink.

Swipe to scroll horizontally

Year

2025

2026

2024

2025

2026

2027

Density

128

NVL72

NVL144

NVL576

GPU Architecture

CDNA 4

CDNA 5

Blackwell

Blackwell Ultra

Rubin

Rubin Ultra

GPU/GPU+CPU

MI355X

MI400X

GB200

GB300

VR200

VR300

Compute Chiplets

256

144

576

GPU Packages

128

144

FP4 PFLOPs (Dense)

1280

1440

720

1080

3600

14400

HBM Capacity

36 TB

51 TB

14 TB

21 TB

147 TB

HBM Bandwidth

1024 TB/s

1,400 TB/s

576 TB/s

936 TB/s

4,608 TB/s

CPU

EPYC ‘Turin’

EPYC ‘Venice’

72-core Grace

88-core Vera

NVSWitch/UALink/IF

–

UALink/IF

NVSwitch 5.0

NVSwitch 6.0

NVSwitch 7.0

NVSwitch Bandwidth

3600 GB/s

7200 GB/s

14400 GB/s

Scale-Out

800G, copper

1600G, optics

Form-Factor Name

OEM/ODM proprietary

Helios

Oberon

Kyber

The real change will occur next year with the first AMD-designed rack-scale system called Helios, which will use Zen 6-powered EPYC ‘Venice’ CPUs, CDNA ‘Next’-based Instinct MI400-series GPUs, and Pensando ‘Vulcano’ network interface cards (NICs) that are rumored to increase the maximum scale-up world size to beyond eight GPUs, which will greatly enhance their capabilities for training and inference. The system will adhere to OCP standards and enable next-generation interconnects such as Ultra Ethernet and Ultra Accelerator Link, supporting demanding AI workloads.

You may like

”So let me introduce the Helios AI rack,” said Andrew Dieckman, corporate VP and general manager of AMD’s data center GPU business. “Helios is one of the system solutions that we are working on based on the Instinct MI400-series GPU, so it is a fully integrated AI rack with EPYC CPUs, Instinct MI400-series GPUs, Pensando NICs, and then our ROCm stack. It is a unified architecture designed for both frontier model training as well as massive scale inference [that] delivers leadership compute density, memory bandwidth, scale out interconnect, all built in an open OCP-compliant standard supporting Ultra Ethernet and UALink.”

From a performance point of view, AMD’s flagship Instinct MI400-series AI GPU (we will refer to it as to Instinct MI400X, though this is not the official name, and we will also call the CDNA Next as CDNA 5) doubles performance from the Instinct MI355X and increases memory capacity by 50% and bandwidth by more than 100%. While the MI355X delivers 10 dense FP4 PFLOPS, the MI400X is projected to hit 20 dense FP4 PFLOPS.

Overall, the company says that the flagship MI400X is 10 times more powerful than the MI300X, which is a remarkable progress given that the MI400X will be released about three years after the MI300X.

“When you look at our product roadmap and how we continue to accelerate, with MI355X, we have taken a major leap forward [compared to the MI300X]: we are delivering 3X the amount of performance on a broad set of models and workloads, and that is a significant uptick from the previous trajectory we were on from the MI300X with the MI325X,” said Dieckman. “Now, with the Instinct MI400X and Helios, we bend that curve even further, and Helios is designed to deliver up to 10X more AI performance on the the most advanced frontier models in the high end.”

Swipe to scroll horizontally

Year

2024

2025

2024

2025

2026

2027

Architecture

CDNA 4

CDNA 5

Blackwell

Blackwell Ultra

Rubin

GPU

MI355X

MI400X

B200

B300 (Ultra)

VR200

VR300 (Ultra)

Process Technology

N3P

4NP

N3P (3NP?)

Physical Configuration

2 x Reticle Sized GPU

2 x Reticle Sized GPUs

2 x Reticle Sized GPUs, 2x I/O chiplets

4 x Reticle Sized GPUs, 2x I/O chiplets

Packaging

CoWoS-S

CoWoS-L

FP4 PFLOPs (per Package)

100

FP8/INT6 PFLOPs (per Package)

5/-

10/?

4.5

INT8 PFLOPS (per Package)

4.5

0.319

BF16 PFLOPs (per Package)

2.5

2.25

TF32 PFLOPs (per Package)

1.12

2.5

FP32 PFLOPs (per Package)

153.7

1.12

0.083

FP64/FP64 Tensor TFLOPs (per Package)

78.6

1.39

Memory

288 GB HBM3E

432 GB HBM4

192 GB HBM3E

288 GB HBM3E

288 GB HBM4

1 TB HBM4E

Memory Bandwidth

8 TB/s

almost’ 20 GB/s

8 TB/s

4 TB/s

13 TB/s

32 TB/s

HBM Stacks

NVLink/UALink

Infinity Fabric

UALink, Infinity Fabric

NVLink 5.0, 200 GT/s

NVLink 6.0

NVLink 7.0

SerDes speed (Gb/s unidirectional)

224G

GPU TDP

1400 W

1600 W (?)

1200 W

1400 W

1800 W

3600 W

CPU

128-core EPYC ‘Turin’

EPYC ‘Venice’

72-core Grace

88-core Vera

The new MI400X accelerator will also surpass Nvidia’s Blackwell Ultra, which is currently ramping up. However, when it comes to comparison with Nvidia’s next-generation Rubin R200 that delivers 50 dense FP4 PFLOPS, AMD’s MI400X will be around 2.5 times slower. Still, AMD will have an ace up its sleeve, which is memory bandwidth and capacity (see tables for details). Similarly, Helios will outperform Nvidia’s Blackwell Ultra-based NVL72 and Rubin-based NVL144.

However, it remains to be seen how Helios will stack against NVL144 in real-world applications. Also, it will be extremely hard to beat Nvidia’s NVL576 both in terms of compute performance and memory bandwidth in 2027, though by that time, AMD will likely roll out something new.

At least, this is what AMD communicated at its Advancing AI event this week: the company plans to continue evolving its integrated AI platforms with next-generation GPUs, CPUs, and networking technology, extending its roadmap well into 2027 and beyond.

Follow Tom’s Hardware on Google News to get our up-to-date news, analysis, and reviews in your feeds. Make sure to click the Follow button.

Source link

June 13, 2025 0 comments

Gaming Gear

AMD unwraps 2027 AI plans: Verano CPU, Instinct MI500X GPU, next-gen AI rack

by admin June 12, 2025

AMD is accelerating its CPU, GPU, and AI rack-scale solutions roadmaps to a yearly cadence, so the company is set to introduce its all-new EPYC ‘Verano’ CPU, Instinct MI500-series accelerators, and next-generation rack-scale AI solution in 2027, the company revealed at its Advancing AI event.

”We are already deep in the development of our 2027 rack-scale solution that will push the envelope even further on performance efficiency and scalability with our next generation Verano CPUs and Instinct MI500X-series GPUs,” said Lisa Su, chief executive of AMD, at the event.

AMD’s 2026 plans for rack-scale AI solutions already look impressive as the company’s first in-house designed Helios rack-scale system for AI will be based on AMD’s 256-core EPYC ‘Venice’ processor (expected to deliver a 70% generation-to-generation performance improvement); Instinct MI400X-series accelerators projected to double AI inference performance compared to the Instinct MI355X; and Pensando ‘Vulcano’ 800 GbE network cards compliant with the UEC 1.0 specification. But the company is set to introduce something even more impressive the following year.

You may like

That would be AMD’s second generation rack-scale system powered by its EPYC ‘Verano’ processors, Instinct MI500X-series accelerators, and Pensando ‘Vulcano’ 800 GbE NICs.

AMD did not reveal any specifications or performance numbers for its second gen rack-scale solution, EPYC ‘Verano’ processors, or Instinct MI500X-series GPUs. However, based on a picture the company provided, the post-Helios rack-scale machine will feature more compute blades, thus boosting performance density. This alone points to higher performance and power consumption, which will come handy as this one will have to rival Nvidia’s NVL576 ‘Kyber’ system based on 144 Rubin Ultra packages (each packing for reticle-sized compute elements).

Production of EPYC ‘Verano’ CPUs and Instinct MI500X-series accelerators in 2027 align perfectly with TSMC’s roll-out of its A16 process technology in late 2026, its first production node to offer backside power delivery, a technology particularly useful for heavy duty datacenter CPUs and GPUs. We do not know whether AMD’s 2027 processors and accelerators will rely on TSMC’s A16, though it isn’t unreasonable to speculate.

Follow Tom’s Hardware on Google News to get our up-to-date news, analysis, and reviews in your feeds. Make sure to click the Follow button.

Source link

June 12, 2025 0 comments

Gaming Gear

AMD supercomputers take gold and silver in latest Top500 as Chinese HPC remains shrouded in secrecy

by admin June 11, 2025

Top500.org on Tuesday released its 65th list of the world’s most powerful supercomputers, revealing the dominance of American AMD-based systems amid a lack of new entries from China, as well as the performance of supercomputers used by AI giants such as xAI and OpenAI.

The latest update to the global high-performance computing rankings places the AMD Instinct MI300A-based El Capitan at the forefront with Rmax performance of 1.7 FP64 ExaFLOPS, followed by the AMD-powered Frontier (1.353 ExaFLOPS) and Intel-based Aurora (1.012 ExaFLOPS) as the top three exascale-class systems, all of which are operated by U.S. Department of Energy laboratories.

The world’s fourth-most most powerful supercomputer — Germany’s Jupiter Booster, based on Nvidia’s GH200 platform — is a new entrant offering Rmax performance of 0.793 ExaFLOPS. Rounding out the top five, Microsoft’s Eagle achieved 0.561 ExaFLOPS.

You may like

El Capitan, located at Lawrence Livermore National Laboratory in California, achieves 1.742 ExaFLOPS on the HPL benchmark using AMD’s fourth-generation EPYC processors and Instinct MI300A accelerators, all within an HPE Cray EX255a framework. It comprises over 11 million cores and uses the HPE Slingshot interconnect.

El Capitan also topped the companion High Performance Conjugate Gradients (HPCG) performance list with 17.1 PetaFLOPS and in the HPL-MxP mixed-precision benchmark with 16.7 ExaFLOPS. As for energy efficiency, it stands at 58.9 GigaFLOPS per watt.

Frontier, the second-place system in HPL, is installed at Oak Ridge National Laboratory in Tennessee and achieved 1.353 ExaFLOPS. It uses AMD’s third-generation EPYC CPUs, Radeon Instinct 250X accelerators, and HPE Cray EX235a infrastructure with the same Slingshot interconnect. It operates with over 8.6 million cores.

Frontier ranks third in the HPCG benchmark (behind Japan’s Fugaku) with 14.05 petaflops. Frontier also took third place in the HPL-MxP benchmark with 11.4 ExaFLOPS. Its energy efficiency works out to 54.98 GigaFLOPS per watt.

Aurora is housed at Argonne National Laboratory in Illinois and recorded 1.012 ExaFLOPS on the HPL benchmark for the bronze medal in HPL. Built on Intel Xeon CPU Max and Data Center GPU Max (aka Ponte Vecchio) components, it relies on an HPE Cray EX installation using the Slingshot interconnect. It secured the second spot in the HPL-MxP benchmark with 11.6 ExaFLOPS, but it’s in fourth place on the HPCG benchmark at 5.6 PFLOPS.

JUPITER Booster is Europe’s first system in this performance class. The machine is installed at the Jülich Supercomputing Centre in Germany and reaches its fourth-place spot with a preliminary 793.4 PFLOPS in HPL. JUPITER Booster uses Nvidia Grace Hopper hardware on Eviden’s BullSequana XH3000 platform with direct liquid cooling and HP’s Slingshot networking. It is currently being brought online.

Microsoft Azure’s Eagle system holds the fifth position in HPL with 561 PFLOPS using Xeon Platinum 8480C processors and Nvidia H100 GPUs.

Although CPUs from AMD power two of the world’s highest-performing processors and five out of Top 10 machines, a detailed analysis of the June 2025 Top 500 list reveals that Intel processors are used in 294 of the 500 systems. AMD follows with 173 supercomputers.

Nvidia systems appear in 13 entries, showing early traction for the Arm-compatible Grace Hopper architecture. Nine systems are based on other Arm processors, such as Fujitsu’s A64FX. An additional six systems use other processor types, including IBM Power9 and China’s Sunway architecture.

The U.S. extended its numerical lead in total systems in the 65th list, while China continued its downward trend as it no longer submits results of its latest systems to Top500.org. As a result, there are 175 American systems, 47 Chinese supercomputers, and 41 German machines in the Top 500 list.

On the energy efficiency front, Germany’s JEDI system leads with 72.73 GigaFLOPS per watt, followed by France’s ROMEO-2025 at 70.91 and Adastra 2 at 69.1. All three use BullSequana XH3000 infrastructure. El Capitan and Frontier ranked 26th and below on energy, reflecting a different balance between performance and efficiency.

Follow Tom’s Hardware on Google News to get our up-to-date news, analysis, and reviews in your feeds. Make sure to click the Follow button.

Source link

June 11, 2025 0 comments

Gaming Gear

GIGABYTE AORUS: the premier choice for an AMD gaming PC

by admin June 9, 2025

The motherboard is the heart of your PC, the nexus into which everything else connects. Using a good quality motherboard is important for any gaming PC build, as choosing second best can lead to lower framerates and less expansion capability.

For AMD’s AM5 processors, such as the latest 9000-series chips including the mighty Ryzen 7 9800X3D, that means the X870 chipset, and the X870E AORUS Master is GIGABYTE’s top-end ATX board, offering the best power and thermal design alongside exceptionally fast PCIe 5.0 and USB4 connectivity.

The X870E AORUS Master is built to be powerful, yet also to run at surprisingly low temperatures. It manages this by designing thermal management into the board at every level. The VRM—the voltage regulator module that helps guide the right amount of power to critical systems such as the CPU and RAM—is enveloped in GIGABYTE’s advanced Thermal Armor system. This uses heatpipes and a thermal pad to prevent this important corner of the motherboard from getting too hot, despite the large amounts of power that pass through it every time you switch your PC on. The IO shield is integrated into the cooling system, so you won’t have to find it in the box and fiddle around installing it the right way up.

(Image credit: GIGABYTE)

Taming the beasts

Your PC’s SSDs can get hot too, especially if they’re the cutting-edge PCIe 5.0 drives supported by the X870E AORUS Master. The motherboard comes with a heatsink that will dissipate the heat they produce, helping to prevent them slowing down as they reach the point of thermal throttling. Earlier generations of M.2 SSDs didn’t need this additional cooling, but if you want the fastest in 2025 you need to take temperature into account. Three of the X870E AORUS Master’s M.2 slots support PCIe 5.0 x4 speeds, with the fourth still speedy at PCIe 4.0 x4, meaning you can fill a PC with a lot of fast storage by taking advantage of all these possibilities.

With all this heat being pumped into the air inside the case, you’ll need to supply fresh air through the use of case fans. The X870E AORUS Master offers a total of eight PWM fan headers, giving you endless intake and exhaust options alongside your choice of CPU cooler, all software controlled so that they only spin up to their full power when they need to, keeping the system quiet while under a reduced load. While you’re plugging things in, there are four aRGB Gen 2 headers and a four-pin light strip socket, to really make your PC your own with a unique look.

(Image credit: GIGABYTE)

The greatest graphics

A top-end motherboard deserves a similarly premium GPU, and GIGABYTE has you covered with its Radeon RX 9070 Gaming OC 16G. With 16 GB of VRAM on a 256bit interface, these cards offer a huge boost in raw performance over the previous generation, and bring with them improvements in AI calculations and ray-tracing too.

All this rendering power requires cooling that can keep up, and GIGABYTE’s Windforce system combines three of GIGABYTE’s innovative Hawk fans capable of spinning in alternate directions to reduce turbulence and increase airflow. The fan blades have been designed taking inspiration from the aerodynamic shape of an eagle’s wing, shaped to reduce air resistance and noise levels, and capable of a 12.5% increase in air volume without extra noise. Inside the fan is graphene nano lubricant, which eases the passage of the fan blades through the casing and can extend the life of the fan by 2.1 times over less sophisticated bearings. It’s quieter too.

Added to this futuristic engineering are heatpipes, a large copper plate in contact with the GPU itself, and topped off with server-grade thermal conductive gel to get temperatures under control. The card is capable of utilising a dual BIOS structure so that it can easily switch between Performance and Silent modes, using less power and needing less cooling—it can even turn its fans off completely when you’re not asking it to create fantastic graphics at phenomenal frame rates.

(Image credit: GIGABYTE)

Build the future

The X870E AORUS Master and Radeon RX 9070 Gaming OC 16G help make building your own PC easier than ever with their innovative building features. The graphics card slips home in the top PCIe slot easily thanks to EZ-Latch Plus with no screws to do up, while there are magnetic attachments in place for the SSD heatsink (EZ-Match) while the drives themselves which pop up at the press of a button and also have a fully screwless design. Even the WI-Fi 7 antennas connect easily to the terminals on the IO shield, with no fiddling or screwing required.

GIGABYTE’s supremely well designed motherboards for the latest generation of PC components are the premier choice for high-end AMD PC builds with X3D processors. Combine one with GIGABYTE’s Radeon RX 9070 Gaming OC 16G graphics card, and you have the basis of a system that will provide the highest frame rates and visual clarity in your games.

Source link

June 9, 2025 0 comments

Gaming Gear

AMD unveils Ryzen AI Z2 Extreme with AI processing NPU, expands family with two new chips for handheld gaming consoles

by admin June 8, 2025

AMD has announced two new additions to its Ryzen Z2 series of processors meant for handheld gaming devices. The company has introduced a new top-of-the-line Ryzen AI Z2 Extreme, which is essentially the Z2 Extreme with a dedicated Neural Processing Unit (NPU). As a quick reminder, the original lineup was announced back in January and included the Ryzen Z2 Extreme, Ryzen Z2, and the Ryzen Z2 Go.

The second chip is the Ryzen Z2 A, which is stacked directly below the Ryzen Z2 Go, featuring fewer GPU cores and a lower configurable TDP (Thermal Design Power). These new additions indicate that we could see a wide range of handheld gaming consoles launch this year, potentially targeting the holiday season.

The company has shared limited information regarding the specifications and performance capabilities. The Ryzen AI Z2 Extreme has a similar configuration to the non-AI variant, including an 8-core/16-thread CPU, 24MB of cache, support for LPDDR5X-8000 memory, a configurable TDP of 15-35 watts, and 16 RDNA 3.5 graphics cores. The new Z2 processor is the first and only processor in the lineup to include a dedicated NPU, which is said to offer up to 50 TOPS of AI compute power and support for Microsoft’s Copilot+ features.

You may like

Image 1 of 2

(Image credit: AMD)(Image credit: AMD)

The Ryzen Z2 A now becomes the entry point into AMD’s Ryzen Z2 lineup, although it is based on the much older Zen 2 architecture. It features a 4-core/8-thread CPU, 8 RDNA 2 graphics cores, 6MB of cache, and supports LPDDR5-6400 memory. This makes it even less powerful than the previously announced Ryzen Z2 Go, but interestingly, it has a lower configurable TDP ranging from 6- 20W, which might be beneficial for battery life.

Swipe to scroll horizontally

Processor

Architecture

Cores/Threads

Graphics architecture

Graphics cores

Cache

NPU TOPS

AMD Ryzen AI Z2 Extreme

Zen 5

8 cores/16 threads

RDNA 3.5

24MB

50 TOPS

AMD Ryzen Z2 A

Zen 2

4 cores/8 threads

RDNA 2

6MB

N/A

AMD hasn’t shared any information on what kind of performance we can expect from these new chips, nor has the company confirmed any new upcoming devices that will be powered by the new chips. The addition of an NPU on the Ryzen AI Z2 Extreme could translate to potential AI-based enhancements like improved system optimization, faster real-time processing, and even power management to some extent. It may also prove beneficial for device manufacturers to integrate software and tools that enable them to tweak system performance in a more efficient manner.

Currently, the Lenovo Legion Go S and Legion Go 2 Prototype are the only two handhelds confirmed to utilize AMD’s Z2 series chips. Asus is another OEM that is expected to announce at least one new handheld this year, likely a successor to the ROG Ally and Ally X.

Follow Tom’s Hardware on Google News to get our up-to-date news, analysis, and reviews in your feeds. Make sure to click the Follow button.

Source link

June 8, 2025 0 comments

Gaming Gear

ASRock tailors budget-friendly AMD A620 mini-ITX motherboard to SFF enthusiasts

by admin June 7, 2025

ASRock (via momomo_us) has introduced a new AMD mini-ITX motherboard for budget-conscious consumers. The A620AI WiFi, powered by the A620 chipset, arrives with native support for Ryzen 7000 (Raphael), Ryzen 8000G (Phoenix), and Ryzen 9000 (Granite Ridge) processors, including the Ryzen 9000X3D variants.

The A620AI WiFi isn’t the first mini-ITX motherboard with the A620 chipset in ASRock’s arsenal. The manufacturer had already released the A620I Lightning WiFi, but it was intended for a different market segment. The A620AI WiFi, on the other hand, is the more affordable version of the A620I Lightning WiFi.

The A620AI WiFi features an eight-layer PCB with an 8+2+1-phase power delivery subsystem and 80A power phases. Supplementary power to the processor is delivered via an additional 8-pin EPS power connector. The A620AI WiFi’s design is more than adequate to handle Ryzen processors of all statures, spanning from entry-level 35W chips to flagship 170W parts.

You may like

Logically, mini-ITX motherboards have some limitations compared to standard ATX ones, such as the number of memory and expansion slots. The A620AI WiFi provides two DDR5 memory slots supporting ECC and non-ECC memory modules beyond DDR5-8200. The maximum capacity supported is 128GB when pairing two 64GB DDR5 memory modules together.

Image 1 of 2

(Image credit: ASRock)(Image credit: ASRock)

The A620AI WiFi offers one PCIe 4.0 x16 expansion slot for discrete graphics cards. There are two M.2 slots: one sticks to PCIe 4.0 x4 and the other to a slower PCIe 3.0 x4 interface. The motherboard also has two SATA III connectors. RAID arrays are on the support list. You can do RAID 0 and RAID 1 with SATA drives and RAID 0, RAID 1, and RAID 10 for M.2 drives.

You can access the internet via either a wired or wireless connection. For the former, you can pick between a 2.5 Gigabit Ethernet port that utilizes the Realtek RTL8125BG controller or a standard Gigabit Ethernet port wired to the Realtek RTL8111 controller. On the other hand, wireless connectivity includes Wi-Fi 6E and Bluetooth 5.2, enabled by the 802.11axe module.

As with the USB mix, the A620AI WiFi delivers two USB 3.2 Gen 2 Type-A ports, one USB 3.2 Gen 2 Type-C port, one USB 3.2 Gen 1 port, and two USB 2.0 ports. If you don’t plan to use a discrete graphics card, there’s a single HDMI 2.1 port capable of 4K (3840 x 2160) output at 120 Hz. As for audio, you receive three 3.5mm audio jacks for 7.1-channel audio support connected to a Realtek ALC897 codec.

ASRock hasn’t disclosed the pricing and availability for the A620AI WiFi motherboard. Since the A620I Lightning WiFi retails for approximately $140, the A620AI WiFi should launch with a lower price tag.

Follow Tom’s Hardware on Google News to get our up-to-date news, analysis, and reviews in your feeds. Make sure to click the Follow button.

Source link

June 7, 2025 0 comments

Product Reviews

ASE adopts AMD CPUs, begins evaluating Instinct MI300-series GPUs for AI

by admin June 6, 2025

AMD this week said in a blog post that ASE Technology, the world’s largest outsourced semiconductor assembly and test (OSAT) provider, has transitioned to EPYC and Ryzen processors across its data centers and client systems, respectively. The transition has resulted in significant performance improvements and energy efficiency gains. However, perhaps more important is that ASE is now evaluating AMD’s Instinct MI300-series processors for AI workloads.

By adopting AMD’s EPYC processors for servers and Ryzen CPUs for client desktop and laptop PCs, ASE achieved a 50% boost in system performance and a 6.5% reduction in power consumption compared to its previous infrastructure, which resulted in a 30% decrease in total cost of ownership, delivering both operational and financial benefits. AMD’s blog does not disclose which processors ASE used before adopting AMD-based solutions, nor does it indicate whether all systems in ASE’s fleet now use EPYC or Ryzen processors. However, the mention of operational and financial benefits points to a substantial adoption of AMD-based systems.

“We need to handle a big volume of data analysis, including leading-edge technology for AI applications and our smart factories,” said Jekyll Chen, Director of IT Infrastructure for ASE. “We work for many semiconductor companies. Our challenges are the need for high performance, low latency, and high core count, in alignment with ASE’s ESG policy. Stability and scalability are two primary goals for us.”

You may like

(Image credit: ASE Technology)

ASE Technology Holdings is the world’s largest outsourced semiconductor assembly and test provider, with packaging facilities in China, Japan, Korea, Malaysia, Singapore, and Taiwan. The company has worked with AMD on advanced 2.5D packaging since 2007, and this largely resulted in the invention of high-bandwidth memory (HBM). However, while ASE does provide packaging services for AMD these days, we are unsure whether ASE packages AMD’s AI GPUs, as Instinct processors utilize TSMC’s CoWoS technology.

AMD says that many companies are adopting or evaluating its Instinct processors for on-prem AI inference, though ASE is probably the first company of this scale to confirm the evaluation of these accelerators. In fact, the confirmation may indicate that ASE is close to adopting Instinct MI300-series GPUs for its internal AI workloads.

“We must perform data processing, run AI algorithms, and make sure everything operates smoothly, efficiently, and with the flexibility needed in our smart factories,” Chen said. “For client PCs, we need to make sure that they meet the needs of engineering design and the high-performance objectives of digital transformation. We also evaluated the performance, stability, core count, efficiency, total cost of ownership, AI speed, and multi-tasking capabilities of the new servers.”

Follow Tom’s Hardware on Google News to get our up-to-date news, analysis, and reviews in your feeds. Make sure to click the Follow button.

Source link

June 6, 2025 0 comments

Product Reviews

AMD RX 9060 XT 16GB flies off shelves, 8GB lingers – GPU launch highlights demand split between variants

by admin June 6, 2025

It’s been a few hours since the AMD RX 9060 XT hit online shelves, and so far, the state of things across the web seems split harshly down the line of the card’s 8GB and 16GB variants. The card is widely available at MSRP in the U.S. and several European countries in both its 8GB SKUs, but 16GB stock is far harder to find than the 8GB variant, now stabilizing at around $40 over MSRP online in the United States and Europe.

The 9060 XT’s 8GB and 16GB models launched today at suggested retail prices of $299 and $349, undercutting Nvidia and finally providing an on-ramp to 1080p gaming in the current generation of GPU releases. The card has many variants from an array of board partners, and our team has had trouble finding 16GB models remain in stock for longer than two hours. That’s not a problem with the 8GB version, though; We’ve had trouble finding sites that don’t offer the 8GB at MSRP.

Newegg currently hosts the greatest number of 9060 XT models for U.S. shoppers, with the site offering many 8GB models in stock at MSRP of $299. Its 16GB models are noticeably scarcer, with several SKUs selling out during the time of writing this article, and only $389 models are currently available.

You may like

Other sites in the U.S., such as Best Buy, seems to be drip-feeding supply throughout the day. Several 8GB models are still available at MSRP on a variety of U.S. retailers, with 16GB now also popping up closer to $390.

Western European retailers have largely raised prices to €349 ($399) for the RX 9060 XT 16GB this morning, but several like Overclockers UK still offer several models at €319, matching the U.S. MSRP after the mandated VAT. Some further inland have complained about limited supply at scalper’s prices even before sales began, indicating another desert in GPU supply in Eastern Europe.

In-person retailers like MicroCenter are also reporting high stock on physical shelves at a variety of locations, and starting at the coveted $349 price tag.

For a constantly-updated look at availability trends on the 9060 XT throughout the day, be sure to refer back to our where to buy article, which will be updated as sales continue to shift throughout the week.

As many online critics suggested pre-release, the 8GB does not seem to be enticing early adopters, indicated by its availability. But if nothing else, it is nice to see a GPU still available at MSRP on its launch day in 2025.

Source link

June 6, 2025 0 comments

Product Reviews

Graphene thermal pad for AMD CPUs promises 17X better conductivity than thermal paste, 2X improvement over Thermal Grizzly

by admin June 5, 2025

Chinese manufacturer Coracer has introduced the GPE-01 graphene thermal pads for AMD’s AM5 processors, which are currently some of the best CPUs you can buy. Previously, the GPE-01 was compatible only with Intel’s LGA1851 and LGA1700 chips.

Traditional thermal paste has existed for decades. While it may not be the most exciting component, it effectively fulfills its purpose. In recent years, however, enthusiasts have developed a preference for alternative materials to fill the gap between the processor and CPU cooler. As a result, options like liquid metal and thermal pads have become more common in the market and on our page of the best thermal paste and TIM we’ve tested.

Segotep introduced the GPE-01 (21 x 44mm) at the end of last year, tailored for the elongated designs of Intel’s LGA1851 and LGA1700 processors. While Segotep is a well-established brand in the Chinese market, we were unaware of Coracer. A Google search brought no results, leaving us uncertain whether Coracer is a new sub-brand of Segotep or if the original manufacturer licensed the GPE-01 to another brand.

You may like

Regardless, the AM5 variant of the GPE-01 measures 32 x 32mm and aligns perfectly with the processor’s integrated heat spreader (IHS). While AM5 chips feature an unconventional design, the GPE-01 adopts a square shape that covers the entire IHS but doesn’t overreach to the corners.

According to Coracer, the GPE-01 utilizes a combination of graphene and silicon to achieve an eye-popping thermal conductivity up to 130 W/m·K. The thermal pad is encased in an isolating material that stops the graphene from creating a short circuit with the processor. Graphene is known for its excellent conductivity. As a result, manufacturers increasingly utilize this material in thermal pads and paste.

GPE-01 Graphene Thermal Pad Specifications

Swipe to scroll horizontally

Brand

Thermal Conductivity

Pricing

GPE-01

130 W/m·K

Thermal Grizzly Conductonaut

73 W/m·K

$12.43

Honeywell PTM7950

8.5 W/m·K

$19.99

Thermal Grizzly KryoSheet

7.5 W/m·K

$18.13

Arctic MX-6

7.5 W/m·K

$7.99

The GPE-01 boasts an impressive thermal conductivity of 130 W/m·K. This figure is nearly 2X higher than Thermal Grizzly’s Conductonaut liquid metal thermal paste. Additionally, it exceeds the thermal conductivity of Thermal Grizzly KryoSheet graphene thermal pads and Arctic MX-6 carbon filler-based thermal paste by more than 17X, and the Honeywell PTM7950 phase change thermal pad by 15X.

As with any claims regarding thermal conductivity, we advise being skeptical of figures provided by vendors. There have been instances where manufacturers inflate the thermal conductivity values to promote their products. While we are not alleging that this is the situation with the GPE-01, it is wise to approach the numbers with caution. Research has shown that graphene is a thermally conductive material that can reach levels up to 4,000 W/m·K, so there may be some credence to the GPE-01’s high conductivity figures.

In addition to its astronomical conductivity values, Coracer states that the GPE-01 can last up to 10 years. This assertion is more credible, given that the performance of thermal paste diminishes over time and ultimately dries out. Thermal pads, in contrast, have a longer lifespan, eliminating the need to repaste your processor every few years like with regular thermal paste. With the GPE-01, users likely won’t need to replace the pads until their next upgrade, unless they keep their system for over a decade.

We haven’t come across any online reviews for the GPE-01, so we are unable to assess the performance of the graphene thermal pad. Coracer has not disclosed the price or availability of the GPE-01, while the Intel version is priced at approximately $15 on the Chinese e-commerce site Taobao. This price is similar to that of Thermal Grizzly’s KryoSheet graphene thermal pads.

Follow Tom’s Hardware on Google News to get our up-to-date news, analysis, and reviews in your feeds. Make sure to click the Follow button.

Source link

June 5, 2025 0 comments