Tag Archives: TSMC

AMD @ Computex 2021 : Everything They Announced!

AMD @ Computex 2021 : Everything They Announced!

AMD unveiled a WHOLE SLEW of new products and technologies at Computex 2021!

Here is a recap of everything AMD announced at Computex 2021!

 

AMD @ Computex 2021 : Everything They Announced!

On 1 June 2021, AMD kicked off their virtual Computex 2021 press conference, announcing a whole slew of new products and technologies!

Let’s take a look at what AMD announced at Computex 2021!

3rd Gen EPYC Beats Intel Xeon Scalable!

AMD President and CEO Dr. Lisa Su showed how the dual-socket 3rd Gen EPYC 7763 system beat the dual-socket Intel Xeon Platinum 8380 system by 50%!

  • 2P Intel Xeon Platinum 8380 delivered 201,334 Java operations per second
  • 2P AMD EPYC 7763 delivered 301,297 Java operations per second

This is because the EPYC 7763 has far more cores than the Xeon Platinum 8380 – 64 cores versus just 40 cores!

AMD RDNA 2 Coming To Automotive + Mobile!

AMD is working to bring the RDNA 2 gaming architecture to industry leaders in the automotive and mobile segments :

Tesla Model S + Model X

The new infotainment systems in the Tesla Model S and Model X are powered by an AMD Ryzen Embedded APU, with an AMD RDNA 2-based GPU capable of AAA gaming!

Samsung Exynos

The next-generation Samsung Exynos SoC will feature custom AMD RDNA 2-based graphics IP, that will introduce raytracing and variable rate shading capabilities to flagship mobile devices!

AMD Ryzen 5000G Desktop APUs

Dr. Lisa Su also announced that they are introducing two AMD Ryzen 5000G series desktop APUs, that will be available to the DIY market starting 5 August 2021.

AMD Ryzen 7 5700G

The AMD Ryzen 7 5700G has an 8-core, 16-thread processor, with a base clock of 3.8 GHz and a boost clock of 4.6 GHz.

It has 4 GB of L2 cache, and 16 GB of L3 cache; and 8-core integrated Radeon graphics clocked at 2.0 GHz.

This 65 watt desktop APU will have a launch price of US$359.

AMD Ryzen 5 5600G

The AMD Ryzen 5 5600G has an 6-core, 12-thread processor, with a base clock of 3.9 GHz and a boost clock of 4.4 GHz.

It has 3 GB of L2 cache, and 16 GB of L3 cache; and 7-core integrated Radeon graphics clocked at 1.9 GHz.

This 65 watt desktop APU will have a launch price of US$259.

AMD Ryzen PRO 5000 Series Desktop CPUs

While not mentioned in the main Computex 2021 press conference, AMD also launched the Ryzen PRO 5000 series desktop processors based on the Zen 3 core!

AMD Ryzen PRO 5000 G-Series (65W)

  • Ryzen 7 5750G : 8 core, 16 thread, 3.8 GHz to 4.6 GHz, 4 MB L2, 16 MB L3
  • Ryzen 5 5650G : 6 core, 12 thread, 3.9 GHz to 4.4 GHz, 3 MB L2, 16 MB L3
  • Ryzen 3 5350G : 4 core, 8 thread, 4.0 GHz to 4.2 GHz, 2 MB L2, 8 MB L3

AMD Ryzen PRO 5000 GE-Series (35W)

  • Ryzen 7 5750GE : 8 core, 16 thread, 3.2 GHz to 4.6 GHz, 4 MB L2, 16 MB L3
  • Ryzen 5 5650GE : 6 core, 12 thread, 3.4 GHz to 4.4 GHz, 3 MB L2, 16 MB L3
  • Ryzen 3 5350GE : 4 core, 8 thread, 3.6 GHz to 4.2 GHz, 2 MB L2, 8 MB L3

You can get the full details in this official AMD deep dive into the Ryzen PRO 5000 series desktop processors!

AMD Radeon RX 6000M Series Mobile Graphics

Scott Herkelman, Vice President & General Manager of AMD Graphics Business Unit, officially announced the introduction of the Radeon RX 6000M series mobile graphics for laptop gaming.

AMD Radeon RX 6800M

Their flagship model has a 2.3 GHz game clock and 12 GB of GDDR6 memory, and is designed to deliver 120 fps gaming at 1440p.

AMD says that it will offer significantly better performance than the competition :

  • 40% to 70% better gaming performance than the GeForce RTX 2070 at 1440p Max Settings.
  • 14% to 40% better gaming performance than the GeForce RTX 3080 while gaming on battery.

AMD Radeon RX 6700M

This model has a 2.3 GHz game clock and 10 GB of GDDR6 memory, and is designed to deliver 100 fps gaming at 1440p.

AMD Radeon RX 6600M

This model has a 2.177 GHz game clock and 8 GB of GDDR6 memory, and is designed to deliver 100 fps gaming at 1080p.

AMD FidelityFX Super Resolution (FSR)

FidelityFX Super Resolution, or FSR for short, is AMD’s spatial upscaling technology that is designed to boost frame rates, while delivering a high-quality gaming experience.

Think of it as AMD’s version of NVIDIA’s DLSS. But while DLSS is proprietary, FidelityFX Super Resolution is open-source, and will be supported on both AMD processors and GPUs as well as NVIDIA GPUs!

More than 10 game developers plan to integrate FSR into their top titles and gamer engines in 2021.

AMD FidelityFX Super Resolution (FSR) will be available starting 22 June 2021.

AMD Advantage Gaming Laptops

Think of AMD Advantage as a marketing programme, based around a “design framework” of premium gaming features.

  • Amplified Performance : AMD Ryzen processor + AMD Radeon RX 6800M graphics + AMD Radeon Software for gaming at 100 fps at 1080p
  • Premium Displays : 144Hz+ with LFC, low latency, AMD FreeSync Premium, 300 nits+ brightness, IPS / OLED panels, 100% sRGB / 72% NTSC minimum
  • Built To Game : NVMe SSD, less than 40°C WASD keyboard temperatures when gaming, >10 hours binge watching on battery

AMD 3D Chiplet Technology

AMD continues to build on their packaging technologies with their new 3D Chiplet technology, developed in collaboration with TSMC.

It enhances their existing chipset architecture with 3D stacking using a hybrid bond approach that provides over 200X the interconnect density of 2D chipsets and more than 15X the density of existing 3D packaging solutions.

This new 3D chiplet technology also consumes less energy than current 3D solutions, and is the most flexible active-on-active silicon stacking technology.

In Computex 2021, Dr. Lisa Su showed off the 3D vertical SRAM cache bonded to an AMD Ryzen 5000 Series processor prototype.

AMD is on-track to begin production on high-end computing products with 3D chiplets by the end of 2021.

 

Help Support My Work!

If you would like to support my work, you can do so via bank transfer /  PayPal / credit card.

Name : Adrian Wong

Credit Card / Paypal : https://paypal.me/techarp
Bank Transfer : CIMB 7064555917 (Swift Code : CIBBMYKL)

Thank you in advanced! 

 

Recommended Reading

Go Back To > Computer | GamingTech ARP

 

Support Tech ARP!

If you like our work, you can help support us by visiting our sponsors, participating in the Tech ARP Forums, or even donating to our fund. Any help you can render is greatly appreciated!


Apple M1 Gaming : Watch It Run The Witcher 3!

Windows gaming on the ARM-based Apple M1 is possible!

Watch how well the Apple M1 runs The Witcher 3 using CrossOver 20!

 

Apple M1 Gaming : Watch It Run The Witcher 3 On CrossOver 20!

AppleGamers successfully ran The Witcher 3 on CrossOver 20, using the 2020 Apple MacBook Pro with the new ARM-based Apple M1 SoC.

Here was their recorded gameplay of The Witcher 3 at the 1080p resolution, with the Medium graphics and post-processing presets, and VSync enabled.

They could not get the FPS counter to work, but it looks smooth and very playable, albeit with some visual artefacts.

Basically – Apple M1 gaming is not only possible, it is possible to play Windows games!

 

Windows Gaming On Apple M1 : Rosetta 2 + Windows API Translation!

That quick gameplay showcase of The Witcher 3 shows that Windows gaming is possible on the Apple M1.

The ability to run The Witcher 3 on CrossOver 20, and at such playable frame rates, is important for two reasons.

Windows App Compatibility

The Witcher 3 is a 64-bit Windows-only game, so you wouldn’t expect it to run well on the ARM-based Apple M1, right?

Running it on the Apple M1 involves translating Windows API calls using CrossOver, and then translating x86 instructions to ARM instructions using Rossetta 2.

Yet it not only worked, it ran pretty well on the Apple M1!

Performance

While AppleGamers was not able to obtain a frame rate, The Witcher 3 appears to run pretty well at the 1080p resolution.

That shows that the Apple M1’s integrated 8-core GPU is fast, and will have no problem handling native ARM games at 1080p, and likely 1440p as well.

 

Apple M1 : A Quick Primer

The Apple M1 is the first ARM-based SoC (System on a Chip) designed by Apple for use in Mac computers.

Packing 16 billion transistors, it is the first chip to be manufactured on the new 5 nm TSMC process technology.

It comes with 4 high-performance Firestorm CPU cores, 4 power-efficient Icestorm cores, an 8-core GPU, and a 16-core Neural Engine.

As the M1 is based on the ARM architecture, you can natively run iPhone and iPad apps on it. However, existing macOS apps will have to be ported over, or translated on-the-fly using Rosetta 2.

It is currently available in these Apple Mac computers :

  • 2020 MacBook Air : US | UK | AU | MY | SG
  • 2020 MacBook Pro 13-inch : US | UK | AU | MY | SG
  • 2020 Mac mini : US | UK | AU | MY | SG

 

Recommended Reading

Go Back To > Gaming | Computer | Home

 

Support Tech ARP!

If you like our work, you can help support us by visiting our sponsors, participating in the Tech ARP Forums, or even donating to our fund. Any help you can render is greatly appreciated!


Fact Check : AMD Ryzen 4000 Will Use 5nm TSMC Process?

Is it true that the upcoming AMD Ryzen 4000 (Zen 3) desktop processors will be fabricated on the 5nm TSMC process, instead of 7nm as announced? Let’s find out!

 

AMD Ryzen 4000 : Zen 2 | Zen 3

At CES 2020, AMD launched the Ryzen 4000 family of mobile processors, which only just kicked off after delays due to the COVID-19 pandemic.

AMD is expected to introduce the Ryzen 4000 family of desktop processors, built on a newer Zen 3 microarchitecture, later this year.

AMD’s naming convention confuses people, because the Ryzen 4000 mobile processors are 3rd Gen Ryzen processors, while the upcoming Ryzen 4000 desktop processors will be 4th Gen Ryzen processors.

To make it easier for you to understand, we created this table to differentiate the two processor families.

Specifications Ryzen 4000
Desktop
Ryzen 4000
Mobile
Ryzen Generation 4th Gen 3rd Gen
Microarchitecture Zen 3 Zen 2
Fabrication Process Improved 7 nm 7 nm
Launch Date Late 2020 Jan 2020

 

DigiTimes : AMD Ryzen 4000 Will Use 5nm TSMC Process

A DigiTimes article, purportedly translated and posted by RetiredEngineer, claimed that the next-generation AMD Ryzen 4000 desktop processors, using the Zen 3 microarchitecture, will be fabricated on the brand-new 5nm process technology, instead of 7nm as announced earlier.

Rumor has it that AMD and TSMC have adjusted their foundry blueprints. The Ryzen 4000 series of desktop processors originally expected to launch at the end of 2020 will now use TSMC’s enhanced 5nm process (5nm Plus) instead of 7nm EUV, clearly demonstrating that AMD is now considered a tier-one customer for TSMC.

This new development confirms earlier news that TSMC, which only recently entered the 5nm era in April, is accelerating its enhanced 5nm process, bringing forward the mass production schedule to 4Q; and the enhanced 5nm Ryzen series processors will also put unprecedented pressure on Intel. Both TSMC and AMD declined to comment on rumors.

In the first quarter of 2020, AMD benefitted from the pandemic, which boosted demand due to the work-from-home economy. PC and server sales exceeded expectations; Ryzen and EPYC series processors shipments were strong, driving first quarter revenue up by 40% compared to the same period in 2019. Earnings were also better than market expectations.

Despite AMD’s conservative outlook for Q2, with revenue estimates between $1.75B and $1.95B, it is still up by 21% compared with the same period in 2019, and up 4% sequentially, still a good result despite the traditional low season for PCs and the impact of the pandemic.

As understood, AMD’s desktop and server shipments have strengthened across the board since the second half of 2019, and demand has exceeded market expectations. Even AMD themselves and TSMC were caught by surprise. This has led to significant improvements in AMD’s profitability. TSMC happily welcomes the growing strength of its chip customers.

According to sources in the semiconductor industry, AMD had announced a comprehensive partnership with TSMC for sub-7nm processes, and has also confirmed mass production schedules for 7nm, 7nm EUV, 5nm and 3nm products. However, beyond expectations, AMD’s performance took a Great Leap Forward over the last one and a half years. Board, PC and server manufacturers significantly increased their ‘weightage’ (adoption of AMD products relative to others), in part due to Intel’s chip shortages. More importantly, the partnership with TSMC resulted in drastically improved product performance and yield, as well as price competitiveness for AMD, leading to rising shipments and market share quarter after quarter.

In 2020, because of the huge increase in sales, AMD has been urgently chasing after TSMC to expedite orders. The size of those orders were not small, becoming the greatest fallback for TSMC, which got caught in the Huawei ban crisis, making demand and production planning extremely difficult. As a result, TSMC is also adjusting (broadening) its services for AMD, and in considering the optimum allocation of production capacity, modified the process plans for AMD’s products.

As understood, TSMC has entered the 5nm era since April. The enhanced version of 5nm will also enter mass production in Q4, ahead of market expectations, and the first customer to adopt it is AMD, with its new Ryzen 4000 series processors.

AMD’s new generation Ryzen 4000 series processor (codenamed Vermeer), originally planned to use 7nm EUV, will be unveiled around Sep-Oct, but in line with the mass production schedule of TSMC’s enhanced 5nm process, will only be launched at the end of the year or during CES in Jan 2021.

According to semiconductor industry players, AMD plonked down a lot of money to enter the enhanced 5nm era, hoping to build upon its success and expand its leadership in advanced process technology before Intel’s 10nm process is fully deployed and 7nm process is yet to debut, once again capitalising on the narrative around introduction of advanced process technology and performance/efficiency improvements, to narrow its market share with Intel.

Judging by AMD’s current momentum and product performance, the enhanced 5nm Ryzen series processors will put unprecedented pressure on Intel. If Intel does not cut prices or accelerate the introduction of 10nm desktop processors, its empire may be gradually eroded by AMD, ushering in the biggest change in the global PC platform competitive landscape in 15 years: AMD’s market share can be expected to reach historical highs.

Separately, Nvidia has finally joined the 7nm bandwagon recently, entering full production in the second half of the year, while 5nm capacity continues to be reserved for Apple and HiSilicon, although the status of HiSilicon’s orders after Q4 is unknown due to the heightened US ban on Huawei.

Now, we have not directly seen the DigiTimes article, so we cannot vouch for its authenticity. We only have this picture to go by :

 

5 Reasons Why AMD Ryzen 4000 Will NOT Use 5nm TSMC Process

We have no idea whether the DigiTimes article exists, or was translated accurately, but we seriously doubt it will happen. Here’s why…

Reason #1 : AMD Unlikely To Delay Ryzen 4000 Desktop Launch To 2021

AMD CEO Dr. Lisa Su confirmed at CES 2020 that Zen 3 will debut in 2020, most likely around October to make it in time for the year-end holiday season.

While the COVID-19 pandemic has delayed the consumer launch of the Ryzen 4000 mobile processors, it is unlikely to delay the launch of the Ryzen 4000 desktop processors until early 2021.

Reason #2 : Desktop Ryzen Processors Don’t “Need” 5nm

While shrinking die size is generally a good thing, desktop processors like the upcoming 4th Gen Ryzen 4000 CPUs do not “need” a die shrink.

The focus will be on achieving high clock speeds at lower costs. That’s why the 3rd Gen Ryzen 3000 desktop processors use a chiplet design – the CPU dies are fabricated on 7nm, while the I/O die is fabricated on 12nm.

A matured and improved 7nm process would allow the 4th Gen Ryzen 4000 processors to deliver higher clock speeds at a much lower cost than a new 5nm process.

Reason #3 : Mobile Ryzen Processors Will Likely Be First To Use 5nm

AMD is strongest in the desktop and HEDT segment, outperforming Intel by sheer brute force, thanks to the higher number of cores in their desktop Ryzen and Threadripper processors.

They are weakest in the mobile market, with Intel mobile processors still controlling the vast majority of the market.

To seriously take on Intel in this critical segment, AMD would probably leverage the costly 5nm process to give their future Ryzen 5000 mobile processors a significant advantage in both performance and power consumption.

Reason #4 : 5nm Not That Advantageous Over Improved 7nm

While a 5nm transistor node sounds like it will offer 28.5% smaller transistors than a 7nm node, the difference in reality is much smaller.

That’s partly because the transistor sizes – 7nm, 5nm – are really marketing terms, not precise engineering definitions – and partly because of diminishing returns.

In any case, the TSMC 5nm process promises to offer 25% better performance over 7nm, but their improved 7nm process will offer at least 10% better performance over 7nm.

In other words, the net performance difference between the TSMC 5nm and improved TSMC 7nm (N7+) is just 13.6%. The cost of 5nm would, no doubt, be far more than improved 7nm.

Reason #5 : 5nm Yields Are Still Poor

Another thing to consider with a new transistor node is yield. This is the problem Intel had with their 10nm process technology – poor yield.

As of December 2019, the 5nm TSMC process has an average yield per wafer of ~80% with a tiny die size of 17.92 mm2. That yield goes down to an abysmal 32% with a 100 mm2 die size.

While no one knows how big the Zen 3 die will be, it is safe to say it will be closer to 100 mm2. The Zen 2 die, for example, is 74 mm2 in size.

Will AMD risk the success of their 4th Gen Ryzen 4000 desktop processors on TSMC hitting reasonable yields on 5 nm? We think not!

 

Recommended Reading

Go Back To > Computer Hardware | Business | Home

Support Tech ARP!

If you like our work, you can help support our work by visiting our sponsors, participating in the Tech ARP Forums, or even donating to our fund. Any help you can render is greatly appreciated!


The Kirin 950 & Kirin 650 Technology Report

In our Kirin 950 and Kirin 650 technology report, you will be able to watch the full technical briefing and check out the performance of the Kirin 955 compared to the Exynos 8890. We will also share with you the presentation slides for your perusal. Enjoy!

 

Kirin 950 & Kirin 650 Technical Briefing

We were invited to an exclusive briefing on the unique features of the Kirin 950 and Kirin 650 family of processors (more accurately, System on a Chip or SOCs). The briefing was conducted by Mr. Zhou Chen, the Director of the Kirin Chipset Solution Planning, HiSilicon Wireless Terminal BU of Huawei.

Mr. Zhou Chen and his colleague also demonstrated the performance advantage of the Kirin 955-powered Huawei P9 Plus smartphone against the Exynos 8890-powered Samsung Galaxy S7 edge smartphone. Check it out :

[adrotate banner=”5″]

 

Key Features Of The Kirin 950 Family Of SoCs

  • Manufactured on the TSMC 16 nm FinFET+ process technology
  • Processor : 4 x ARM Cortex-A72 cores (2.3 GHz to 2.5 GHz) + 4 x ARM Cortex-A53 cores (1.8 GHz)
  • Graphics Processor : ARM Mali-T880 GPU (900 MHz)
  • Motion Co-Processor : i5 sensing co-processor
  • Memory Support : LPDDR3 or LPDDR4 SDRAM
  • Dual 14-bit ISPs (960 MP/s) with standalone DSP
  • HiFi audio DSP
  • 4K video decoder (supports H.265)
  • 5 Mode LTE Cat6 modem, with RF band of 450 MHz ~ 3.5 GHz

 

Key Features Of The Kirin 650 Family Of SoCs

  • Manufactured on the TSMC 16 nm FinFET+ process technology
  • Processor : 4 x ARM Cortex-A53 cores (2.0 GHz) + 4 x ARM Cortex-A53 cores (1.7 GHz)
  • Graphics Processor : ARM Mali-T830 MP2 GPU (900 MHz)
  • Motion Co-Processor : i5 sensing co-processor
  • Memory Support : LPDDR3 SDRAM
  • LTE Cat7 modem, with support for VoLTE and pseudo base station protection
  • Features SPLC intelligent voice enhancement technology

Next Page > Kirin 950 & Kirin 650 Briefing Slides

 

Support Tech ARP!

If you like our work, you can help support our work by visiting our sponsors, participate in the Tech ARP Forums, or even donate to our fund. Any help you can render is greatly appreciated!

Kirin 950 & Kirin 650 Briefing Slides

Huawei has kindly provided us with the slides for their Kirin 950 and Kirin 650 technical briefing, which we sharing with you here. They should allow you to have a clearer picture of the video presentation above.

[adrotate banner=”5″]

 

Support Tech ARP!

If you like our work, you can help support our work by visiting our sponsors, participate in the Tech ARP Forums, or even donate to our fund. Any help you can render is greatly appreciated!