You are currently viewing The AMD Computex 2024 Keynote Live Blog (6:30pm PT/01:30 UTC)

The AMD Computex 2024 Keynote Live Blog (6:30pm PT/01:30 UTC)

Computex keynote season is kicking into high gear this morning with the show’s leading keynote, which is being delivered by AMD. Company CEO Dr. Lisa Su will be presenting a keynote entitled “The future of high-performance computing in the AI era,” and with a run time of 90 minutes, we’re expecting AMD to have a whole host of product announcements covering their full spectrum of product categories.

The big expectation here is fresh news around AMD’s Zen 5 CPU core architecture, and the chips built around it. AMD’s most recent Zen 5 roadmap has it slated to deliver all three flavors of Zen 5 by the end of this year, and we’re coming up on the two-year anniversary of the Zen 4 architecture launch.

Along with client chips, AMD has been pushing their server CPUs hard, and they’ve previously told investors that the next-gen EPYC Turin CPU is “looking great”. So we’ll likely hear about both client and server Zen 5 product plans during this keynote.

On the GPU/accelerator side of matters, AMD is mid-cycle (at best) with their Instinct MI300 series accelerators. With the company’s sales repeatedly beating their own expectations, AMD doesn’t seem to need much help moving this premium silicon right now. But with AI being the operative buzzword of this year’s Computex (and indeed, the computing industry as a whole), it would be weird for AMD to not have something to say about their rapidly growing AI accelerator product line.

09:30PM EDT – Our roving reporter Gavin Bonrshor is at this year’s show soaking in a wealth of information (not to mention the infamous Taipei humidity). He won’t be short of things to do over the next few days

09:31PM EDT – And if you needed any guesses as to what the theme is for this keynote, AMD’s logo should tell you everything you need to know

09:37PM EDT – Hello all, and we have a preview of what to expect to Computex prior to the opening keynote being delivered by Dr. Lisa Su

09:42PM EDT – We’re still waiting on TAITRA’s Computex preview speech to wrap up. AMD’s live stream is also on hold until then

09:43PM EDT – Currently watching a pre recorded video, although it is hard to tell if it is AI or actually real (a nod to generative AI)

09:46PM EDT – And the AMD keynote is close to kicking off, just a quick introduction before Dr. Lisa Su joins the stage

09:47PM EDT – And straight into a video highlighting AMD’s commitment to advancing AI

09:47PM EDT – And the moment we have been waiting for, Dr. Lisa Su has entered the stage to open Computex 2024

09:49PM EDT – Dr. Lisa Su looking vibrant in her red/pink blazer, perhaps a sign of plenty of vibrant products

09:49PM EDT – Currently recapping AMD’s varied range of solutions including computing, but their focus has quickly shifted to AI

09:50PM EDT – AMD’s 3 priorities,broad portfolio of training and inference compute engines

09:50PM EDT – Open and proven soiftware capability

09:50PM EDT – And third, an AI ecosystem with deep co-innovation

09:51PM EDT – Introducing a number of new products, including the Zen 5 core, XDNA 3.5 for mobile

09:51PM EDT – And first up, gaming PCs. Not just desktops, but consoles and handhelds

09:52PM EDT – Introducting AMD’s Ryzen 9000 series for desktop. PCIe 5.0 and DDR5 inclusive

09:52PM EDT – AMD Zen 5 has a few updates over Zen 4, including a dual pipeline front end, improving branch prediction accuracy

09:53PM EDT – Also Zen 5 has a wider pipeline, although AMD hasn’t disclosed the finer details, these can be expected closer to launch

09:53PM EDT – Dr. Lisa Su is claiming 16 % gains in IPC, and Lisa is showing off the Ryzen 9 9950X, 16C/32T, 5.7 GHz Boost, and at a 170 W TDP

09:54PM EDT – Ryzen 9 9950X has a 64 MB cache, much like Zen 4’s Ryzen 9 7950X, as well as 1 MB of L2 cache per core

09:54PM EDT – Some basic performance figures from AMD, but no actual data to go with it

09:55PM EDT – Dr. Lisa Su is quicky recapping the AM4 platform, and boosting support for AM5 into 2027, instead of the previous 2025 timeline

09:56PM EDT – The Ryzen 9 9950X (16C/32T), Ryzen 9 9900X (12C/24T), Ryzen 7 9800X (8C/16T) and Ryzen 5 9600X (6C/12T), coming in July

09:56PM EDT – AMD currently switching to their mobile portolio. ALso introducing the 3rd Gen Ryzen AI NPU, developed with Xilinx

09:57PM EDT – Introducting Strix Point, the Ryzen AI 300 series, new nomenclature, with XDNA 3.5, Zen 5, 3rd Gen NPU, up to 50 TOPS

09:58PM EDT – Note: That’s 5 TOPS more than the Qualcomm Snapdragon X Elite chip

09:58PM EDT – Ryzen AI 9 HX 370 with 12C/24T of Zen 5 cores

09:59PM EDT – Also the Ryzen AI 9 365 which has 10C/20T with Zen 5, RDNA 3.5

10:00PM EDT – AMD’s new third generation NPU looks to bring a new AI floating point data type, the Block FP16, the speed of INT8 at the accuracy of FP16, without quantization penalties

10:01PM EDT – Generative AI is a huge market, so no surprises AMD is going all out to stand out above Inten and Qualcomm. Block FP16 could be a pivotal move in Gen AI performance

10:01PM EDT – 3 cakes, only 2 look edible

10:02PM EDT – AMD also touching more on their Microsoft Copilot + partnership. Enter Pavan Davuluri from Microsoft to the stage

10:03PM EDT – Microsoft just recapping their commitment to leveraging AMD’s new 3rd Gen Ryzen AI NPU within the Ryzen AI 300 series for mobile

10:03PM EDT – And a quick video from Microsoft and Copilot+

10:05PM EDT – All focusing on the NPU, which until now, has been pretty mediocre

10:05PM EDT – Microsoft just recapping their software using the NPU, including Recall, Cocreator and Live Captions

10:06PM EDT – Microsoft highlighting their own runtime, and comitting to having responsible fundamentals surrounding AI

10:07PM EDT – Pavan is essentally saying AI running on-chip and off the cloud is cheaper, which we already knew. The 40 TOPS minimum for Copilot+ has been surpassed by AMD’s 3rd Gen NPU with 50 TOPS

10:08PM EDT – Dr. Lisa Su being pushed to deliver more TOPS, which is easier said than done. Having a minimum standard brings a better overall ecosystem

10:09PM EDT – Getting to 50 TOPS is a “quantum leap”

10:09PM EDT – And back to Dr. Lisa Su, recapping their long list of partners in software, with over 150+ ISVs in 2024

10:10PM EDT – AMD is welcoming HP CEO to the stagem Enrique Lores to the stage

10:11PM EDT – Another conversation about AI, which is the main theme of Computex 2024

10:11PM EDT – The AI PC is the biggest change in computing in decades (according to everyone selling AI PCs)

10:12PM EDT – HP is also one of AMD’s main ecosystem partners in the mobile space, along with the likes of Dell and Lenovo

10:13PM EDT – And HP is introducing a new notebook, the HP Omnibook, with AMD’s Ryzen AI 300, also the first device to have the new chips with the 3rd Gen Ryzen AI NPU

10:13PM EDT – And HP is introducing a new notebook, the HP Omnibook, with AMD’s Ryzen AI 300, also the first device to have the new chips with the 3rd Gen Ryzen AI NPU

10:15PM EDT – And a demo of Stable Diffusioin based Gen AI demo on the HP Omnibook with the new Ryzen AI NPU

10:15PM EDT – Using the new Block FP16 data poiint

10:16PM EDT – No mention of specifications or pricing though. And back to Dr. Lisa Su. AMD comparing Ryzen AI 300 to Intel, Apple M3, and Qualcomm Snapdragon X Elite

10:16PM EDT – No mention of specifications or pricing though. And back to Dr. Lisa Su. AMD comparing Ryzen AI 300 to Intel, Apple M3, and Qualcomm Snapdragon X Elite

10:17PM EDT – And bringing Luca Rossi to the stage, President of IDG at Lenovo. Expect a lot of vendors and partners to come on stage announcing new notebooks with Ryzen AI 300 series SoCs

10:18PM EDT – Luca just recapping their partnership with AMD, including Threadripper. But today is about Ryzen 9000 and Ryzen AI 300 mobile, with the latter being the key focal point today. Zen 5

10:18PM EDT – Wiating for Luca to bring out a Lenovo laptop with Ryzen AI 300, he’s bound to!

10:19PM EDT – The Lenovo Yoga with Ryzen AI 300, coming later this year, but no time frame given. For commercial, SMEs with a Thiinkbook. No specific models announced, just the series

10:20PM EDT – Announcing Lenoivo AI Now, additional software leveraging the Ryzen AI NPU

10:21PM EDT – More Generative AI features, which is the main use for AI on chip within the AI PC

10:22PM EDT – Lenovo also announcing Smart Connect, bringing the Lenovo AI Ecosystem together

10:23PM EDT – Lenovo showing off a new notebook, straight from R+D, the new Yoga Laptop powered by Ryzen AI 300 series.

10:23PM EDT – No specifications, but promising a significant jump in AI performance from Lenovo devices

10:24PM EDT – Dr. Lisa Su now welcoming her next guest, Jonney Shih, the Chairman of ASUS

10:24PM EDT – Dr. Lisa Su being told she’s the ‘Pride of Taiwan’

10:25PM EDT – ASUS recapping their use of AMD’s Ryzen AI 3rd gen NPU. This is likely to be a common theme for the next 10-15 minutes. Showcasing AI use cases from AMD’s close partners

10:26PM EDT – Ubiquitous AI, ASUS’s new phrase for AI powered notebooks

10:27PM EDT – A lot of co-oinnoivation between ASUS and AMD. ASUS at 4pm (Taipei time) is launching new notebooks across a range of their typical families with the 3rd Gen Ryzen AI SoC and NPU

10:27PM EDT – New Vivobook, ProArt, Zenbook, ROG Zepheryus and TUF Gaming notebooks

10:28PM EDT – ASUS notebooks with Ryzen AI 300 series, coming July, also the first to market.

10:29PM EDT – Jonney is so excited, he’s shouting about it, incredible passion from the ASUS chairman

10:29PM EDT – Recapping some ASUS specific features, ASUS StoryCube, driven by AI

10:30PM EDT – Lisa is recapping the AMD and ASUS partnership over the years between the two companies

10:31PM EDT – Jonney mentioning the ROG Ally, the first gaming handheld from ASUS (not the first). ASUS leveraging AMD across multiple portfolios from desktop, to mobile, to server, to handhelds

10:32PM EDT – And back to Dr. Lisa Su, first notebooks coming in July with Ryzen AI 300, with more than 100 models to come over the next few months

10:32PM EDT – Dr. Lisa Su is pivoting to its AI platforms, for research, medical, automotive, ADAS, etc.

10:33PM EDT – AMD XDNA and Adaptive Compute for the Edge, AMD is claiming

10:33PM EDT – to be number 1*

10:33PM EDT – Onto the unsung area of AMD’s portfolio, their Versal AI solutions, including a run down of use cases

10:34PM EDT – “AI at the edge is a hard problem”

10:35PM EDT – AMD with the Versal AI Edge Gen2 series, not new, but AMD is announcing early access to the chip to develop new technologies

10:35PM EDT – And now we come to the data center, with a quick recap of the market, use cases and enterprise

10:36PM EDT – AMD recapping EPYC use cases in the clouds, with partnerships with the world’s largest hyperscalars

10:37PM EDT – EPYC was launced in 2017, and the performance gains from then to now is just incredible over the years

10:37PM EDT – AMD announcing a 33% market share in the data center, it’s growing slower over the last year or so, but it’s still growing

10:37PM EDT – AMD comparing EPYC performance to Intel, with claimed gains of 1.5X (50%)

10:38PM EDT – New solutions can reduce rackspace by 80%

10:39PM EDT – AMD is previewing the 5th Gen of AMD EPYC CPUs, codenamed Turin, including a chip in hand

10:39PM EDT – 192 Cores, 384 threads, 13 chiplets on package, support for the SP5 socket

10:39PM EDT – So Turin can drop into exisiting Genoa platforms

10:40PM EDT – AMD showcasing some preliminary performance figures, 3.1X more performance than Intel in STMV

10:40PM EDT – Rolling a demo of Llama 2, showing off the AI capabilities of AMD EYPC 5th Gen (Turin)

10:41PM EDT – The difference is night and day in AI performance, but EPYC has double the cores

10:41PM EDT – Turin is also the go to for Chatbot, with 5.4x the token generation.

10:41PM EDT – AMD EPYC 5th Gen is on track for 2H 2024

10:42PM EDT – And moving onto Instinct, with the MI300X, recapping partners who adopted Instinct

10:43PM EDT – AMD recapping ROCm 6 software, with improved performance in Meta Llama-3

10:44PM EDT – AMD want to offer choice in the industry, with Open AI with Triton using AMD Instinct

10:45PM EDT – Introducing Stability.ai CEO and CTO, Christian Laforte to the stage

10:45PM EDT – Discussing improving performance using LLMs and new models, designed to create applications faster

10:47PM EDT – Launching Stable Diffusiion 3 to the public, coming June 12th, available for everyone to download

10:48PM EDT – Currently demoing some Stable Diffusion improvements

10:49PM EDT – Suitable fpr beginners and the more advanced

10:50PM EDT – Comparing Stable Diffusion 3 to the previous XL 1.0 version launched last year

10:50PM EDT – Finger on the left, looks badly broken. SD3 looks much better

10:51PM EDT – Upscaling with the MI300X, with 100MP, compared to NVIDIA H100 at 32MP?

10:52PM EDT – Stability.Ai asking for more compute, memory and bandwidth from Dr. Lisa Su.

10:52PM EDT – More Microsoft guests?

10:53PM EDT – AMD is bringing Microsoft chairman Satya Nadella via a pre recorded video

10:53PM EDT – Recapping Ryzen AI 300 with Copilot, and first cloud to enable AMD’s accelerators.

10:54PM EDT – Quick and short. AMD Instinct MI300X is being deployed across the cloud by Microsoft. And rolling a demo of ChatGPT-4 AI Assistant Demo

10:55PM EDT – Who’s giving the opening Keynote? Dr. Lisa Su of AMD of course!

10:55PM EDT – Creating an itinerary fior a trip to Taipei…

10:56PM EDT – AMD introducing the Instinct MI325X, for 2024, refreshing the MI300, also bringing MI350 with CDNA4 in 2025. ALso CDNA Next MI400 coming in 2026

10:57PM EDT – Also leveraging HMB3E memory, with the MI325X, with up to 288 GB of memory, 6TB/s bandwidth

10:57PM EDT – Also uses the same infrastructure of MI300, essentially refreshing the technology for cross platform support

10:58PM EDT – Comparing to H200, 2x the memory, 1.3x the bandwidth

10:58PM EDT – Next year, AMD will announce the MI350 series, with up to 288GB of HBM3e memory, FP4/FP6 support

10:59PM EDT – AMD showcasing huge generational leaps. 35X increase in inference performance to CDBA3

10:59PM EDT – AMD is talking about one more topic, the field of networking and connectivity.

11:01PM EDT – The Ultrs Accelerator Link, partnering for scalable accelerators. Coming later this year

11:01PM EDT – Scaling up and scaling out into the data center, where interconnect and bandwidth is important

11:01PM EDT – AMD is also part of the Ultra Ethernet Consortium, with HPE, Intel, Meta, Oracle, Microsoft among others

11:02PM EDT – UA link combined with the UE Consortium. EPYC with Instinct and UA Link feeding the needs of the data center

11:03PM EDT – Recapping and wrapping up

11:03PM EDT – Ryzen 9000 desktop CPUs coming July, 4 chips, up to 16C/32T of Zen 5

11:03PM EDT – Ryzen AI 300 mobile SoCs, coming July, 3rd Gen Ryzen AI NPU with RDNA 3.5 graphics and Zen 5 of course!

11:04PM EDT – Also Turin EPYC 5th Gen, so much to announce. No technical disclosures however

11:04PM EDT – And that wraps up AMD’s opening keynote. Zen 5 being the focal point, with Ryzen AI 3rd Gen

11:05PM EDT – Thanks for joining us!

11:05PM EDT – And that’s a wrap

11:05PM EDT – And that’s a wrap

Source