#Hotchips
LightNews LightNews
flow-computing.bsky.social
Different workloads need different compute. @ #HotChips the #Intel presentation showed how #CPU #GPU & #NPU support each other. For us, that is central to CPU evolution. We're not replacing it. We're adding complementary capabilities alongside it. Learn more: flow-computing.com/news/hot-chi...
#HPC
Graphic for Flow Computing’s Hot Chips 2026 mini series on Intel Wildcat Lake. Text reads: “Different workloads need different compute. CPU, GPU & NPU don’t need to do the same job. They need to do the right job. That’s heterogeneous computing.” Flow and Hot Chips logos appear at the bottom.
September 24, 2026 at 8:10 AM
raendragon.bsky.social
yeah i'm so #girlbossing and eating my #hotchips ngl bestie
September 8, 2026 at 9:34 AM
pedro.vza.net
Ahem x.com/cdleary/arti...
Compilers 2.0: AI as stochastic optimizer
There has been a lot of discussion following the presentation of the Jalapeño MLA kernel at HotChips and subsequent commentary by SemiAnalysis. As OpenAI’s hardware team, we just barely touched on
x.com
September 3, 2026 at 3:05 AM
flow-computing.bsky.social
#HotChips2026 may be over, but the conversation isn’t. 👀 We’ll be taking a closer look at some #CPU presentations from this year + sharing our perspective on where CPU architecture & performance are heading. Coming soon. For now, see at the poster we shared: flow-computing.com/hotchips-pos...
#HPC
Flow Computing graphic announcing an upcoming Hot Chips 2026 content series. Flow and Hot Chips logos appear at the top above the headline “Hot Chips. Our take.” Text explains that over the next few weeks, Flow will examine selected CPU presentations from Hot Chips 2026 and share its perspective on what they mean for the future of CPU performance. A call to action invites readers to explore Flow’s Hot Chips poster.
September 1, 2026 at 1:48 PM
pulseofnations.lol
First benchmarks show OpenAI's custom inference chip delivers 1.5-1.9x the performance per watt of Nvidia's flagship at Hot Chips 2026

#AiChip #Broadcom #HotChips #Inference #Jalapeno #Nvidia #OpenAI #Semiconductor
OpenAI Jalapeño Chip Beats Nvidia on Efficiency
First benchmarks show OpenAI's custom inference chip delivers 1.5-1.9x the performance per watt of Nvidia's flagship at Hot Chips 2026
pulseofnations.lol
August 27, 2026 at 2:37 PM
flow-computing.bsky.social
We’ve had great conversations & we’re looking forward to more before we head back home. If you didn’t get a chance to see our poster, you can still look here: flow-computing.com/hotchips/ Thank you, #HotChips, and thank you to everyone who made our time here worthwhile.🦾
#CPU #HPC #DeepTech #FlowPPU
August 27, 2026 at 9:07 AM
flow-computing.bsky.social
That’s a wrap of the #HotChipsSymposium! We’ve been in the US for 2 weeks exchanging ideas, & talking about the future of #CPU performance. #HotChips was a highlight. Thank you to everyone who to talk w/ us, saw our poster, or took the time to meet w/ us, & to Fudo for being part of the journey.
Black-and-white selfie of three Flow Computing team members wearing Flow shirts and sunglasses in front of Hoover Tower at Stanford University, with dramatic rippled clouds overhead. Black-and-white selfie of three Flow Computing team members wearing Flow shirts and Hot Chips conference badges outside a Stanford University building. A Hot Chips sign is partially visible in the foreground.
August 27, 2026 at 9:04 AM
pulseofnations.lol
Diamond Rapids brings 256 cores, PCIe Gen6 and drops hyperthreading, but Intel pushes launch back as it targets HPC and agentic AI workloads.

#AiCompute #DiamondRapids #HotChips #Hpc #Intel #PcieGen6 #Server #Xeon
Intel Unveils 256-Core Xeon 7 at Hot Chips, Delays Launch to 2027
Diamond Rapids brings 256 cores, PCIe Gen6 and drops hyperthreading, but Intel pushes launch back as it targets HPC and agentic AI workloads.
pulseofnations.lol
August 26, 2026 at 10:20 PM
instlatx64.bsky.social
#Intel released the HotChips 2026 conference #CrescentIsland presentation:
cdrdv2-public.intel.com/926893/Cresc...
August 26, 2026 at 5:42 PM
pcmasterinsider.bsky.social
Intel unveiled its new Crescent Island AI accelerator architecture at Hot Chips, focusing on massive FLOPS-per-watt efficiency.

https://pcmasterinsider.com/intel-crescent-island-ai-accelerator-hot-chips/

#Intel #HotChips #AIAccelerator #Hardware
Intel Details Crescent Island AI Accelerator for High-Efficiency Workloads
Intel unveiled its Crescent Island AI accelerator at Hot Chips, delivering high FLOPS-per-watt performance for edge and inference tasks.
pcmasterinsider.com
August 26, 2026 at 12:03 PM
instlatx64.bsky.social
#Intel released the HotChips 2026 conference #DiamondRapids presentation:
cdrdv2-public.intel.com/926846/diamo...
#APX #AMX #AVX10 #PantherCove
August 26, 2026 at 9:47 AM
pulseofnations.lol
OpenAI debuts benchmark results for its first custom inference chip, showing 1.5-1.9x more AI work per watt than Nvidia's latest hardware.

#Broadcom #Chip #CustomSilicon #HotChips #Inference #Jalapeno #Nvidia #OpenAI #Tsmc
OpenAI Jalapeño Chip Beats Nvidia on Efficiency at Hot Chips 2026
OpenAI debuts benchmark results for its first custom inference chip, showing 1.5-1.9x more AI work per watt than Nvidia's latest hardware.
pulseofnations.lol
August 26, 2026 at 8:18 AM
hnws.bsky.social
HotChips 2026 Samsung LPDDR5X-PIM (processing-in-memory)
comments · posted on 2026.08.25 at 13:21:12 (c=1, p=4)
August 25, 2026 at 5:57 PM
news.karthihegde.dev
HotChips 2026 Samsung LPDDR5X-PIM (processing-in-memory)
Discussion | hackernews | Author: cmrdporcupine

#SystemsProgramming
HotChips 2026 Samsung LPDDR5X-PIM (processing-in-memory)
Samsung has a new generation of its LPDDR5X-PIM solution at Hot Chips 2026. This processing in memory solution looks more polished this generation
www.servethehome.com
August 25, 2026 at 5:57 PM
hn100.bsky.social
HotChips 2026 Samsung LPDDR5X-PIM (processing-in-memory)

Discussion
August 25, 2026 at 5:54 PM
cevregazetesi.bsky.social
IBM, Z ve LinuxONE İçin Çift Mimarili Yeni Nesil İşlemciyi Tanıttı

IBM, hem IBM Z hem de LinuxONE sistemlerinde çalışacak ilk çift mimarili ana bilgisayar işlemcisini tanıttı.

yalovacevre.com/ibm-z-ve-lin...

#IBM #HotChips #ARM #zArchitecture #AArch64Architecture #Linux
August 25, 2026 at 9:20 AM
jangray.bsky.social
#HotChips
ZFLOPS/100MW, a bold new hyperbole unit for the "AI Factory" (unfortunate term)
Table in an NVIDIA Rubin slide: Specification 100 MW AI Factory* NVFP4 Inference 2 ZFLOPS NVFP4 Training 1.4 ZFLOPS HBM4 Memory 11 PB HBM4 Memory Bandwidth 800 PB/s
August 25, 2026 at 12:00 AM
jamesaltonsanders.com
I managed to get a selfie with Satoshi Matsuoka, the director of RIKEN (and also the co-programmer of Pinball on the NES) at #HotChips!
August 24, 2026 at 9:22 PM
mobiledom.eurosky.social
oh before i forget @jangray.bsky.social if you happen to run into Dr. Ian Cutress at HotChips, thats my boss. he's usually pretty happy to take selfies.
August 24, 2026 at 9:18 PM
computerbase.de
Intel Diamond Rapids: 256 Kerne, 16 CPU-Tiles, vier Cache-Dies und zwei I/O-Tiles #Intel #DiamondRapids #Hotchips
Intel Diamond Rapids: 256 Kerne, 16 CPU-Tiles, vier Cache-Dies und zwei I/O-Tiles
Intel enthüllt Diamond Rapids: 256 Kerne werden durch 16 CPU-Tiles in 18A-P realisiert, die auf vier Cache-Base-Tiles in Intel 3-T sitzen.
www.computerbase.de
August 24, 2026 at 5:15 PM
xchatter.techmeme.com
This tweet appeared under this Techmeme headline:

Ben Pouladian / @benitoz:

HotChips warm up: NVIDIA just posted the slide. Groq 3 LPX. 3,431 tokens a second. 4x the next public endpoint. Gemma 4 31B at 100,000-token context. In full production. Faster than Cerebras 😮🔥 $NVDA
August 24, 2026 at 4:22 PM
pulseofnations.lol
Stanford symposium hosts simultaneous reveals of NVIDIA Rubin, AMD MI400 and Intel Crescent Island as AI chip race intensifies

#AiGpu #AmdMi400 #HotChips #Intel #NvidiaRubin #Semiconductors #Stanford
NVIDIA, AMD, Intel Unveil New GPUs at Hot Chips
Stanford symposium hosts simultaneous reveals of NVIDIA Rubin, AMD MI400 and Intel Crescent Island as AI chip race intensifies
pulseofnations.lol
August 24, 2026 at 2:53 PM
jangray.bsky.social
#FPGA #HotChips #OoO #EDGE
This just came up again. Good times!
arxiv.org/pdf/1803.06617
Floorplan of an FPGA-efficient “parallel” OoO instruction scheduler. Floorplan of an FPGA-efficient OoO instruction window including teo “incremental” dataflow instruction schedulers.
August 24, 2026 at 5:23 AM
jamesaltonsanders.com
On my way to #HotChips
August 22, 2026 at 8:16 PM