So a laptop without the 200Gbps nic and only 64GB of RAM could maybe start at $2.5K?
Or it could all just go down the drain since Nvidia said they're gonna make GPU chiplets for Intel CPUs.
Or it could all just go down the drain since Nvidia said they're gonna make GPU chiplets for Intel CPUs.
November 8, 2025 at 3:01 AM
So a laptop without the 200Gbps nic and only 64GB of RAM could maybe start at $2.5K?
Or it could all just go down the drain since Nvidia said they're gonna make GPU chiplets for Intel CPUs.
Or it could all just go down the drain since Nvidia said they're gonna make GPU chiplets for Intel CPUs.
New technical papers recently added to Semiconductor Engineering’s library
semiengineering.com/chip-industr...
#semiconductor #MRAM #AI #processors #DRAM #chiplets #FeRAM
semiengineering.com/chip-industr...
#semiconductor #MRAM #AI #processors #DRAM #chiplets #FeRAM
November 4, 2025 at 4:21 PM
New technical papers recently added to Semiconductor Engineering’s library
semiengineering.com/chip-industr...
#semiconductor #MRAM #AI #processors #DRAM #chiplets #FeRAM
semiengineering.com/chip-industr...
#semiconductor #MRAM #AI #processors #DRAM #chiplets #FeRAM
We are in the age of RISC-V chiplets, apparently.
Still testing the new DeepComputing DC-ROMA Mainboard II, with 8 SiFive P550 cores, Imagination GPU, and separate NPU chip, with 32 GB of shared RAM. The way it works is... interesting.
Stay tuned for a video. Might take a little longer than usual.
Still testing the new DeepComputing DC-ROMA Mainboard II, with 8 SiFive P550 cores, Imagination GPU, and separate NPU chip, with 32 GB of shared RAM. The way it works is... interesting.
Stay tuned for a video. Might take a little longer than usual.
October 16, 2025 at 1:54 AM
We are in the age of RISC-V chiplets, apparently.
Still testing the new DeepComputing DC-ROMA Mainboard II, with 8 SiFive P550 cores, Imagination GPU, and separate NPU chip, with 32 GB of shared RAM. The way it works is... interesting.
Stay tuned for a video. Might take a little longer than usual.
Still testing the new DeepComputing DC-ROMA Mainboard II, with 8 SiFive P550 cores, Imagination GPU, and separate NPU chip, with 32 GB of shared RAM. The way it works is... interesting.
Stay tuned for a video. Might take a little longer than usual.
SemiEngineering has over 500 Tech Talk videos to learn from.
www.youtube.com/user/Sperlin...
Hot topics included challenges involving chiplets/ heterogeneous integration, AI, data management, DRAM, manufacturing and test & much more. Let us know what other topics are of interest.
#semiconductor
www.youtube.com/user/Sperlin...
Hot topics included challenges involving chiplets/ heterogeneous integration, AI, data management, DRAM, manufacturing and test & much more. Let us know what other topics are of interest.
#semiconductor
October 11, 2025 at 4:36 PM
SemiEngineering has over 500 Tech Talk videos to learn from.
www.youtube.com/user/Sperlin...
Hot topics included challenges involving chiplets/ heterogeneous integration, AI, data management, DRAM, manufacturing and test & much more. Let us know what other topics are of interest.
#semiconductor
www.youtube.com/user/Sperlin...
Hot topics included challenges involving chiplets/ heterogeneous integration, AI, data management, DRAM, manufacturing and test & much more. Let us know what other topics are of interest.
#semiconductor
ADVANCED MICRO DEVICES, INC: “we have invented a new kind of chiplet that’s made out of other chiplets. this enables us to create the new EPYC 9666X3D with 32GB of L2 cache. topology information will be available to system software via a new CPUID leaf that instantly kills you.”
October 8, 2025 at 1:20 AM
ADVANCED MICRO DEVICES, INC: “we have invented a new kind of chiplet that’s made out of other chiplets. this enables us to create the new EPYC 9666X3D with 32GB of L2 cache. topology information will be available to system software via a new CPUID leaf that instantly kills you.”
Nvidia is investing $5 billion into Intel to jointly develop PC and data center chips. The investment will see Intel help Nvidia build x86 chips that integrate RTX GPU chiplets. Full details 👇 www.theverge.com/news/780792/...
Nvidia invests $5 billion into Intel to jointly develop PC and data center chips
Intel will help build x86 chips with Nvidia RTX GPU chiplets
www.theverge.com
September 18, 2025 at 11:38 AM
Nvidia is investing $5 billion into Intel to jointly develop PC and data center chips. The investment will see Intel help Nvidia build x86 chips that integrate RTX GPU chiplets. Full details 👇 www.theverge.com/news/780792/...
now that chip makers want even more bandwidth for less power, BGA is dying and we're seeing them literally fuse metal together inside the chips instead of doing BGA solders
we can see this with the new ryzen 9000 series, which the chiplets are no longer connected using BGA
we can see this with the new ryzen 9000 series, which the chiplets are no longer connected using BGA
December 7, 2024 at 10:05 AM
now that chip makers want even more bandwidth for less power, BGA is dying and we're seeing them literally fuse metal together inside the chips instead of doing BGA solders
we can see this with the new ryzen 9000 series, which the chiplets are no longer connected using BGA
we can see this with the new ryzen 9000 series, which the chiplets are no longer connected using BGA
#Chiplets: because cramming more transistors onto a single chip is so last decade. Modular mini-chips are the new black in the #tech world's fight against Moore's Law.
www.technologyreview.com/2024/01/08/1...
www.technologyreview.com/2024/01/08/1...
Chiplets: 10 Breakthrough Technologies 2024
Chipmakers are betting that smaller, more specialized chips can extend the life of Moore’s Law.
www.technologyreview.com
December 26, 2024 at 2:57 PM
#Chiplets: because cramming more transistors onto a single chip is so last decade. Modular mini-chips are the new black in the #tech world's fight against Moore's Law.
www.technologyreview.com/2024/01/08/1...
www.technologyreview.com/2024/01/08/1...
@Forbes: Chipping In: Billionaire-Founded Unicorn Startup Makes Semiconductor ‘Chiplets’ For The AI Boom https://t.co/jHnhzIYKCm
February 17, 2024 at 12:30 AM
@Forbes: Chipping In: Billionaire-Founded Unicorn Startup Makes Semiconductor ‘Chiplets’ For The AI Boom https://t.co/jHnhzIYKCm
this is just an optical network to reduce communication times between chiplets and it hasn't been implemented yet
I'm excited about this tech too, but it's *way* too early and they haven't disclosed costs either
I'm excited about this tech too, but it's *way* too early and they haven't disclosed costs either
October 7, 2024 at 2:41 PM
this is just an optical network to reduce communication times between chiplets and it hasn't been implemented yet
I'm excited about this tech too, but it's *way* too early and they haven't disclosed costs either
I'm excited about this tech too, but it's *way* too early and they haven't disclosed costs either
Lovely work, @esperanto_ai!
Notably the new generations have coherent shared memory across chiplets (256 and 512 cores / caches!). 👍
Notably the new generations have coherent shared memory across chiplets (256 and 512 cores / caches!). 👍
November 18, 2024 at 10:38 PM
Lovely work, @esperanto_ai!
Notably the new generations have coherent shared memory across chiplets (256 and 512 cores / caches!). 👍
Notably the new generations have coherent shared memory across chiplets (256 and 512 cores / caches!). 👍
When the bag has been reduced to chiplets. If there are no longer full size intact chips to pull out I am not sticking my hand in that bag.
July 20, 2025 at 4:45 PM
When the bag has been reduced to chiplets. If there are no longer full size intact chips to pull out I am not sticking my hand in that bag.
Intel managed to somewhat match what AMD was offering — at the cost of producing a lot of heat and having extreme power consumption. However, AMD had managed to break its CPU into modular units called chiplets, which made them significantly easier to produce, at little performance impact.
August 29, 2025 at 1:04 AM
Intel managed to somewhat match what AMD was offering — at the cost of producing a lot of heat and having extreme power consumption. However, AMD had managed to break its CPU into modular units called chiplets, which made them significantly easier to produce, at little performance impact.
Lilbits: Intel chips with NVIDIA GPU chiplets and GPD Win 5 performance previews (AMD Strix Halo in a handheld) #Chiplet, #Chips, #GpdWin5, #HandheldGamingPcs, #Igpu, #Intel, #Lilbits, #Nvidia, #SteamDeck, #SteamDeckLcd, #StrixHalo, #Zimaos
Lilbits: Intel chips with NVIDIA GPU chiplets and GPD Win 5 performance previews (AMD Strix Halo in a handheld)
Intel has made some big investments in GPUs in recent years, and while the company hasn't exactly become a major player in the desktop graphics space, it's made big strides in the performance of the integrated GPUs that are part of Intel's latest mobile chips. But that doesn't mean there's not room for better performance. So it's interesting to see that while Intel plans to continue developing its own iGPUs for the foreseeable future, …
liliputing.com
September 22, 2025 at 10:44 PM
Lilbits: Intel chips with NVIDIA GPU chiplets and GPD Win 5 performance previews (AMD Strix Halo in a handheld) #Chiplet, #Chips, #GpdWin5, #HandheldGamingPcs, #Igpu, #Intel, #Lilbits, #Nvidia, #SteamDeck, #SteamDeckLcd, #StrixHalo, #Zimaos
I've been watching $NXPI and note NXP must marry 3nm GAA compute to 5nm EUV I/O chiplets via TSV/RDL. BEOL CDU/LER yield risk; 30% perf/W upside vs 7nm monolith, mispriced.
November 6, 2025 at 4:06 PM
I've been watching $NXPI and note NXP must marry 3nm GAA compute to 5nm EUV I/O chiplets via TSV/RDL. BEOL CDU/LER yield risk; 30% perf/W upside vs 7nm monolith, mispriced.
BIOS-Updates für Ryzen 9000 senken Latenzen
Bei AMDs Ryzen 9 9000 war die Latenz zwischen zwei Chiplets bisher ungewöhnlich hoch. Die neue AGESA-Version 1.2.0.2 senkt sie […]
Bei AMDs Ryzen 9 9000 war die Latenz zwischen zwei Chiplets bisher ungewöhnlich hoch. Die neue AGESA-Version 1.2.0.2 senkt sie […]
Original post on social.heise.de
social.heise.de
September 18, 2024 at 4:39 PM
BIOS-Updates für Ryzen 9000 senken Latenzen
Bei AMDs Ryzen 9 9000 war die Latenz zwischen zwei Chiplets bisher ungewöhnlich hoch. Die neue AGESA-Version 1.2.0.2 senkt sie […]
Bei AMDs Ryzen 9 9000 war die Latenz zwischen zwei Chiplets bisher ungewöhnlich hoch. Die neue AGESA-Version 1.2.0.2 senkt sie […]
arXiv:2502.16633v1 Announce Type: new
Abstract: Chiplets are modular integrated circuits that can be combined to form a larger system, offering flexibility and performance enhancements. However, their dense packing often [1/7 of https://arxiv.org/abs/2502.16633v1]
Abstract: Chiplets are modular integrated circuits that can be combined to form a larger system, offering flexibility and performance enhancements. However, their dense packing often [1/7 of https://arxiv.org/abs/2502.16633v1]
February 25, 2025 at 6:13 AM
arXiv:2502.16633v1 Announce Type: new
Abstract: Chiplets are modular integrated circuits that can be combined to form a larger system, offering flexibility and performance enhancements. However, their dense packing often [1/7 of https://arxiv.org/abs/2502.16633v1]
Abstract: Chiplets are modular integrated circuits that can be combined to form a larger system, offering flexibility and performance enhancements. However, their dense packing often [1/7 of https://arxiv.org/abs/2502.16633v1]
Why AMD's Chiplets Work https://youtu.be/-x9nGo0Ge70?si=Tz8GO_b1Uejk0fv7
Why AMD's Chiplets Work
For the first six decades of the semiconductor industry, manufacturers improved performance by cramming more transistors onto a single die. It wasn't easy, b...
youtu.be
December 4, 2023 at 10:50 AM
Why AMD's Chiplets Work https://youtu.be/-x9nGo0Ge70?si=Tz8GO_b1Uejk0fv7
Finding the correct balance between customization and standardization is proving to be more difficult than initially thought.
semiengineering.com/chiplet-trad...
#chiplets
semiengineering.com/chiplet-trad...
#chiplets
Chiplet Tradeoffs And Limitations
Multi-die assemblies offer more flexibility, but figuring out the right amount of customization can have a big impact on power, performance, and cost.
semiengineering.com
April 15, 2025 at 3:17 PM
Finding the correct balance between customization and standardization is proving to be more difficult than initially thought.
semiengineering.com/chiplet-trad...
#chiplets
semiengineering.com/chiplet-trad...
#chiplets
Chip Industry Technical Paper Roundup: May 20 Reducing stress in chiplets; CFETs beyond 3nm; RISC-V eGPUs for TinyAI; LLM for VHDL MPU design; cache side-channel attacks on LLMs; memory prefetching for HPC processors; EUV scatterometry on 2D interconnec...
| Details | Interest | Feed |
| Details | Interest | Feed |
Origin
semiengineering.com
May 20, 2025 at 8:05 AM
Watching Texas Instruments $TXN: 180-65nm BCD + 28nm RDL chiplets, ~20% perf/W analysiquantatativ
September 19, 2025 at 3:06 PM
Watching Texas Instruments $TXN: 180-65nm BCD + 28nm RDL chiplets, ~20% perf/W analysiquantatativ
Exclusive: companies that make chip design software, like Cadence & Synopsys, offer AI tools to improve on what humans can do. But Intel has proprietary AI tools of its own. They're critical to squeezing the most out of Meteor Lake, a processor made of multiple "chiplets" in one package. #Intel #AI
AI Helps Chipmakers Design the Very Processors That Speed Up AI
Artificial intelligence helps Intel squeeze the most out of the design and manufacturing of its new Meteor Lake processor that arrives Thursday.
www.cnet.com
December 13, 2023 at 3:20 PM
Fujitsu flaunts massive 144-core Monaka Arm chip — 2nm and 5nm chiplets, 3D-stacked CPU cores over memory | Tom's Hardware https://www.tomshardware.com/pc-components/cpus/fujitsu-flaunts-144-core-monaka-cpu-2nm-and-5nm-chiplets-soic-and-cowos-packaging
December 12, 2024 at 2:19 AM
Fujitsu flaunts massive 144-core Monaka Arm chip — 2nm and 5nm chiplets, 3D-stacked CPU cores over memory | Tom's Hardware https://www.tomshardware.com/pc-components/cpus/fujitsu-flaunts-144-core-monaka-cpu-2nm-and-5nm-chiplets-soic-and-cowos-packaging
Intel: This chiplet thing won't work, it's stupid glue after all.
AMD: Ryzen just out performs Intel for few generations.
Intel: Fuccccck. (eventually gets chiplets out.)
AMD: Ryzen just out performs Intel for few generations.
Intel: Fuccccck. (eventually gets chiplets out.)
September 20, 2023 at 10:41 AM
Intel: This chiplet thing won't work, it's stupid glue after all.
AMD: Ryzen just out performs Intel for few generations.
Intel: Fuccccck. (eventually gets chiplets out.)
AMD: Ryzen just out performs Intel for few generations.
Intel: Fuccccck. (eventually gets chiplets out.)
I mean, you've moved to chiplets now, you're moving to 3D cache.
Never a leader, always a follower.
I'm not sure you can do anything and bring something out first.
Never a leader, always a follower.
I'm not sure you can do anything and bring something out first.
December 26, 2023 at 12:10 PM
I mean, you've moved to chiplets now, you're moving to 3D cache.
Never a leader, always a follower.
I'm not sure you can do anything and bring something out first.
Never a leader, always a follower.
I'm not sure you can do anything and bring something out first.