Old Optane powers 2.8T AI on a CPU 💨
A homelabber reports running a 2.8-trillion-parameter Kimi K3 AI model on a CPU, using hot, discontinued Optane memory for capacity.
A homelabber reports running a 2.8-trillion-parameter Kimi K3 AI model on a CPU, using hot, discontinued Optane memory for capacity.
November 8, 2026 at 5:34 PM
Old Optane powers 2.8T AI on a CPU 💨
A homelabber reports running a 2.8-trillion-parameter Kimi K3 AI model on a CPU, using hot, discontinued Optane memory for capacity.
A homelabber reports running a 2.8-trillion-parameter Kimi K3 AI model on a CPU, using hot, discontinued Optane memory for capacity.
Hobbyist draait Kimi K3 op Xeon met Optane-geheugen
Een hobbyist draait het AI-model Kimi K3 lokaal op een Xeon-processor met 40 cores en Optane PMem 200-geheugen, zonder GPU voor de inferentie.
Lees op techspeeltuin.nl
#technieuws #AI
Een hobbyist draait het AI-model Kimi K3 lokaal op een Xeon-processor met 40 cores en Optane PMem 200-geheugen, zonder GPU voor de inferentie.
Lees op techspeeltuin.nl
#technieuws #AI
Hobbyist draait Kimi K3 op Xeon met Optane-geheugen
Een hobbyist draait het AI-model Kimi K3 lokaal op een Xeon-processor met 40 cores en Optane PMem 200-geheugen, zonder GPU voor de inferentie.
techspeeltuin.nl
November 8, 2026 at 5:53 PM
Hobbyist draait Kimi K3 op Xeon met Optane-geheugen
Een hobbyist draait het AI-model Kimi K3 lokaal op een Xeon-processor met 40 cores en Optane PMem 200-geheugen, zonder GPU voor de inferentie.
Lees op techspeeltuin.nl
#technieuws #AI
Een hobbyist draait het AI-model Kimi K3 lokaal op een Xeon-processor met 40 cores en Optane PMem 200-geheugen, zonder GPU voor de inferentie.
Lees op techspeeltuin.nl
#technieuws #AI
Homelab Runs 2.8 Trillion Parameter Model on CPU with Intel Optane Memory
Why Intel Optane Persistent Memory Matters for Large Models Traditional server memory, such as DDR4, reaches its
https://bloggersminds.com/post/homelab-runs-2-8-trillion-parameter-model-on-cpu-with-intel-optane-memory-10413
Why Intel Optane Persistent Memory Matters for Large Models Traditional server memory, such as DDR4, reaches its
https://bloggersminds.com/post/homelab-runs-2-8-trillion-parameter-model-on-cpu-with-intel-optane-memory-10413
Homelab Runs 2.8 Trillion Parameter Model on CPU with Intel Optane Memory
Why Intel Optane Persistent Memory Matters for Large Models Traditional server memory, such as DDR4, reaches its limits when a model requires more than a few hundred gigabytes of RAM. Intel’s Optane Persistent Memory (PMem) offers a larger addressable space while keeping latency lower than typical storage solutions.
bloggersminds.com
November 8, 2026 at 12:32 PM
Homelab Runs 2.8 Trillion Parameter Model on CPU with Intel Optane Memory
Why Intel Optane Persistent Memory Matters for Large Models Traditional server memory, such as DDR4, reaches its
https://bloggersminds.com/post/homelab-runs-2-8-trillion-parameter-model-on-cpu-with-intel-optane-memory-10413
Why Intel Optane Persistent Memory Matters for Large Models Traditional server memory, such as DDR4, reaches its
https://bloggersminds.com/post/homelab-runs-2-8-trillion-parameter-model-on-cpu-with-intel-optane-memory-10413
Homelabber Runs 2.8-Trillion-Parameter AI Model on CPU With Optane Memory
A Reddit user ran Moonshot AI's Kimi K3 model on a 40-core Xeon CPU using second-gen Optane PMem 200 memory, achieving 0.85 tokens per second with 1-bit quantization.
A Reddit user ran Moonshot AI's Kimi K3 model on a 40-core Xeon CPU using second-gen Optane PMem 200 memory, achieving 0.85 tokens per second with 1-bit quantization.
Homelabber Runs 2.8-Trillion-Parameter AI Model on CPU With Optane Memory
A Reddit user ran Moonshot AI's Kimi K3 model on a 40-core Xeon CPU using second-gen Optane PMem 200 memory, achieving 0.85 tokens per second with 1-bit quantization.
hermes-vector.buzz
November 8, 2026 at 12:49 PM
Homelabber Runs 2.8-Trillion-Parameter AI Model on CPU With Optane Memory
A Reddit user ran Moonshot AI's Kimi K3 model on a 40-core Xeon CPU using second-gen Optane PMem 200 memory, achieving 0.85 tokens per second with 1-bit quantization.
A Reddit user ran Moonshot AI's Kimi K3 model on a 40-core Xeon CPU using second-gen Optane PMem 200 memory, achieving 0.85 tokens per second with 1-bit quantization.
Moonshot AI's 2.8-trillion-parameter Kimi K3 ran CPU-only in a homelab on second-gen Optane PMem 200. A 1-bit quant produced 0.85 tokens per second.
November 8, 2026 at 12:31 PM
Moonshot AI's 2.8-trillion-parameter Kimi K3 ran CPU-only in a homelab on second-gen Optane PMem 200. A 1-bit quant produced 0.85 tokens per second.
ALCF recently announced that all 800 (of 1024 total) DAOS servers are now in production on Aurora. The hardware was delivered in 2021, and Intel cancelled Optane (the enabling hardware technology) in 2022. The remaining 224 servers are for spares and development.
www.alcf.anl.gov/news/aurora-...
www.alcf.anl.gov/news/aurora-...
Aurora’s expanded DAOS storage system targets data demands of exascale science and AI | Argonne Leadership Computing Facility
www.alcf.anl.gov
November 6, 2026 at 7:38 PM
ALCF recently announced that all 800 (of 1024 total) DAOS servers are now in production on Aurora. The hardware was delivered in 2021, and Intel cancelled Optane (the enabling hardware technology) in 2022. The remaining 224 servers are for spares and development.
www.alcf.anl.gov/news/aurora-...
www.alcf.anl.gov/news/aurora-...
Oh noes! Altman needs your RAM!
Can multiple Optane drives in a stripe setup be enough to replace it?
youtu.be/Y7CHZwoTE0Q
Can multiple Optane drives in a stripe setup be enough to replace it?
youtu.be/Y7CHZwoTE0Q
Oh no! More Optane! Can multiple ones in RAID 0 replace your RAM?
YouTube video by Ex-IT guy
youtu.be
November 5, 2026 at 9:35 AM
Oh noes! Altman needs your RAM!
Can multiple Optane drives in a stripe setup be enough to replace it?
youtu.be/Y7CHZwoTE0Q
Can multiple Optane drives in a stripe setup be enough to replace it?
youtu.be/Y7CHZwoTE0Q
Several people pointed to discontinued bets like Optane as proof of the rot—arguing persistent memory could have been a superpower for AI inference workloads instead of an abandoned project. Instead, both legacy giants watched Nvidia run away with the AI boom. #hardware 2/4
November 5, 2026 at 3:00 AM
Several people pointed to discontinued bets like Optane as proof of the rot—arguing persistent memory could have been a superpower for AI inference workloads instead of an abandoned project. Instead, both legacy giants watched Nvidia run away with the AI boom. #hardware 2/4
ALCF recently announced that all 800 (of 1024 total) DAOS servers are now in production on Aurora. The hardware was delivered in 2021, and Intel cancelled Optane (the enabling hardware technology) in 2022. The remaining 224 servers are for spares and development […]
Original post on mast.hpc.social
mast.hpc.social
November 6, 2026 at 7:38 PM
ALCF recently announced that all 800 (of 1024 total) DAOS servers are now in production on Aurora. The hardware was delivered in 2021, and Intel cancelled Optane (the enabling hardware technology) in 2022. The remaining 224 servers are for spares and development […]
💻🔥 ¡OFERTA FLASH ASUS! ⚡
✨ ZenBook 14 Ultralight
🚀 Intel Core i7 + 16 GB RAM
💾 SSD 512 GB + Optane 32 GB
🖥️ Pantalla Full HD de 14"
🪶 Diseño ligero y elegante
💥 ¡Un portátil para llevar a todas partes!
⚠️ Sin sistema operativo.
👉 link.amazon/B08y6Ljix
#ASUS #ZenBook #Amazon #Ofertas #Amazonfins
✨ ZenBook 14 Ultralight
🚀 Intel Core i7 + 16 GB RAM
💾 SSD 512 GB + Optane 32 GB
🖥️ Pantalla Full HD de 14"
🪶 Diseño ligero y elegante
💥 ¡Un portátil para llevar a todas partes!
⚠️ Sin sistema operativo.
👉 link.amazon/B08y6Ljix
#ASUS #ZenBook #Amazon #Ofertas #Amazonfins
November 4, 2026 at 6:43 AM
💻🔥 ¡OFERTA FLASH ASUS! ⚡
✨ ZenBook 14 Ultralight
🚀 Intel Core i7 + 16 GB RAM
💾 SSD 512 GB + Optane 32 GB
🖥️ Pantalla Full HD de 14"
🪶 Diseño ligero y elegante
💥 ¡Un portátil para llevar a todas partes!
⚠️ Sin sistema operativo.
👉 link.amazon/B08y6Ljix
#ASUS #ZenBook #Amazon #Ofertas #Amazonfins
✨ ZenBook 14 Ultralight
🚀 Intel Core i7 + 16 GB RAM
💾 SSD 512 GB + Optane 32 GB
🖥️ Pantalla Full HD de 14"
🪶 Diseño ligero y elegante
💥 ¡Un portátil para llevar a todas partes!
⚠️ Sin sistema operativo.
👉 link.amazon/B08y6Ljix
#ASUS #ZenBook #Amazon #Ofertas #Amazonfins
本日の「自作PC」業界ニュース
・NVIDIAがカスタムラップ付きRTX 5090を11月12日まで抽選配布
・64GB構成のシェア急落とDDR4プラットフォームの反動回復
・OptaneメモリとXeon CPUのみで2.8兆パラメータAIモデルを推論
・iBuyPower Slate RTX 5060搭載PCが500ドル値下げで限定販売
・Diamond Foundryが200mm単結晶ダイヤモンドウェーハを生産
https://648news.com/pc_build/2026/11/09-rt5090-giveaway-ram-diamond-cooling
#自作PC
・NVIDIAがカスタムラップ付きRTX 5090を11月12日まで抽選配布
・64GB構成のシェア急落とDDR4プラットフォームの反動回復
・OptaneメモリとXeon CPUのみで2.8兆パラメータAIモデルを推論
・iBuyPower Slate RTX 5060搭載PCが500ドル値下げで限定販売
・Diamond Foundryが200mm単結晶ダイヤモンドウェーハを生産
https://648news.com/pc_build/2026/11/09-rt5090-giveaway-ram-diamond-cooling
#自作PC
自作PCニュース：NVIDIAが『ホグワーツ・レガシー』対応RTX 5090を抽選配布
NVIDIAがカスタムラップ付きRTX 5090を11月12日まで抽選配布、64GB構成のシェア急落とDDR4プラットフォームの反動回復、OptaneメモリとXeon CPUのみで2.8兆パラメータAIモデルを推論、iBuyPower Slate RTX 5060搭載PCが500ドル値下げで限定販売、Diamond Foundryが200mm単結晶ダイヤモンドウェーハを生産
648news.com
November 8, 2026 at 10:35 PM
本日の「自作PC」業界ニュース
・NVIDIAがカスタムラップ付きRTX 5090を11月12日まで抽選配布
・64GB構成のシェア急落とDDR4プラットフォームの反動回復
・OptaneメモリとXeon CPUのみで2.8兆パラメータAIモデルを推論
・iBuyPower Slate RTX 5060搭載PCが500ドル値下げで限定販売
・Diamond Foundryが200mm単結晶ダイヤモンドウェーハを生産
https://648news.com/pc_build/2026/11/09-rt5090-giveaway-ram-diamond-cooling
#自作PC
・NVIDIAがカスタムラップ付きRTX 5090を11月12日まで抽選配布
・64GB構成のシェア急落とDDR4プラットフォームの反動回復
・OptaneメモリとXeon CPUのみで2.8兆パラメータAIモデルを推論
・iBuyPower Slate RTX 5060搭載PCが500ドル値下げで限定販売
・Diamond Foundryが200mm単結晶ダイヤモンドウェーハを生産
https://648news.com/pc_build/2026/11/09-rt5090-giveaway-ram-diamond-cooling
#自作PC
We need to nationalize Intel and Micron
make them bring back 3D XPoint SSDs (Optane)
make them bring back 3D XPoint SSDs (Optane)
"A poll I recently commissioned from the Center for Strategic Politics found that a supermajority of Democratic voters support public ownership and operation of healthcare and the electrical grid." peoplespolicyproject.org/2026/10/30/p...
Popularist Public Ownership
Democratic voters overwhelmingly support the public ownership and operation of several key industries.
peoplespolicyproject.org
October 31, 2026 at 4:33 PM
We need to nationalize Intel and Micron
make them bring back 3D XPoint SSDs (Optane)
make them bring back 3D XPoint SSDs (Optane)
💻🔥 ¡OFERTA FLASH ASUS! ⚡
✨ ZenBook 14 Ultralight
🚀 Intel Core i7 + 16 GB RAM
💾 SSD 512 GB + Optane 32 GB
🖥️ Pantalla Full HD de 14"
🪶 Diseño ligero y elegante
💥 ¡Un portátil para llevar a todas partes!
⚠️ Sin sistema operativo.
👉 link.amazon/B08y6Ljix
#ASUS #ZenBook #Amazon #Ofertas #Amazonfins
✨ ZenBook 14 Ultralight
🚀 Intel Core i7 + 16 GB RAM
💾 SSD 512 GB + Optane 32 GB
🖥️ Pantalla Full HD de 14"
🪶 Diseño ligero y elegante
💥 ¡Un portátil para llevar a todas partes!
⚠️ Sin sistema operativo.
👉 link.amazon/B08y6Ljix
#ASUS #ZenBook #Amazon #Ofertas #Amazonfins
Amazon.com: ASUSF ASUS ZenBook 14 Ultra-Slim Laptop 14” FHD NanoEdge Bezel Display, Intel Core i7-1165G7, NVIDIA MX450, 16GB RAM, 512GB SSD, ScreenPad 2.0, Thunderbolt 4, Windows 10 Pro, Pine Grey, UX...
Amazon.com: ASUSF ASUS ZenBook 14 Ultra-Slim Laptop 14” FHD NanoEdge Bezel Display, Intel Core i7-1165G7, NVIDIA MX450, 16GB RAM, 512GB SSD, ScreenPad 2.0, Thunderbolt 4, Windows 10 Pro, Pine Grey, UX...
link.amazon
November 4, 2026 at 6:12 AM
💻🔥 ¡OFERTA FLASH ASUS! ⚡
✨ ZenBook 14 Ultralight
🚀 Intel Core i7 + 16 GB RAM
💾 SSD 512 GB + Optane 32 GB
🖥️ Pantalla Full HD de 14"
🪶 Diseño ligero y elegante
💥 ¡Un portátil para llevar a todas partes!
⚠️ Sin sistema operativo.
👉 link.amazon/B08y6Ljix
#ASUS #ZenBook #Amazon #Ofertas #Amazonfins
✨ ZenBook 14 Ultralight
🚀 Intel Core i7 + 16 GB RAM
💾 SSD 512 GB + Optane 32 GB
🖥️ Pantalla Full HD de 14"
🪶 Diseño ligero y elegante
💥 ¡Un portátil para llevar a todas partes!
⚠️ Sin sistema operativo.
👉 link.amazon/B08y6Ljix
#ASUS #ZenBook #Amazon #Ofertas #Amazonfins
Optane のキャッシュメモリあるだけで、巨大なハードディスクへのアクセスが早くなり、めっちゃ快適。
ハードディスクですら価格がゴミみたいに上がる今、大容量ハードディスクを快適かつ長く使うのにうってつけでは。
･･･生きていたらの話だがな。
ハードディスクですら価格がゴミみたいに上がる今、大容量ハードディスクを快適かつ長く使うのにうってつけでは。
･･･生きていたらの話だがな。
November 2, 2026 at 2:10 PM
Optane のキャッシュメモリあるだけで、巨大なハードディスクへのアクセスが早くなり、めっちゃ快適。
ハードディスクですら価格がゴミみたいに上がる今、大容量ハードディスクを快適かつ長く使うのにうってつけでは。
･･･生きていたらの話だがな。
ハードディスクですら価格がゴミみたいに上がる今、大容量ハードディスクを快適かつ長く使うのにうってつけでは。
･･･生きていたらの話だがな。
SSDは書き込み上限あるって聞くと、HDDオンリーのパソコンにoptane入れてる今のパソコンと同じ構成のが良いんだけど(SSDのが早いのは認めた上で)、どこも作らなくなったからなぁ
定期的に変えるつもりならいいんだが・・・
スマホと同じ感覚で良いなら気にしなくて良いのかなぁ?
定期的に変えるつもりならいいんだが・・・
スマホと同じ感覚で良いなら気にしなくて良いのかなぁ?
October 31, 2026 at 10:43 AM
SSDは書き込み上限あるって聞くと、HDDオンリーのパソコンにoptane入れてる今のパソコンと同じ構成のが良いんだけど(SSDのが早いのは認めた上で)、どこも作らなくなったからなぁ
定期的に変えるつもりならいいんだが・・・
スマホと同じ感覚で良いなら気にしなくて良いのかなぁ?
定期的に変えるつもりならいいんだが・・・
スマホと同じ感覚で良いなら気にしなくて良いのかなぁ?
I hate Intel optane drives with a passion 😤
October 25, 2026 at 1:51 AM
I hate Intel optane drives with a passion 😤
_sighs_
Well all the reboots for CVEs and such apparently ate an Optane drive, that's slightly obnoxious.
Well all the reboots for CVEs and such apparently ate an Optane drive, that's slightly obnoxious.
October 22, 2026 at 8:03 PM
_sighs_
Well all the reboots for CVEs and such apparently ate an Optane drive, that's slightly obnoxious.
Well all the reboots for CVEs and such apparently ate an Optane drive, that's slightly obnoxious.
little girls deserve at least 64gb of low latency ddr5
also wouldn’t mind some intel optane either but that’s not as important rn
also wouldn’t mind some intel optane either but that’s not as important rn
October 21, 2026 at 9:01 PM
little girls deserve at least 64gb of low latency ddr5
also wouldn’t mind some intel optane either but that’s not as important rn
also wouldn’t mind some intel optane either but that’s not as important rn
intel optane < sodimm to pci-e adapter for the optimally wasteful ram disk
October 22, 2026 at 2:02 AM
intel optane < sodimm to pci-e adapter for the optimally wasteful ram disk
Homelabbers are desperate to buy a 4-year-old SSD that barely exists anymore
Homelabbers are desperate to buy a 4-year-old SSD that barely exists anymore
Intel gave up on Optane. Homelabbers didn't.
ift.tt
October 17, 2026 at 7:00 PM
Homelabbers are desperate to buy a 4-year-old SSD that barely exists anymore