#dgx — Public Fediverse posts

Spark em ação: DeepSec V4 Flash a 60 tokens/s

Quer ver como a nova metodologia Spark já está mudando a velocidade das IAs? 🤯

- O que está rolando:
• Pesquisadores e desenvolvedores estão retrenando modelos com a metodologia chamada Spark 🔥
• A comunidade já tenta executar o DeepSec V4 Flash usando duas unidades DGX Spark

- Resultado prático mencionado no trecho:
• Execução do DeepSec V4 Flash com 2 DGX Spark atingindo ~60...

#IA #Spark #DeepSec #DGX #MachineLearning #MorningCrypto

中央通訊社 2026-06-25 10:05:00 CST

Two Qwen3 models on one DGX Spark: the residency math

https://www.devashish.me/p/two-qwen3-models-on-one-dgx-spark

#HackerNews #Qwen3 #DGX #Spark #AI #residency #math #deep #learning

Two Qwen3 models on one DGX Spark: the residency math

https://www.devashish.me/p/two-qwen3-models-on-one-dgx-spark

#HackerNews #Qwen3 #DGX #Spark #AI #residency #math #deep #learning

Two Qwen3 models on one DGX Spark: the residency math

https://www.devashish.me/p/two-qwen3-models-on-one-dgx-spark

#HackerNews #Qwen3 #DGX #Spark #AI #residency #math #deep #learning

Two Qwen3 models on one DGX Spark: the residency math

https://www.devashish.me/p/two-qwen3-models-on-one-dgx-spark

#HackerNews #Qwen3 #DGX #Spark #AI #residency #math #deep #learning

Two Qwen3 models on one DGX Spark: the residency math

https://www.devashish.me/p/two-qwen3-models-on-one-dgx-spark

#HackerNews #Qwen3 #DGX #Spark #AI #residency #math #deep #learning

https://www.europesays.com/pl/503415/ ITReseller | Destination AI Gdańsk: „Przechodzimy przez falę od Agentic AI do Physical AI” – podkreśliła Małgorzata Gniech, NVIDIA #blackwell #DestinationAI #dgx #gdańsk #MałgorzataGniech #nvidia #PhysicalAI #PL #Poland #Polish #Polska #Polski #TDSYNNEX

Тестируем NVIDIA HGX B300 — инференс-сервер с 8 GPU и 2,3 ТБ VRAM на DeepSeek, Qwen и MiniMax

Итак, вы внедрили ИИ в свой сервис и решили ехать в продакшен, где у вас много пользователей. Закономерно возникает вопрос — а на чем запустить инференс, чтобы и пользователи были довольны скоростью работы, и бизнес не разорился. Привет! На связи Никита, системный архитектор Читать далее →

https://habr.com/ru/companies/selectel/articles/1035066/

#selectel #инференс #llm #gpu #nvidia #dgx #hgx_b300

Тестируем NVIDIA HGX B300 — инференс-сервер с 8 GPU и 2,3 ТБ VRAM на DeepSeek, Qwen и MiniMax

Итак, вы внедрили ИИ в свой сервис и решили ехать в продакшен, где у вас много пользователей. Закономерно возникает вопрос — а на чем запустить инференс, чтобы и пользователи были довольны скоростью работы, и бизнес не разорился. Привет! На связи Никита, системный архитектор Читать далее →

https://habr.com/ru/companies/selectel/articles/1035066/

#selectel #инференс #llm #gpu #nvidia #dgx #hgx_b300

Тестируем NVIDIA HGX B300 — инференс-сервер с 8 GPU и 2,3 ТБ VRAM на DeepSeek, Qwen и MiniMax

Итак, вы внедрили ИИ в свой сервис и решили ехать в продакшен, где у вас много пользователей. Закономерно возникает вопрос — а на чем запустить инференс, чтобы и пользователи были довольны скоростью работы, и бизнес не разорился. Привет! На связи Никита, системный архитектор Читать далее →

https://habr.com/ru/companies/selectel/articles/1035066/

#selectel #инференс #llm #gpu #nvidia #dgx #hgx_b300

https://www.youtube.com/watch?v=VWZB-W_H7VE

#youtube.com #dgx station #matroska

https://www.youtube.com/watch?v=VWZB-W_H7VE

#youtube.com #dgx station #matroska

https://www.youtube.com/watch?v=VWZB-W_H7VE

#youtube.com #dgx station #matroska

https://www.youtube.com/watch?v=VWZB-W_H7VE

#youtube.com #dgx station #matroska

Nvidia veröffentlicht den Open-Source-Stack NemoClaw zur Absicherung von OpenClaw-Agenten in Firmennetzen.

Die OpenShell reglementiert den Daten- und API-Zugriff der Systeme restriktiv. Für das lokale Offline-Training bietet Nvidia die DGX Station mit dem GB300-Chip, 20 Petaflops und 748 GB RAM an. Der Code wechselt später ohne Aufwand auf RZ-Server.

#Nvidia #NemoClaw #OpenClaw #DGX #News
https://www.all-ai.de/news/news26top/nvidia-nemoclaw-openclaw

Nvidia veröffentlicht den Open-Source-Stack NemoClaw zur Absicherung von OpenClaw-Agenten in Firmennetzen.

Die OpenShell reglementiert den Daten- und API-Zugriff der Systeme restriktiv. Für das lokale Offline-Training bietet Nvidia die DGX Station mit dem GB300-Chip, 20 Petaflops und 748 GB RAM an. Der Code wechselt später ohne Aufwand auf RZ-Server.

#Nvidia #NemoClaw #OpenClaw #DGX #News
https://www.all-ai.de/news/news26top/nvidia-nemoclaw-openclaw

Nvidia veröffentlicht den Open-Source-Stack NemoClaw zur Absicherung von OpenClaw-Agenten in Firmennetzen.

Die OpenShell reglementiert den Daten- und API-Zugriff der Systeme restriktiv. Für das lokale Offline-Training bietet Nvidia die DGX Station mit dem GB300-Chip, 20 Petaflops und 748 GB RAM an. Der Code wechselt später ohne Aufwand auf RZ-Server.

#Nvidia #NemoClaw #OpenClaw #DGX #News
https://www.all-ai.de/news/news26top/nvidia-nemoclaw-openclaw

#Nvidia's #N1/#N1X chips leak once again, this time tipped for release in first half of 2026 — hotly-anticipated chips to debut on #Dell and #Lenovo #laptops
N1 and N1X chips are Arm #SoC from Nvidia, purportedly featuring up to 20 CPU cores and rumored #RTX5070-level integrated GPU. Jensen Huang confirmed that #GB10 Superchip powering #DGX #Spark is actually based on N1 silicon, so it's already out there... just not with the gaming-focused slant we expect from the N1.
https://www.tomshardware.com/pc-components/cpus/nvidias-n1-n1x-chips-leak-once-again-this-time-tipped-for-release-in-first-half-of-2026-hotly-anticipated-chips-to-reportedly-debut-on-dell-and-lenovo-laptops

#Nvidia's #N1/#N1X chips leak once again, this time tipped for release in first half of 2026 — hotly-anticipated chips to debut on #Dell and #Lenovo #laptops
N1 and N1X chips are Arm #SoC from Nvidia, purportedly featuring up to 20 CPU cores and rumored #RTX5070-level integrated GPU. Jensen Huang confirmed that #GB10 Superchip powering #DGX #Spark is actually based on N1 silicon, so it's already out there... just not with the gaming-focused slant we expect from the N1.
https://www.tomshardware.com/pc-components/cpus/nvidias-n1-n1x-chips-leak-once-again-this-time-tipped-for-release-in-first-half-of-2026-hotly-anticipated-chips-to-reportedly-debut-on-dell-and-lenovo-laptops

#Nvidia's #N1/#N1X chips leak once again, this time tipped for release in first half of 2026 — hotly-anticipated chips to debut on #Dell and #Lenovo #laptops
N1 and N1X chips are Arm #SoC from Nvidia, purportedly featuring up to 20 CPU cores and rumored #RTX5070-level integrated GPU. Jensen Huang confirmed that #GB10 Superchip powering #DGX #Spark is actually based on N1 silicon, so it's already out there... just not with the gaming-focused slant we expect from the N1.
https://www.tomshardware.com/pc-components/cpus/nvidias-n1-n1x-chips-leak-once-again-this-time-tipped-for-release-in-first-half-of-2026-hotly-anticipated-chips-to-reportedly-debut-on-dell-and-lenovo-laptops

#Nvidia's #N1/#N1X chips leak once again, this time tipped for release in first half of 2026 — hotly-anticipated chips to debut on #Dell and #Lenovo #laptops
N1 and N1X chips are Arm #SoC from Nvidia, purportedly featuring up to 20 CPU cores and rumored #RTX5070-level integrated GPU. Jensen Huang confirmed that #GB10 Superchip powering #DGX #Spark is actually based on N1 silicon, so it's already out there... just not with the gaming-focused slant we expect from the N1.
https://www.tomshardware.com/pc-components/cpus/nvidias-n1-n1x-chips-leak-once-again-this-time-tipped-for-release-in-first-half-of-2026-hotly-anticipated-chips-to-reportedly-debut-on-dell-and-lenovo-laptops

#Nvidia's #N1/#N1X chips leak once again, this time tipped for release in first half of 2026 — hotly-anticipated chips to debut on #Dell and #Lenovo #laptops
N1 and N1X chips are Arm #SoC from Nvidia, purportedly featuring up to 20 CPU cores and rumored #RTX5070-level integrated GPU. Jensen Huang confirmed that #GB10 Superchip powering #DGX #Spark is actually based on N1 silicon, so it's already out there... just not with the gaming-focused slant we expect from the N1.
https://www.tomshardware.com/pc-components/cpus/nvidias-n1-n1x-chips-leak-once-again-this-time-tipped-for-release-in-first-half-of-2026-hotly-anticipated-chips-to-reportedly-debut-on-dell-and-lenovo-laptops

Тестируем B200 от NVIDIA: живые бенчмарки с GLM-4.7

Если вы занимаетесь обучением или тюнингом больших языковых моделей, используете инференс в режиме реального времени или выполняете сложные HPC-симуляции, то наверняка задавались вопросом: «а каково это будет на одном из лучших в мире чипов»? Как только мы получили B200, графический процессор, который по заявлениям производителя открывает новые грани производительности, гибкости и масштабируемости, то сразу побежали его тестировать. Сегодня я и мои коллеги из

https://habr.com/ru/companies/cloud_ru/articles/993498/

#b200 #hgx #a100 #h100 #h200 #dgx #ml #glm47

Тестируем B200 от NVIDIA: живые бенчмарки с GLM-4.7

Если вы занимаетесь обучением или тюнингом больших языковых моделей, используете инференс в режиме реального времени или выполняете сложные HPC-симуляции, то наверняка задавались вопросом: «а каково это будет на одном из лучших в мире чипов»? Как только мы получили B200, графический процессор, который по заявлениям производителя открывает новые грани производительности, гибкости и масштабируемости, то сразу побежали его тестировать. Сегодня я и мои коллеги из

https://habr.com/ru/companies/cloud_ru/articles/993498/

#b200 #hgx #a100 #h100 #h200 #dgx #ml #glm47

Тестируем B200 от NVIDIA: живые бенчмарки с GLM-4.7

Если вы занимаетесь обучением или тюнингом больших языковых моделей, используете инференс в режиме реального времени или выполняете сложные HPC-симуляции, то наверняка задавались вопросом: «а каково это будет на одном из лучших в мире чипов»? Как только мы получили B200, графический процессор, который по заявлениям производителя открывает новые грани производительности, гибкости и масштабируемости, то сразу побежали его тестировать. Сегодня я и мои коллеги из

https://habr.com/ru/companies/cloud_ru/articles/993498/

#b200 #hgx #a100 #h100 #h200 #dgx #ml #glm47

Đang chuẩn bị mua 2× DGX Spark, lo ngại kết nối chỉ 1 cáp 200 Gbps gây băng thông giới hạn so với bộ nhớ thống nhất ~275 Gbps. Thêm cáp thứ hai (dual‑link) có thể thu hẹp khoảng cách. Cáp khuyên dùng: QSFP56 200G (0.5 m) hay QSFP112? Người dùng muốn cổng Ethernet Mellanox để nối thẳng ZFS 7450 Pro. #DGX #AI #InfiniBand #Networking #CôngNghệ #CôngNghệAI

https://www.reddit.com/r/LocalLLaMA/comments/1qquinu/bottlenecked_dgx_spark_by_network/

Bucking

Today at work I got to enjoy another round of “Operation: Donkey Punch”, a term I picked up from my hacking friends. It’s where, after fighting with a device to make a change through standard means and failing repeatedly because the standard means are absurdly dumb or limited by policy or design, you tear it apart, make the change while it’s not looking, and then put it back together.

In this example, my department got a pair of Dell DGX devices — basically a set of ARM cores and a high-end NVIDIA GPU with lots of memory. They run a custom version of Ubuntu and are sold as desktop AI accelerators. They’re about the size of a Mac Mini. Team wants to try running models locally.

Thing is, the Setup Wizard expects you, the customer who just shelled out $5000 each, to be in a SOHO office environment, and it makes bad assumptions that don’t work in the Enterprise. My office has a MITM web filter that decrypts web traffic, sniffs it, and encrypts it with our own CA certificate. Every device that needs web access (which is mostly HTTPS these days) must have this cert installed or the device will trust nothing.

Despite being on the lab wired network, the Setup Wizard kept giving me the prompt to select a WiFi AP; that’s because even with correctly configured Ethernet, when it tries to call home to see if it’s truly on the Internet, it fails the cert trust and falls back to demanding WiFi. We can’t use WiFi in the lab; security policy.

There’s no widget to add a cert. Can’t even login on ssh or an alternate terminal. Completely locked out until Setup Wizard finishes. I tried every way to make it work.

Frustrated, I decided it was time to void the warranty. I opened the case, removed the storage, attached it to my workstation, copied the cert file to the right folder, simulated what update-ca-certificate does to “install” the cert, reinstalled the storage into the DGX, and powered it up. Restarted the Setup Wizard, and at the point where it would’ve asked for WiFi, it went directly to downloading system updates and finishing.

Fist up. Big ol’ punch to the head. Take it, bitch.

Vượt ngoài hỗ trợ của NVIDIA, một lập trình viên đã cụm hóa 3 DGX Sparks bằng cách tự viết plugin NCCL với 1500 dòng code C, đạt tốc độ suy luận phân tán trên 8 GB/s.

#NVIDIA #DGX #HPC #AI #Tech #Programming
#CôngNhệ #TríTuệNhânTạo #LậpTrình

https://www.reddit.com/r/LocalLLaMA/comments/1q8hqgd/i_clustered_3_dgx_sparks_that_nvidia_said_couldnt/

Bisschen late to the party, aber dennoch: Ich freue mich auf die erste echte #OnPrem LLM-Inference mit unserem neuen #NVIDIA #DGX #Spark.

Bisschen late to the party, aber dennoch: Ich freue mich auf die erste echte #OnPrem LLM-Inference mit unserem neuen #NVIDIA #DGX #Spark.

Bisschen late to the party, aber dennoch: Ich freue mich auf die erste echte #OnPrem LLM-Inference mit unserem neuen #NVIDIA #DGX #Spark.

Bisschen late to the party, aber dennoch: Ich freue mich auf die erste echte #OnPrem LLM-Inference mit unserem neuen #NVIDIA #DGX #Spark.

Dell's version of the DGX Spark fixes pain points

https://www.jeffgeerling.com/blog/2025/dells-version-dgx-spark-fixes-pain-points

#HackerNews #Dell #DGX #Spark #AI #Innovation #Technology #PainPoints

Dell's version of the DGX Spark fixes pain points

https://www.jeffgeerling.com/blog/2025/dells-version-dgx-spark-fixes-pain-points

#HackerNews #Dell #DGX #Spark #AI #Innovation #Technology #PainPoints

Dell's version of the DGX Spark fixes pain points

https://www.jeffgeerling.com/blog/2025/dells-version-dgx-spark-fixes-pain-points

#HackerNews #Dell #DGX #Spark #AI #Innovation #Technology #PainPoints

Dell's version of the DGX Spark fixes pain points

https://www.jeffgeerling.com/blog/2025/dells-version-dgx-spark-fixes-pain-points

#HackerNews #Dell #DGX #Spark #AI #Innovation #Technology #PainPoints

Dell's version of the DGX Spark fixes pain points

https://www.jeffgeerling.com/blog/2025/dells-version-dgx-spark-fixes-pain-points

#HackerNews #Dell #DGX #Spark #AI #Innovation #Technology #PainPoints

Người dùng đã thử nghiệm Spark với mô hình Nemotron3 Nano 30B, đạt tốc độ xử lý batch ấn tượng ~1300 token/giây với 200 yêu cầu đồng thời. Hiệu suất này rất hứa hẹn so với thế hệ trước và B200. Bạn nghĩ sao về việc so sánh với cấu hình 4x 3090?

#AI #HieuNang #XuLyBatch #DGX #Spark #Nemotron3 #GPU #Performance #BatchProcessing

https://www.reddit.com/r/LocalLLaMA/comments/1ptp8lq/dgx_spark_and_batch_processing/

🎯 AI
===================

Executive summary:
A consolidated NVIDIA security update for DGX Spark GB10 addresses a set of 14 vulnerabilities, the most severe being CVE-2025-33187 (CVSS 9.3) in the SROOT (Secure Root) component. The vendor states that this issue can allow an attacker with privileged OS access to pivot into SoC-protected areas, enabling arbitrary code execution, data tampering, and privilege persistence beyond normal OS controls.

Technical details:

CVE-2025-33187 – SROOT privilege pivot (CVSS 9.3)
CVE-2025-33188 – Hardware resource tampering (CVSS 8.0)
CVE-2025-33189 – Firmware out-of-bounds write (CVSS 7.8)
CVE-2025-33191 – Memory read error
CVE-2025-33197 – NULL pointer dereference

The SROOT vulnerability specifically impacts the trust boundary between the host OS and System-on-Chip protected regions. The firmware and hardware resource control flaws raise the risk of data corruption, denial-of-service, and potential code execution via firmware OOB writes. Affected installations are DGX OS versions on DGX Spark GB10 prior to OTA0.

Analysis:
The worst-case scenario described involves an attacker already holding privileged OS credentials (for example, a compromised root account) leveraging SROOT to escape OS-enforced protections and access or manipulate model weights, training data, or firmware-controlled resources. Hardware resource tampering and firmware OOB writes increase the attack surface for persistent or destructive modifications that evade standard OS detection.

Detection:
Monitor for anomalous access attempts to SROOT-exposed interfaces, unexpected firmware write operations, and any integrity violations of SoC-protected storage. Correlate privileged user sessions with unexplained firmware actions or hardware resource reconfigurations.

Mitigation (as reported):
NVIDIA released a consolidated patch bundle; affected DGX Spark GB10 units running DGX OS prior to OTA0 are listed as vulnerable. The vendor recommends applying the OTA0 release to remediate the 14 identified issues.

References:
The advisory enumerates 14 vulnerabilities; prominent IDs include CVE-2025-33187, CVE-2025-33188, and CVE-2025-33189.

🔹 NVIDIA #DGX #CVE-2025-33187 #firmware #SROOT

🔗 Source: https://securityonline.info/critical-patch-nvidia-dgx-spark-flaw-cve-2025-33187-cvss-9-3-exposes-ai-secrets-to-takeover/

NVIDIA DGX Spark, el nuevo mini PC con el chip GB10 Blackwell, es capaz de ejecutar Cyberpunk 2077 y juegos emulados de PS3 y Xbox https://blog.elhacker.net/2025/11/nvidia-dgx-spark-el-nuevo-mini-pc-con-el-chip-gb10-blackwell-es-capaz-de-ejecutar-cyberpunk-2077-emulador-box64-arm.html #Blackwell #emulador #Windows #gaming #juegos #nvidia #spark #mini #arm #dgx

NVIDIA DGX Spark, el nuevo mini PC con el chip GB10 Blackwell, es capaz de ejecutar Cyberpunk 2077 y juegos emulados de PS3 y Xbox https://blog.elhacker.net/2025/11/nvidia-dgx-spark-el-nuevo-mini-pc-con-el-chip-gb10-blackwell-es-capaz-de-ejecutar-cyberpunk-2077-emulador-box64-arm.html #Blackwell #emulador #Windows #gaming #juegos #nvidia #spark #mini #arm #dgx

NVIDIA DGX Spark, el nuevo mini PC con el chip GB10 Blackwell, es capaz de ejecutar Cyberpunk 2077 y juegos emulados de PS3 y Xbox https://blog.elhacker.net/2025/11/nvidia-dgx-spark-el-nuevo-mini-pc-con-el-chip-gb10-blackwell-es-capaz-de-ejecutar-cyberpunk-2077-emulador-box64-arm.html #Blackwell #emulador #Windows #gaming #juegos #nvidia #spark #mini #arm #dgx

NVIDIA DGX Spark, el nuevo mini PC con el chip GB10 Blackwell, es capaz de ejecutar Cyberpunk 2077 y juegos emulados de PS3 y Xbox https://blog.elhacker.net/2025/11/nvidia-dgx-spark-el-nuevo-mini-pc-con-el-chip-gb10-blackwell-es-capaz-de-ejecutar-cyberpunk-2077-emulador-box64-arm.html #Blackwell #emulador #Windows #gaming #juegos #nvidia #spark #mini #arm #dgx

Just in case anyone out there is interested, the #dgx_spark does about 12min for the top 2bil passwords on an MD5 crypt hash. Sure that's not what it's meant for but come on...
#hashcat
#hashcat7
#dgxspark
#dgxsparkgb10
#dgx

Để mở rộng quy mô trên 2 DGX Sparks trong một cluster, bạn có thể tận dụng dual 100Gbit QSFP28 links và cấu hình ROCE v2 với layer 3 links. Điều chỉnh NCCL variables để sử dụng ROCE v2 và kết nối cả hai cổng CX7 với switch để tăng băng thông.
#LocalLLaMA #NVIDIA #DGX #clusters #AI #vietnam

https://www.reddit.com/r/LocalLLaMA/comments/1oieip0/theoretically_scaling_beyond_2_dgx_sparks_in_a/

Nvidia DGX Spark: When Benchmark Numbers Meet Production Reality

https://publish.obsidian.md/aixplore/Practical+Applications/dgx-lab-benchmarks-vs-reality-day-4

#HackerNews #Nvidia #DGX #Spark #Benchmark #Production #Reality #AI #Technology #HPC

Nvidia DGX Spark: When Benchmark Numbers Meet Production Reality

https://publish.obsidian.md/aixplore/Practical+Applications/dgx-lab-benchmarks-vs-reality-day-4

#HackerNews #Nvidia #DGX #Spark #Benchmark #Production #Reality #AI #Technology #HPC

Nvidia DGX Spark: When Benchmark Numbers Meet Production Reality

https://publish.obsidian.md/aixplore/Practical+Applications/dgx-lab-benchmarks-vs-reality-day-4

#HackerNews #Nvidia #DGX #Spark #Benchmark #Production #Reality #AI #Technology #HPC