@XeleraTech profile picture
Xelera Technologies GmbH @XeleraTech

The Xelera Suite software accelerates data center and cloud workloads.

Joined June 2023
  • Tweets 135
  • Following 78
  • Followers 21
86 Photos and videos
We launched a 3-part series on GBT inference latency in electronic trading, based on STAC-ML. Part 1 by Andrea Suardi explored how model structure shapes inference latency. ๐Ÿ“–๐Ÿ”— ๐—ฅ๐—ฒ๐—ฎ๐—ฑ ๐˜๐—ต๐—ฒ ๐—ณ๐˜‚๐—น๐—น ๐—ฏ๐—น๐—ผ๐—ด ๐—ต๐—ฒ๐—ฟ๐—ฒ: xelera.io/post/gradient-boosโ€ฆ

Gradient Boosted Tree Inference is not what you think it is

GBT inference speed is not limited by arithmetic; it is limited by memory access. Two structural choices made at training time, tree depth and ensemble-to-core mapping, determine whether latency is...

xelera.io
6
In HFT, inference speed matters. Xelera Silva offers sub-microsecond CPU-Only performance and deterministic FPGA acceleration at scale, all through one product and one API.
8
๐—ซ๐—ฒ๐—น๐—ฒ๐—ฟ๐—ฎ ๐—ฆ๐—ถ๐—น๐˜ƒ๐—ฎ, pushing the limits on @blackcoretechโ€™s overclocked servers, delivered strong benchmark results. Proud of this partnership and excited to share the findings with the trading community. ๐Ÿ”—blackcoretech.com/labs-centeโ€ฆ
9
In data centers, the real question is simple: what does this technology improve? At Xelera, we pair very fast hardware with software that turns performance into practical value.
1
6
Xelera AP helps remove security bottlenecks, speed up AI inference, and improve network performance while offloading work from servers. Built for scale. ๐Ÿ“–๐Ÿ”—xelera.io/post/xelera-buildiโ€ฆ
6
When every microsecond matters, inference speed matters. Xelera Silva delivers deterministic microsecond-level AI inference for HFT, time-critical network analytics, and fraud detection. ๐Ÿ”—linkedin.com/feed/update/urnโ€ฆ

Xelera Silva: Redefining Ultra-Low Latency AI Inference | Xelera Technologies

๐ƒ๐จ ๐ฒ๐จ๐ฎ ๐ง๐ž๐ž๐ ๐€๐ˆ ๐ˆ๐ง๐Ÿ๐ž๐ซ๐ž๐ง๐œ๐ž ๐ซ๐ž๐ฌ๐ฎ๐ฅ๐ญ๐ฌ ๐ข๐ง ๐š ๐ฌ๐ข๐ง๐ ๐ฅ๐ž ๐ฆ๐ข๐œ๐ซ๐จ๐ฌ๐ž๐œ๐จ๐ง๐ ๐จ๐ซ ๐ฅ๐ž๐ฌ๐ฌ?ย  Does your business requireย predictable latenciesย to ensure that you get results...

linkedin.com
8
Proud to collaborate with Blackcoreยฎ Technologies as an official Blackcore Labs Partner. Together, we combine Xelera Silva with overclocked ACE servers for ultra-low latency AI inference in HFT. ๐Ÿ”—blackcoretech.com/labs-centeโ€ฆ

Xelera Technologies & Blackcore Technologies

Xelera Technologies specializes in high-speed network technology and machine learning software, addressing data rate and latency bottlenecks with a performance-optimized software stack. Its platform...

blackcoretech.com
11
A major milestone for Xelera! ๐Ÿš€ Weโ€™ve raised โ‚ฌ1.5M in seed funding led by @Hiinov & @EUeic Fund to accelerate AI & network security workloads across data centers, cloud & edge. ๐Ÿ’ก #AI #DeepTech
22
Xelera joined an event by @PlugandPlayTC Munich in partnership with @AirbusDefence @AirbusSpace. ๐—๐˜‚๐—น๐—ถ๐—ฎ๐—ป ๐—žรค๐˜‚๐˜€๐—ฒ๐—ฟ pitched ๐—ซ๐—ฒ๐—น๐—ฒ๐—ฟ๐—ฎ ๐—ก๐—ฒ๐˜ โ€” our SmartNIC for secure, high-speed networking. #CyberSecurity #Sovereignty
36
๐Ÿ“ก At the 5G Consortium @TUDresden, Alexander Lange presents Xelera DYRA โ€” our Open-RAN 5G RAN system built with 5G-OPERA. Scalable, energy-efficient, and ready for Industry 4.0. #5G #OpenRAN
28
Xelera joined @AMD for a webinar on accelerated computing! ๐—™๐—ฒ๐—น๐—ถ๐˜… W๐—ถ๐—ป๐˜๐—ฒ๐—ฟ๐˜€๐˜๐—ฒ๐—ถ๐—ป and ๐— ๐—ถ๐—ฐ๐—ต๐—ฎ๐—ฒ๐—น ๐— ๐—ฐ๐—š๐˜‚๐—ถ๐—ฟ๐—ธ showcased the AMD Alveoโ„ข V80 & ultra-low-latency AI with Xelera Silva. #AI #FPGA
1
29
Xelera Silva powers AI inference in microseconds โ€” supporting XGBoost, LightGBM, CatBoost & LSTM. Optimized for speed, built for real-world demands. Proud to partner with @AMD #AIinference
30
We had a great time celebrating 10 years of @blackcoretech at their Summer Social in London! Thanks for the warm hospitality and unforgettable evening. ๐ŸŽ‰๐Ÿฅ‚ #HighPerformanceComputing #Blackcore #Xelera #Partnership
39
๐—ซ๐—ฒ๐—น๐—ฒ๐—ฟ๐—ฎ ๐—ฆ๐—ถ๐—น๐˜ƒ๐—ฎ ๐—”๐—œ ๐—œ๐—ป๐—ณ๐—ฒ๐—ฟ๐—ฒ๐—ป๐—ฐ๐—ฒ ๐—ฃ๐—น๐—ฎ๐˜๐—ณ๐—ผ๐—ฟ๐—บ, built for industries where microseconds matterโ€”high-frequency trading, cybersecurity, real-time analytics. Ultra-low latency: 1โ€“3ยตs. #HighFrequencyTrading #AMDAlveo
16
๐Ÿš€ Ultra-Low Latency ML for HFT ๐Ÿš€ Weโ€™re excited to share new benchmark results for ๐—ซ๐—ฒ๐—น๐—ฒ๐—ฟ๐—ฎ ๐—ฆ๐—ถ๐—น๐˜ƒ๐—ฎ ๐—œ๐—–๐—–: an FPGA-based AI solution delivering sub-2ยตs inference for real-time trading. #HFT #AI #UltraLowLatency
1
35
On ICC VEGA systems, Silva achieves up to 20x faster inference vs CPUs. ๐Ÿ”— Full results here: โ–ช๏ธ ๐—œ๐—–๐—– ๐—ฉ๐—˜๐—š๐—” ๐Ÿญ๐Ÿญ๐Ÿฒ๐—œ: cdn.prod.website-files.com/6โ€ฆ โ–ช๏ธ ๐—œ๐—–๐—– ๐—ฉ๐—˜๐—š๐—” ๐—ฅ-๐Ÿญ๐Ÿญ๐Ÿด๐—ถ: cdn.prod.website-files.com/6โ€ฆ
17
On June 11, join Xelera & @AMD to explore ultra-low latency AI Inference with ๐—ซ๐—ฒ๐—น๐—ฒ๐—ฟ๐—ฎ ๐—ฆ๐—ถ๐—น๐˜ƒ๐—ฎ on the AMD Alveoโ„ข V80 โ€” built for high-speed trading. ๐Ÿ”—webinar.amd.com/Using-the-AMโ€ฆ
1
23
Learn how ๐—ซ๐—ฒ๐—น๐—ฒ๐—ฟ๐—ฎ ๐—ฆ๐—ถ๐—น๐˜ƒ๐—ฎ delivers sub-ฮผs inference for HFT with AMD Alveo V80. โฌ‡๏ธRegister Now๐Ÿ”—webinar.amd.com/Using-the-AMโ€ฆ #XeleraSilva #LowLatency #FPGAs #AI #AlveoV80
20
Xelera Technologies GmbH retweeted
GITEX AI EUROPE@GITEXEUROPE
15 May 2025
Optimising Data Infrastructure for the GenAI Era AI growth is putting new pressure on Europeโ€™s data centres. @XeleraTech improves performance and energy efficiency by accelerating data flows and maximising existing infrastructure.
0:32
1
1
1
49
Xelera Technologies GmbH retweeted
European Innovation Council
@EUeic
5 Mar 2025
We are at Day 3 at @MWCHub 2025! Reflecting on an amazing week filled with innovation and exciting panels. Make sure to stop by tomorrow for more! ๐Ÿ“ Hall 8.1, Booth B52, in the @4YFN_MWC area Meet our 15 innovative European companies ๐Ÿ‘‰ tinyurl.com/4nh2nja4 #EICatMWC2025
1
1
10
534
Load more