Powering Efficient AI: A continuous profiling platform for CPU and GPU workloads, that provides actionable insights to optimize perf & debug faster
Joined December 2024
- Tweets 42
- Following 1
- Followers 154
- Likes 22
12 Photos and videos
zymtrace retweeted
May 29
๐ข May episode is out! ๐ข
OpenTelemetry Comes of Age: CNCF Graduation, Profiling and eBPF
With @horovits hosting @israel_ogbole, CEO & Co-founder @zymtrace, OTel Profiling SIG member.
๐ง open.spotify.com/episode/5bEโฆ
๐บ youtu.be/Eo4g38QYX68
recap post:
๐ medium.com/p/c66dc2cd7d3c
1
3
3
123
Zymtrace automatically detects MIG topology and attributes GPU performance metrics to each MIG instance, correlated with full-stack GPU flamegraphs.
zymtrace.com/article/mig
34
๐ Weโre thrilled to welcome Abdullah Eryuzlu (@KernelNomad_ ) to zymtrace.
Abdullah is a systems engineer with a deep passion for low-level programming, kernels, and performance optimization.
1
3
7
277
Join @zymtrace and help us design agentic workflows and user experiences to optimize AI workloads.
zymtrace.com/career/zymtraceโฆ
2
6
137
๐ Learn more about Timโs background and why he joined @zymtrace here:
zymtrace.com/article/welcomeโฆ
153
zymtrace retweeted
Mar 19
My friends at @zymtrace are looking for Rust engineers ๐ฆ
They're building the platform that makes GPUs go brr โ fully remote, early stage, real impact
DM me for an intro, or email team@zymtrace.com
I can vouch that the team is top notch.
10
27
112
9,608
Big news: today we are announcing that we have raised $12.2M!
We are building an agentic, profile-guided opt platform that continuously profiles execution from CPU code to CUDA kernels in production to help teams maximize throughput per GPU.
zymtrace.com/article/zymtracโฆ
1
5
25
9,973
Big news: today we are announcing that we have raised $12.2M!
We are building an agentic, profile-guided opt platform that continuously profiles execution from CPU code to CUDA kernels in production to help teams maximize throughput per GPU.
zymtrace.com/article/zymtracโฆ
1
5
25
9,973
zymtrace retweeted
The @zymtrace folks are going *really deep* into what your GPUs are actually doing (and waiting for) when you're running your most expensive workloads... ๐ค
@MeetAnamAI achieved 250% faster inference and 90% higher throughput with $0 spent on new GPUs โก
Most teams solve GPU performance by throwing money at more @nvidia GPUs. Anam took the performance optimization approach.
1
10
1,095
zymtrace retweeted
Feb 6
The @zymtrace folks are killing it - this article is really quite a great read:
zymtrace.com/article/anam-zyโฆ
1
5
20
2,682
@MeetAnamAI achieved 250% faster inference and 90% higher throughput with $0 spent on new GPUs โก
Most teams solve GPU performance by throwing money at more @nvidia GPUs. Anam took the performance optimization approach.
1
5
1,244
The best part? They're feeding @zymtrace GPU profiles into AI coding agentsโautomatically creating PRs with performance fixes ๐
zymtrace.com/article/anam-zyโฆ
2
124
๐ Learn more about Matti's background and why he joined zymtrace here:
zymtrace.com/article/welcomiโฆ
1
2
1,334
๐ Lots of progress with #zymtrace in Sept: GPU CPU metrics, absolute CPU % mode, Python 3.13 support, Zing unwinding & major performance improvements.
Also contributing Python unwinding improvements to the upstream OTel eBPF profiler ๐ PR #837
zymtrace.com/article/sep-202โฆ
1
4
243
Metrics surface the WHAT, while profiles uncover the WHY at the GPU and Host level.
- GPU metrics: Utilization, memory, power, tensor cores, per-process usage
- Host metrics: Collected via #eBPF (super low-overhead)
๐ Learn more: zymtrace.com/article/gpu-hosโฆ
1
98