leighton (@uneventomato) header
leighton (@uneventomato) profile picture

Applied math, supercomputing, trains, banjos, county courthouses, college football, county courthouses, Steely Dan, "HPC," mismatched, black cats

BHM/ Bmore/ A2/ STL
Joined August 2021
Photo by leighton (@uneventomato)Photo by leighton (@uneventomato)Photo by leighton (@uneventomato)Photo by leighton (@uneventomato)Photo by leighton (@uneventomato)Photo by leighton (@uneventomato)Photo by leighton (@uneventomato)Photo by leighton (@uneventomato)Photo by leighton (@uneventomato)Photo by leighton (@uneventomato)
This GTA was originally going to be set in Dothan Alabama but due to graphical limitations on current consoles they couldn’t effectively render the majesty of the peanut festival
51
713
41
8,808
112,346
Super excited for @juliechoi new role. As Chief Ecosystem Officer, she will work with me to evangelize the power of fast inference and expand @Cerebras' place in the AI ecosystem.... Congratulations to Julie...
After 3 incredible years as CMO at @cerebras, I’m excited to step into the role of Chief Ecosystem Officer! I’ll focus on expanding our global AI ecosystem, forging strategic partnerships & bringing the fastest AI to more of the world. Grateful to @andrewdfeldman and the best team ever as we enter this next chapter of growth. Same mission Bigger impact We got this! 🚀
12
5
111
5,883
After 3 incredible years as CMO at @cerebras, I’m excited to step into the role of Chief Ecosystem Officer! I’ll focus on expanding our global AI ecosystem, forging strategic partnerships & bringing the fastest AI to more of the world. Grateful to @andrewdfeldman and the best team ever as we enter this next chapter of growth. Same mission Bigger impact We got this! 🚀
24
4
2
124
14,591
It took us 3 years to build our first piece of working hardware. JP, Gary, Sean, Michael and I, spent many late nights in our makeshift office in downtown Los Altos. The building wasn’t exactly designed for hardware. To keep the machine cool, we left the windows open and blew a hole in the wall to bring in a chiller from outside. Now, a decade on, we’re deploying our fourth generation. 2x the speed of anything we’ve served before. A look into the people and magic that make this possible:
Fearless Engineering: The CS-4 is our first engineering documentary. Through lab footage and conversations with the engineers behind each breakthrough, the film follows the creation of the CS-4, our latest wafer-scale AI system. 0:00 introducing the engineering team 1:09 Wafer-scale history 2:28 Powering the wafer 4:00 The cooling problem 4:18 Designing the cold plate 5:11 Holes in the wafer 6:33 the alignment problem 7:47 The CS-4 PRD 8:35 Designing for deployability and serviceability 11:01 Removing the motherboard 11:25 Removing internal cooling loops 11:41 inventing the power rack 12:35 A one-person lab deployment 14:03 Backpack: 50% fewer components 14:55 Connecting the compute backpack to cooling manifold 17:58 shifting Cerebras’s focus from training to inference 19:09 Building the Cerebras Cloud 20:01 CS-3 vs. CS-4 performance 20:49 3D Stacked DRAM
6
19
2
244
10,169
Always start singing this to myself when I remember I'm a Southerner in the Midwest and don't really have any idea what's going on over there in the Bay youtube.com/watch?v=FPJx3G5L…
5
342
This is a thoughtful explanation of how @cerebras manages parametric yield at wafer scale. For those interested in the nitty gritty of chip design, i think you will enjoy it.
7
18
1
149
7,569
Extremely excited to see us roll out Kimi K3.1, and proud I got to implement and own one of the backing wafer kernels for this model. One of the great things about @cerebras is we're all empowered to wear lots of different hats. Very grateful to contribute to the inference stack alongside other Cerebras initiatives
Moonshot just dropped Kimi K3.1 Pro. We used it to speed run online shopping. One prompt → monitor found → checkout → order confirmed. 11.5 seconds. Watch @Kimi_Moonshot K3.1 Pro in action, accelerated by Cerebras.
1
23
1,141
Pretty torn up on the Tachyum bankruptcy news. RIP
1
439
Replying to @OracleCloud
@OracleCloud will be offering @cerebras in OCI for high-speed inference. The best companies offer the best choices to their customers. So excited to be working with @Oracle.
OCI customers will have another high-speed inference option for building AI agents that can respond quickly and tackle demanding tasks. Cerebras AI Inference on OCI is planned for 2027: social.ora.cl/6016BHoFum #AIWorld
1
3
22
445
leighton retweeted
How we deploy multi-trillion parameter models at scale Today, let's break down the math for how @cerebras: 1. Fits the full model into on-chip memory 2. Maps the model across multiple chips 3. Handles a high # of concurrent requests
89
319
37
3,042
304,197
leighton retweeted
Yesterday, Oracle and Cerebras announced that Oracle Cloud Infrastructure (OCI) will offer Cerebras’ fast inference to AI labs, enterprises, and developers. After more than a year of extensive evaluation of Cerebras technology, Oracle selected Cerebras to bring the world's fastest AI inference to its cloud infrastructure. We're excited to bring blazingly fast inference to more AI labs, enterprises, and developers around the world. Together with Oracle.
The world’s fastest AI is coming to @Oracle Cloud Infrastructure.
28
30
9
358
32,647
Trying to channel my inner @MilksandMatcha and learn how to leverage AI to make really cool content so here's an illustration of mapping RF emulation to the @cerebras wafer-scale engine
5
3
38
2,391
All the data was taken from a real run. If you work backwards from the 3D positions in space of the transmitters and receivers, you can derive the 2D pattern you see on the wafer of which cores are actively computing over time.
1
123
It's so weird that amidst the 21st century tech boom, the Bay Area has actually *lost* pro sports teams since apparently none of these tech geniuses is normal enough to want to be a local hero and own a baseball team.
138
61
45
2,125
243,526
Rust belt city in the middle of the sunbelt
27
14
6
1,273
87,734
leighton retweeted
Your AI agent has a to-do list. Waiting shouldn’t be the longest item. Cerebras’ Andy Hock joins theCUBE to discuss why inference speed matters for reasoning models and AI agents. Where would less waiting make the biggest difference? Watch the full conversation with Andy Hock and theCUBE: youtube.com/watch?v=H2GnroPz…
4
15
79
10,272