Uday ๐จโ๐ป๐
13.5K posts

Uday ๐จโ๐ป๐
@u_plus_k
In a default consumptive state | Early stage venture investor |



Blue Owl says their software portfolio is "pristine" and has "largely green flags" Public markets disagree (-24% YTD, 52-wk lows). 15% redemptions also disagree. But the marks agree. That's what matters.















$META Reels Monetization Timeline: - Q3 20': Launched in August 2020 - Q2 22': $1 billion run rate - Q3 22': $3 billion run rate - Q2 23': $10 billion run rate - Q3 25': "Reels now has an annual run rate of over $50 billion".

1/ why Quantum is BULLSHIT "paving a path towardsย potential future" like saying: 'Working towards possibility of maybe a thing that might one day be a breakthrough if a breakthrough were to maybe occur' Here I assemble just 2005 to 2015 annual headlines or this BULLSHIT


The quality of AMD software now is totally different from when we started deeply using summer 2024.ย In 2024, we were running into many ROCm specific bugs. Today, the frequency in running ROCm bugs is orders of magnitude lower. AMD hardware is pretty good & the software is getting better every night. On Llama3 70B FP8 reasoning workloads at frontier lab volume pricing, MI300X vLLM offers 5-10% lower perf per TCO than H100 vLLM from our benchmarking across all interactivity levels (tok/s/user) and competitive perf per TCO on MI325X vLLM vs H200 vLLM and GPTOSS MX4 weights 120B Mi355 vs B200. Of course there is also various workloads in InferenceMAX where AMD software is currently losing too. The point of InferenceMAX is that there is nuance and we benchmark every night so that we are able to track the software improvements. visit inferencemax dot ai to see the full set of nuanced nightly results.




















