Eng @brewlabshq_ maintainer @dynosolpool oss @blueshift Member @SuperteamIN Fellow @solana

Joined April 2020
Bringing the Solana fam together in Mumbai
Two ecosystems, one city, and a room full of people pushing the industry forward. Excited to bring everyone together with @Astralaneio in Mumbai this November. See you at Sangam 🍸
4
201
Maybe we can finally go back to writing code like normal humans : )
My diagnosis of the Solana compiler is that it’s bad at optimizing normal code, so Solana devs end up going out of their way to instrument the compiler and deviate from normal best practices. Then the compiler further optimized for those patterns, reinforcing this very Solana-specific style of coding: - sprinkle #[inline(always)] and #[inline(never)] everywhere - break up enum-handling paths by hand - drop into asm blocks And now the code is harder to read, harder to maintain, and full of optimization boundaries that would get in the way of a normally functioning compiler. I’ve shown one example of this with the memcmp libcall. Here’s another one: xitter.cf/clairefxyz/status/2104… btw couldn't have made this without @rishavmehraa he spent countless nights to narrow down the issue!
6
516
Rishav Mehra retweeted
My diagnosis of the Solana compiler is that it’s bad at optimizing normal code, so Solana devs end up going out of their way to instrument the compiler and deviate from normal best practices. Then the compiler further optimized for those patterns, reinforcing this very Solana-specific style of coding: - sprinkle #[inline(always)] and #[inline(never)] everywhere - break up enum-handling paths by hand - drop into asm blocks And now the code is harder to read, harder to maintain, and full of optimization boundaries that would get in the way of a normally functioning compiler. I’ve shown one example of this with the memcmp libcall. Here’s another one: xitter.cf/clairefxyz/status/2104… btw couldn't have made this without @rishavmehraa he spent countless nights to narrow down the issue!
1
7
1
50
3,964
Upgraded to v4.3.0 Way to AlpenGlow 🏔️
🚨 Mainnet-beta validators: Agave v4.3 is the recommended version for all validators, upgrade now.
8
436
letsss goooo sol_multi3 as the first SVM JIT intrinsic 10k u128 muls 450k -> 110k CU : ) native ops where it matters, portable BPF everywhere else
7
406
We’re ready for Alpenglow⚡ BrewlabsHQ is already running on the Solana Alpenglow community test cluster 81.7K+ SOL activated stake and producing/voting on slots Seeing what Alpenglow is capable of at 183K TPS and 25ms finalization makes this even more exciting
When ever i doubt my life decisions up to now I go to the @solana alpenglow community test cluster and look at how sometimes 183k transactions finalize in 25ms. Be ready for it! Check out the ⛰️ migration guide! Link to the cluster website is in there. solana.com/upgrades/alpenglo…
2
3
1
27
2,170
Rishav Mehra retweeted
Validators! We need more stake on v4.3.0-rc.1. Please upgrade today. We're hoping to have 25% adoption going into the weekend.
4
11
1
51
3,044
One provider went down Solana didn’t : ) That’s the point of decentralization
All the SOL haters treating terraswitch outage as a knock against Solana... but like, you're completely missing the point. It didn't take Solana down. Because there is no one provider that has enough of stake. That's a win not a failure
14
1,282
p90 at 0.9 slots across regions is pretty wild. consistent sub-slot landing matters more than headline TPS
Transaction landing should happen within one slot. P90 ping-thing is stable at 0.9 slots now across all regions. Our transaction sending Cascade included with every subscription @triton_one. No tips. No extra fees.
3
394
Research > hype
new research article just dropped merch coming soon.
2
356
This one’s special Been part of the development for a while, and it feels great to finally see it live on main stage
1/ Solana onchain governance is live🗳️ Validators can now propose, support, and decide core protocol decisions via Solana Governance Proposals (SGPs) These are fully onchain, stake-weighted, and verified by Merkle proof 👇
7
449
Rishav Mehra retweeted
Finally it’s out, been involved in the development for a while for this, it’s been fun and finally it’s ship to prod 🫡
1/ Solana onchain governance is live🗳️ Validators can now propose, support, and decide core protocol decisions via Solana Governance Proposals (SGPs) These are fully onchain, stake-weighted, and verified by Merkle proof 👇
5
9
1
49
3,586
happy birthday @deanmlittle may all your CFG edges be valid and all your functions reachable : )
happy birthday @deanmlittle ! sbpf also just added control-flow graph and graph engines for writing CFG analyses! It also comes with graph dump utilities ✨ Here's a reachability analysis built with the worklist engine, along with the CFG before/after a dead-func-elimination pass. now go write your own compiler optimization pass and wish him a happy birthday! (I promise I didn't plan to publish this on his birthday)
2
6
748
upgraded to v4.1 ✅ The road to Alpenglow starts here 100M CU blocks and 200ms slots are coming closer
Agave v4.1 is now recommended for mainnet validators and sets the stage for an exhilarating v4.2. Introducing prerequisites for: - Alpenglow - 100M CU blocks - 200ms slot times
1
4
497
Agreed, SIMD needs a real name
they need real names brethren The Burn SIMD The Disinflation SIMD The 2x200 SIMD The BR Share SIMD
5
284
W🔥
1/ Some personal updates. I've been a bit inactive over the past few weeks, and I really appreciate everyone who reached out to check in on me. I’ve been focused on improving my skills, working on personal projects, learning new things, contributing to open source, and looking for a role that truly fits what I want
2
303
Soul healed
i just need sol to hit 70 then my soul will be healed
1
5
567
DEVS CARRYING THE ENTIRE ECOSYSTEM ON THEIR BACKS
CAN THE DEVS DO SOMETHING?!? (yes) THE DEVS ARE GOING TO INCREASE BANDWIDTH AND REDUCE LATENCY THE DEVS ARE GOING TO FIX THE TOKENOMICS THE DEVS ARE GOING TO CREATE A MANSION OF A HOUSE FOR ALL OF FINANCE THE DEVS ARE GOING TO KEEP SOLANA GREAT
2
203
Fast enough to remove the bottleneck
Most Solana teams optimize RPC. We decided to remove it. At @brewlabshq_ , our latency requirements pushed us beyond Yellowstone and toward consuming validator data directly from the source. Today we're open-sourcing part of that work: Qlaster — Shared-Memory Data Streaming for Colocated Solana Services.🧵
6
529