A Democratic AI Frontier Governance Lab based in San Francisco. Notes from the frontlines of inventing @thestorecloud and multi-agentic governed inference.
Joined March 2021
- Tweets164
- Following5
- Followers200
- Likes2
STORE Research retweeted
🔥
We have successfully onboarded with Anchorage.
In the last 60 days, institutional support has rallied around the project. We've onboarded with:
- Anchorage
- Circle
- Gemini
Previously, we've onboarded with Bitgo (who will custody $STORE).
🔥
✅ v0.1 of the STORE Oracle -- an incentivized, governed, composable, and programmable network of APIs giving instructions to @thestorecloud
STORE Research retweeted
✅ We have successfully upgraded the Store Chain test network. All 17 Raspberry Pi's have 128 GB of storage, 8GB of RAM, and two ethernet ports.
Next: The network will simulate Block miners validating transactions and Cloud miners storing all blockchain data on @thestorecloud.
We are now upgrading capacity in @thestorecloud to increase our DDoS test from 50k RPS to 500 RPS and 2MM RPS.
The "attack" sent 83 times (50k/600) the normal capacity of the VM and 33 (50k/1,500) times its breaking point. We learned that we can actually go over 50k RPS, which improves these numbers even further.
Note: When Solana was recently attacked, the load was about 4 times more.
Next, we added NGINX with rate limits at 600 RPS to attempt to hit the server through NGINX from JMeter with the speed of 50K RPS.
Results: System is stable with CPU at 65-75% usage while system load is at around 200%.
We then used JMeter for client replicas. We hit the server directly without NGINX at the speed of 1.5K RPS.
CPU hit 100% while system load increased to 600+% (screenshot above).
Moving the STORE Chain test environment from 20 VMs to 90 inside @thestorecloud.
Research update: On STORE Chain, we are going to simulate the scenario that slowed down Solana with our DDoS technical implementation. Will publish results. theblockcrypto.com/post/1312…
Upgrading our STORE Chain test network to 17 Raspberry PI's, each with 128 GB of storage, 8GB RAM, and two ethernet ports.
Upgrading our STORE Chain test network to 17 Raspberry PI's, each with 128 GB of storage, 8GB RAM, and two ethernet ports.
STORE Research retweeted
Do we have any idea the average number of Ethereum validators running per machine? Median is probably 1. What is mean? 3? 5? 10? @prylabs
Currently studying the fee markets of various layer ones to understand the relationship between the infrastructure costs (compute + storage) of a settlement-layer transaction and the minimum fee charged by miners to:
a) pay for the infrastructure costs
b) incentivize out DDoS
Upgrading the sync of an Ethereum archival node to two separate 2 TB SSD machines on @thestorecloud. Using RAID0 to combine them into a logical view.
We're bringing L2 load balancer and DDoS mitigation into @thestorecloud.
This new environment first powers our R&D with BlockfinBFT. It will then extend to all miners running Clouds and Chains on Store.
STORE Research retweeted
Probably the golden age of being a researcher in the space.
Opportunity is abundant for the researchoor.
📢 We're looking to add a protocol researcher to our team @StoreResearch empirically study other Web3 protocols. It's a paid, part-time contract role with token upside. To apply, send us a short message answering the question "Why you, why now?".
📧 [email protected]