mhtirogla
Home
Systems
API
Pricing
About
Log in
← Back
Home
›
Systems
›
Simulation and scientific computing
mpi-cluster
HPC MPI cluster
supercomputers
System
Techniques · 5
Primitives · 29
mpi-cluster
Backfill scheduling
distributed · medium: Run small jobs in gaps ahead of reserved large jobs without delaying them
Halo exchange
distributed · medium: Each rank stores its block plus a border copy of its neighbours' edge cells, refreshed by small messages every step; communication scales with block surface, not volume.
›
Parallel file system striping
distributed · medium: Striping files across object storage targets (Lustre, GPFS); collective I/O and metadata servers
Checkpoint and restart
hardware-os · medium: Periodically save enough state to resume a computation after a crash, preemption or move, with an interval that balances save cost against lost work.
›
Priority and preemption
systems · medium: Priority classes, quota and evicting lower-priority work to admit higher
Eager rendezvous protocol
distributed · low: MPI point-to-point: eager sends for small messages, rendezvous handshake for large ones
Rdma
networking · low: RDMA verbs over InfiniBand/RoCE: queue pairs, memory registration, one-sided ops
Numa aware allocation
hardware-os · low: Allocate memory and pin threads on the local NUMA node; first-touch and interleave policies
Roofline model
hardware-os · low: Bound a kernel's speed by min(peak FLOP/s, arithmetic intensity × memory bandwidth) to know whether to optimize compute or data movement.
›
Virtual memory and the TLB
hardware-os · low: How page tables and TLBs translate every address, what TLB reach means, and when 2 MiB or 1 GiB huge pages pay off on Linux.
›
Collective communication
ml-training · low: The family of group exchanges (broadcast, reduce, all-gather, reduce-scatter, all-to-all) and the latency-plus-bytes cost model that picks ring, tree or direct algorithms by message size and topology.
›
Ring all-reduce
ml-training · low: Sum a buffer across N workers via reduce-scatter then all-gather around a ring; each link carries 2(N−1)/N of the buffer, which is bandwidth-optimal.
›
Kernel-bypass networking
networking · low: User space (DPDK, AF_XDP) or an in-driver eBPF hook (XDP) handles packets straight from NIC rings, skipping sockets, for 10-40 Mpps per core.
›
Network topologies
networking · low: Fat-tree and Clos, dragonfly, torus and rail-optimized topologies: bisection bandwidth and diameter
Interval tree
data-structures · low: Index of ranges that answers "which intervals overlap this point or range?" in O(log n + k); static sorted-array form is the compact, cache-friendly choice.
›
Zero-copy I/O
networking · low: Move bytes between files, sockets and NICs by passing page references instead of copying through user space: sendfile, splice, MSG_ZEROCOPY, kTLS.
›
Leases
distributed · low: A lock that expires on its own: bounded-time ownership that survives crashes and partitions, given bounded clock drift and fencing tokens on writes.
›
Range partitioning
distributed · low: Cut the sorted key space into contiguous half-open ranges kept in a boundary map; split hot or big ranges, merge small ones, move ranges between nodes.
›
Copy-on-write fork
hardware-os · low: fork() gives the child the parent's whole address space by sharing pages read-only and copying one only when someone writes it; vfork and posix_spawn skip even that.
›
Dirty page tracking
hardware-os · low: Write-protection or PML logs of modified guest/process pages
mmap and the page cache
hardware-os · low: How read() and mmap reach file data through the kernel page cache, what readahead and madvise do, and why databases use their own buffer pool instead.
›
+ 13 more primitives
System
Techniques · 5
Primitives · 29
mpi-cluster
Backfill scheduling
distributed · medium: Run small jobs in gaps ahead of reserved large jobs without delaying them
Halo exchange
distributed · medium: Each rank stores its block plus a border copy of its neighbours' edge cells, refreshed by small messages every step; communication scales with block surface, not volume.
›
Parallel file system striping
distributed · medium: Striping files across object storage targets (Lustre, GPFS); collective I/O and metadata servers
Checkpoint and restart
hardware-os · medium: Periodically save enough state to resume a computation after a crash, preemption or move, with an interval that balances save cost against lost work.
›
Priority and preemption
systems · medium: Priority classes, quota and evicting lower-priority work to admit higher
Eager rendezvous protocol
distributed · low: MPI point-to-point: eager sends for small messages, rendezvous handshake for large ones
Rdma
networking · low: RDMA verbs over InfiniBand/RoCE: queue pairs, memory registration, one-sided ops
Numa aware allocation
hardware-os · low: Allocate memory and pin threads on the local NUMA node; first-touch and interleave policies
Roofline model
hardware-os · low: Bound a kernel's speed by min(peak FLOP/s, arithmetic intensity × memory bandwidth) to know whether to optimize compute or data movement.
›
Virtual memory and the TLB
hardware-os · low: How page tables and TLBs translate every address, what TLB reach means, and when 2 MiB or 1 GiB huge pages pay off on Linux.
›
Collective communication
ml-training · low: The family of group exchanges (broadcast, reduce, all-gather, reduce-scatter, all-to-all) and the latency-plus-bytes cost model that picks ring, tree or direct algorithms by message size and topology.
›
Ring all-reduce
ml-training · low: Sum a buffer across N workers via reduce-scatter then all-gather around a ring; each link carries 2(N−1)/N of the buffer, which is bandwidth-optimal.
›
Kernel-bypass networking
networking · low: User space (DPDK, AF_XDP) or an in-driver eBPF hook (XDP) handles packets straight from NIC rings, skipping sockets, for 10-40 Mpps per core.
›
Network topologies
networking · low: Fat-tree and Clos, dragonfly, torus and rail-optimized topologies: bisection bandwidth and diameter
Interval tree
data-structures · low: Index of ranges that answers "which intervals overlap this point or range?" in O(log n + k); static sorted-array form is the compact, cache-friendly choice.
›
Zero-copy I/O
networking · low: Move bytes between files, sockets and NICs by passing page references instead of copying through user space: sendfile, splice, MSG_ZEROCOPY, kTLS.
›
Leases
distributed · low: A lock that expires on its own: bounded-time ownership that survives crashes and partitions, given bounded clock drift and fencing tokens on writes.
›
Range partitioning
distributed · low: Cut the sorted key space into contiguous half-open ranges kept in a boundary map; split hot or big ranges, merge small ones, move ranges between nodes.
›
Copy-on-write fork
hardware-os · low: fork() gives the child the parent's whole address space by sharing pages read-only and copying one only when someone writes it; vfork and posix_spawn skip even that.
›
Dirty page tracking
hardware-os · low: Write-protection or PML logs of modified guest/process pages
mmap and the page cache
hardware-os · low: How read() and mmap reach file data through the kernel page cache, what readahead and madvise do, and why databases use their own buffer pool instead.
›
Ptrace process control
hardware-os · low: Ptrace: attach, read and write registers and memory, single-step, intercept signals
d-ary heap
data-structures · low: Array heap with d children per node (d = 4 typical): half the binary heap's depth, siblings in one cache line, faster priority queues.
›
Resource usage accounting
hardware-os · low: Cgroup/procfs/GPU counters sampled into per-workload usage
Graceful draining
systems · low: On a termination signal, stop taking new work, finish in-flight work until a deadline, tell clients to reconnect elsewhere, then exit; clients resume from a token.
›
False sharing
concurrency · low: Independent variables written by different cores on one cache line force the line to bounce between cores; pad or regroup hot data to fix it.
›
Memory ordering
concurrency · low: What Relaxed, Acquire, Release and SeqCst atomics actually guarantee, what they cost on x86 and ARM, and which one each pattern needs.
›
SPSC ring buffer
data-structures · low: Bounded wait-free queue for exactly one producer and one consumer thread, built from an array and two indices with acquire/release ordering.
›
Fencing tokens
distributed · low: A number that grows with every lock grant and is checked by the protected resource, so a paused ex-holder's late write is rejected.
›
CPU caches and the memory hierarchy
hardware-os · low: Sizes, line widths and latencies of L1/L2/L3/DRAM on modern x86 and Apple M-series, and the data-layout rules that follow from them.
›
Huge pages
hardware-os · low: 2 MB/1 GB pages to cut TLB misses; THP vs explicit hugetlbfs
SIMD programming
hardware-os · low: Process 4-16 lanes per instruction with AVX2, AVX-512 or NEON via Rust std::arch, with runtime dispatch and a portable fallback.
›
Ebpf xdp
networking · low: EBPF programs at the XDP/tc hooks for in-kernel packet processing at line rate
Receive side scaling
networking · low: NIC hashing of flows to RX queues (RSS) and symmetric hashing for per-core processing
Copy for AI
Similar systems
Share the most building blocks with this one.
training-fabric
30 %
scale-up-fabric
17 %
lbm-solver
15 %