I/o Is No Longer the Bottleneck? (2022)
Key topics
The provocative claim "I/O is no longer the bottleneck" sparked a lively debate, with some commenters pointing out that the assertion isn't entirely new, as a similarly titled post had appeared just a week prior by a different author. As the discussion unfolded, it became clear that the original post was actually a retort to the earlier article, and commenters began to unpack the nuances of I/O bottlenecks in modern computing. Some noted that I/O was never really about sequential reads, while others highlighted the blurring of lines between memory and I/O with advancements in PCIe/CXL technology, sparking speculation about potential revolutions in computer architecture driven by AI funding. The conversation was marked by a mix of confusion, clarification, and insightful observations, making for a fascinating exploration of the evolving landscape of computing performance.
Snapshot generated from the HN discussion
Discussion Activity
Very active discussionFirst comment
1h
Peak period
62
0-6h
Avg / period
18.9
Based on 132 loaded comments
Key moments
- 01Story posted
Jan 5, 2026 at 7:02 PM EST
5d ago
Step 01 - 02First comment
Jan 5, 2026 at 8:30 PM EST
1h after posting
Step 02 - 03Peak activity
62 comments in 0-6h
Hottest window of the conversation
Step 03 - 04Latest activity
Jan 9, 2026 at 3:18 PM EST
1d ago
Step 04
Generating AI Summary...
Analyzing up to 500 comments to identify key contributors and discussion patterns
Want the full context?
Jump to the original sources
Read the primary article or dive into the live Hacker News thread when you're ready.
Edit: Ah, I saw @wmf has edited their comment to include "related"
Technically there's nothing stopping me from mmaping an entire multi-terabyte nvme at the block level except for the part where I don't want to reimplement a filesystem from scratch in addition to needing to share it between lots of different programs.
You can’t really optimise your code for faster sequential reads than the IO is capable of, so the only thing really worth focusing on is how to optimise everything that isn’t already sequential.
When you see advertised numbers like '200 GB/s' that is total memory bandwidth, or all cores combined. For individual cores, the limit will still be around 6 GB/s.
This means even if you write a perfect parser, you cannot go faster. This limit also applies to (de)serializing data like JSON and Protobuf, because those formats must typically be fully parsed before a single field can be read.
If however you use a zero-copy format, the CPU can skip data that it doesn't care about, so you can 'exceed' the 6 GB/s limit.
The Lite³ serialization format I am working on aims to exploit exactly this, and is able to outperform simdjson by 120x in some benchmarks as a result: https://github.com/fastserial/lite3
Samsung is selling NVMe SSDs claiming 14 GB/s sequential read speed.
Yes, those numbers are real but only in very short bursts of strictly sequential reads, sustained speeds will be closer to 8-10 GB/s. And real workloads will be lower than that, because they contain random access.
Most NVMe drivers on Linux actually DMA the pages directly into host memory over the PCIe link, so it is not actually the CPU that is moving the data. Whenever the CPU is involved in any data movement, the 6 GB/s per core limit still applies.
That's not actually an option for NVMe devices, is it? The actual NAND storage isn't available for direct memory mapping, and the only way to access it is to set up DMA transfers.
its phrased a bit weird.
As for SSD, for most drives, it's true true that they cannot sustain writes indefinitely. They often write in SLC mode then have to rewrite, re-pack things into denser storage configurations that takes more time to write. They'll do that in the background, given the chance, so it's often not seen. But write write write and the drive won't have the time.
Thats very well known, very visible, and most review sites worth a salt test for it and show that sustained write performance. Some drives are much better than others. Even still, an Phison E28 will let you keep writing at 4GB/s until just before the drive is full full full. https://www.techpowerup.com/review/phison-e28-es/6.html
Drive reads don't have this problem. When review sites benchmark, they are not benchmarking some tiny nanosliver of data. Common benchmark utilities will test sustained performance, and it doesn't suddenly change 10 seconds in or 90 seconds in or whatever.
My finger is hovering over the bozo bit here man. C'mon.
In Linux you can use direct IO or RWF_UNCACHED to avoid paying extra for unwanted readahead.
it's why GPUs have their memory chips in a circle around the GPU chip.
If you want schema, it must be done by the application through runtime type checking. All messages contain type information. Though I do see the value of adding pydantic-like schema checking in the future.
Do you have any thoughts/recommendations for someone if they were to try making a pydantic interface layer for lite3?
Doing this by definition requires processing an entire message which can be a little slow, so it is best done in C and exposed through CPython bindings / CFFI.
That sounds a bit silly..
> All messages contain type information.
That would (partially) enable what you were suggesting in the sentence I quoted first. But that's orthogonal to being schema-full or schema-less.
Disclosure: I am the author.
I suppose that in real life such ideal condition do not occur, but it shows how badly the CPU is limited by its memory bandwidth for streaming tasks. Its maximum memory-read bandwidth is 768 bits per clock. only 60% of its peak bit-crunching performance. DRAM bandwidth is even more limiting. And this is a single core of at least 12 (and at most 64).
That must be the reason, why EPYC 9175F exists. It is only 16-core CPU, but all 16 8-core CCDs are populated and only one core on each is active.
The next gen EPYC is rumored to have 16 instead of 12 memory channels (which were 8 only 4-5 years ago).
If your workload is mostly about DMAing large chunks of data around between devices and you still want to examine the chunk/packet headers (but not touch all payload) on the CPU, this could be a good choice. You should have the full PCIe/DRAM bandwidth if all CCDs are active.
Edit: Worth noting that a DMA between PCIe and RAM still goes through the IO Hub (Uncore on Intel) inside the CPU.
Are those numbers also measured while skipping the CPU cache?
A single threaded Zen2 program could very well take 1 second to scan through your RAM, during which it's entirely trivial to read stuff from disk, so the modern practice of keeping a ton of stuff in RAM might be actually hurting performance.
Algorithms, such as garbage collection, which scan the entire heap, where the single-threaded version is probably slower than the naive zen 2 memcpy might run for more than a second even on a comparatively modest 16GB heap, which might not be even acceptable.
https://chipsandcheese.com/p/intels-clearwater-forest-e-core...
In the case of append, the old value still lives inside the buffer but is zeroed out. This means that if you keep replacing variable-sized elements, over time the buffer will fragment. You can vacuum a message by recursively writing it from the root to a new buffer. This will clear out the unused space. This operation can be delayed for as long as you like.
If you are only setting fixed-size values like integers or floats, then the buffer never grows as they are always updated in-place.
Btw: in my project README I have benchmarks against Cap'N Proto & Google Flatbuffers.
One primary difference is that Rkyv does not support in-place mutation. So any modification of a message requires full reserialization, unlike Lite³.
What makes M-series have 3x the bandwidth (per core), over x86?
The tradeoff is that it's non-upgradeable, but (contra some people who claim this is only a cash-grab by Apple to prevent RAM upgrades) it's worth it for the bandwidth.
That, and if you come at it from the phone / tablet or even laptop angle: most people are quite ok just buying their computing devices pre-assembled and not worrying about upgrading them. You just buy a new one when the old one fails or you want an upgrade.
Similar to how cars these days are harder to repair for the layman, but they also need much less maintenance. The guy who was always tinkering away with his car in old American sitcoms wasn't just a trope, he was truth-in-television. Approximately no one has to do that anymore with modern cars.
Apple are able to push a wider bus at higher frequencies because they aren't limited by signal integrity problems you encounter trying to do the same over sockets and motherboard traces. x86 CPUs like the Ryzen AI Max 395+, when packaged without socketed memory, are able to push equally wide busses at high frequencies.
The ability to mutate serialized data allows for 2 things:
1) Services exchanging messages and making small changes each time e.g. adding a timestamp without full reserialization overhead.
2) A message producer can keep a message 'template' ready and only change the necessary fields each time before sending. As a result, serializing becomes practically free from a performance perspective.
FWIW the other use case I was expecting is something like database queries where you’d select one particular column out of a json-like blob and want to avoid deserialization.
That being said storing trees as serializable flat buffers is definitely useful, if only because you can release them very cheaply.
Which file formats allow partial parsing?
Imagine the following object:
In JSON, if you want to extract the "name" property, you need to parse the whole "attributes" array. However, if you encoded the size of the "attributes" array in bytes, a parser could look at the key "attributes", decide that it's not interested, and jump past it without parsing anything.You'd typically want some kind of binary format, but for illustration purposes, here's an imaginary XML representation of the JSON data model which achieves this:
If this was stored in a file, you could use fseek to seek past the "attributes" array. If it was compressed, or coming across a socket, you'd need more complicated mechanisms to seek past irrelevant parts of the object.You still have to load a 64-byte cache line at a time, and most CPUs do some amount of readahead, so you'll need a pretty large "blank" space to see these gains, larger than typical protobufs.
> If however you use a zero-copy format, the CPU can skip data that it doesn't care about, so you can 'exceed' the 6 GB/s limit.
Of course the "skipping" is by cachelines. A cacheline is effectively a self-contained block of data from a memory throughput perspective, once you've read any part of it the rest comes for free.
The real joke is on me though, some of these GPU servers actually have 2 TB of RAM now. Crazy engineering!
Although... may be impossible to buy 2TB of RAM later in 2026! ;)
On a more serious note, the performance of hardware today is mind boggling from what we all encountered way back when. What I struggle to comprehend though is how some software (particularly Windows as an OS, instant messaging applications etc.) feel less performant now than they ever were.
Both Telegram and FB messenger are snappy; I didn't use anything else seriously as of late. (Especially not Teams, nor the late Skype.)
The problem is sloppy programming. We knew how to make small, fast, programs 20+ years ago that would just scream on modern hardware. But now everything is bloated and slow. CPUs can retire billions of instructions per second. Discord takes 10+ seconds to open. I’m simply not creative enough to think up how to keep the cpu busy that long opening IRC.
And you are right, that we got good UI latency even back in the 1980s. You just have make sure that you do the absolute minimum amount of work possible in the UI 'thread' and do gradual enhancement as more time passes.
As an example, the Geos word processor on the C64 does nice line breaks at the end of words only. But if you type really fast, it just wraps lines when you hit exactly x letters, and later when it has some time to catch up, it cleans up the line breaks.
That way it can give you a snappy user experience even on a comically underpowered system. But you can also see that the logic is much more complicated, than just implementing a single business logic for where line breaks should be. Complication means more bugs, more time spent writing and debugging and documenting etc.
they are forcing me to use the web client...
https://www.rtings.com/tv/tests/inputs/input-lag
A CRT's latency starts at ZERO, depending on what's driving it and when the input is received ("racing the beam").
even more so when considered in the context of dev 'remote workstations'. I benchmarked perf on AWS instances that was at least 5x slower than an average m1 macbook, and cost hundreds of dollars a dev per month (easily), and the macbook was a sunk cost!
What if you could take it for granted that mmap()ing a file has the exact same performance characteristics as malloc(), aside from the data not going away when you free the address space? What if arbitrary program memory could be given a filename and casually handed off to the OS to make persistent? A lot of basic software design assumptions are still based on the constraints of the spinning rust era...
It's just that mmap is slower than using read/write, because the kernel knows less about your data access patterns and thus has to guess for how to populate caches etc.
Instead, imagine if I could just state in one line of system-agnostic code "give me a pointer to /home/user/abc" and it does the right thing--assuming there was some way around mmap's current set of caveats. Imagine if I could turn a memory buffer into a file in one line of code and it Just Worked. Imagine if the OS treated my M.2 SSD as just another chip on the bus instead of still having a good amount of code on the hot path that assumes I'm manually sending bytes to a mechanical drive.
I suspect in some cases it will be more convenient and in other cases it will be less convenient to use. The write interface isn't so bad for some use cases like appending to logs. It's also not bad for its ability to treat different types of objects (files, pipes, network) generically.
But if you want to do all manipulation through pointers it should be doable. To support appending to files you'd probably need some way to explicitly grow the file size and hand back a new pointer. Some relevant syscalls would probably be open, ftruncate, and mmap.
fsync() is still slow, and you need that for real persistence. It's not just about spinning rust, there's very good reasons for wanting a different treatment of clearly ephemeral/scratchpad storage.
Some other participants said that they measured 0 difference in runtime between pshufb+eq and eqx3+orx2, but i think your problem has more classes of whitespace, and for the histogram problem, considerations about how to hash all the words in a chunk of the input dominate considerations about how to obtain the bitmasks of word-start or word-end positions.
close
The performance bottleneck is whatever resource hits saturation first under the workload you actually run: CPU, memory bandwidth, cache/allocations, disk I/O, network, locks/coordination, or downstream latency.
Measure it, prove it with a profile/trace, change one thing, measure again.
In the optimized case:
0.196 + 0.117 = 0.313, so we used 313ms of CPU time but the entire command took 395ms, with the CPU idle for 82ms.In other words: yes, the author managed to beat the speed of the disk subsystem. With two caveats:
1. not by much - similar attention to tweaking of I/O parameters might improve I/O performance quite a bit.
2. the I/O path is CPU-bound. Those 117ms (38% of all CPU cycles) are all spent in the disk I/O and file system kernel code; if both the disk and your user code were infinitely fast, the command would still take 117ms. (but those I/O tweaks might reduce that number)
Note that the slow code numbers are with a warm cache, showing 48ms of system time - in this case only the ext4 code has to run in the kernel, as data is already cached in memory. In the cold cache case it has to run the disk driver code, as well, for a total of 117ms.