←back to thread

420 points speckx | 1 comments | | HN request time: 0s | source
Show context
dylan604 ◴[] No.44533476[source]
SSD speeds are nothing short of miraculous in my mind. I come from the old days of striping 16 HDDs together (at a minimum number) to get 1GB/s throughput. Depending on the chassis, that was 2 8-drive enclosures in the "desktop" version or the large 4RU enclosures with redundant PSUs and fans loud enough to overpower arena rock concerts. Now, we can get 5+GB/s throughput from a tiny stick that can be used externally via a single cable for data&power that is absolutely silent. I edit 4K+ video as well, and now can edit directly from the same device the camera recorded to during production. I'm skipping over the parts of still making backups, but there's no more multi-hour copy from source media to edit media during a DIT step. I've spent many a shoot as a DIT wishing the 1s&0s would travel across devices much faster while everyone else on the production has already left, so this is much appreciated by me. Oh, and those 16 device units only came close to 4TB around the time of me finally dropping spinning rust.

The first enclosure I ever dealt with was a 7-bay RAID-0 that could just barely handle AVR75 encoding from Avid. Just barely to the point that only video was saved to the array. The audio throughput would put it over the top, so audio was saved to a separate external drive.

Using SSD feels like a well deserved power up from those days.

replies(8): >>44533735 #>>44534375 #>>44535266 #>>44535471 #>>44536311 #>>44536501 #>>44539458 #>>44539872 #
bob1029 ◴[] No.44536311[source]
The latency of modern NVMe is what really blows my mind (as low as 20~30 uS). NVMe is about an order of magnitude quicker than SAS and SATA.

This is why I always recommend developers try using SQLite on top of NVMe storage. The performance is incredible. I don't think you would see query times anywhere near 20uS with a hosted SQL solution, even if it's on the same machine using named pipes or other IPC mechanism.

replies(3): >>44536919 #>>44540452 #>>44542264 #
Numerlor ◴[] No.44536919[source]
Then there's optane that got ~10us with. The newest controllers and nand is inching closer with randoms but optane is still the most miraculous ssd tech that's normally obtainable
replies(3): >>44537013 #>>44537056 #>>44538546 #
teaearlgraycold ◴[] No.44537056[source]
Eventually we'll have machines with unified memory+storage. You'll certainly have to take a bit of a performance hit in certain scenarios but also think about the load time improvements. If you store video game files in the same format they'd be needed at runtime you could be at the main menu in under a second.
replies(3): >>44537445 #>>44538862 #>>44543140 #
inkyoto ◴[] No.44538862[source]
The separation into RAM and external storage (floppy disks, magnetic tapes, hard drives and later SSD etc) is the sole consequence of technology not being advanced enough at the time to store all of the data in memory.

Virtual memory subsystems in operating systems of the last 40+ years pretty much do exactly that – they essentially emulate infinite RAM that spills over onto the external storage that backs it up.

Prosumer grade laptops are already easily available, and in 2-3 years there will be ones with 256-512 Gb as well, so… it is not entirely incoceivable that in 10-20 years (maybe more, maybe less) the Optane style memory is going to make a comeback and laptops/desktops will come with just memory, and the separation into RAM and external storage will finally cease to exist.

P.S. RAM has become so cheap and has reached such large capacity that the current generation of young engineers don't event know what a swap is, and why they might want to configure it.

replies(1): >>44538908 #
numpy-thagoras ◴[] No.44538908{3}[source]
I have a feeling (and it's just a feeling) that many SoC-style chips of the future will abandon Von Neumann Architecture entirely.

It's not that much of a stretch to imagine ultra dense wafers that can have compute, storage, and memory all in one SoC.

First, unify compute and memory. Then, later, unify those two with persistent storage so that we have something like RAM = VRAM = Storage.

I don't think this is around the corner, but certainly possible in about 12 years.

replies(1): >>44539180 #
1. inkyoto ◴[] No.44539180{4}[source]
I am also of the opinion that we are heading towards the convergence, although it is not very clear yet what the designs are going to converge on.

Pretty much every modern CPU is a hybrid design (either modified Harvard or von Neumann), and then there is SoC, as you have rightfully pointed out, which is usually modified Harvard, with heterogenuous computing, integrated SIMT (GPU), DSP's and various accelerators (e.g. NPU) all connected via high-speed interconnects. Apple has added unified memory, and there have rumours that with the advent M5 they are going to change how the memory chips are packaged (added to the SoC), which might (or might not) lay a path for the unification of RAM and storage in the future. It is going to be an interesting time.