Bit of a blast from the past, I did own one of these and certainly it felt faster than anything I had used before. Am I right in saying 10k RPM is about the upper limit for spinning disks?
20,000 RPM disks were trialed but I don’t know if they ever saw production. Part of the problem is keeping the platters together at those speeds; and about that time SSDs took over.
20k rpm is relatively clunky F1 combustion engine territory.
I find it hard to believe spinning an electric motor and a relatively light and small disc that fast is much of a challenge to keep together, especially in such a coddled environment.
Even back in the early 90s my RC10 had a 38k rpm "modified" motor; the Motown Missile. That thing lived through hell...
Those motors don’t need to keep minuscule tracks lined up with increasingly small read/write heads. A dremel spins at 40k RPM but even though seek times are low total storage is also low.
This is 100% not true. The engines are sealed by the FIA and replacing parts of the engine incurs gird penalties. You have a limit of 4 engines a year you can swap without a penalty. In the last 2 races both Lewis and Valtteri took penalties for swapping out the ICE part of the system. When Max had the crash at Silverstone with Lewis, RedBull was not 100% sure of the state of the engine as they are not allowed to disassemble it. They had to use fiber optic cameras to look inside. Even then they got it wrong and Max took a new engine in Turkey.
Being pedantic, the modern engines that you're talking about don't run at 20,000 RPM either, they are limited to 15,000 RPM and I believe they basically never actually reach that limit, usually topping out at 12,000-13,000 instead.
When the engines did run closer to 20,000 they were indeed rebuilt much more often. I am not well versed in F1 regulation history but Wikipedia claims that before 2005 engines were not required to last for two race weekends[1], meaning you could rebuild the engine between each race weekend if you wanted to. At that time there was no RPM limit[2] for the engines and the iconic Ferrari F2004 supposedly maxed out at 19,000 RPM[3].
So maybe the comment you are replying to is referring to pre-2005 F1 engines :) I have no idea myself if a modern F1 engine could run at 20,000 and still be as durable as the current engines or if running at such high RPM inherently means bad durability.
We're talking about a tiny spinning disc on a brushless hub motor. There's basically a single moving part, maybe a few more if riding on ball bearings.
Do you have any understanding of what is going on inside a many-cylinder internal combustion piston engine spinning at 20k rpm? We can view the flywheel as the hard disk platter equivalent, the real madness is at the reciprocating mass being flung back and forth at the same rate.
Edit: Here's another useful reference point to help put RPM numbers into perspective: a turbocharger's rotating assembly spins on the order of 200-300k RPM without flying apart. A minute is a pretty long time.
You're underselling what HDDs do. The seeks they have to do are so precision, if you made the hard drive the size of the earth, the "head" would still only be a couple meters from the ground and it would have to go any square meter on the entire earth in 1/100 of a second. It's absolutely incredible that the tiny SATA bay in my computer holds an 18TB drive. That's 18 * 8 trillion bits of data, or if the hard drive had the surface area of the earth, 282 bits per square meter.
This precision structure has to be maintained at 10k RPM. Can it be maintained at 20k RPM? Maybe not so easily. Let's not undersell the technology.
okay, i guess in my mind I automatically translate “coming apart at those speeds” to “the microstructure stretches to a degree it’s impossible to read/write to the drives”. Nothing catastrophic.
Yes I had a 52x drive in the early 00s shatter a cheap writable CD (I think rated at 24x or something along those lines) and I was picking bits out of the drive to get it working again. But then again, a CD is an insanely cheap piece of plastic, so how fast should we expect it to reliably spin at?
A 50x CD drive actually reaches 10,000RPM, and that's with a cheap and wide piece of plastic. So I really doubt holding together is the limiting factor for these drives.
Not as SATA or IDE? The Cheetah is SAS/SCSI I think. I have (somewhere) a 15k RPM 143GB SCSI drive which I used in a home built Windows Server machine for a while. At the time, I had never seen Windows boot that fast. Ahhh.. simpler times.
No. There's a tradeoff. Faster drives means more power, more heat, more vibrations, more noise, less durability. Spinning disk is mainly about price per byte at good performance - there's simply no market.
100%. This is why when I ordered my home NAS I pick 5400RPM NAS drives (FreeNAS Mini). The WD Red drives in my current system have been spinning since 2014, 4x4TB in a mirrored stripped set for a whopping 8TB of space + 2x120GB Evo write caches (mirrored).
The key to speed is having lots of drives, RAM and SSD write cache.
I plan on getting the new TrueNAS Mini XL this year with 8x14TB.
Hey, I'm building a computer with lots of storage, you sound like someone who knows what they're doing... maybe you could help me... currently the plan is:
3x Firecuda 4tb NVME
2x QVO 8tb SATA SSD
4x WD Ultrastar 18tb.
My use case is I need tons of storage that operates fast and is used for all sorts of things (creative work, gaming, archiving, job processing), sometimes all at the same time. It's a threadripper pro workstation pc, so I have spare PCIE slots to upgrade later.
I'm thinking of replacing a couple of the QVO SSDs with Ultrastars and using a couple of the FireCudas as cache drives for the platter drives. Good idea or bad idea? Would I be making a meaningful tradeoff or should I just go for the extra space?
Depends on what OS you are running. In my case I am using TrueNAS so it is made for being a NAS and you can just tell it, hey here is a cache, here is a log drive, etc.
If you are talking about local storage for a workstation then I am not sure. Depends on your workflow. If you have a "work on this on the fast stuff, then when I am done I can move it to the spinning rust" then you might want to figure out largest project size for fast vs long term storage of the projects.
