620GB/platter for a 3GB unit

"suggest that the capacity is actually measured at 1000B/KB instead of 1024 bytes/KB, which would make the drive capacity about 1.86TB and not 2TB"

Hasnt this always been the case with harddrives?

If you check the capacity of any drive (and I mean before partitioning/formatting) you will find that the size is closer to the SI unit than the computer unit. A 500GB drive will be closer to 500,000,000,000 (500 x 10^9) bytes than 536,870,912,000 (500 x 2^30) bytes.

Hard drive manufacturers have used SI units for years. This is why there is confusion. Hence, many refer to 1024 bytes as a KiB (kibibyte). So, in these terms 2TB=1.86TiB.

I think the author is being nostalgic.

Back in the day, pre-1995 or so, hard drive manufacturers all used 1K = 1024, 1M=1024^2, and 1G=1024^3. Gigabytes were gigabytes.

Around the mid-90's some marketer noticed that bigger capacities could be claimed by changing the unit from 1024 to 1000. It was borderline dishonest and rather deceptive, and some people (like me for instance) are still irritated by it.

If you've been buying disks only since about 1995 you may not know about this.

Imagine if tomorrow they decided to quote disk capacity in GB defined as, oh I dunno, 976^3 bytes. You might find that to be questionable.

Contrary to common belief, power-of-10 prefixes are much more commonly used than power-of-2 prefixes in the computer industry. The only few places where the latter are used are to refer to RAM capacities and file sizes, whereas power-of-10 prefixes apply to most other areas and all units (not "only bitrates", as some claim): storage capacity, clock frequency, stream bandwidth, baud, pixel numbers, data throughput, processing power, etc.

o A 8 GByte USB flash drive is 8 * 10^9 byte (power of 10)

o A 50 GB dual-layer Blu-ray Disc is 50 * 10^9 byte (power of 10)

o A 4.7 GB single-layer DVD is 4.7 * 10^9 byte (power of 10)

o A 2.5 GHz processor is 2.5 * 10^9 Hz (power of 10)

o A PC3200 (as in 3200 MByte/s) memory stick is 3200 * 10^6 byte/s (power of 10)

o A 25.6 GFLOPS CPU core is 25.6 * 10^9 FLOPS (power of 10)

o A 128 kbit/s audio stream is 128 * 10^3 bit/s (power of 10)

o An 8 kbaud V.92 modem is 8 * 10^3 baud (power of 10)

o A 6 Mpixel digital camera is 6 * 10^6 pixel (power of 10)

o A 4000 MB/s HyperTransport link is 4000 * 10^6 byte/s (power of 10)

o A 480 Mbit/s USB2 link is 480 * 10^6 bit/s (power of 10)

o A 2.5 Gbit/s PCI-E lane (after 8b/10b encoding) is 2.5 * 10^9 bit/s (power of 10)

o A 250 MB/s PCI-E lane (before 8b/10b encoding) is 250 * 10^6 byte/s (power of 10)

o A 1 Gbit/s ethernet card is 1 * 10^9 bit/s (power of 10)

o A 54 Mbit/s 802.11g network is 54 * 10^6 bit/s (power of 10)

o A 3.0 Gbit/s SATA link (after 8b/10b encoding) is 3.0 * 10^9 bit/s (power of 10)

o A 300 MB/s SATA link (before 8b/10b encoding) is 300 * 10^6 byte/s (power of 10)

o A 6 Mbit/s DSL line is 6 * 10^6 bit/s (power of 10)

o Curiosity: a 1.44 MByte floppy disk is 1.44 * 1000 * 1024 byte (mix of power of 10 and 2)

o And of course, a 750 GByte hard disk drive is 750 * 10^9 byte (power of 10)

2Tb will bring about yet another reduction in the I/O density (IOPs and MB/s per GB). It's in the geometry - the Greeks would have understood. Capacity scales by area, sequential read speed linearly and random IOPs scarcely at all...

I think you'll find drive storage has been given the base 10 spin for many many years, even back to the days of 100K floppy drives in the early-mid 80s... That's as far back as my memory goes, I wouldn't be in the least surprised if it was even older.

Back in those days the small missing 2.4% wasn't noticed, however now it's been compounded by the rounding of first Kilo, Meg, Gig... 2.4 x 2.4 x 2.4... and people notice 13% missing!

Speaking only of 3.5 inch IDE and SCSI hard disk drives, I clearly remember the unit changing in approx 1995. Disks were available in the several hundred megabyte range, a megabyte was 1024^2. When the unit was changed fine print stating that 1MB=1000000 was added to the packaging for retail disks. At the same time the stated capacities did a modest jump. IIRC Quantum did it first and the others shortly followed. This was commented upon in the PC press and by hard drive manufacturers.

I have no comment about floppies.

IBM disk were sold till 2000 in heavy Gb (1024^3 B.)

The stopped that as I believe they were the last ones and loosing business form it. I do believe I still have 120 GB drives in Heavy GB while the 250 are if I remember right light Gb (10e9 B)