On Wednesday, 5 December 2018 at 19:59:46 UTC, Joakim wrote:
On Wednesday, 5 December 2018 at 09:04:49 UTC, Walter Bright
wrote:
#4 on HackerNews front page!
https://news.ycombinator.com/
33 points at the moment!
Now one of the top-voted links on the front page of HN.
I'd just like to point out that Andrei put Liran and I together
to do this interview in summer '17- though I had emailed Liran
about doing one in '15 and never got a response- and he finally
got some time to respond this summer.
Now I just need Andrei to finally do an interview, which he's
been putting off for even longer. :)
Awesome work!
Question about one of the comments, because I don't understand
the hardware aspect of things. this one:
https://news.ycombinator.com/item?id=18607297
Pasted here:
"Weka is a name of a Machine learning product from New Zealand.
The IBM system had 24xHGST N200 SSD, 830 000 random read, 200 000
random write. that is 24 * 200 000 IOPS=4.8Million random writes.
The Matrix system had 64x1.2TB Micron 9100 SSDs. 750 000 random
read, 300 0000 random writes. 64 * 300 000=19.2 Million IOPS
Ok so you have benchmarked hardware with 4.8 Million write ops vs
19.2 Million write ops."
Anyone know if that last line is correct?