Search
Tuesday, 3rd of December 2019, 6:48:47 UTC
7:01:53
loke`
beach: Would you be able to create a redirct for lispos? I have posted links to that PDF in many places.
7:23:17
beach
Looks too hard. I'll just copy the file.
7:26:12
beach
By the way, there is a new section about checkpointing using log-structured secondary memory.
7:27:03
beach
Also, I recently read that SSDs are getting cheaper fast, so there may be less concern with disk performance.
7:27:55
loke`
There is almost some progress is new memory types. I can't remember what it's called, but it's sersiustent, and (almost) as fast a DRAM
7:28:02
loke`
So basically like good old core memory
7:28:52
beach
Yes, there is good progress.
7:29:59
beach
3D Xpoint looked dead for a while, but I just read an article in CACM that claims that it is getting better.
7:31:34
no-defun-allowed
ACTION reads the wikipedia article on 3D Xpoint
7:32:49
no-defun-allowed
The name "Optane" is familiar. I had to disable that on my new laptop to get dual-booting Linux and Windows working, sadly.
7:35:26
beach
I believe it's Intel's name for its 3D XPoint product.
7:35:34
no-defun-allowed
Intel's name for storage devices that use that 3D Xpoint stuff. (Micron calls theirs QuantX.)
7:37:42
no-defun-allowed
But I don't have any of that to my memory (as I didn't have to sell any kidneys for this laptop). I think it's just their name for anything that caches a slower storage device with a faster one.
7:38:06
jackdaniel
beach: thank you. I might have asked that in the past, but do you have an index of lisp articles on metamodular?
7:39:24
jackdaniel
understood, no worries
7:43:45
nirved
disk blocks might need additional checksums for reliability
7:45:01
beach
nirved: I took that into account in section 7.2.
7:52:28
nirved
beach: sha should be the very first thing in header, and done over the rest of the header as well
7:57:58
nirved
if it's easier, then the final checksum might be sha(sha(rest-of-header) | sha(data)), "|" here is concatenate
7:59:55
beach
I am sure you are right. I don't know enough about that stuff.
8:00:24
beach
I don't think I elaborated that stuff in detail in the document.
8:01:50
nirved
this is just an idea, I think the headers should have checksum(s) as well, just for added reliability
8:13:14
loke`
ck_: No. Memristors don't exist.
8:14:03
loke`
Basically I'm talking about future generations of the Xpoint stuff that no-defun-allowed mentioned.
8:14:23
loke`
I think eventually they'll be as fast as DRAM.
8:15:56
loke`
It seems NVDIMM is a term that is used.
8:16:34
loke`
NVDIMM-P specifically
12:11:20
loke`
How do I change the default text style for a pane?
12:36:53
scymtym
loke`: maybe (reinitialize-instance PANE :text-style TEXT-STYLE)?
12:42:29
loke`
I came up with a better solution
Tuesday, 3rd of December 2019, 18:48:47 UTC