@bwDraco Only because of SLC caching. And we've only barely gotten to the point of TLC being "acceptable"... bringing on the QLC sounds like a repeat of early TLC :P
My desktop has it because "longevity" (again, Astaroth was designed to be a long-term platform), but 98% of today's consumers will see no benefit from MLC NAND. 3D TLC NAND has gotten to the point where it's satisfactory for the vast majority of consumer uses.
@Bob That's... basically what they're for, yeah :-P
During the first phase of 4G deployment here, all the legacy 2G microcells got turned off but some are slowly being reactivated on 4G to deal with concentrated load.
A couple operators make a big deal of deploying (read: tech demo'ing) some advanced microcell technologies in certain flagship venues (read: O2 in the O2 Millenium dome, EE in Wembley Stadium, etc.)
Wembley tends to be where EE shows off their latest LTE CA tech (450Mbps was first deployed there, 800Mbps was first tested there, dunno what they're doing now, probably gigabit class LTE)
I think QLC NAND will be sold aside cheap DRAMless SSDs for capacity. 3D MLC NAND of the sort Samsung makes will still be around for some time but it'll eventually disappear. Their MLC drives commands a big premium (look at the 970 PRO) and yet they're still selling reasonably well on Amazon.
Some do a better job than others... Some have a habit of putting illegal masts up and ignore planning laws...
@Bob If it's a single global firmware then there shouldn't be any differences in the modem binary between different areas. As far as I know there are no configuration options within any CSC files that change/alter/disable/enable any modem functionality. Except for fast dormancy, which was known to be a buggy clusterfuck for a while
There are funky advanced tweaking options in servicemode menus, that are not network specific (I can change my LTE UE category from 9 to 10, it's officially sold as a Cat 9 device), as well as reported 3GPP RRC revision and HSPA capabilities/revision. But as I say, that's basically engineering tweaks and not available to networks via CSCs
@Bob I don't think it does, I've not seen any spec differences on the S9 for the dual sim version. In the UK it's advertised by samsung as the default, "Hybrid SIM" version. Both the regular and hybrid (dual-sim) are advertised as 1.2Gbps LTE
@Bob I wonder if any of mine are TLC? I have a ton of SATA 850 Pros, two new SATA 860 Pros (4 TB), and a much faster but smaller 970 EVO... I bet the latter is TLC but boy does it fly
If you don't use your Sony much you should temporary root it and run NSG on it (P.S. are you still using an MVNO for your main SIM? I think you mentioned they may have a lower limit on those, they certainly do here)
QLC/TLC is probably fine for game / media storage drives or other predominantly WORM workloads - no spinning rust, a very low proportion of the drive gets rewritten per day on average, and demand is for capacity rather than top-end speeds
@allquixotic That's slightly weird, but not unprecedented. 10 year warranties on storage devices are ludicrious anyway, and remember when HDD manufacturers reduced warranties from 5 years as standard down to 3 years then 1 year (then started charging premium prices for "Pro" or "Server" drives just to restore the 3/5 year warranty)?
it's always interesting to me that consumer SSDs don't come in 3.5" form factor - they could use old and partly obsolete fabs and just print more chips to fill up the extra space and have crazy capacity... like, 2 years from now when a 970 EVO is nothing special, they could take the NAND from those PCBs and fit 4 of them in a 3.5" for 8 TB and sell it on the cheap because the fab is well-debugged for that fab size and just cranking out volume is desirable so they recoup their investment
@allquixotic TLC isn't that bad. I've got/had an 850 Evo 1TB as my main drive for several years. Performs like a beast, despite being TLC, still beats a lot of latest-gen low end and mid-range SSDs (even NVMe ones, if disregarding STR)
@allquixotic It's largely the controller chips. They generally don't contain enough channels or capability to drive sufficient NAND chips. That's why all the high-capacity 3.5" and PCIe drives use multiple controllers in RAID
Now there's an idea for a kickstarter... a 3.5" enclosure that just RAID's 4x 970 Evo NVMe drives >_>
> This is largely thanks to Samsung’s mid-range drive adopting nearly the same 64-layer MLC V-NAND technology as its higher-end Pro brothers. Previously, the 960 Evo was built with a more affordable, but slower, form of TLC V-Nand. The updated Pro and Evo drives also share a newly designed Phoenix controller.
@allquixotic All the Samsung TLC drives are fast with smaller loads cause they have a decently large SLC cache. But with a sufficiently large sustained write you'll see speeds drop. That said, even after the drop Samsung still maintains >300 MB/s on SATA, IIRC. Even higher (600?) on NVMe. Much better than most others.
@Bob Yeah, the 970 Evo will sustain 600MB/sec throughout the whole drive, after starting at 1800MB/sec for the first few gigabytes
My SATA 850 Evo sustains about 300-400MB/sec, whereas my Toshiba NVMe starts at 1.2GB/sec and drops to 200MB/sec after a few seconds before overheating and turning off.
960 Pro sustains 1.8GB/sec throughout the whole drive though o_0
@Bob it's odd that the drive wouldn't contain some internal "BBU" like a RAID card does to make sure the DRAM completes write to the persistent storage
Dell issued a firmware update about six months after release that "fixed" it, now it just throttles and stalls when it overheats instead of disconnecting and crashing your system
@Bob Intel and Micron announced they'd be parting ways for next-gen 3DXpoint, because Micron wanted to focus on consumer targets more than Intel was happy with
haha, in 5-10 years Louis Rossmann could reinvent his business as an SSD repair shop for non-controller, non-NAND failures like random caps on the board or something
once the warranty runs out on these drives and people want to keep using them because they're nowhere near their lifetime writes
@Bob yeeeeah, it needs to approach 860 Pro 4 TB prices for consumer interest, otherwise you're looking at ~2011 SSD prices for pure XP drives of similar capacity, and they probably wouldn't sell anything >1TB either.
The main issue with SSDs is cost per GB. Performance is, for most consumers, a solved problem. Most consumers do not need super-fast performance or high endurance.
Tbh you'd have to have a really bad SSD for it to drop below 100MB/sec. My NVMe one is a pretty shit and pathetic cheap-ass piece of OEM crap, but when it's not overheating it'll sustain 150-250MB/sec steady state even when full
@allquixotic Eh... depends on the game. Quite a few are still only so big because they decide to ship every language's audio. IIRC GTAV had 10-20 GB of audio within its 50 GB.
@Bob Yet somehow the NVMe SSD Dell puts in their top of the range gaming machines is considerably worse... I let it get to 49% of 20GB this time just like Techhive's test on the Trion 100:
@Bob When it's not busy overheating, yeah...
(It used to be it'd passively overheat if I was gaming on my laptop and the fans weren't on max)
@RegularGDPR I guess I can say I've never had the Trions overheat. Then again they're sitting in open air and 2.5" form factor, so that's enough (passive) cooling.
Also I feel like an idiot... I just realised I've had a 16 GB page file sitting on an old Intel SSD (320? 120 GB)
@RegularGDPR Ha. I have 32 GB of RAM ... and a 50+ GB page file.
Fucking memory leaks.
Discord with 200,000 handles open comes to mind.
Still annoys me that I can go to the trouble of identifying the specific handles open (registry keys for MMDevices, btw) but they won't accept the bug report because they (meaning: testing users) can't repro
And I say this as a dev who's also reported 3 other major handle leaks, all fixed: to a dev, knowing which handle is being leaked is usually enough to find the cause. And is usually the hard part to isolate, with how inconsistent reproduction can be.
@Bob I've never seen it happen before today, and I've dd'd into C:\ and D:\ from non-elevated CMD's at least once a month
So after manual garbage collection and setting the fans to max to let it cool down a bit, it gets close(r). Bear in mind this thing's Bitlocker encrypted so garbage collection is a little... inconsistent
82-83'c seems to be the throttling temperature, it got to 81'c in that screenshot and then hit 82'c a few seconds later, then began bouncing between 50 and 200MB/s
Wonder what'd happen if I did a file copy straight from RAM cache
Being able to copy 10GB files straight from RAM because you haz lots of RAM... hah I remember the old days of only having 4GB
Hah that was some lolz... 1.7GB/sec 1.5GB/sec 1.3GB/sec 0.8GB/sec 0.002GB/sec 0.3GB/sec
Seems to have an initial 1-2GB SLC cache at >1GB sec, then sustains about 200-250MB/sec before the themal throttle kicks in and it hovers around 125MB/sec. With fans on max, though the fans are nowhere near the drive.
Like I said, this is the sort of crap Dell puts in their top of the range gaming machines -_-
OCZ Triton not looking so bad now huh
Wonder what happens if I let it cool down fully. To 53'c which seems to be the minimum powered-on temperature
Funny thing is my SATA 850 evo, over USB, can sustain a higher write speed while copying internally on the drive itself.
So that's about the best case scenario, from 15+ minutes idle, garbage cleaned, writing sequentially, taking about 2 minutes to write 20GB. Would be nice to have a 32GB SLC cache like on the Samsung drives, so doing a hibernate could be completed in 15 seconds instead of 150 seconds...
Apparently this is the clean-state performance you get on mostly empty 1TB new drive, writing compressible unencrypted data:
This on the other hand, is more like what I get, on a smaller, fuller, encrypted, overheatier drive. Bearing in mind the read speed is still pretty good, so I put games and OS on it for theoretically better load times, but haven't really noticed the difference because as soon as anything starts writing in the background the whole thing grinds to a halt
Thanks to some amazing efforts and collaboration between our veteran users and quite a few new faces that jumped in to help us with our inclusion efforts, our brand new Code of Conduct (CoC) is rolling off the press and going into effect across the network today. For those of you that haven’t been following along with announcements we’ve been making on our Meta site, we’re replacing our current ‘Be nice’ policy with a formal, far less ambiguous and way more informative Code of Conduct.
As far as our rules go, nothing really changes: we’re just clarifying that we don’t have space for belittling …
The code of conduct announcement blog just went out, yet the link to the code of conduct page in it yields a 404 - page not found.
Seems like a bug.
https://stackoverflow.com/conduct
This is on latest version of google chrome / windows 8.1
Should probably sell all my old PC parts, they've been rotting away depreciating for years
Maybe I should stick to investing in companies and markets I actually know about and understand. There's a reason every time I invest in AMD I make £200 in a day
Everything else is just random shit based on dumbass clickbait articles
I have the tab open for some reason, probably because someone mentioned it in here, but I haven't a clue why or what it means. I blame @Bob, because he's the only guy smart enough for me to bother paying attention to :-P
@RegularGDPR Erm. Event listeners by default pause event handling by the browser until the listener returns, to allow a preventDefault. A passive listener does not pause, but means preventDefault does not work. Passive listeners are a new thing; all events are not passive by default. So, a few months back, Chrome made touch event listeners passive by default in the name of performance ... and broke half the web.
Also, it looks like CTIL are going to be installing a/their first second generation Elara mast on SGC tomorrow between 10:00 and 16:00. I'm tempted to drive out there and watch!
@Bob South Gyle Crescent. A road across town that has a recently updated mast already, but is being replaced by a new one
They're replacing all the standard 4G street poles with slightly larger 4G street poles, as well as some internal electronics upgrades in the base cabinets.
I'm curious as to what the exact changes are, since both the masts and cabinets are covered by plastic shells/covers/shrouds to prevent you seeing inside (unless a very strong storm blows them off)
My guess is it's either enabling LTE2600 or LTE2300. Or LTE800+1800+2100, though that'd be relatively useless
But I've never had advanced notice before of work being done so it's rare for me to actually catch them in the act, except by accident which has only happened once or twice.
Heck, I only found out by accident cause I was driving by there the other day to do some speed tests, and saw the advance notice of road closure sign.
Sign says road closure for "Cherry picker operation". Thought "Hmm, wonder what that could be..." Lookup the traffic regulation/authorisation on the council website: "Telecomms company
Also got a new vape mod, which takes 3x18660 LiIon cells (I find it pretty stupid and irresponsible the ecigarette industry is the only one that not only allows but encourages consumers to directly handle bare, unprotected Lithium Ion cells), but the shop sells them for £10 each so I figured I'd just rip some out of this old laptop battery I got for a fiver that contains six 18650's
@Bob I deliberately redialled because the first time it didn't come up as a HD voice call... I just had to listen to my doorbell in HD Voice obviously... priorities...
It's one of these ^^ wonder if it'll actually... well... work
No idea why it's not oneboxing...
A bit overkill, but I figured the USB-C charging passthrough might be useful for a future tablet or something.
Don't need the card reader, but all the ones without a card reader cost more or lacked the display output
Not sure if I mentioned - the reason I want it running off the Thunderbolt port is cause there's something awfully wonky about my regular southbridge USB ports
There'll be times when I walk in the room, and the vibration alone makes one USB device disconnect/reconnect, and in doing so every USB device on every port also disconnects and resets...
Or I'll disconnect a USB drive from a hub connected to one port, and every device on every other port not connected to the hub resets. Which is awfully not useful when half of them are USB hard drives with write cache on
Ah shit just realised my heat gun/soldering iron is out of gas, so I won't be able to cut through this laptop battery casing or desolder the batteries. Fuck.
Interesting. The HDMI port works, but it seems the USB-C display output is hardwired to the Intel IGP.
Which is weird because all the other display outputs are hardwired to the dGPU
The code of conduct announcement blog just went out, yet the link to the code of conduct page in it yields a 404 - page not found.
Seems like a bug.
https://stackoverflow.com/conduct
This is on latest version of google chrome / windows 8.1
