@tereško Phones aren't very nomadic; your phone IP roams with you across towers.

Anything built on top of IP doesn't really have to care that it's a phone (except it's probably CGN'd)

Bottom mounted PSUs, fan up or down?

@MichaelFrank Up.

I do have a bottom grill, but this PC sits on carpet.

Well, facing into the case usually.

Usually the PSU fan is part of the air extraction of your computer.

So it pulls air down into th.. right

If you have lots of other outflow fans and are concerned about the PSU being exposed to higher temps from inside the case, you might want to point its intake out, but usually it's in.

@fixer1234 If you're around sometime we can try to take a look at that script more interactively.

@Bob, the plot thickens. I just retested the same site in Firefox, and now that works. If I remember right, the throttling message from before popped up when I loaded the SU page. We may never know what happened. I'll play with some sites in Firefox just to confirm everything's working. I started by installing the script in Tampermonkey in Vivaldi (my primary browser). I just reinstalled for good measure. In Vivaldi, the script still has no effect. Want to spin off a room to investigate?

@fixer1234 sure, lemme just find a vm to chuck vivaldi onto..

@fixer1234 probably fixed, try now

(everything comes up as pending when not logged in)

@Bob, yup. It looks like it's working in Vivaldi now. Nice job!. I was looking for one with a pending review and couldn't lay my hands on one. So I looked at your image and finally figured out that it was the same one. Then couldn't figure out how you went back in time to test it before the review was complete. Then noticed the thing about not being logged in. Shouldn't the status show whatever the status is, regardless of whether you're logged in? Seems like that could get confusing.

@fixer1234 Unfortunately, the timeline doesn't show review completed when not logged in

Though the link to the review works and I could in theory go there and scrape that - but that's an additional request.

Alternatively I could try to detect not logged in and show a different message...

@fixer1234 Actually, I'm not sure pending reviews show at all.

I tried a few StackOverflow ones and even if it appears in the review queue it won't appear in the timeline until the review is completed.

@fixer1234 anything that works in chrome should work in vivaldi ;p

@Bob, just caught the embedded link to the actual review summary. Nice touch.

@Bob, is there a simple way to check log-in status? If so, it might be better to just not show the message if not logged in. Actually, if the timeline data doesn't support displaying the pending message, the simplest solution might be to just leave that option out.

@fixer1234 Login can be checked by looking for the login button, if nothing else :P

I'm not sure what actually happens with a pending request

maybe the ones I looked at had 0 votes so they didn't come up?

@Bob, the review queue data is hidden to protect the process if you haven't voted yet (guess they don't want people basing their vote on what other people have done). You can't see any pending votes before you vote yourself. I wonder if that's what's happening with your access to the pending data. Probably need to find a question in the reopen queue with just a few votes that you haven't voted on yet. Test the script. Then vote and test the script again.

Of course even if that's what's going on, it complicates displaying the status with the script.

@fixer1234 turns out, it's not just logged in vs not, it's anyone without review privs

I can see the reopen event, but not the completed status

hm

but even without review privs it's possible to click through the link and see the stats ... except that's not in static JS, the review queue loads the stats dynamically

which makes it much harder to scrape

@Bob, I've been having trouble finding a pending reopen case. Are you aware of any on SU?

I'm not really in the mood for reverse engineering whatever requests the page JS makes

@fixer1234 No, I went to SO because they have more

Gonna grab lunch first and then I'll probably distinguish "completed" and "unknown status"

maybe just have a message to view the review in case of unknown

@Bob, the SE UI is insanely complicated. Even trying to do seemingly trivial things isn't trivial. :-)

@fixer1234 updated

You might want to wait a bit for GitHub cache to refresh so 1.1.1 comes down

1.1.0 doesn't work (...I may have introduced a bug :P)

There we go.

@Bob, updated. Not being logged in works as you show it. New information. No signal on screen was recently closed and untouched. I edited it, which bumped it into the reopen review queue: superuser.com/review/reopen/889536. So I know it's in the cue. The review queue did not show up in the question's timeline, and the script didn't detect it.

Then I went back to the review queue and voted, so not having voted was not an issue. The question still doesn't show the review in the timeline and the script doesn't detect it. The timeline shows the different reviews, like First Post and Closure, with a status of "Completed", as if it might show a review that is pending, but that doesn't appear to be the case. It looks like it reports on queues only after completion.

But that's a question with a known pending reopen review if you want to check whether that status is accessible.

BTW, I suspect that question will never be reopenable, which is why I picked it. But if it does get turned into a gem that should be reopened, I'll recruit the votes.

@bob, I was thinking of letting people know about your gem. Any problem with my posting some kind of notice about it, maybe on Meta SU?

@fixer1234 Sure, go for it. Not sure how much interest there actually will be but I certainly don't mind people using it :P

Lawn mowed, now all sweaty.

@djsmiley2k Why is your lawn sweaty?

:D

I should really go and cut my grass

Jeeves, fetch the scythe!

lol

well, technically it gives off a load of moisture when cut

Fair point :P

Mine needs cut too

One of the cats that hangs out in my garden killed a small mouse the other day

Couldn't find it, either they ate it, it ran off or it's too well hidden

Don't really want to run over a mouse with lawnmower

Probably wouldn't do either of them a whole lotta good

i doubt the lawn mower would notice

and if the mouse is still around, it's not going to be in any state to notice...

tomshardware.com/reviews/… (@JourneymanGeek)

@bwDraco the only complaints I have about my 2018 is 1) they didn't have black when I got it 2) If it had a matching footstool I could sleep in it its so good 3) It ruins any lesser office chairs for me ;p

I might actually get a new one when I move out, and leave this for dad or hand it down to hypothetical future mrs geek

@djsmiley2k I have plastic blades that seem to break at the slightest provocation :P

ah pffffffft/

$360 for the Omega?

I'm sure when I checked it was 500 Queen's pounds

@djsmiley2k In my defence, I have had three hand-me-down lawn mowers, and this is the only one that is still functional :P

Ah, the Omega is currently £299 here

So... Ryzen 3000 series reviews are going up in an hour. And the Women's World Cup final is in four hours.

It's going to be an eventful day.

I haven't had a huge amount of rest last night ._.

i feel like i've not slept this weekend

wonder if there's a leak on my mask :/

saying that I've got a sore throat this morning, so I likely just getting ill.

@bertieb SL literally never sells it at RRP

There's one sale or another

There's other vouchers and stuff you can claim but we just claim amazon vouchers every time

anandtech.com/show/14605/…

The embargo has been lifted. Let the 7nm revolution begin!

extremetech.com/computing/…

pcworld.com/article/3405567/…

I've just started to read the reviews, but multithreaded performance is utterly unsurprising.

That is not how it works. That is not how any of this works

Engadget is generally not my first source for tech reviews.

These kinds of errors are precisely why.

I tend to stick with OC3D and GamerNexus

AnandTech, Tom's Hardware, TechSpot, PCMag, and ExtremeTech.

The advantage is you can get better (full?) performance with fewer lanes for things like multigpu and crossfire

@Aibobot every time I read that sentence it makes less and less sense

@Bob ;p

THATS NOT HOW IT WORKS. THAT IS NOT HOW ANY OF THIS WORKS

Gaming performance continues to fall short of Intel's best, at about 90% of the i9-9900K, but this is primarily an issue of clock frequency. The Zen 2 architecture still struggles a bit with memory latency, which games are very picky about. Faster clock frequencies and larger caches have helped but it's not quite enough.

like, ok, more bandwidth ... but load times? NVMe? ok I suppose you can get faster NVme driv... hold on weren't you talking about GPUs, not the CPU/mobo??

lol.

they confused GPUs with storage

but, eh, I'm still 2-3 years off even beginning to plan an upgrade

I suspect the big advatages of PCIe 4.0 and 5.0 is bigger numbers when you have more devices

I just built a PC ;p

slightly annoyed I chose kaby as the gen to go on, but ... I honestly haven't noticed any CPU-realted latency so... bleh

@Aibobot not really, but what they did is is imply that current GPUs can saturate PCIe3

I want to do some incremental upgrades but I don't see anything particularly useful there

@tereško PRETTY sure they don't saturate even x8

@tereško last I checked you barely notice a slowdown running a 1080Ti on PCIe 2.0 x4

let alone x8, x16, 3.0 or any combination of those

In future, sure. But there's going to be a point where its diminishing retutrns per device

I'm mostly interested in either faster PCIe or more lanes for expansion (SAS cards :D)

but more lanes is more useful

because most expansion cards don't support the higher speeds

@Bob for most filthy casuals people that's less of an issue

heck, I still haven't found a SATA card that can do PCIe 3, rather than just 2. and the cheaper SAS cards are 2 only as well

I think on my builds I have 1x x16, and at most x4 NVMe lanes in use

there's a couple SAS cards out there that do 3 at a decent price, second-hand

@Aibobot yea, for most people PCIe 4.0 honestly means absolutely nothing

and I have no spinning rust at the moment, that will go into a future storage box

and if they weren't chasing more numbers you could stick a GPU on 3.0 x8 and have no problems at all

@Bob its a bigger number ;p

Hmm... AnandTech seems to be getting closer to 95-100% on their review.

@Bob I'd also say, sensibly done, and with a few minor improvements, most folks can manage with the sort of computing power we had... 5 years ago

@Aibobot 5 years? my dev box at work is mostly 2013 hardware and honestly does fine

I wish I had more RAM, but the ivy bridge CPU is performing well

heh

@Bob high end systems would be a ... snap

I will be looking to upgrade when the next AMD socket comes to market

(and the RAM thing is more budget than technical... the system supports 32, I'm on 16)

@Aibobot FWIW, I've been pleasantly surprised by the performance of my Surface Go and its slow Pentium Gold 4415Y. It's memory that's more of a limiting factor these days.

@Bob only moved off Ivy cause shiny... and possible hardware issues

@bwDraco I got my chinese tablet dialed in

@Aibobot I'd still be on Sandy if it wasn't for wanting more expansion and a more reliable mobo :P

the big limiting factor is the emmc

I'm actually not entirely sure the Kaby vs Sandy CPU diff affects me much

(and the being annoyed at picking Kaby is more of a ooh shiny thing than a real need)

@Bob I think I was having odd driver issues. And possibly overheating. I redid the heatsink, and need to do a fresh install to test

@Bob water cooled core i9.... I think that's pretty much the definition of "OOH SHINY THING" )

@Aibobot Sandy was on the pre-on-PCH-USB 3.0 gen, so the mobo had a Renesas controller ... one of the buggy ones

that's a good chunk of the reason I wanted to upgrade

Other than my new obession with those wierd chinese motherboards I need to get over ;)

the other being the 4 SATA 2 ports, rather than 3

peripherals and storage!

I don't need a quarter terabyte of ram and 20 cores.... even if I could likely speed up drying laundry with how much heat they generate

@Bob current build is all M.2 SSD - I currently have my old HDD on my desk 'in case'

I am debating a mini itx build for just storage eventually

I could buy a NAS but... boring

@Bob though, I suspect that's also cause PCIe 3 is overkill for most storage

Fastest you'd need is 10gbps SAS and that's at most... 4 lanes worth?

@Aibobot You probably know my storage problem :P

@Bob yup

@Aibobot I've been debating building a NAS but I'd want a bigger case just to shove all the HDDs in

@Aibobot Nah, problem is when you try to conserve lanes

@Bob well, that's one option

the most common free slot is a x1

PCIe 2.0 x1 is slower than SATA 3

if you had a x2 or x4 it's fine

I kinda was looking at 4u cases on aliexpress ;p

PCIe 3.0 x1 is also fine

pricy + they're just expensive enough that I'd need to pay sales tax ._.

I've got my PCIe 3.0 x8 SAS card in a x4/physical x16 slot

ah

which has enough bandwidth for the 4 SSDs I hang off it

simultaneously

would a HEDT or workstation board work better?

if it were PCIe 2.0, I'd need x8 if I wanted to run all 4 SSDs at once (and that's ignoring that it supports 8 SATA drives without expanders)

ah

@Aibobot yea, given you'd have more lanes available

but PCIe 3.0 more or less hits the SATA 3 sweet spot

PCIe 4.0 might be useful if you wanted to cram more on a x1, or wanted to do NVMe

but NVMe is expensive

@Aibobot problem with HEDT/workstation is the cost

Well, yes, and you'd probably be looking at U.2 not M.2

@Bob I was going to say ... ;p

I got my SAS card and cables for around 100 AUD

Unless you have dave's dumpster

for that price (+ my high-mainstream CPU + mobo) I might be able to go HEDT CPU + mobo ... of 2 gens ago

might

ah yeah

3 gens ago if we keep in mind HEDT is usually a gen behind

@Aibobot 'tis a magical place

about 20 mins drive that-a-way

honestly I think my dumpster dive potential's decent. Then I see him pull out 2-3 recent AIOs, a oldschool cheesegrater mac...

so far in the past decade I think I've come across 2-3 PCs worth fixing, one of which I picked up. Couple of displays/TVs...

office parks, man

not your boring residential dumpsters :P

lol. My brother and I have been joking about getting a flatted factory unit as a workshop

@Bob singaporean residencial dumpsters have a few advantages

High person density per dumpster

and asian dumpster etiquette

If its dangerous, you cut the cable. If its usable, you put it near, not in the dumpster...

and unless you're looking at credit card embossers...

Overall performance vs Intel 9th Gen: Gaming 90-100%, multithreaded workstation 125-150%.

I think some optimization work may still be needed to account for the Zen 2 processor's heavy reliance on cache. Workloads that fit in cache excel on this architecture. While AMD has done a lot of work to hide memory latency, individual memory accesses have actually gotten slower.

All told, can't say I'm disappointed.

I'd want to think that games can be optimized to perform better than this. Memory latency has always been an issue with the Zen architecture but the larger cache and improved memory subsystem, including the ability to efficiently handle lots of simultaneous pending memory accesses (memory-level parallelism), offer a lot of room for software optimization.

64 MB of L3$, anyone?

In other news, I repotted a load of my cacti

Looks much better now

@bwDraco Linus Tech Tips shows that if you restrict Battlefield V to running on the threads of a single CCX (instead of all of them) it can beat the 9900K, but otherwise Intel crushes AMD on that particular game, so clearly infinity fabric "2" is still a bottleneck if you have threads banging on the same data across multiple CCXes

how quickly/easily can you limit programs like that

it's literally faster to run Battlefield V on 4 cores instead of 8 or 12 for Ryzen 3

it almost sounds like windows understanding of NUMA is still broken

@djsmiley2k tech-recipes.com/rx/37272/…

you just have to know which cores correspond to which CCXes on your CPU

but is there ever a case where you wouldn't want this to happen?

yes, it depends entirely on the data access pattern of the threads

I feel ike, windows and maybe linux's, understand of 'multicore' is now almost a problem

because they view all cores as equal

(by default)

if you have 12 threads running on unrelated data, distributing them across all 12 cores (3 CCXes) is completely fine and you'll saturate your CPUs without using much infinity fabric at all

I meant, for a perticular program/thread usage

I understand that in general, you want to spread stuff as much as possible.

if you have 12 threads constantly accessing the same data (and updating it in a very tight loop), you might just want to put it all on one CCX

there might be some sort of compiler hint about thread data locality that would give the C runtime some sort of idea of whether it should run the threads in the same CCX or try to spread them out... if there isn't, there should be

@allquixotic This is the kind of thing I'm wondering about

do schedulers understand this, yet.

because, manually limiting something as generic as BF to certain CCX's seems...... lame

that sorta stuff should happen automatically, at least be hinted etc, etc.

tomshardware.com/reviews/…

> AMD carves the Threadripper 2990WX into four NUMA domains that cannot be altered. As such, the processor does not have a local memory toggle for its Game Mode feature. Instead, the processor simply flips into "1/4" mode, which disables all but one die and effectively creates an 8C/16T CPU. Ryzen Master also has "1/2" and "Off" options that expose 16 cores and 32 threads, or 32 cores and 64 threads.

so they've obviously been thinking about this; "Ryzen Master" is their proprietary Windows software that lets you tweak your CPU

that's for Zen+ threadripper and not Ryzen 3, but the concept is probably the same

I'm guessing LTT didn't test "Game Mode" if that's still an option for Ryzen 3

either that or it's exposed to the system as a UMA chip, which seems unlikely given the chiplet architecture

that would be a bad idea imo because the performance definitely does not resemble a UMA chip

the problem is, even if the OS knows it's a NUMA chip, the performance of different workloads will range from: horribly unoptimized (with lots of threads updating and reading the same data), to decent (with some shared data and some non-shared), to amazing (with each thread not sharing any data with any other thread)

and the frequency of shared data update basically dictates how frequently the infinity fabric has to sync data across the CPUs' L3 caches

and you end up wasting computation on speculative execution when it turns out that the shared data needs to be updated

these are general problems with multithreaded workloads in general, but on Intel and non-Infinity Fabric AMD CPUs, all cores shared the same L3 cache, and data transfer between cores was not really a factor in performance

@allquixotic hasn't that generally been an issue with AMD's CCXes? EVen back to bulldozer?

@djsmiley2k The working theory is that it's better to schedule a thread to run on a free core, even if on a different NUMA node, than force it to wait for a slot to open on the same core it was previously running on.

This is generally true.

But having to worry about NUMA in general-purpose computing is weird.

something something bigger numbers ;p

It's also a case where you can't simply throw more threads at a problem.

In theory this can be dealt with using thread affinities but the current infrastructure around that is fairly sledgehammery.

Because this hasn't been a problem until recently.

(It's a bit like running these games on a 2x socket server...)

@djsmiley2k I have no TV license, so don't think I'd be eligible

@Bob Do Don't do what I did and spread the drives over two cases :D

lol

@allquixotic Already talked at length about this. It's just like a 4P server with memory attached to only two sockets.

news.ycombinator.com/item?id=20373872

@JourneymanGeek so I'm going to explore thinkpads a bit.

@rahuldottech why do you think low key travel is bad for programming / lots of typing? I think I type just as fast, if not faster, on a modern MBP keyboard than a nice mechanical keyboard (and I have some really nice ones)

also maybe it's me but Cmd+Left / Cmd+Right / Cmd+Tab / Cmd+C / Cmd+V are easier, more convenient shortcuts than Home/End/Alt-Tab/Ctrl+C/Ctrl+V

for one thing, I have to reach back to the left behind my left hand to reach the ctrl key, but the Cmd key on a Mac is right there, easily accessible by thumb

something something vim keys

@allquixotic my l;ittle finger reaches my ctrl key easily but alt (or in your cae cmd) involves my thumb moving backwards, whcih is unconformable. -- horses for courses.

@allquixotic funny thing is I completely cannot type on a mac keyboard, least the older ones

A good chunk of it is personal preference - but I don't really like the modern thinkpad keys compared to the classic scalloped keys

@DavidPostill can't remmeberi f you watch TV either

as @bertieb is out :/

@djsmiley2k Not really.

@DavidPostill a post you hammered close is brought up on MSE

Maybe ask the resident doggy to migrate it ...

@JourneymanGeek ^^^

Done

ah for the love of 'too localised'

@allquixotic Also... on the Ryzen 9 3900X, two cores are disabled on each CCD. I wonder if workloads that use more than six cores will behave better on the fully-enabled Ryzen 9 3950X...

Overall, it does look like some optimization is still going to be necessary to make the most of these processors. But no matter how you look at it, this is a serious competitor to Intel's top-dollar processors for practically any workload.

The Ryzen 7 2700X performed at about 75-85% of the i9-9900K. Here, we're looking at 90-100%.

Hey folks, you've probably see this already. but if you know any teachers, they can get Zachtronics games for free for educational purposes. zachtronics.com/zachademics

@djsmiley2k i don't watch enough TV

does it track streaming services/DVDs?