[X] My Assistant

Someone on reddit offered to mail me a 1060 Ti for free.

So I'm gonna wait and see if that happens, but it'd be great if it does, just for the GPGPU stuff. Still would like a Radeon and say goodbye to Nvidia eventually, though.

In the past I've gotten a motherboard and CPU this way as well, although there I felt bad and offered to fix a game boy for them.

Might have to upgrade my PSU if I do get it, though. This thing apparently is designed for double the wattage of my 750 Ti (120W, versus 60W). So it'd probably make the most sense if I didn't use it most of the time and saved on power by using the 750 Ti instead. The VRAM upgrade would be nice, however (2GB -> 6GB).

I think my PSU as it sits would likely handle it alright, but I just wouldn't feel too comfortable with it. Even with my extra cooling exhaust fan ziptied to the back of the PSU.

Wish I could find a 4GB 750 Ti card somewhere. Maybe some day.

Or a Radeon card that could take the place of my 750 Ti, even if the power budget has to go up for it.

This post has been edited by dragontamer8740: Jan 1 2022, 03:42

Someone on reddit offered to mail me a 1060 Ti for free.

seemslegit.jpg

Might have to upgrade my PSU if I do get it, though. This thing apparently is designed for double the wattage of my 750 Ti (120W, versus 60W). So it'd probably make the most sense if I didn't use it most of the time and saved on power by using the 750 Ti instead. The VRAM upgrade would be nice, however (2GB -> 6GB).

have you checked with a killawatt or something similar to see how much your setup actually pulls in reality? like on paper I should be maxing out my shitty 600W psu as well but I only ever pull ~400W at peak

seemslegit.jpg

In the past it actually has been. So I'd say there's a 30% chance it is this time.

have you checked with a killawatt or something similar to see how much your setup actually pulls in reality? like on paper I should be maxing out my shitty 600W psu as well but I only ever pull ~400W at peak

Nope, I haven't.

I know I'm not stressing it too hard, but this is also an old PSU and it always has run warm.

I don't have a killawatt, but I do have a fluke multimeter and a jewel bearing Weston AC ammeter (model 155) from around 1920 that might be useful for this purpose.

Also my PSU is just over 400W IIRC.
Edit: yeah, 430W.

PC has:

• Asrock H77M motherboard
• i5-3470 CPU (TDP around 77W, but I don't use integrated graphics and the similar i5-3350p only draws 69W, so I'm going to bullshit guess 72W).
• 750 Ti, which apparently draws up to 38W according to nvidia-smi
• 7 3.5" hard disks ranging from 2007 to 2021, median age 2009
• 1 SSD
• Audigy 2 ZS
• Some random C-Media card because it can emulate an OPL3 pretty well
• (these PCI cards are on a PCIe->PCI bridge provided on the motherboard)
• ASMedia 1064 sata PCIe controller card

Sometimes one of the PCI sound cards gets swapped for a video capture card or an adaptec PCI-X ultra320 parallel SCSI controller.
If the datasheets are to be believed, I think my HDD that draws the most current can peak at 33.6 watts(!?) (12V * 2.8 amps=33.6 watts). I highly doubt that that happens often in practice.
That's the ST3500630AS, my boot HDD.
[www.seagate.com] https://www.seagate.com/docs/pdf/datasheet/...uda_7200_10.pdf
Average being more like 13 watts.

My guess is in reality I'm under 200W most of the time. In the 150-200 range.

Thought:
Did you know that a lot of Kirby's Adventure for NES was actually written in C (or something very similar to it)? Looks that way, based on the disassembly, at least. Which might explain why it slows down so easily (even for an NES game).
But Kirby's Adventure also did make a lot of use of the NES/Famicom's capabilities; it's a gorgeous title.

Thought again:
I think I like playing NES most on my Apple IIe monitor, even though it's technically the lowest end out of all my color CRT's. I think the dot pitch looks just about perfect for it.

This post has been edited by dragontamer8740: Jan 1 2022, 06:40

I now have a microwave with grill function.

A sausage cooked in microwave mode should be pierced to not explode.
A sausage cooked in grill mode should not be pierced, for the skin to retain the heat and allow proper cooking.
In what state should be the Schrodinger's sausage, in microwave+grill mode?

Seriously, keywords to search how to use this kind of microwave constantly lead to unrelated stuff, I'm not sure how to reliably use the damn thing.

Let me know if any good. I bought an air fryer instead last time as people said the combo units were a bit slow to cook. Also it was peak covid so it was hard to buy the models I wanted to a price I was willing to pay.

I got out of this by mashing buttons, but I have no idea what I actually did to get out. There's no room for running squat jumps here.
I can't figure out what in the code would cause this or let you escape this.
[i.imgur.com] (IMG:[i.imgur.com] https://i.imgur.com/zwGUj25g.jpg)

On another note:
does anyone know a nintendo controller cord, or something that feels like a nintendo controller cord, that is somewhat long and has six or more conductors inside? I need a cord with one wire more than an NES controller usually uses to make a longer cable for player II on my famicom (the microphone gives it an extra wire).

I have player 1 taken care of courtesy of an american NES controller already.

This post has been edited by dragontamer8740: Jan 1 2022, 07:43

Indicator number 5891 that hardware is going to shit. I just popped my WD Black virginity, got a corrupted file on it today. Whilst my 15 years old Seagate is still kicking and in perfect shape.

(I'm annoyed because I'm careful with those disks, a decent journaling filesystem on each and I even hdparm the disks to stop spinning everytime I need to move them. That minimizes the chances of random damage.)

I got out of this by mashing buttons, but I have no idea what I actually did to get out. There's no room for running squat jumps here.

Its possible to go through blocks with absurd pixel perfect timing or mashing luck.

Some of the speed runners use the technique but reliability is the problem.

Indicator number 5891 that hardware is going to shit. I just popped my WD Black virginity, got a corrupted file on it today. Whilst my 15 years old Seagate is still kicking and in perfect shape.

Would yours also happen to be a ST3500630AS? Because between mine and one other guy I know who has double the hours on his, they seem to be rock steady drives. It has one reallocated sector, which it already had six years ago. Oh, and before I got it it had been running hot but I solved that in my machine by giving it a little airflow.

One of my favorite things about that disk is the sounds it makes when it's working. It's always made them, but it's just a sort of loud drive by today's standards. Maybe a bit louder than my black.

My only WD black is from 2015, 3TB. WD3003FZEX according to hdparm. So far so good on that one… but I never spin them down unless shutting the computer off.

I also have a WD green with over one million head parks logged and I keep it out of morbid curiosity as to how much longer it can last.

(I'm annoyed because I'm careful with those disks, a decent journaling filesystem on each and I even hdparm the disks to stop spinning everytime I need to move them. That minimizes the chances of random damage.)

You spin them down and park the heads?
I only do manual head parks on my oldest disks (from 1991 and 1996).

Thought: I've noticed my deskstars give me a lot more SMART information than any of my other drives.
Also, interestingly, my SSD seems to run hotter than any of my HDD's are. Maybe that's just due to fan positioning. To my knowledge, all of my HDD's are configured to never spin down when idle. Though the green parked quite often before I did that.

CODE

# for file in /dev/sd[a-z]; do echo "$file"; (hdparm -i "$file"|grep Model=|sed 's/SerialNo=.*/SerialNo=REDACTED/'); (printf ' '; smartctl --all "$file" |grep Celsius);  (printf ' '; smartctl --all "$file" |grep Airflow); echo; done
/dev/sda # Barracuda 500GB
Model=ST3500630AS, FwRev=3.AAK, SerialNo=REDACTED
194 Temperature_Celsius     0x0022   036   066   000    Old_age   Always       -       36 (0 16 0 0 0)
190 Airflow_Temperature_Cel 0x0022   064   034   045    Old_age   Always   In_the_past 36 (Min/Max 35/36)

/dev/sdb # WD Green 500GB
Model=WDC WD5000AAVS-00ZTB0, FwRev=01.01B01, SerialNo=REDACTED
194 Temperature_Celsius     0x0022   111   091   000    Old_age   Always       -       36

/dev/sdc # DeskStar 500GB
Model=Hitachi HDP725050GLA360, FwRev=GM4OA5CA, SerialNo=REDACTED
194 Temperature_Celsius     0x0002   187   187   000    Old_age   Always       -       32 (Min/Max 14/60)

/dev/sdd # WD Red "Plus" 6TB (non-SMR drive)
Model=WDC WD60EFZX-68B3FN0, FwRev=81.00A81, SerialNo=REDACTED
194 Temperature_Celsius     0x0022   120   114   000    Old_age   Always       -       32

/dev/sde # WD Black 3TB
Model=WDC WD3003FZEX-00Z4SA0, FwRev=01.01A01, SerialNo=REDACTED
194 Temperature_Celsius     0x0022   116   102   000    Old_age   Always       -       36

/dev/sdf
Model=SanDisk SSD PLUS 480GB, FwRev=UG5000RL, SerialNo=REDACTED
194 Temperature_Celsius     0x0022   062   061   000    Old_age   Always       -       38 (Min/Max 18/61)

/dev/sdg # DeskStar 320GB
Model=Hitachi HDT725032VLA380, FwRev=V54OA73A, SerialNo=REDACTED
194 Temperature_Celsius     0x0002   171   171   000    Old_age   Always       -       35 (Min/Max 18/56)

Another thing:
My laptop has this problem where every few months the wireless card starts spitting out DMA reset failure errors. Rebooting or soft-resetting the card via sysfs will temporarily fix things, but it will happen more and more frequently until I open the computer up and remove, clean, and reseat the card. Then it won't do it for a few months.

Wish I knew why I had to do this. Ugh. Minor inconvenience (about eight screws, plus a risk of cracking brittle plastic parts; palmrests in particular love to self destruct).

This post has been edited by dragontamer8740: Jan 1 2022, 21:32

Linus Torvalds dislikes Intel's AVX-512?

Linus Torvalds dislikes Intel's AVX-512?

seems like it, but for good reasons:
[www.realworldtech.com] https://www.realworldtech.com/forum/?thread...urpostid=193190

I hope AVX512 dies a painful death, and that Intel starts fixing real problems instead of trying to create magic instructions to then create benchmarks that they can look good on.

I hope Intel gets back to basics: gets their process working again, and concentrate more on regular code that isn't HPC or some other pointless special case.

I've said this before, and I'll say it again: in the heyday of x86, when Intel was laughing all the way to the bank and killing all their competition, absolutely everybody else did better than Intel on FP loads. Intel's FP performance sucked (relatively speaking), and it matter not one iota.

Because absolutely nobody cares outside of benchmarks.

The same is largely true of AVX512 now - and in the future. Yes, you can find things that care. No, those things don't sell machines in the big picture.

And AVX512 has real downsides. I'd much rather see that transistor budget used on other things that are much more relevant. Even if it's still FP math (in the GPU, rather than AVX512). Or just give me more cores (with good single-thread performance, but without the garbage like AVX512) like AMD did.

I want my power limits to be reached with regular integer code, not with some AVX512 power virus that takes away top frequency (because people ended up using it for memcpy!) and takes away cores (because those useless garbage units take up space).

Yes, yes, I'm biased. I absolutely destest FP benchmarks, and I realize other people care deeply. I just think AVX512 is exactly the wrong thing to do. It's a pet peeve of mine. It's a prime example of something Intel has done wrong, partly by just increasing the fragmentation of the market.

Stop with the special-case garbage, and make all the core common stuff that everybody cares about run as well as you humanly can. Then do a FPU that is barely good enough on the side, and people will be happy. AVX2 is much more than enough.

Yeah, I'm grumpy.

Linus

all the searches for this are from july last year though, shouldnt come as a surprise now.

in other thoughts unrelated to any of this, since my old amd card will be arriving this week, wonder how much of an issue non dkms drivers are these days. since kernels do update a lot and I remember it being somewhat of a pain with nvidia until I found the -dkms, but theres no video-ati-dkms. then again I dont really remember having any issues until I started using nvidia hardware all that time ago either, maybe its not even an issue that occurs anymore these days

I've been living without dkms for years for nvidia.

Lynx is my friend.

I've been living without dkms for years for nvidia.

Lynx is my friend.

the issue mightve stopped existing, I remember every other kernel update the drivers would start complaining or not working, until I got to the dkms nvidia drivers. then again proprietary nvidia drivers, especially in the beginning... not the best thing to go off probably
remeberance edit: 99% of the time it was the drivers werent updated fast enough and they'd be very 'for kernel x'. I cant see the opensource ones ever doing something like that

This post has been edited by cate_chan: Jan 2 2022, 17:12

Correct. The free software ones with upstream kernel code should be fine.

Thought:

Until I can make myself buy a new battery for it, I am setting up my non-tablet X201 (backup laptop) as a streaming device connected to my TV via VGA (displayport's only on the docking station, and I don't have a spare dock). I'm installing OpenBSD on it.

新的一年，继续加油！

I just spent around four hours taking apart and washing every switch on an old Dell keyboard with Alps switches.

It was the black alps kind (usually considered the worst of the "complicated alps" designs), so probably not worth my effort, but I'm letting them dry now. The one that I dried totally and re-assembled is feeling a lot better. I'd never seen alps switches get that dirty before.

Thinking I might leave the click leaves out and make it a linear keyboard. Might need an extra spring or something to support the spacebar weight. Or a stiffer one, if I can find one of the right size (probably can't).

This post has been edited by dragontamer8740: Jan 3 2022, 11:22

新的一年，继续加油！

translate tells me this is happy newyear, probably. so a 新年快乐 to you as well, young lady

Thinking I might leave the click leaves out and make it a linear keyboard. Might need an extra spring or something to support the spacebar weight. Or a stiffer one, if I can find one of the right size (probably can't).

are the alps designed in a such a way the click spring mattered for keeping up the weight of keys? I dont have any but I always assumed that if anything they were more stiff/hardier than cherry so linears should deal fine with the spacebar.

Thought: OpenBSD is working out surprisingly nicely; my wlan card even supports 40mhz "wide band" 802.11n on the latest snapshots (yeah, I know it's 2022). Graphics work nicely, trackpoint was configurable, and pretty much everything works out of the box. Even sound.

translate tells me this is happy newyear, probably. so a 新年快乐 to you as well, young lady

are the alps designed in a such a way the click spring mattered for keeping up the weight of keys? I dont have any but I always assumed that if anything they were more stiff/hardier than cherry so linears should deal fine with the spacebar.

You're correct, the tactile leaf does add some more "weight" to the keys.

I'd say that since linearizing them they're lighter than cherry blacks, but maybe somewhat heavier than reds? They're pretty close to reds by my estimate, though.

Off the top of my head, reds are supposed to be about 45g. That's nine nickels (5g each). Without a keycap, it takes me ~10-11 nickels to fully depress a linearized black Alps switch. So I'd guess these are around 50-55g.

With the normal click leaf, people say it's a 70g switch. So just slightly heavier than MX blue/black, but lighter than MX green/grey. I can say that subjectively it feels easier to press a key on a model M keyboard than on black alps.

I actually ended up bending the "hooks" on one of the click leaves straight to make it linear and keep the weight.

The spacebar will work without that, but doesn't feel good to me. On my last one of these I used some little "sideways spring" (IDK what they're called, but they're the springs that pinch things or spread things apart; scissor-like; I hooked it on the stabilizer to exert more force up). But that was from a Fujitsu FKB4700 and I don't have another one of those for a donor spring.

I have a whole second AT101W's worth of black alps tactile leaf springs here, so I didn't feel too bad about flattening one out.

Not counting the time it took for the water to evaporate off of them, the entire process of taking the caps off (without a proper puller; alps key caps are often very tough to pop off), taking the tops off of every switch on the board (they can be disassembled without removing from the plate, unlike most Cherry switches), bathing the plastic parts of the switches and the keycaps in warm soapy water, rinsing, putting them out to dry on a towel, and then re-assembling probably took around five hours over a two day period. But it feels great now.

I'd heard Alps switches got really bad when dirty but had never seen it before; I can confidently say that a good wash fixes everything though. So if you see an alps board you want for cheap, but it's dirty, and you're okay with making it into a weekend project, it can turn out really nicely.

I recommend basically every other alps switch over black alps, though. It feels okay, and if you linearize them they're really, really smooth, but the stock switch is just "okay." If you manage to find orange or "salmon" pink alps, those are much nicer tactile switches IMO (and a little lighter than blacks). And even white alps, probably the other most common kind alongside black alps, are quite nice clicky switches.

Blues are apparently a lot like oranges for feel, but have a click like white alps instead of just a tactile bump. I'd like some but don't think I want to spend anything more than $25 on them.

Well, I think i'm keyboarded out for a little while now. Got the board back together.

This post has been edited by dragontamer8740: Jan 4 2022, 05:45

My MX silent blacks are fine for feel proper- force seems about right. I think they have too much travel and noise though. Currently using blue o-rings and cherry profile keycaps.

Digikey order arriving today. I can see if it's just the OP-Amps that were dead (I didn't bother trying to test the ones I pulled, there's a non-trivial chance I killed them in the process anyways). I suppose the mixer is the next big suspect, but assuming the op-amps don't fix it I'll need to probe around in it with RF on and see if I can make sense of it. There's a splattering of 2N2222 transistors that could be suspect as well. All the small diodes seem fine. There's a TVS diode array chip I also need to look at..

My MX silent blacks are fine for feel proper- force seems about right. I think they have too much travel and noise though. Currently using blue o-rings and cherry profile keycaps.

Digikey order arriving today. I can see if it's just the OP-Amps that were dead (I didn't bother trying to test the ones I pulled, there's a non-trivial chance I killed them in the process anyways). I suppose the mixer is the next big suspect, but assuming the op-amps don't fix it I'll need to probe around in it with RF on and see if I can make sense of it. There's a splattering of 2N2222 transistors that could be suspect as well. All the small diodes seem fine. There's a TVS diode array chip I also need to look at..

clean linearized black alps are ultra-smooth. And mine sound pretty much silent (until they bottom out of course).

If you linearized a dampened Alps switch like "dampened cream" or "dampened white" (dampened white should be identical to black alps apart from the dampening), maybe you'd enjoy that. Their dampening is internal to the switch, so key travel is not impacted as it is with O-rings. And it catches both upwards and downards motion, instead of downwards-only.

Alps's key travel is also slightly shorter than MX; 3.5mm instead of 4mm. But if that's not enough of a reduction, there's something else you can do (and you don't even have to solder to do it). Since alps switches actuate pretty high up in the keypress, you could probably 3d print some little things to fill up a little space at the bottom of the switch interiors and reduce travel. Switch tops can be removed without desoldering; I typically use a pair of knives to pull the clips open and pull the switch top up.

MX switches have longer travel than both Alps switches and IBM buckling springs, in my experience. Although IBM's are closer to MX switch travel than to Alps.

Dampening is courtesy of these little black rubber thingies in the slider on the left.

[deskthority.net] (IMG:[deskthority.net] https://deskthority.net/wiki/images/thumb/d/df/Alps_SKCM_Cream_Damped_vs_White_--_slider_stops.jpg/790px-Alps_SKCM_Cream_Damped_vs_White_--_slider_stops.jpg)

You can find those dampened switches in a lot of [deskthority.net] Apple Extended Keyboard II's. Also [deskthority.net] some SGI keyboards.

I'm partial to AEK 1 which has no dampening and is guaranteed to have Alps switches, though.
AEK 2 can also have [mousefan.telcontar.net] mitsumi switches, which feel great but aren't really dampened or anything.

Actually, you may enjoy the Mitsumi switches. I think they feel wonderful. And despite not being dampened, they are actually pretty quiet. And they're almost (but not quite) entirely linear. When typing at normal speeds they certainly feel linear.

Didn't you say you like scissor switches?

Thought: I am an hour and a half into calibrating and color profiling my plasma screen, and I have no idea when it's going to finally end.

Update: It finished at around an hour and fifty minutes, and the result looks like shit. God dammit.

I suspect the automatic brightness reduction "feature" is interfering, but this is a relatively low end plasma so I don't think I can turn it off.

Update again:
[i.imgur.com] (IMG:[i.imgur.com] https://i.imgur.com/99LXaxUg.jpg)

Also, I learned none of the BSD's are supported by Argyll CMS. It can build, but it won't detect any USB colorimeters and the code for the BSD's is incomplete. That's a shame.

It's because Argyll decided to re-invent the wheel and use its own USB library instead of something like libusb that actually does work on the BSD's and that most other Unix/windows cross platform software uses.

Yet another thought: I like C, but dislike malloc/calloc/realloc. It all makes total sense why it's designed like it is, but I just don't like it anyway.

I've been dealing with memory allocation problems today. I think I have them figured out and I fixed an additional memory leak in the process.

This post has been edited by dragontamer8740: Jan 5 2022, 06:39

sober thoughts about that radeon 6970, everything is great
plugged it in threw away the nvidia drivers, grabbed xf86-video-ati and its all sunshine, no tearing, no xorg issues, no weird configs, no annoying drivers
and now I have a whole 4 display ports left:

CODE

DVI-0 connected primary 1920x1080+0+0 (normal left inverted right x axis y axis) 510mm x 290mm
   1920x1080     60.00*+
   1680x1050     59.88  
   1280x1024     75.02    60.02  
   1280x960      60.00  
   1152x864      75.00  
   1024x768      75.03    60.00  
   832x624       74.55  
   800x600       75.00    60.32    56.25  
   640x480       75.00    59.94  
   720x400       70.08  
DisplayPort-0 disconnected (normal left inverted right x axis y axis)
DisplayPort-1 disconnected (normal left inverted right x axis y axis)
DisplayPort-2 disconnected (normal left inverted right x axis y axis)
DisplayPort-3 disconnected (normal left inverted right x axis y axis)
DVI-1 connected 1920x1080+1930+0 (normal left inverted right x axis y axis) 510mm x 290mm
   1920x1080     60.00*+
   1680x1050     59.88  
   1280x1024     75.02    60.02  
   1280x960      60.00  
   1152x864      75.00  
   1024x768      75.03    60.00  
   832x624       74.55  
   800x600       75.00    60.32    56.25  
   640x480       75.00    59.94  
   720x400       70.08