ZA Furries

Kievvu wrote: ↑Sun Apr 12, 2020 12:10 pm I've attempted getting in contact with "old friends" which has been somewhat disappointing since the response is mostly several days of silence and/or blocking me after I just say Hi.
Probably wasn't the best idea tho

Did things end on bad terms the last time?

So frustrated! I'm trying to get protein sequences for work but I don't know which ones to get. Databases do not seem to have labelling to the level that I need and previous papers are very vague. "Yes we used representative sequences from databases." Well which ones? What did you align against!? The one paper teased me by saying the data for the figure was available and it was just the final tree structure, not what they used to align. How am I supposed to work with that?

Lets hope my laptop doesn't melt overnight. I've got it trying to construct a complicated phylogenetic tree because the cluster is too outdated to run the programme. Not the only reason though. But in the last five hours, it's calculated 4 of 20 starting trees. Plus I think there will be more calculations after it's done with all that. At least its not using all the cores so it cycles between them and it doesn't actually run too hot.

Rakuen Growlithe wrote: ↑Tue Apr 21, 2020 8:16 pm Lets hope my laptop doesn't melt overnight. I've got it trying to construct a complicated phylogenetic tree because the cluster is too outdated to run the programme. Not the only reason though. But in the last five hours, it's calculated 4 of 20 starting trees. Plus I think there will be more calculations after it's done with all that. At least its not using all the cores so it cycles between them and it doesn't actually run too hot.

Now. I'm no mechanical engineer.
But from an engineering standpoint. That makes no sense? Cycling through the cores I mean.
it will make one core get hot, and the others not, causing hot and cold spots, which in turn causes mechanical stress.
I'd rather it do all cores simultaneously, and finishing the job quicker. The CPU will make sure that it doesn't overheat by throttling.

I mean. Have you seen Tocsys streams? His GPU temps in the corner are terrifying. But thats fine. The hardware was designed to handle that.

Edit:
As a side. I have been hitting my processor of both my desktop, Work PC and laptop and GPU 24/7 with 100% loads for Protein folding simulations for the Folding@home project.

I dunno. I can see from the system monitor that it does occasionally switch cores. The pre-test step recommended 4-5 cores to maximise the speed. If it uses all of them it starts wasting time transferring information between them and setting them up. I need to actually decide what I'll do with it. It apparently did the main task in ~17 hours and now it's supposed to be running 1000 bootstraps but if that's going to take too long I might just have to cut that or do it later. I probably have to repeat this at some point anyway because there's a missing sample that I'm still waiting for.

Rakuen Growlithe wrote: ↑Wed Apr 22, 2020 12:32 pm I dunno. I can see from the system monitor that it does occasionally switch cores. The pre-test step recommended 4-5 cores to maximise the speed. If it uses all of them it starts wasting time transferring information between them and setting them up.

Depending on the CPU and the task it may work better if you restrict it to certain cores. For example, Adagio's CPU does not support Hyper-Threading (the number of cores equals the number of threads/logical processors), so he will get good scaling by using all of them. But Inkbunny's do support it; so a query on our secondary server actually runs fastest if we restrict the database it to eight "CPUs", not 16 - because all eight are truly independent ones. However, that's just true for one query - if we have more than one at once, it makes sense to use the hyperthreads, even if it risks making a specific query run by itself slower, it results in higher throughput.

My Surface Pro has a HyperThreaded dual-core CPU, so scaling is limited after 50% load; it will go up to 100% but that is a relative increase of ~20-30% more, not 100% more. And it may even be less, because if I run BOINC on it (as I do), the CPU/iGPU uses more power than it can dissipate and throttles back.

You have to be careful setting thread affinity, though. On Windows, on a HyperThreaded dual-core, quad-thread system, CPU0 and CPU1 are the first core, and CPU2 and CPU3 are the second core. On Linux, it is more likely than CPU0 and CPU2 are the first core, and CPU1 and CPU3 are the second core.

An operating system may skip around to spread the thermal load, if not all cores are utilized. You want all parts of a CPU about the same temperature, because that maximises thermal dissipation through an ideal heatsink. However, it might focus on keeping things on the same core, in the hope that relevant data is on the cache. It depends on what goals the scheduling algorithm has been given and what else is running - maybe a higher-priority foreground process preempted it, in which case when resumed it could switch to another core.

Oh, and make sure you don't have dust in your heatsinks, that will cause performance to tank. If, with the CPU fully-loaded, the CPU speed on the left of the Windows Task Manager performance tab is lower than the official base speed on the right, it's a good sign that your thermal solution has issues. Or, your hardware vendor is limiting performance (which you may be able to adjust through power settings).

@GreenReaper.
This is all interesting!
I'm not sure who you are refferring to whith the dusty heatsing part?
If it's me, my computer is barely breaking 60C.
Definable not being thermal throttled.

You do have a point though. His processor; Ryzen 7 3700U, has 4 cores and 8 threads. So limiting it to physical processors would be a good idea.
Though, as a counter argument. With a heavily paralleled workload like simulations, isn't that where multi threading would actually be advantageous?

I'm pretty sure the dust things is just a general point. It looks like GreenReaper is trying to just do one post per year! :p

Anyway, this is a well known procedure which takes a long time. I would ideally run it on a cluster so I don't have to stress my poor laptop but the one I have access to is too outdated to run the software. (The Linux kernel version went out of support like 3 years ago!) Next week, I will try get access to a better updated cluster.

Adagio wrote: ↑Sat Apr 25, 2020 7:10 pmYou do have a point though. His processor; Ryzen 7 3700U, has 4 cores and 8 threads. So limiting it to physical processors would be a good idea.
Though, as a counter argument. With a heavily paralleled workload like simulations, isn't that where multi threading would actually be advantageous?

Most likely, simultaneous multi-threading will provide a benefit, or no great degradation. But it's the sort of thing that can only be confirmed by testing; disabled in the BIOS to free up partitioned CPU buffers and ensure that the OS does not apply additional mitigations for recent shared-core vulnerabilities. The software may have been designed more for Intel architectures rather than AMD's Ryzen.

I guess Rakuen is not in the position to upgrade his cluster's kernel himself; but if he can, there might be drop-in backports/hardware-enablement kernels that could be applied to raise the kernel version while the distribution remained otherwise the same. Could be a suggestion to raise with IT/support, if nothing else - the issue probably impacts others.

As for posts: I may make another in a different thread soon; but it depends on the response I get from Microsoft Azure, which is currently operating on lockdown protocols in its South African datacenters.

Here's a headscratcher for you.
I have a monitor that Nanukk gave to me cause he thought it was broken.
It had some dark areas is you looked at a plain white screen. And a line down the screen where it fell against something sharp.
It also had really weird colors.

But all this went away, after I overclocked it.
I mean. The line is still there. But so feint I can't see it without looking for it.
Colors look amazing, and the weird dark areas are gone.

I set up ssh so I can access my pc from my phone or laptop on the home network. I need to fix up some folders for sharing media still though. But now I can make my pc update during the night without even getting out of bed.

YOYOYO! WHUDITIZ MUTHERFUUKERS? why this place so quiet? as for what i'm doing right now, I JUST TOOK A HUGE SHIT! UwU just kidding, i'm adding in some scripted scenes in my game. its almost to the point where i can release a beta.

Yeah, it's been exceptionally quiet. Personally, I've just had nothing to say.

I honestly thought people would be MORE active over the lockdown

Adagio wrote: ↑Wed May 20, 2020 9:40 pm I honestly thought people would be MORE active over the lockdown

Me too. I guess it means people have less to say and do now. Or there's too much stress about other things to worry about that are more important.

I'd go option two tbh. I find myself bogged down with so much work now because things tend to get away from me.

on an academic note: why does sexual intercourse cause such feelings of dehydration afterwards? is it the fluid dynamics involved in the action or is there something more? i mean i have just identified a large portion of the afterglow to be satiety, specifically in terms of feeling parched then drinking. is this feeling of thirst a snap back effect from the pseudo quenching effect? i mean i'd think drinking would then extend the feeling of afterglow however it seems to pretty much kill it.

Animew wrote: ↑Fri May 22, 2020 12:04 am on an academic note: why does sexual intercourse cause such feelings of dehydration afterwards? is it the fluid dynamics involved in the action or is there something more? i mean i have just identified a large portion of the afterglow to be satiety, specifically in terms of feeling parched then drinking. is this feeling of thirst a snap back effect from the pseudo quenching effect? i mean i'd think drinking would then extend the feeling of afterglow however it seems to pretty much kill it.

We probably shouldn't go too much farther here as this borders on NSFW, but my thought would be that it's primarily the energy expenditure involved.
As for what I'm doing, heading to see family out of state for the weekend today.

Look what I got to keep me warm and cosy. It's even got a humidifier!
I need to get an extension cord though to bring it nearer my desk.

I feel like Rakuen is overcompensating...

It was 13-14 degrees in the morning last week! And we're not even all the way into winter.

I'm still wearing shorts...granted, I'm also wearing socks and fingerless gloves and a hoodie, but shorts nonetheless

And we're working again! Updated my PC but new Linux kernel took startup arguments differently so nothing booted. Then I spent around 3 hours trying to figure out the problem and fix it. A decent portion of that was me having found the right solution but typing it in incorrectly! x.x Missed quotation marks...

oi, firedog! UwU/ the PM's are broken, i cant delete stuff. keeps telling me "The submitted form was invalid. Try submitting again."
i mean normally i wont care but ive got 2 PMs thats on hold so i gotta delete 2 things to get at them.

Animew wrote: ↑Mon Jun 15, 2020 3:15 am oi, firedog! UwU/ the PM's are broken, i cant delete stuff. keeps telling me "The submitted form was invalid. Try submitting again."
i mean normally i wont care but ive got 2 PMs thats on hold so i gotta delete 2 things to get at them.

This is not the place to report bugs. In any case, I fiddled around and it seems to be a problem with phpbb itself and should be patched whenever the next version is released. It might be possible to edit the installed files but I just slightly increased the maximum number of allowed PMs.

My first attempt at making something from Alpha_Ki's cookbook!

Really simple but I had to make some adjustments because I had eaten some of the chocolate already. Also my scale is not really accurate for the small measurements (especially after I adjusted the recipe) and I nearly never bake, so it doesn't look as good as the pics in the book. I think it needed more mixing. Still tastes pretty good though! Next time will be better.

My roommate severely underestimated how much pasta we would be receiving from a "pasta shipping experiment" and now we have 160 lbs of pasta in 8 giant boxes in our living room (about 72.5 KG!).
We're still not entirely sure what we're going to end up doing with it all, haha.

Trace wrote: ↑Thu Jun 18, 2020 9:25 pm My roommate severely underestimated how much pasta we would be receiving from a "pasta shipping experiment" and now we have 160 lbs of pasta in 8 giant boxes in our living room (about 72.5 KG!).
We're still not entirely sure what we're going to end up doing with it all, haha.

O.O That is a lot of pasta! Although, if there was such a dramatic underestimate, you must have got a great deal!

We had another box show up yesterday, bringing our total to 180 lbs of linguine. XD
The company that was doing it is a deal site that likes to play around with the format. We were warned that it would be an undisclosed amount of pasta, and they delivered, haha.

Your roommate is the person the prophecies in my math textbook warned about.