George.

It has also to do with how much cache is used, how much the cpu has to hit the ram, and what instruction sets are used.
Your program is so small it may fit inside the level 2 cache of your cpu and basically have minimal ram access and use only the basic instruction sets.
Blender will use AVX and other instruction sets, and pi is very optimized.

I am most definitely not an expert but I believe this is due to each load having different length instruction pipelines and the instruction (for example, AVX is notorious for it) used in the process. Longer it works without a wait state = more heat it generates. I too would like to learn more if anyone knows more or if I am wrong.

Generally gaming experience on Linux is worse because it has to run games via emulation most of the time. That being an issue to game developers, the OS itself is actually lighter, therefore faster with the proper drivers present.
If a game is perfectly supported by both Linux and Windows, then it's probably the same or even better since it has less overhead. (more available RAM and less VRAM occupied by the UI). So if theoretically a game is supposed to run on both natively, the worse the hardware, the bigger overhead Windows has, therefore Linux could be better. Realistically, however, it will probably come down to drivers. Nvidia is kinda weird about Linux drivers while AMD is doing fine I think. You would have to tweak driver settings so that they are the same on both OSes tho. Not all features are present on Linux either afaik.
If you just wanna play games, just stick with Windows. It's the superior gaming experience across a huge number of games. If you're like me, however, enjoying the process of setting up and running games just to see how they run and not really playing them (apart from WoW and cs go sometimes), then you would probably have more fun on Linux haha DIY is cool if you have the spare time. 
 
tldr: in a perfect scenario Linux can be better, but for now in most games it is worse.
 
My personal experience:
- everything running via Proton (wine + lutris) is running suprisingly well considering the real-world DX11 to Vulkan calls translation and emulation of Win-exclusive libraries and whatnot, HOWEVER, it is still a bit worse than Windows where things run natively.
- Games that require a lot of RAM are running better on Linux for me because my laptop has only 8 gigs and some games max it out on Windows, whereas I still have some available on Linux, because my Ubuntu setup uses 1GB at idle up to 2 with a bunch of things open (few browser tabs and some light apps), whereas on Windows (even after decrapyfing it) I get 2,5-3 gb RAM usage at idle. That 1,5-2 gb of extra ram really does help when you need it.
My temps are also a bit lower on Linux since my memory isn't maxed out and I guess the CPU has a few spare clock cycles to chill.
- CS:GO runs straight up faster on my Ubuntu setup in comparison to Windows.
- WoW Shadowlands runs a bit worse, but loading times are actually faster and the overall performance is actually good so far.
- Rocket league (via Epic) is OK, but a bit lower framerate.
- Games in Steam are running well enough, either the same fps or slightly lower.
- System responsiveness while playing a game is actually better at times, probably because of the extra available RAM +/- different process scheduling mechanisms, idk.

Check memory for errors first. 
Could also be an OS related issue. I've had Windows goof up like that before. You can run "sfc  /scannow" in cmd as admin. 
While waiting for the scan, open the system logs in event viewer and check for errors. They are separated in different categories depending on the type of error and what caused it. Look into critical and error submenus. If something nasty happened soon-ish it should be there. 
Could try reinstalling everything if you have the time or boot from a different drive and run some stress tests. 

200 MHz aren't going to make a noticeable difference. The latency however might. I'd go with the 3000 CL16 instead of 3200 cl20. 

200 MHz aren't going to make a noticeable difference. The latency however might. I'd go with the 3000 CL16 instead of 3200 cl20. 

People, the chip responsible for video encoding/decoding is an asic apart from the actual GPU core that renders your game. Basic streaming using the GPU won't affect the performance in games. With that said...
Your GPU has an older NVENC, and the quality isn't that good when compared to CPU encoding at medium/fast presets. Give it a try and see if it's good enough for you. If so, great! If not, try CPU encoding with different presets until you achieve a nice balance between quality and usage/game impact.

yes, like @Grabhanem said, when idle you're not supposed to hit max frequency. Try loading it up with sth heavy and if it doesn't hit max frequency at 100% utilization then you can proceed to investigate, if it does - everything's normal.
 
P.S. afaik AVX-512 instruction set extension runs at sub-nominal frequency, so try loading it up with sth basic at first (CPU Stress Test - An online tool to test your CPU stability | CPU Expert (cpux.net)  - like this thing - just set it to the number of threads you have, power at max and hit run) 

Classical SSDs do not have the write endurance but stuff like intel's Optane memory do. This costs a lot more than an SSD per GB but costs a lot less than DDR and has the advantage that you can connect to it over PCIe. I would not expect them to use a NAD flash but possibly something closer to Optane. 
 
Not just a lot smaller but also they draw less power probationally to the performance the provide. IceStore core performance provider about 1/4 of the performance (depends on the task due to different caches sizes and much smaller IO bandwidth) of a firestorm core but draw less than 1/8 the power. 
 

I realize that the thread has already moved on in terms of speculating what Apple is up to... Uh, guys? Wasn't it obvious. It's the multi-chip "chiplet" thing, like AMD. Apple's chiplet is called Jade, and it lets Apple make even more money by just cranking up the volume on their one, massive production line of 5nm chiplets.
 
Look, here's the list again:
Jade C-Chop: 8 firestorm, 2 icestorm cores, 16 GPU cores, up to 64 GB RAM Jade C-Die: 8 firestorm, 2 icestorm cores, 32 GPU cores, up to 64 GB RAM Jade 2C-Die: 16 firestorm, 4 icestorm cores, 64 GPU cores Jade 4C-Die: 32 firestorm, 8 icestorm cores, 128 GPU cores It looks to me like they don't have a separate I/O die the way AMD uses a Global Foundries 14nm process for their PCI-E and DDR4 controllers. And in fact Apple is listening to us! They moved from the M1 (4 firestorm, 4 icestorm) to 8 firestorm and 2 icestorm cores. By default it comes with 32 GPU cores too.
 
C-Chop: chop out some GPU cores. (These are probably lower-binned chips or chips with defects.)
C-Die: just one Jade for you.
2C-Die: you get 2 x Jades
4C-Die: you get 4 x Jades
 
Simple as that!

The power usage and for the die area they use icestorem cores compared to the performance is better than firestorm cores.
The limiting factor is the single threaded speed. So as long as you have a test that is not urgent it is better to run it on the ice storm cores than it is on a firestorm core.

If you put those tasks on the main firestorm cores you end up drawing more power (thus reducing the total power draw aka number of cores) and you use up more die space but also when switching to these tasks you might end up evicting data from the L2 and L1 cache of that core.  This data eviction can have a big performance hit if you have lots of background tasks jumping in all the time momentary disrupting your hard work (video render etc).
 
It is much better to use the space that 1 firestorm core would use and put 4 or even 6 ice storm cores in that space and they are perfect for those tasks that are labels with priority background apple have been agresivngly pushing developers to label tasks within their code with these priority flags for years now. I expect the desktop options to have 2 Icestorm cores per chiplet (having them spread out across the chiplets will help with data sharing between background and userInteractive tasks).
 
Tasks labels with background and utility are as the name suggested not urgent and will be computed as and when there is time, on intel systems the OS also prefers to run these on cores already running at a lower frequency so that the other cores do not suffer a drop in frequency and intentionally fills them with lots of small pauses so that the power levels in the intel cpus do not get triggered. Also the chips clocks are profited from entering turbo modes, so that if important task ( userInteractive or userInitiated ) needs to be worked on all of a sudden on that core the core is ready to run at the full turbo for the time limit that intel set in their firmware.

This type of careful power management and very fine scheduling but also using that fact that devs have been good at labelling the parts of their applications with priority is why when you run macOS on the 16" MBP it gets a good few % points better scores than when running windows on the same hardware. In the windows space while there are such labels you can apply very very very few applications every make use of them so almost all apps run assuming all tasks (even background cleanup stuff) is userInteractive most important must happen right now.

On macOS and iOS userInteractive tasks tend to be things that would result in a scroll or other UI gesture being interrupted and dropping frames normal tasks are userInitiated that means the system will priorities UI interactions (scroll and gestures) above them this makes for a much smother UI and helps ensure even when an app is working hard the UI stays responsive. 
 
 
With respect to clock speeds apple will not not move the cores to run outside of the optimal perf/W level they are at. Since they know the exact power envelope of every stem they are targeting (unlike intel and AMD) it is much better to put in more cores (or build a wider larger core they could do this) than just increase the clocks speeds of the current cores since a soon as you leave the optimal frequency for your node/doping/design the power increase if very non linear with respect to the performance you get out. 

They should honestly make it more modular...there's so much junk most people don't ever need that's shipped with the system...even if you're a power user, you still stick to some kind of power usage, not a single person needs absolutely everything that's enabled and part of the OS by default.
The OS isn't as light as it used to be and if they cut down things, make them easily installable via powershell or the "add additional features" module it'd be a good thing imo.

My solution to his problem is to just hold the phone closer to my face.  Have to take my glasses off.  Who ever thought being near sighted would be handy.

No. do not upgrade. 3200mhz is already good enough. you would likely see no difference in games and very little difference in editing.

A few vendors come to mine:
Dell
HPE (as you mentioned)
Lenovo
Cisco
 
Depending on what you need, pre-built or heavy customization, it could sway your decision but most of the above have configuration tools you can play with as well to spec things out.

HPE and DELL are good but most expensive.
Fujitsu and Lenovo are just as good in reality, but cheaper usually.

If you want a more budget-friendly one, you can look at SuperMicro as they make their own servers as well as supply tons of companies with their servers as OEM devices for the companies to brand as theirs.

That being said, you would probably be best off if you used a 1U server as the "head" of the system, it should be equipped with SSDs, and use a budget-friendly storage array for data. Something like a Lenovo/IBM Storwize V3000-series (V3700 for example) with another disk shelf if you need even more space would do the job quite well.
You connect that array to the server directly and you're golden.

To build the best system possible, you need to determine which part of the system is the most heavily used (storage or compute) and then go from there. If fast and/or reliable storage is important, a dedicated storage array would be ideal.

Yeah, I'd be bottlenecked way before nvme speed caps by my external storage medias and LAN,WAN link speeds haha.
 

Thanks for this answer. That was the answer that I was looking for!

the charging control thing seems like a fast charging switch - where the driver gains low-level control and turns on/off fast charging.
the battery check thing I assume only reads battery status data and reports it to the user...could be disabled, but needs to be researched in this exact case.
igfx tray is the tray icon for the Intel Integrated graphics app, it should still work if you disable it, just won't appear among the tray icons.
Rainmeter would probably reduce your booting time considerebly, but I assume you want it running for customization purposes.
 
Best thing you could do is press right mouse key on each of them and select "search online" then you will be redirected to a list of sites that explain the exact purpose of the exact task with the exact process name.

the charging control thing seems like a fast charging switch - where the driver gains low-level control and turns on/off fast charging.
the battery check thing I assume only reads battery status data and reports it to the user...could be disabled, but needs to be researched in this exact case.
igfx tray is the tray icon for the Intel Integrated graphics app, it should still work if you disable it, just won't appear among the tray icons.
Rainmeter would probably reduce your booting time considerebly, but I assume you want it running for customization purposes.
 
Best thing you could do is press right mouse key on each of them and select "search online" then you will be redirected to a list of sites that explain the exact purpose of the exact task with the exact process name.

AFAIK iCloud manages storage automatically and purges local cache on its own when storage gets low. At least if the app natively supports iCloud integration and it's not a "connect to iCloud, download file locally into app" affair.

depending on the games and your HW you could definitely play some less demanding titles.
However, in the case of very old games, check the driver and hw compatibility because I skipped that once and it costed me my whole day haha.
So my gf wanted to play a very old strategy game that didn't run on anything past xp (and vista with patches) on her laptop and I did everything vm-wise (xp service pack 3), installed all drivers I could find, even applied some tweaks in the config file of the game and it would just keep crashing every single time, no matter in what mode I would run it in (compatibility, as admin, with disabled enchancements, etc.), Literally dug into the OS configs, edited the registry and whatnot....
 
Turns out (and It's pretty obvious when you think about it) the game doesn't support intel gpus (integrated graphics) because they didn't exist back when it was made.   felt like a moron after everything I did that day. 
 

/8 for megabytes
and say a modern phone/camera creates a 10mb jpeg file, that'd be 2,225,000 jpegs.

/8 for megabytes
and say a modern phone/camera creates a 10mb jpeg file, that'd be 2,225,000 jpegs.

Thanks everyone, I'm excited to get back to it again!