News Which PC Boots Up and Shuts Down Faster: AMD or Intel?

[TerryLaze]

I wanted to add, for booting an OS, I would argue the higher IOPS is better, not the one who has the faster raw bandwidth. The stuff that gets loaded is all over the place on the drive in small files and they're not even loaded at once unless you're waking up from hibernating or Fast Boot. Which in that case, you're not even loading those files anymore.

From windows 8.1 onwards they changed the boot process into hybrid boot which is basically hibernation but with the additional step of closing all user apps.
I don't even think you can do a full normal boot anymore.
https://docs.microsoft.com/en-us/wi...ering-a-great-startup-and-shutdown-experience

 

[hotaru.hino]

From windows 8.1 onwards they changed the boot process into hybrid boot which is basically hibernation but with the additional step of closing all user apps.
I don't even think you can do a full normal boot anymore.
https://docs.microsoft.com/en-us/wi...ering-a-great-startup-and-shutdown-experience

You can turn that off and it only kicks in if you do a Shut Down. If you do Restart, it does the normal boot process.

Judging by the posts here, people rarely do an actual shut down. They either put the computer to sleep or do a restart because Windows needs to update.

 

[cryoburner]

or those that think PCIe 4.0 x4 SSDs boot faster because of their sequential performance is so much faster than the PCIe 3.0 x4 Optane 905P, it's not quite true yet but close. Head to head Optane boots 0.2-0.3 seconds faster on average on AMD X570 platforms with Fast Boot enabled. Optane's random performance advantage makes up for the difference. It is minor, but NAND is still slower at boot. Things may change with the next-gen PCIe 4.0 SSDs though.

The problem isn't that a PCIe 4.0 drive would be faster than an Optane drive, but that realistically Optane isn't practical for the vast majority of systems, so it's providing a skewed view of what percentage of the boot process is determined by storage performance relative to BIOS initialization.

It's not particularly useful information to say that the Intel system boots around 40-50% faster, when that difference only amounts to around four seconds, and there were no tests measuring how long it took to get past the BIOS screen, compared to the actual loading of Windows on the kinds of storage devices that people are far more likely to be using today.

 

[CountMike]

I can't believe this article made it past the editor's desk.

Likewise, 2 examples of thousands if not millions and that doesn't include different BIOS versions on each system let alone startup programs and drivers in each OS and the order in which they start which in Windows 10 can be quite different, even from start to start..
There's another thing author missed, No data about POST time, to OS desktop or when everything quiets down to idle or when OS becomes usable.

 

[seanwebster]

The problem isn't that a PCIe 4.0 drive would be faster than an Optane drive, but that realistically Optane isn't practical for the vast majority of systems, so it's providing a skewed view of what percentage of the boot process is determined by storage performance relative to BIOS initialization.

It's not particularly useful information to say that the Intel system boots around 40-50% faster, when that difference only amounts to around four seconds, and there were no tests measuring how long it took to get past the BIOS screen, compared to the actual loading of Windows on the kinds of storage devices that people are far more likely to be using today.

The focus of the article was just to show that AMD is slower than Intel at booting at the time since that is what I experienced with my latest test benches. They have some work to do to speed things up. Regardless of the drive used, the difference IS still be there. I mention % in speaking about it, but the fact is that AMD systems boot slower does NOT change. Graphs show time as that matters more than percent. Again, it's about looking at the overall picture. I just thought additionally stating the % would better support that fact. It is also why I added in Ubuntu results.

Likewise, 2 examples of thousands if not millions and that doesn't include different BIOS versions on each system let alone startup programs and drivers in each OS and the order in which they start which in Windows 10 can be quite different, even from start to start..
There's another thing author missed, No data about POST time, to OS desktop or when everything quiets down to idle or when OS becomes usable.

I tested various BIOSes and many UEFI configs and drivers on each. I have that data. I have over 200 data points on just booting on one system alone... But again, the focus on this wasn't that, rather the overall experience difference. I mean, if you want that type of article, the previous one I did goes into more detail on those types of changes. I could probably do another if there is enough interest. Even the motherboard form factor can make a difference. Regardless, the fact remains that AMD UEFIs boot slower.

Have you used an SSD or Optane? I say this because while HDDs aren't normally usable right after the desktop loads, with flash that changes. With my Optane SSD 905P, basically everything is 100% usable immediately.

 

[nings]

You're adding fuel to the fire when it comes to fanboyism posts.

The article wasn't biased at all, it's people's views that are biased when see see it as detrimental towards AMD, which it isn't purposefully.

I'll repeat it again '90+% of users do not care about a 6 second boot time/shutdown difference'

It has total irrelevance to actual system performance once inside the OS.

Probably not have understand but the true war it's windows VS linux and the misunderstanding of Windows hidden mechanical by "computer specialist"..
AMD have already win not need to speak anymore about that. A true test should be neutral and neutral is removing POST time due to different manufacturer and POST processing. Tomshardware are just Intel fanboy it's a fact and write wrong news a lot of time and that's just sad. AMD win actually but all can change in some years nothing know.

So two big mistake :
first not know how windows 7,8,8.1,10 work compared to XP, Vista. Tomshardware go directly into microsoft newbie trap
second not give a neutral testing by measuring from boot loader until OS FULLY loaded

So fake news and useless but tomshardware can take back this article and solve all issues.

 

[nings]

Windows XP, Vista and 7? They are all irrelevant in 2020. Why bother with an HDD ofr OS/VM usage? They are slow and flash is affordable.

When testing my boot speeds, the resource manager showed little to no activity after booting for minutes after booting. Again, I'm using one of the world's fastest SSDs for this. Not archaic tech that takes a few ms to respond to a request.

I'm sorry but it's just wrong. Windows 10 not respect anything sometime you even haven't the network when you reach desktop. All services boot and process after show desktop for cheat about boot time that slow down all even with faster NVME drive. Why you have message than can't contact AD server on login sometime. You are just too fast Windows not have the time to take an IP and contact AD server before prompt you login screen.
Launch a VM who ask all resources directly at startup and it's just slow down all. Try with an HDD it's a way better for view this big issue and this Microsoft bad cheat introduced since W7 due to Vista slow boot time. Microsoft change nothing from Vista to W7 just launch all after show desktop.

Systemd isn't distribution related is just a Windows like services management who slow down ALL even the boot process can take 2m boot time due to this crap on some computers. So Windows not beat Linux and Linux can be faster without systemd why this article is a big joke for a computer specialist sorry.
Why you not graph REAL boot time without POST and when OS is just fully loaded (all services up ?) you can check all on events viewer.
People still have HDD, SSD, cheap QLC NVME drive and a very low people have the most expensive faster NVME drive. Windows 10 computer run on older computer than Windows 7 that's what Microsoft ads say. So try with an HDD launch on of their 6 moths update and look by yourself what is really this OS

So increase the quality of this and next article please. I don't care I know but a lot of people not know how it's work so your job is show bad OS cheat hidden to people...

 

[seanwebster]

I'm sorry but it's just wrong. Windows 10 not respect anything sometime you even haven't the network when you reach desktop. All services boot and process after show desktop for cheat about boot time that slow down all even with faster NVME drive.

I compared to OSes at stock settings - latest upates . I only changed the hardware config. I monitor Windows activity for hours a day almost every day for years reviewing SSDs. My experience is drastically different than yours.

Why you have message than can't contact AD server on login sometime. You are just too fast Windows not have the time to take an IP and contact AD server before prompt you login screen.

Most don't use AD servers...

Launch a VM who ask all resources directly at startup and it's just slow down all. Try with an HDD it's a way better for view this big issue and this Microsoft bad cheat introduced since W7 due to Vista slow boot time. Microsoft change nothing from Vista to W7 just launch all after show desktop.

My virtual machines all run off NVMe SSDs and I don't experience any slowdown, but then again, I have responsive systems. I haven't touched a HDD in years other than for my NAS or to upgrade with an SSD.

Systemd isn't distribution related is just a Windows like services management who slow down ALL even the boot process can take 2m boot time due to this crap on some computers. So Windows not beat Linux and Linux can be faster without systemd why this article is a big joke for a computer specialist sorry.

The focus is not the Windows vs Ubuntu, rather Intel vs AMD booting. Ubuntu was added as support for this.
If you want a Windows vs Linux article, feel free to leave some suggestions and we will consider it. Both OSes were simply installed, updated, and then testing began - only change was to the UEFI config. I don't use Ubuntu or Linux all that much - more so off and on over the years, so if you want me to config Ubuntu to boot faster, I am open to suggestions. Maybe I can do a Windows vs Ubuntu showdown next.

Why you not graph REAL boot time without POST and when OS is just fully loaded (all services up ?) you can check all on events viewer.

Because 99% of resources are available to me to use at my will the moment the desktop loads. Any background tasks take up 1-5% of resources at most after I hit the desktop. It's fully functional. Again, this is something NVMe SSD experience related, so if you are using a HDD, you wouldn't notice this type of responsiveness, unfortunately.

People still have HDD, SSD, cheap QLC NVME drive and a very low people have the most expensive faster NVME drive. Windows 10 computer run on older computer than Windows 7 that's what Microsoft ads say. So try with an HDD launch on of their 6 moths update and look by yourself what is really this OS

So increase the quality of this and next article please. I don't care I know but a lot of people not know how it's work so your job is show bad OS cheat hidden to people...

Again, the OS performance and storage isn't the point of the article. The focus was to point out the differences that I noticed in boot times between my AMD and Intel hardware. The facts still remain. Stock for stock, Windows is capable of booting faster than Ubuntu due to its Fast Boot feature and Intel UEFIs POST faster than AMD.

 

[hotaru.hino]

All services boot and process after show desktop for cheat about boot time that slow down all even with faster NVME drive.

Is this something you can actually prove via documentation or tools that I can run on Windows itself to verify this?

I ain't going to take your word based on anectdotal evidence, mind you.

EDIT: If you're going to say "look at Event Viewer", that could simply be a product of Event Viewer being late to the party and having a pile of messages to deal with at once, not that every one else is late to the party.

 

[nings]

Is this something you can actually prove via documentation or tools that I can run on Windows itself to verify this?

I ain't going to take your word based on anectdotal evidence, mind you.

EDIT: If you're going to say "look at Event Viewer", that could simply be a product of Event Viewer being late to the party and having a pile of messages to deal with at once, not that every one else is late to the party.

Take a noisy HDD install Windows 10 1607 (or even ltsb version) go update to another version and just use your ears. When HDD stop krrrkrrrkrrr your windows is ready for launch software and it's take couple of minutes even on a 7200rpm drive. I have try on a notebook with HDD that need minutes before can launch any applications and use it. But it's the same with an SSD or NVME not have full power when you reach desktop and if you use really your computer it's visible. After take Vista or XP and again when reach desktop your computer has full performance available.
After you install linux (no systemd if possible) and made the same. One time reach desktop you can use directly your computer full performance available.

If you not want to do that install virtualbox (free) and launch another OS directly at startup (shortcut for start vm in startup folder) and look the time needed for boot and reach desktop on the VM. And other case you wait 5min after Windows 10 reach desktop and try the same by booting virtualbox machine and look time difference.

Everyone can view that and if you use a lot of computer resources directly at startup all is pretty slow due to I/O since W7 so not a new issue. But most people not even view that I don't understand how it's possible and almost all review do not take this into account.

 

[TerryLaze]

Take a noisy HDD install Windows 10 1607 (or even ltsb version) go update to another version and just use your ears. When HDD stop krrrkrrrkrrr your windows is ready for launch software and it's take couple of minutes even on a 7200rpm drive. I have try on a notebook with HDD that need minutes before can launch any applications and use it. But it's the same with an SSD or NVME not have full power when you reach desktop and if you use really your computer it's visible. After take Vista or XP and again when reach desktop your computer has full performance available.
After you install linux (no systemd if possible) and made the same. One time reach desktop you can use directly your computer full performance available.

If you not want to do that install virtualbox (free) and launch another OS directly at startup (shortcut for start vm in startup folder) and look the time needed for boot and reach desktop on the VM. And other case you wait 5min after Windows 10 reach desktop and try the same by booting virtualbox machine and look time difference.

Everyone can view that and if you use a lot of computer resources directly at startup all is pretty slow due to I/O since W7 so not a new issue. But most people not even view that I don't understand how it's possible and almost all review do not take this into account.

Sure windows only loads up all the things that everybody needs to get to desktop and then loads up all the stuff that are not loaded up by everybody ,things that you can enable and disable.

What does this change? The CPU that is faster in loading things before the desktop will also be faster in loading things after the desktop,it's not going to change anything while it is going to introduce a lot more variables that will make the test harder to perform without yielding any better information.

 

[nings]

I compared to OSes at stock settings - latest upates . I only changed the hardware config. I monitor Windows activity for hours a day almost every day for years reviewing SSDs. My experience is drastically different than yours.

What is my experience like you know me ?
Where I work ?

Most don't use AD servers...

It's not a reason just a proof than something is wrong with Windows 10 "fast" startup.

My virtual machines all run off NVMe SSDs and I don't experience any slowdown, but then again, I have responsive systems. I haven't touched a HDD in years other than for my NAS or to upgrade with an SSD.

When you launch your VM with a shortcut in "startup" folder. It's boot in exactly same time than after 5min on desktop ? You really do/measure that ?
SSD are more expensive than HDD and a lot of people buy first price notebook who have just recently change to all SSD/NVME drives now. An OS is JUST an applications launcher nothing more and shouldn't take all computer resources but only people working under linux can understand that.

The focus is not the Windows vs Ubuntu, rather Intel vs AMD booting

The main issue was than Intel or AMD not produce motherboard but only CPU and chipset. If you want compare both manufacturer you need several motherboard manufacturer each BIOS/UEFI are differents. My current BIOS version of an Asus mainboard start reboot and start again that's just linked to this bios version never have that before with same parameters. Why it's important to push button until desktop (really booted) and from bootloader until desktop fully loaded for comparing CPU manufacturer and OS boot process. POST too dependent of motherboard manufacturer.

Because 99% of resources are available to me to use at my will the moment the desktop loads. Any background tasks take up 1-5% of resources at most after I hit the desktop. It's fully functional. Again, this is something NVMe SSD experience related, so if you are using a HDD, you wouldn't notice this type of responsiveness, unfortunately.

I can't use a VM at windows startup even with a sata SSD take couple minutes for boot instead of 20s but test please and look by yourself.
NVME help even more than sata SSD due to more I/O but can't save all. Launch heavy applications/VM directly in the startup folder or "run" registry key.

Again, the OS performance and storage isn't the point of the article. The focus was to point out the differences that I noticed in boot times between my AMD and Intel hardware. The facts still remain. Stock for stock, Windows is capable of booting faster than Ubuntu due to its Fast Boot feature and Intel UEFIs POST faster than AMD.

I'm sorry but with this article I can't say anything. Probably bad booting finish time for Windows used for compare with Ubuntu. Not enough motherboard for compare AMD/Intel POST difference. No graphic for view bootloader until desktop time who can compare CPU only.

 

[nings]

Sure windows only loads up all the things that everybody needs to get to desktop and then loads up all the stuff that are not loaded up by everybody ,things that you can enable and disable.

What does this change? The CPU that is faster in loading things before the desktop will also be faster in loading things after the desktop,it's not going to change anything while it is going to introduce a lot more variables that will make the test harder to perform without yielding any better information.

For example I have no network up and ready on the login screen so you can't login on AD server. User have time to type his password and received an error message before it's up even with SATA/NVME OEM drive so not best quality and best I/O. Bad user experience finally and microsoft should wait network and AD reply before show login screen for not have this issue.

When I reach desktop I want to be able to use 100% resources of my computer. An OS is "just" an applications launcher nothing more. Mainly issue isn't the CPU but the drive and I/O may be even memory because Windows use a lot of memory. A computer isn't really well support multitasking even now. Launch an application who use 100% of the drive and 100% of CPU your machine is just unresponsive but a true multitasking computer should continue to work without major slowdown.

Main issue is cheating everywhere... Browser cheat by using insane amount of memory all go directly in memory for show pages faster. OS like Windows cheat with boot time and finish boot process when reach desktop. NVME QLC drive cheat with some Go faster than the most of the drive capacity and after it's just deadly slow. I prefer have constant performance for drive (old samsung pro drive slower but constant) or wait 30sec more at startup but can use my computer like any other time during the day.
Now you transfer 100go of files on your SSD go pretty fast and just slow down at 50mo/s at the end because you fulfill cache and go to crap QLC cells. Like FPS for graphic card who care to have 60 or 66fps that's change nothing you not see these 6 more fps... most important thing it's never go under 30/40fps because in this area you can see real slow down. Same for DPI on monitor all want 300dpi but human eyes can't view more than about 120dpi for a computer monitor may be 200 on a smartphone. TV at 8k are just useless or provide new eyes too in the TV package for view 4x more pixels than 4k ?
Human eye can probably view a little more than 1920x1080 depending size of the TV but less than 4K TV even at 70''. Who buy 8k and for what our eyes not support this resolution.
Smartphone same issue CPU/GPU/battery should power a screen of billions of pixels for zoom 200/300/400% just useless and loss of cpu/gpu power and battery duration.