Basically:
Macrium Reflect on the PC.
That writes an Image out to a folder tree (or single folder) on the NAS.
Scheduling in Macrium makes it easy to denote how long to keep, and what days and time of day.
A Full (and then Incremental) in Macrium does NOT need to be a sector by sector copy.
A full drive image is indeed everything needed to recover that to a new drive, or overwrite a faulty OS.
Yes, I've tested this, and yes, I've used it.
its a little bit confusing to use the first time, but once you get used to it, then it is great,
it allows wireless or USB cable backing up of texts, photos, apps etc to the PC, where you have Vibosoft software on the PC, you run this, then select connect wirelessly, then run the vibosoft software on the android smartphone, scan the QR code from the PC Vibosoft, and now the smartphone is connected, and you can select eg texts, and it will backup all texts from all contacts, with different options, the one I use is 1 html file per contact, where all texts ever appear as one long webpage per contact, which look just like on a smartphone. which is much easier to manage than on a smartphone! I had some 3000 texts with one contact, where I was buying a lot of collector stuff over a year. and it takes 10 years to get to the earliest ones on the dubious Oukitel X1 smartphone.
They all write to the same folder tree in the NAS. Each system, and then each physical drive, gets it own subfolder in that top level folder called "Backups".
This is my current C drive, from Today going back to Mar 9. (there are more)
With Macrium (and the other similar tools), you simply locate the Image, and tell it which drive to apply it to.
It will overwrite everything on the drive, and poof...an exact duplicate of the drive, as it was the day you created that Image.
No preformatting needed.
With Macrium (and the other similar tools), you simply locate the Image, and tell it which drive to apply it to.
It will overwrite everything on the drive, and poof...an exact duplicate of the drive, as it was the day you created that Image.
No preformatting needed.
in the other direction for creating the image, will it deal with the fact that the system drive changes ongoingly?
does it freeze the drive whilst making the copies?
ie the problem of disk sectors changing whilst it is being backed up, which could cause an incoherent disk image.
when I manually backup the Windows drive, I do that from Linux to ensure the drive doesnt change whilst copying, because otherwise the copy can be incoherent if some data changes before being copied.
if its dealing with the disk image, I assume then it will deal with deleted files, where if you restore the disk you can undelete these?
does it also allow for file level backing up, where it just backs up the files, rather than the sector image?
also there are some drives that I would want to only sometimes make a sector based disk image, would I be able to do one off disk images?
these drives are generally usually not connected to the computer, only connected when I am working on the stuff. eg I have a disk which I digitise some of my vinyl and cassettes to, but I only connect that drive when either digitising or playing back some music, which isnt often.
No.
It does a Volume Shadow copy, and proceeds from there.
Things happening during the Image creation may not be reflected in the Image, but the Image creation is quite fast. Especially if you are into the Incremental or Differentials.
MoBo manual page 53 (or 55 in total) shows all the fan headers on your MoBo.
Since you have 7 case fans in total, two most common connection methods;
SYS_FAN5 - rear exhaust fan
SYS_FAN4 - bottom intake
SYS_FAN3 - two front intake fans using Y-splitter, to connect two fans on single header
SYS_FAN2 - one of the top exhaust fans
SYS_FAN1 - remaining two top exhaust fans using Y-splitter, to connect two fans on single header
Note: using Y-splitter will make both fans run in sync and individual fan control is impossible.
Or if you want individual control over each and every case fan, then;
SYS_FAN5 - rear exhaust fan
SYS_FAN4 - bottom intake
PUMP_FAN2 - one front intake fan
SYS_FAN3 - remaining front intake fan
SYS_FAN2 - one of the top exhaust fans
SYS_FAN1 - 2nd top exhaust fan
PUMP_FAN1 - 3rd top exhaust fan
Note: If you use this connection method, do go to BIOS on power on and change the fan mode on PUMP_FAN headers from PWM to DC. Since else-ways, those two Noctua fans connected to PUMP_FAN headers will operate at 100% at all times.
with the CPU and PUMP fans with the Noctuas, should I select "Auto" or is "DC" better?
I connected up the front panel leds, and the "Power Switch", "HDD LED", and "Reset Switch", but the phantek manual says "Power LED is replaced by RGB lighting with PH_ES614PTG model", and they appear to have not supplied RGB lighting.
I suddenly remembered I bought also a further external panel! so will have to wire that up also.
I just dont know whether all the USB C sockets have fast charging, or whether its just JUSB2 if you connect up PD_PWR1?
the power switch at the top of the Phantek tower is a great idea, as one can access it from any direction, whereas my 2010 PC has this at the front panel which is tricky to access, where nowadays I boot by pressing any key of the keyboard.
the problem is say a person enters their current info on a website, where it goes to a file.
the existing data is 1st Jan 2024, weight 80kg, age 50, height 6'.
it is now 26th Jan 2024, their weight is now 81kg, age 51, same height.
this new data is being written to disk,
but the backing up recorded 26th Jan 2024 after this was written, but records 80kg age 50 height 6' before these new values got written. thus the backup now says 26th Jan 2024, weight 80kg, age 50, height 6',
but this data is incoherent.
its similar to an artist drawing a moving subject, at any moment in time he draws whatever he sees right now which should be at the point of the canvas he has reached, but the subject is moving all round the place, and he gets chaos!
eg the subject is a fly, flying around, and the picture has 100 flies, when there was just 1, but it kept moving around!
No.
It does a Volume Shadow copy, and proceeds from there.
Things happening during the Image creation may not be reflected in the Image, but the Image creation is quite fast. Especially if you are into the Incremental or Differentials.
one way to do coherent copies, is that when the OS boots up, before the first disk write if any, a disk is coherent.
once the OS is running, you have the problem of someone editting a file, and the text editor writing to the disk, but not closing the file for hours!
a file is only coherent when it is shut. eg say you copy a 1 Gig file, and 200MB has been copied so far, that destination file is incoherent and usually you cannot access that file.
if via the OS, for an app on request for the next boot up, for a requested range of segments
with the next boot up, for any writes to the disk, the OS checks if this is the first rewrite of a requested sector, if it is, it first reads the original contents and caches the original contents to disk, and then does the write. with time less and less sectors will be cached. it only caches the original contents for the first rewrite.
what the app does is request sectors from the OS, and the OS first checks the cache of original contents, if the requested sector is in the cache, it sends the cached version, otherwise it reads from the disk directly and sends that.
that would enable a backdated coherent copy to be gradually built up. eg you switch off your computer at midnight 26th March.
when you next boot up, the app will now gradually build up a coherent copy of how the disk was at midnight 26th March. it might complete this on 28th March.
eg say just sector 100 and 105 are rewritten to, the OS caches the original contents of 100 and 105.
but if 100 and 105 are written to again, it doesnt cache again.
the app requests sectors 0, 1, 2, 3, ...., but when it reaches 100 the OS finds this in the cache, and sends the cache version, 101 it copies from the disk, similarly 102, 102, 104, but 105 it finds in the cache and uses the cache version.
normally a cache has the freshest version of the data, but this is an inverted cache which caches the stalest versions of data!
the app wont allow you to use the image until it is complete.
when it does the next backup, with each sector, it can check if its the same as the earlier backup, most will be the same, so it only needs to record the ones which arent.
It can only do an Image/Clone/Copy of what is in there currently, at this moment.
Basically, don't be doing a whole lot of stuff while you are making this Image.
The system does not have to be OFF, but don't expect it to make moment by moment saves.
Not only i remembered the PSU tester being the old one, but i also looked at the product image. Since product image is relatively small, it was difficult to make sense if the PSU tester included was the old one or the new, 90 degree one. So, it was mainly because the product image being too small to make a diff between the two.
But i digress.
If you enable SMART fan mode, you can make your own custom fan curve. E.g:
40C - 1000 RPM (50%)
60C - 1500 RPM (75%)
80C - 2000 RPM (100%)
Now, i can see from your Hardware monitor that most of your case fans are spinning 1200-1300 RPM and CPU fan is spinning ~660 RPM. But the two fans connected to PUMP1 and PUMP2 are running at full tilt (2000 RPM) and this is the source of your noise. So, look at those and their profile. You should be able to set custom profile to those two fans as well.
It is not MSI thing, it is thing with EVERY MoBo manufacturer out there. That, and also the fact that Micro$oft also pushes UEFI, namely Win11 and onwards is not compatible with Legacy/BIOS.
Meaning that when you have Win11 installed under UEFI and in GPT partition table and you change BIOS mode to CSM, it essentially bricks your Win11, since Win11 (namely GPT) won't boot in CSM (Legacy/BIOS) mode.
Win10 can be installed under Legacy/BIOS and in MBR partition table, as well as under UEFI and in GPT partition table. So, Win10 is more versatile in that regard.
Without PD_PWR1, JUSB2 still operates fine as USB type-C port, but is able to deliver up to 15W of power.
So, when you don't need up to 60W of power from type-C port, you don't need to connect the PD_PWR1 power cable.
Well, it changes the JUSB2 connector max power output from 15W to 60W.
So, when you have a USB type-C device that needs to operate at 15W or more, but less than 60W, it has it's merit. Otherwise, No.
Neither of the two SATA power connections needed to be connected, but to inform you what they are needed for:
* One is to power the PWM fan hub, which you can't use.
* Another is to power the LED button on front I/O panel and any connected Phanteks LED products (e.g LED strip or Halos fan frame), which you also won't use.
If you have it "Auto", MoBo should be able to tell that you have 3-pin fans and adjust accordingly. But you can put it into "DC" to be sure. Since when it is in "Auto" and MoBo thinks you have 4-pin fans, MoBo will put 3-pin fans spinning at full tilt. So, better to have it in "DC" mode.
Other USB type-C ports should be at 15W. Now, MoBo manual doesn't state the wattage limit on other USB type-C ports (including those at back I/O) but common is max 15W.
but I found that USB 3 and USB4 look like USB2 sockets, the only available mobo sockets are in any case USB2, namely JUSB5 or JUSB6. USB1 and USB2 are both USB3 sockets via a twin cable. and the USB C socket is via JUSB1 or JUSB2,
where the photo at the above URL shows USB1 USB2 USB3 USB4, those arent USB versions! but are just a numbering of the sockets!
I also managed to boot Linux Mint without any hard drives installed!
I always wondered if that would work! it does!
I found there was a problem with the keyboard, where it wasnt functioning from the USB to PS2 splitter, even though it functions with my 2007 laptop and 2010 PC. but I exchanged it with my laptop's more ancient PS2 keyboard and now that functions via the splitter.
now as regards installing the M.2 drive, which is this one:
there is the question of which M.2 socket to use. you probably said which in the earlier topic, but it will be work to locate the advice! should I go for M2_1 which is connected to the CPU?
the other 3 will also support a 2280 and are via the chipset, I dont know if any of those 3 will be at the cost of something else. currently I am just aware of the PCI_E3 being at the cost of USB3 C socket 2 of the back panel on document p23. luckily that isnt the video USB3 C.
am I right in thinking I could for example have 4 of those 2280 M.2 drives in the above URL?
would there be any speed penalty with any of them?
Not only i remembered the PSU tester being the old one, but i also looked at the product image. Since product image is relatively small, it was difficult to make sense if the PSU tester included was the old one or the new, 90 degree one. So, it was mainly because the product image being too small to make a diff between the two.
But i digress.
the human mind isnt what we think! interpretations override specifics with memory. has happened to me where I revisited something, eg a film and it wasnt what I remembered, the memory was of the assumption. where there is no assumption, you will remember more than you interpreted. there is a great B movie about memory called "extraction":
as there are several unrelated DVDs with the same title, it is this one:
but I run into a new problem, that if you look at the right hand y-axis, that is RPM, but it goes all the way to 15000, and so 0 to 2000 are just covered by 0 1500 3000, I have no idea if I have set it to 2000.
I dont understand the graph, because on the left y-axis it gives temperatures, and on the right RPM,
but I would expect the x-axis to be temperatures, and the y-axis rpm.
now when I move a data point, eg say the leftmost one, it does show an x-axis of temperature, see this photo:
Now, i can see from your Hardware monitor that most of your case fans are spinning 1200-1300 RPM and CPU fan is spinning ~660 RPM. But the two fans connected to PUMP1 and PUMP2 are running at full tilt (2000 RPM) and this is the source of your noise. So, look at those and their profile. You should be able to set custom profile to those two fans as well.
ok, I did a custom curve for each, guessing a lot, but trying to conform to the 3 data points you gave, and now it is silent, all fans stopped except the CPU cooler ones which are implausibly quiet.
Now I havent installed any temperature probes, so how does it know the temperatures?
It is not MSI thing, it is thing with EVERY MoBo manufacturer out there. That, and also the fact that Micro$oft also pushes UEFI, namely Win11 and onwards is not compatible with Legacy/BIOS.
Meaning that when you have Win11 installed under UEFI and in GPT partition table and you change BIOS mode to CSM, it essentially bricks your Win11, since Win11 (namely GPT) won't boot in CSM (Legacy/BIOS) mode.
Win10 can be installed under Legacy/BIOS and in MBR partition table, as well as under UEFI and in GPT partition table. So, Win10 is more versatile in that regard.
you should only go to legacy mode to run legacy things,
the legacy mode is just a disposable launchpad, once launched, the software side at least is identical whether you go legacy or UEFI,
even if you used UEFI, that also is a disposable launchpad, its all part of the "bootstrapping" problem.
UEFI and legacy are both just scaffolding, by the end of the bootstrap the OS will control the hardware directly via its own drivers. legacy in fact forces this as the legacy BIOS is 16 bit.
in the 1980s a lot of machines just had the OS in the ROM, and no booting. ironically the Amiga 1000 booted part of the OS with a "kickstart" disk, then an OS boot disk called Workbench. but with the next Amiga, the Amiga 500, the kickstart was now in the ROM!
with my Amiga 500, the ROM was Kickstart 1.2, there was a software upgrade to 1.3. eventually I decided to upgrade to Kickstart 2.0, which is what the next Amiga, the Amiga 600 had. but that required a ROM chip to be soldered, I found someone to solder it for me. I wish I had never upgraded it, because many of the early games only worked with Kickstart 1.x.
via legacy mode, the boot process begins in 16 bit mode, it then promotes itself to 32 bit, eg 32 bit Windows would do this, and then the 32 bit mode can promote itself to 64 bit.
eg 64 bit Linux does this.
I need legacy mode to run some of the code I wrote which runs without OS, where it boots from a floppy disk, where it will also boot from a USB floppy via firmware emulation of the ancient floppy hardware.
I'll try that, if it fails, I have to study the UEFI system, I will in any case have to study the UEFI system for future things. I think UEFI was around when I coded, BUT I wanted my code to run on all systems, so I didnt work with it.
if you program, you need to program for the lowest common denominator, otherwise the code wont work on many systems. the multicore code I wrote, is JUST for AMD non hyperthreaded multicores. I can run it on Intel, but by only running the initial core.
now it might in fact run on AMD hyperthreaded multicores, I would have to study the code. the problem with Intel hyperthreading, is the hyperthreaded cores share certain model specific registers, this is very bad engineering! with software, it has been known since last century that shared variables lead to bugs. where one core could set some hardware register, and then another core change it without its permission. you have to do a lot of unnecessary extra study to be sure that any changes of shared variables are coherent, it is total incompetence by Intel to do this! Also Intel uses different model specific registers from AMD for multicore. the general architecture is identical.
hyperthreading just causes confusion, better just to have lots of single cores, much cleaner system. I read somewhere that they dont even go beyond hyperthreading 2 because of complexity bloat! they shouldnt be doing this in hardware, the entire point of multitasking is to allow one CPU to pretend to be 10 or 20 or 25 etc. hardware assist will just go at cross purposes, and in any case is unviable to do more than 2, what a fiasco!
with the original Intel architecture, they put all kinds of things to "help" programmers, but with the development of RISC, they studied which of that help programmers actually used, and found it was virtually none of the facilities, so they removed all that junk, and arrived at eg ARM, which is super efficient.
the way things work is the hardware boots up with just 1 core, and this has to wake up the other cores, and this is quite involved. I havent looked at this literally for more than a decade, but if I remember rightly the 1 cpu has to communicate with another cpu as if it was some peripheral hardware for the other cpu! AMD makes the caches of the different cores always coherent, eg if the one core writes data which just is in its memory cache, and the other core reads that memory location, the AMD hardware ensures it reads the fresher data in the other core's cache.
I will try to get it to run on the new machine, but I have to first troubleshoot if and where it fails with the legacy boot, and then to try and get it to run with UEFI. I think I may have to discard a lot of the early boot stuff for UEFI, as UEFI has probably done that by its own MO.
access to other OSes can be lost, and it can obliterate other installed stuff.
has happened to me when installing both Ubuntu Linux and different versions of XP on the same machine, where in some cases I'd get a nice list of OSes to boot from, and in other cases I lost all access to the earlier OSes. I sometimes had 2 versions of Linux, or 2 versions of XP. I think in one case I installed both 32 bit and 64 bit Ubuntu Linux.
when you install the new OS, you need to be really sure about the partition you install to. I never install more than 1 version of Windows to the same partition, as that just leads to clutter and confusion.
there are some OSes out there, where if you install, they will obliterate anything that used to be on the disk.
thus before I install to an in use disk, I will do a sector by sector backup, that way I can revert back to the original disk if things go wrong.
repairing a Windows install is also risky, the OS can become irretrievable. best to do a sector backup of the entire problem disk before trying this!
there is also some general danger of whether you have allowed enough disk space for the install.
I dont like keeping user data in the Windows partition, so I dont want it too huge.
right now with this Windows 10 install which I installed in June 2021, ie almost 3 years ago, the OS partition is 122 Gig, and there is 39.4G free, so the OS including installed apps etc, uses some 83Gig. I judged that one correctly, where there is a nice amount of free space for further installs but not too much.
if I had set 200Gig that would be way too much.
the OS partition is the most accessed, and so it will probably be the first disk to wear out.
I had a Samsung IDE system drive, where the first sector eventually wore out, had to junk the drive after salvaging the data.
it is bad engineering for the OS to be unable to work from any partition, old era or new era. I can understand them wanting a more advanced filesystem, but they can do a new era filesystem in an old era partition. a partition is ultimately just a zone of storage to do whatever you want, and is kind of a virtual disk.
Without PD_PWR1, JUSB2 still operates fine as USB type-C port, but is able to deliver up to 15W of power.
So, when you don't need up to 60W of power from type-C port, you don't need to connect the PD_PWR1 power cable.
Well, it changes the JUSB2 connector max power output from 15W to 60W.
So, when you have a USB type-C device that needs to operate at 15W or more, but less than 60W, it has it's merit. Otherwise, No.
I refuse as a matter of principle to use a computer to charge up a smartphone!
most smartphone users dont have PCs, so there will always be a nice convenient way
without needing some hulking technology!
the day will come when they bring in 120Watt charging, and all the existing fast charging
will become obsolete junk!
what they do, is wait till the market for a technology is saturated, ie everyone already bought in, then they release a new technology to force everyone to upgrade. its all about money, not about making life better!
Neither of the two SATA power connections needed to be connected, but to inform you what they are needed for:
* One is to power the PWM fan hub, which you can't use.
* Another is to power the LED button on front I/O panel and any connected Phanteks LED products (e.g LED strip or Halos fan frame), which you also won't use.
If you have it "Auto", MoBo should be able to tell that you have 3-pin fans and adjust accordingly. But you can put it into "DC" to be sure. Since when it is in "Auto" and MoBo thinks you have 4-pin fans, MoBo will put 3-pin fans spinning at full tilt. So, better to have it in "DC" mode.
Other USB type-C ports should be at 15W. Now, MoBo manual doesn't state the wattage limit on other USB type-C ports (including those at back I/O) but common is max 15W.
"to achieve USB PD 60W fast charging for JUSB2, the PD_PWR1 connector needs to be connected to the power supply unit".
interpretation 1:
the other USB C sockets do fast charging, but JUSB2 doesnt unless you connect PD_PWR1
interpretation 2:
only JUSB2 does fast charging, and only if PD_PWR1 is connected.
you have to be careful that with UK consumer law, traps and pitfalls need to be explicitly stated up front. anything which isnt stated should be assumed to be favourable.
ie according to UK consumer law, their wording means all the other USB C sockets are fast charging, just JUSB2 isnt unless you connect PD_PWR1, otherwise it is classed as "unfair trading". ie interpretation 1 is the expected interpretation. interpretation 2 is unfair trading, and is sharp trading.
with UK consumer law, ambiguities legally are always interpreted in favour of the consumer.
Of course it works, since entire OS is loaded into RAM.
I have essentially the same backup method in use, whereby i have bootable Linux Mint on my USB thumb drive and even when ALL storage drives fail, i can still boot to Linux Mint to access my PC and use it.
there is the question of which M.2 socket to use. you probably said which in the earlier topic, but it will be work to locate the advice! should I go for M2_1 which is connected to the CPU?
The beauty of custom curve is, that you can define the fan profile. Be it relaxed profile (low noise), aggressive profile (better cooling) or somewhere in the middle.
What i wrote in the example is linear profile and falls between relaxed and aggressive profiles.
Which, with 4 data points, would be;
20C - 500 RPM
40C - 1000 RPM
60C - 1500 RPM
80C - 2000 RPM
but I run into a new problem, that if you look at the right hand y-axis, that is RPM, but it goes all the way to 15000, and so 0 to 2000 are just covered by 0 1500 3000, I have no idea if I have set it to 2000.
I dont understand the graph, because on the left y-axis it gives temperatures, and on the right RPM,
but I would expect the x-axis to be temperatures, and the y-axis rpm.
they should change the name from "smart fan mode" to "confusing fan mode"!
Hence the reason why i, personally, don't like the BIOS fan control profiles. Those are hard to understand and also difficult to set up as needed. To remedy that, i'm using fan controller, that is completely separate the MoBo and i don't have to deal with BIOS limitations. Dedicated fan controller is FAR more convenient.
Most DC fans won't spin until they get 5V. 6V usually is enough to keep the fans quiet.
Your fans RPM range is 500-2000 RPM. So, if given that 5V = 500 RPM and 12V = 2000 RPM, you can calculate the RPMs at different voltage levels, which i leave it to you to calculate. And once you've calculated the RPMs at different voltage amounts, you can then decide from my above example, which fan profile to use (relaxed, middle, aggressive) or make your own custom profile.
Because this is how all software fan profiles are able to be configured. You can't make e.g: 20C - 1000 RPM, 30C - 500 RPM, 40C - 1500 RPM. Every dot to the right must be either equal or higher compared to the dot on the left.
Control points are from left to right, where: step #1 (the leftmost one), then step #2, then step #3 and lastly step #4 (the rightmost one).
ok, I did a custom curve for each, guessing a lot, but trying to conform to the 3 data points you gave, and now it is silent, all fans stopped except the CPU cooler ones which are implausibly quiet.
Better to have case fans spinning for airflow. Otherwise case fans are dead weight. Also, what use is a fan, when it doesn't spin? Might as well uninstall all of them, if you don't use them.
Yes, MoBo have several temp probes. You can tell which they are from BIOS fan control profile; CPU, System, MOS, Chipset A, Chipset B, T_SEN1, T_SEN2.
Latter two, are dedicated temp probes, that you can connect directly to the MoBo. MoBo manual, page 52.
There are actually far more temp probes in MoBo, but those 5 built-in and 2 add-on ones are the ones you can use to make your fans to follow the temp.
Of course, there is some merit behind hardware compatibility as well. You can not have a platform (e.g MoBo) that can support e.g 10 years worth of hardware innovation, without ever changing the platform fundamentally.
When you have dedicated drive for each OS, you just unplug all other drives when you make the OS install. Once new OS is bootable, then reconnect all other drives and you're golden. So, no issue there.
has happened to me when installing both Ubuntu Linux and different versions of XP on the same machine, where in some cases I'd get a nice list of OSes to boot from, and in other cases I lost all access to the earlier OSes. I sometimes had 2 versions of Linux, or 2 versions of XP. I think in one case I installed both 32 bit and 64 bit Ubuntu Linux.
Were you trying to install all those OSes on the single drive? With partition splitting? If so, then yes, partition splitting can cause such issues. Hence why i and others advise against it.
Yes, there are two interpretations, but as far as i've seen things to be, as long as it isn't explicitly stated that it can do X, one can safely assume that it can not do it.
Whereby, it is explicitly stated that JUSB2 can do 60W if power cable is connected. Since no such info is said about all other USB ports, it is safe to assume that all other USB ports aren't capable of what JUSB2 is, with power connector. Since else-ways, same would be said about other USB ports as well. OR there would be overall statement of: "all USB ports are capable of 60W charging".
in theory it should, but I have used my own computers since 1988, and used ones of school and uni since 1979, and with my own ones this is the first time I have booted an OS without a disk. now some of the school and uni ones could be run without disk, eg the BBC computer and Sinclair Spectrum, you just switched them on and ready to use. I have a Z88 by Cambridge Computer also from long ago which has an LCD display panel and no disk.
The guy in the next room at uni in 1984 had a Sinclair Spectrum, and the hall of residence had just 1 phone for approx 300 students. One day he emerged from the phone box with a cassette recorder and microphone, he told me he had been recording a free game from a magazine onto the cassette, later he showed me how it works, loading the game from the cassette recorder attached to his Sinclair Spectrum! The game was stored as sounds on the cassette, a bit like the old modem sounds. That computer was by Clive Sinclair who invented all kinds of stuff egthe C5 car
I think this Ryzen CPU could be used as a computer system with eg just 1MB of memory, because it has 128MB L3 cache. you just need physical memory for reading and writing disks and possibly other hardware as peripheral hardware cannot access the caches.
where you shunt data from the disk to the 128MB L3 cache via the 1MB physical memory. I bought a Yamaha SCSI CDRW drive in the late 1990s, and it said "minimum system requirements 16MB, Pentium, ...". I used it from my 68030 Amiga 1200 with 6MB!
I have essentially the same backup method in use, whereby i have bootable Linux Mint on my USB thumb drive and even when ALL storage drives fail, i can still boot to Linux Mint to access my PC and use it.
M2_1 slot uses CPU lanes, so best to install OS drive into there. The rest of the sockets use chipset lanes.
ok, what I am thinking now is whether to deliberately not use the best slot, as this M.2 drive is an experiment, where later I might get a drive of a different size for the best slot. that would also give me an idea of how it fares with the worse slots.
sometimes I get the cheapest worst option, as a learning experience. I then later make an "informed" purchase, where I now know what to go for. and eg the current plan is to not use the graphics card, to understand how the system fares without graphics card.
then maybe after some weeks to install the graphics card.
my first PC in 2004, the HP Pavilion, didnt have a graphics card. I eventually got an ATi Radeon, and the improvement was major!
before that PC, I was using an Amiga 1200 (approx 1993), and before that an Amiga 500 (approx 1988).
this is what you might call a calibration MO, and can be cheaper in the long run, because the informed decision is just the right amount of extra features. eg I was shown a higher end Canon camera, and it was seriously heavy, for stability, too heavy to be any use. Even my current Canon is too heavy, next camera will be a light Nikon. so for the next camera I will scrutinise the weight and also size! this is where my Samsung Galaxy Note 4 was great, as the camera pictures were often better quality than the Canon and much smaller and lighter machine, but my current Oukitel X1 is dubious quality.
now where the cheapest option is expensive, and better options are not much more expensive, I will probably go for a better option, eg getting the 2T M.2 drive. the problem with buying too huge a capacity, is the technology will become obsolete long before you have used say 20% of the drive! this is a problem with a $10000 system, that it ought to last 10 years, but after 3 years, you could maybe buy a $2000 system which is better! thus it is better long term to buy lower high end stuff, and upgrade more often. eg the $10000 invested, can realistically earn $2400 interest in 3 years, where you can keep upgrading every 3 years forever on the interest! some people do this with cars, where they buy on finance a 2024 car, and then within 2 years, eg in 2026, they refinance to a brand new 2026 car, etc, they always drive an almost brand new car. Peugeot say you can buy on 0% finance if you pay 40% up front.
I am considering using one of the chipset ones, just to test that out and experiment, then later maybe get a carefully chosen size for the cpu one. eg maybe I will arrange 1 drive just for Win11, and another drive with XP, Win10 and maybe Linux. I dont want a proliferation of drives.
The beauty of custom curve is, that you can define the fan profile. Be it relaxed profile (low noise), aggressive profile (better cooling) or somewhere in the middle.
What i wrote in the example is linear profile and falls between relaxed and aggressive profiles.
Which, with 4 data points, would be;
20C - 500 RPM
40C - 1000 RPM
60C - 1500 RPM
80C - 2000 RPM
I was a bit mystified as to why these each are different temperature choices, eventually I realised its because the independent variable is the RPM, where its always 500, 1000, 1500, 2000, and the variable dependent variable is the temperature. so really the rpm ought to be the x-axis, and the temperature the y-axis!
Hence the reason why i, personally, don't like the BIOS fan control profiles. Those are hard to understand and also difficult to set up as needed. To remedy that, i'm using fan controller, that is completely separate the MoBo and i don't have to deal with BIOS limitations. Dedicated fan controller is FAR more convenient.
Most DC fans won't spin until they get 5V. 6V usually is enough to keep the fans quiet.
Your fans RPM range is 500-2000 RPM. So, if given that 5V = 500 RPM and 12V = 2000 RPM, you can calculate the RPMs at different voltage levels, which i leave it to you to calculate.
And once you've calculated the RPMs at different voltage amounts, you can then decide from my above example, which fan profile to use (relaxed, middle, aggressive) or make your own custom profile.
Because this is how all software fan profiles are able to be configured. You can't make e.g: 20C - 1000 RPM, 30C - 500 RPM, 40C - 1500 RPM. Every dot to the right must be either equal or higher compared to the dot on the left.
Control points are from left to right, where: step #1 (the leftmost one), then step #2, then step #3 and lastly step #4 (the rightmost one).
Better to have case fans spinning for airflow. Otherwise case fans are dead weight. Also, what use is a fan, when it doesn't spin? Might as well uninstall all of them, if you don't use them.
Yes, MoBo have several temp probes. You can tell which they are from BIOS fan control profile; CPU, System, MOS, Chipset A, Chipset B, T_SEN1, T_SEN2.
Latter two, are dedicated temp probes, that you can connect directly to the MoBo. MoBo manual, page 52.
There are actually far more temp probes in MoBo, but those 5 built-in and 2 add-on ones are the ones you can use to make your fans to follow the temp.
Planned obsolescence.
Of course, there is some merit behind hardware compatibility as well. You can not have a platform (e.g MoBo) that can support e.g 10 years worth of hardware innovation, without ever changing the platform fundamentally.
it can depend how well designed the hardware is, eg with good programming, you should allow for future advances.
you have the concept of "future compatibility" and the concept of "backwards compatibility".
backwards compatibility, is where you arrange so that older era things will still function, eg using a USB2 flash drive from a USB3 A socket.
future compatibility is where you arrange that future things will work with the same system, and it means the future hardware can handle older era things by clever design of the old era.
eg I bought two SCSI-2 adapters for my amiga, the Blizzard SCSI and the Squirrel SCSI, the Blizzard much faster as the Squirrel was via the PCMCIA slot. but when I got a SCSI-3 CDRW drive by Yamaha, it only worked with the Squirrel SCSI. because it had better future compatibility. SCSI-3 is something which didnt exist when the 2 adapters were designed.
using them, I could have 14 SCSI devices! 7 with the one system 7 with the other! but in fact I only had maybe 4 devices.
bad programming doesnt allow for future advances, and you have to reprogram from scratch in the future.
there is an entire system of ideas and MO relating to this problem for software.
eg with sockets, to have some unused wires, for future use. where cables of today wont use the extra wires, and thus would be thinner. but eventually you use another 2, then even later a further 2. if you design everything tight, where everything is used, then it wont last into the future. those 4x2 and 3x2 PSU sockets are interesting, as you could extend them to say 5x2, 6x2 etc, and even say 4x3, 4x4, etc. that is really good design, which allows for future advances. also interesting is the shapes of each wire's socket, that this prevents a plug being attached to the wrong socket.
that's an example of "best practice", USB2 is an example of "worst practice"!
where with USB they had to totally redesign the socket because they ran out of road.
with Unix programming, right from the get go, they got future advances built in, BECAUSE Unix was designed to run on ANY hardware, so right at the start it had to run on incomparably different CPUs and memory systems, some even I think were based on 9 bit bytes, where eventually the standard became 8 bit. You can run Unix on Motorola 68030 and on Intel CPUs. some CPUs are left endian addressing, and some are right endian addressing.
if
byte 0 is $12,
byte 1 is $34,
byte 2 is $56,
byte 3 is $78
with a little endian CPU such as Intel, the first 32 bit word with address 0 is $78563412, because the bytes are interpreted as being arranged as ....., $78, $56, $34, $12
whereas with a big endian CPU such as Motorola, the first 32 bit word with address 0 is $12345678, because the bytes are interpreted as being arranged as $12, $34, $56, $78, .....
mathematically the Intel scheme is better, because it is like decimal digits: 1234, digit 0 is 4, digit 1 is 3, digit 2 is 2, digit 3 is 1, better for maths. the digits are ....3,2,1,0
the central reason it is better is binary digits within bytes are numbered right to left, so if you also number the bytes right to left, this is synergetic with the bit numbering.
well written software will work on both systems, it requires programming skill to do this.
now with the PC world, a lot of programmers only program x86, and with the Amiga, a lot of programmers only programmed the Amiga. Unix and Linux are unusual that right from the get go, you have to program for any hardware.
anyway, because Unix was designed to work on diverse hardware, they designed it to work on ANY hardware, this is why Linux and Unix run on most smartphones, Linux as Android, and iphones via Unix.
Unix is from 1971. its an example of where they did enable a system from 1971 to last all the way till 2024, which is 53 years! on technology nobody could imagine even in 2010!
so it can be done, it all depends on how clever or stupid the engineering decisions are, usually stupid. a few cases clever.
When you have dedicated drive for each OS, you just unplug all other drives when you make the OS install. Once new OS is bootable, then reconnect all other drives and you're golden. So, no issue there.
my MO is to install an OS from a partition, its better organisation of data, the OS should just be for the engine, it shouldnt be for the passengers and luggage! the passengers are the apps, those ought to be installed on a different partition, and the luggage is the data, eg emails, photos, documents, audio recordings, etc. on yet further partitions. if properly organised, the OS and app installs are disposable, because you just reinstall from the download binaries or DVDs.
where you just reinstall, and reconnect to the data on other partitions. I learnt this kind of MO from the Amiga where its the advised pathway, and is in fact the best pathway. with Windows the system disk is a soup of everything, chaos when you want to upgrade or want to use a laptop when travelling, and a tower at home: you have to extricate your user data.
I mitigate the problem of overwriting or losing access to an OS by backing up fully the entire disk, before trying to install an OS. if problems arise, to then reinstate the original disk from the backup. I might then consider installing it to a different disk. I think with my earlier PCs, the BIOS boot options were just "hard disk, optical disk, floppy disk, etc", I couldnt specify which disk, so I had no option but to have all OSes on the same disk!
Were you trying to install all those OSes on the single drive? With partition splitting? If so, then yes, partition splitting can cause such issues. Hence why i and others advise against it.
with my earlier PCs, no other option, as the boot order is just "hard disk", I think with later ones it also has "USB disk", I cant narrow down to a specific drive. I also dont want to disconnect and reconnect, as this wears out the sockets and plugs.
where the guy's dad a wealthy businessman returns back to college, for his astronomy option he brings in someone from NASA to do the coursework for him!
I am considering using one of the chipset ones, just to test that out and experiment, then later maybe get a carefully chosen size for the cpu one. eg maybe I will arrange 1 drive just for Win11, and another drive with XP, Win10 and maybe Linux. I dont want a proliferation of drives.
Looks like you can adjust the voltage from BIOS by 0.01V. At least this much is indicated (e.g 0.36V). Fan motor can adjust the RPM by the value of 1, given that you can fine-tune the voltage as much.
Few calculations;
RPM range of 500-2000 gives 1500 individual RPM steps.
Voltage range of 5V-12V gives 7V range, which when considering 0.01V steps, gives 700 individual voltage steps.
Whereby, every voltage step of 0.01V increases the fan RPM by 2.
Do note that this is general calculation, without considering variables, which are:
* fan RPM rotational speed (+/- 10%)
* minimum voltage required for fan operation
Since i do not have the same fans as yours, the following may not be considered, but for my Corsair ML140 Pro red LED fan voltage and RPM values are:
3.3V - fan stops
3.4V - 600 RPM
4V - 700 RPM
5V - 900 RPM
6V - 1100 RPM (default value where i keep my fans)
7V - 1400 RPM
8V - 1600 RPM (semi-audible)
9V - 1700 RPM (audible)
10V - 1800 RPM
11V - 1900 RPM
12V - 2000 RPM (nice whoosh)
While my fan controller can adjust voltage by 0.1V steps, it shows fan RPM by 100 steps. So, even when fan is running e.g 1050 RPM, my fan controller shows the RPM to jump back and forth from 1000 RPM and 1100 RPM.
Well, that is doable, but it may produce annoying event.
Namely, when fan is turned off (0 RPM) and is started up, many systems feed full 12V to the fans for few seconds, before returning to the set profile. This is to test fan operational state.
So, imagine, you've set up your fans where step #1 is 0 RPM and step #2 at 40C is ~600 RPM. Once the system reaches 40C and starts the fans, they all may blast away for full tilt for few seconds, before returning to set profile. This sudden noise form PC can be startling. And once temps drop below 40C, fans stop again. That is, until next time the 40C is reached, then another full blast comes for few seconds and so forth.
I don't think you like to hear this every time the set temp threshold (e.g 40C) is reached.
Now, this isn't certainty, but instead possibility. You need to test it out by yourself.
Backwards compatibility is a thing but future compatibility is not. Since there is no telling what the future hardware is, since it hasn't been made. So, one can not make their hardware "future compatible".
Most what one can do, with e.g connector ports, is to design plenty of pins into it, whereby rev 1 uses some of it but most of the pins are reserved for future revisions.
Best example is PCI-E x16 slot, which started out in 2003 with PCI-E 1.0 x16, with speed of 4 GB/s. Currently, there are PCI-E 3.0 and 4.0 GPUs out there, while PCI-E 7.0 is in development. PCI-E 7.0 x16 slot is expected to deliver 242 GB/s speeds. That's 60.5 times faster speeds than initial PCI-E 1.0 x16 was capable of. While the physical connector is still the same.
Not so good example is PCI, which was in use for 10 years but since PCI is parallel, it has severe restrictions regarding future developments. Hence why PCI was replaced by PCI-E, which is serial and gives plethora of options regarding future developments. Same is with Parallel ATA (aka PATA/IDE) and Serial ATA (aka SATA).
can you remind me of some info covered in the earlier topic, where if you summarise it again it would also help people just looking at this topic. Rather than trying to locate that one subtopic in the earlier topic, I have decided to reread that entire topic again, making some notes, that way I can remind myself of all the subtopics, so far just completed rereading p1!
I think what you said is about how the chipsets have a bottleneck on the total bandwidth, which is less than the bottleneck via the CPU. if you can remind me what the numbers are, and to what extent that makes the M2_1 slot better than the other M2's.
I am sure your reply comment next about PCI-E 5.0 relates to that. if you could supply the bigger picture for this specific mobo and for this specific M.2 drive. eg what is the bandwidth limit for this M.2 drive, and the bandwidth limits for each of the M2 slots.
I think your next comment is implying the other 4 are all the same?
Looks like you can adjust the voltage from BIOS by 0.01V. At least this much is indicated (e.g 0.36V). Fan motor can adjust the RPM by the value of 1, given that you can fine-tune the voltage as much.
Few calculations;
RPM range of 500-2000 gives 1500 individual RPM steps.
Voltage range of 5V-12V gives 7V range, which when considering 0.01V steps, gives 700 individual voltage steps.
Whereby, every voltage step of 0.01V increases the fan RPM by 2.
Do note that this is general calculation, without considering variables, which are:
* fan RPM rotational speed (+/- 10%)
* minimum voltage required for fan operation
Since i do not have the same fans as yours, the following may not be considered, but for my Corsair ML140 Pro red LED fan voltage and RPM values are:
3.3V - fan stops
3.4V - 600 RPM
4V - 700 RPM
5V - 900 RPM
6V - 1100 RPM (default value where i keep my fans)
7V - 1400 RPM
8V - 1600 RPM (semi-audible)
9V - 1700 RPM (audible)
10V - 1800 RPM
11V - 1900 RPM
12V - 2000 RPM (nice whoosh)
how did you obtain this correspondence between voltage and rpm?
by linear extrapolation and or interpolation, ie guesstimates, or these are genuine numbers?
I can see RPM and temperature given in the UEFI interface, but havent located a voltage measurement.
I know this could be expensive, but have you or Tom's hardware tried testing hardware without fans, in different usage scenarios to see if and where the machine becomes permanently kaput?
it would be too expensive for me to try this, although with my 2010 PC, I removed the external fans, the only fans are the one on the CPU and I think the PSU probably has one. and it has lasted 14 years, now I dont do GPU intensive things. most intensive thing is the astro geminii screensavers, some of which are pixel perfect eg their solar system one. https://www.astrogemini.com/solar-system.html
with my Amiga 500, I was interested in 3D graphics, because in those days full "realtime" 3D graphics was beyond the scope of the machines, and the "Virus" game by David Braben was revolutionary in being able to do realtime 3D graphics just with the approx 7MHz Motorola 68000 CPU, via clever programming tricks.
the URL shows some screenshots, by realtime I mean without latency. I had a flight simulator for the Amiga 500, but with that it might be 1 frame per second, and there was a bit of inertia. with Virus it was sufficiently many frames per second to be as if for real. which even today doesnt seem possible for that CPU. its the only realtime 3D software I am aware of done for the Amiga 500.
that program was written as the flagship graphics software for promoting the Acorn Archimedes computer, which had the original ARM chip, where ARM = "Acorn Risc Machine". the predecessor of the Archimedes was the BBC computer, by Acorn, which was done in conjunction with the BBC TV + radio station, which is the UK government broadcasting, but which is run independently of the government, so is kind of a QUANGO (quasi autonomous non governmental organisation, organisations set up by the government but independent of the government, which are kind of fake NGOs). But the Archimedes didnt do well, and Braben eventually ported it to 68000,namely the Amiga and Atari ST. where we then saw the impressiveness wasnt the hardware impressiveness, but impressive programming.
At our school in 1979 or 1980, they had an even earlier Acorn computer, the Acorn Atom, but it was a bit deficient! I personally used a "Cambridge" computer which booted from a large size floppy disk, (larger than 3.5"). I tried an Acorn Atom and it was no good!
the problem is that even in 2004, which amazingly is 20 years ago, as it seems like yesterday!, 3D graphics had gone way beyond what I wanted, eg that ATi Radeon Chimp + butterfly demo, to the extent that I have lost interest in it now, no interest at all. it only interested me when it was near impossible, and one needed clever tricks.
the Archimedes was strange in having the BASIC programming language in the ROM, the CPU was some 4x as fast as other CPUs of the same technology, via more sophisticated arrangement of the same technology enabling it to work on I think 4 instructions in parallel.
While my fan controller can adjust voltage by 0.1V steps, it shows fan RPM by 100 steps. So, even when fan is running e.g 1050 RPM, my fan controller shows the RPM to jump back and forth from 1000 RPM and 1100 RPM.
Well, that is doable, but it may produce annoying event.
Namely, when fan is turned off (0 RPM) and is started up, many systems feed full 12V to the fans for few seconds, before returning to the set profile. This is to test fan operational state.
So, imagine, you've set up your fans where step #1 is 0 RPM and step #2 at 40C is ~600 RPM. Once the system reaches 40C and starts the fans, they all may blast away for full tilt for few seconds, before returning to set profile. This sudden noise form PC can be startling. And once temps drop below 40C, fans stop again. That is, until next time the 40C is reached, then another full blast comes for few seconds and so forth.
I don't think you like to hear this every time the set temp threshold (e.g 40C) is reached.
Now, this isn't certainty, but instead possibility. You need to test it out by yourself.
I suppose the mitigation is to go for the minimum viable RPM until say the 40C is reached.
I am used to this sudden noise problem with my air conditioning, where the condensation water pump will suddenly activate, even after they are switched off! at first it was startling, but after some months I got used to it and it wont wake me up!
I only need those in the hotter summer months, they can heat as well, I tried that but not keen on it.
now I dont mind the sudden activation, as long as it isnt too often, where the temperature reaches 40C and it sets in, the temperature drops to 39.5 it switches off, it rises again to 40, it sets in, etc, where the switching on and off is say every minute or something like that!
so I guess it may be certain temperatures that are better, where it is unlikely to oscillate at.
Backwards compatibility is a thing but future compatibility is not. Since there is no telling what the future hardware is, since it hasn't been made. So, one can not make their hardware "future compatible".
there are all kinds of things which people thought werent possible, until someone made them possible!
you can definitely get limited future compatibility,
you could do a calculation of the hypothetical limit of a technology, eg for memory, a kilogram of matter is the mass of say 2^88 protons. we can probably assume a memory module wont weigh more than 1kg, so even if we could store 1 bit per proton, that would be some 2^85 bytes, thus hypothetically we probably will never need more than 85 bits for a memory module.
we wouldnt need a ton of memory, as it would be too heavy to lift, unless you plan to use robotic arms to install and move your memory modules!
the RAM probably needs power to maintain each bit of memory, power constraints would probably limit this a lot further.
if you calculate the limits decided by the laws of physics, you can probably design a format which will cover the future indefinitely.
another limit is the time it would take to write that memory, if it takes a year, then its not really viable.
with CPU machine code instructions, they usually leave some opcodes unused, for future use. where a future version of the CPU has extra instructions, and the new bigger set of instructions also leaves some unused.
Most what one can do, with e.g connector ports, is to design plenty of pins into it, whereby rev 1 uses some of it but most of the pins are reserved for future revisions.
Best example is PCI-E x16 slot, which started out in 2003 with PCI-E 1.0 x16, with speed of 4 GB/s. Currently, there are PCI-E 3.0 and 4.0 GPUs out there, while PCI-E 7.0 is in development. PCI-E 7.0 x16 slot is expected to deliver 242 GB/s speeds. That's 60.5 times faster speeds than initial PCI-E 1.0 x16 was capable of. While the physical connector is still the same.
Not so good example is PCI, which was in use for 10 years but since PCI is parallel, it has severe restrictions regarding future developments. Hence why PCI was replaced by PCI-E, which is serial and gives plethora of options regarding future developments. Same is with Parallel ATA (aka PATA/IDE) and Serial ATA (aka SATA).
you dont even need to load an OS into RAM to run it from a read only medium,
you just need to design it carefully, eg you could boot the Amiga from a write protected floppy.
provided you dont upgrade the system, there is no need to write to the system disk other than to keep a record of what was done,
the thing to do is to only record such to a redefinable destination directory, where for a write only medium,
you set the destination directory to nil: where eg with the Amiga the nil: directory just discards anything sent to it!
for upgrades, you also just need a redefinable destination directory, where you use a memory based temporary disk,
if Linux loads the OS into RAM, that may explain why the Linux Mint booting takes so long!
this is incredibly inefficient, as most of the OS will not be used for most sessions!
I think my Amiga 500 booted from a write protected floppy faster than Linux Mint from an optical disk!
The M2_1 is on the CPU side and thus fastest. I'm not familiar with your motherboard but on most motherboards using a NVMe drive in this slot will downgrade the PCIe x16 GPU slot from 16x to 8x.
By leaving the M2_1 slot open, you free up all the lanes for the GPU.
A choice must be made whether it's more important to have the fastest SSD performance or to leave all lanes for the GPU. Default choice is to use the NVMe drive with the OS in that slot.
Next generation of intel chipsets will no longer have this problem.
The three M2 slots on the chipset side are "slower" but we're really splitting hairs here. NVMe drives that don't boot Win 11/10 should be installed in those slots.
The numbers 2260, 2280, and 22110 refer to size in mm. NVMes are 22mm wide and 60, 80 or 120mm long.
Obviously you cannot fit a long drive in a short slot.
It's important to pay attention to compatibility but it has nothing to do with performance in this case. It's just physical size.
Using M2_2 or M2_3 makes sense. It leaves M2_4 open for a longer drive, though it's not certain you will ever install such a drive. But may as well leave the door open.
I think what you said is about how the chipsets have a bottleneck on the total bandwidth, which is less than the bottleneck via the CPU. if you can remind me what the numbers are, and to what extent that makes the M2_1 slot better than the other M2's.
CPU PCI-E lanes are faster than chipset PCI-E lanes since CPU PCI-E lanes are hardwired directly to CPU. While chipset PCI-E lanes go to MoBo chipset and take a long way around, before reaching CPU.
The PCIe lanes on a motherboard originate either from the processor itself or the motherboard chipset.
Generally, the processor lanes are reserved exclusively for the graphics card x16 slots and M.2 slots for high-speed SSDs, as they require to move data without being bottlenecked by the chipset.
On the other hand, chipset lanes connect to onboard USB, other M.2 and PCIe slots, and SATA. The chipset itself transfers data to the processor via a dedicated 4-lane PCIe bus.
So, all devices connected via PCIe lanes to the chipset will have a cap on their maximum bandwidth leading to bottlenecks.
While choosing a motherboard, you must ensure that the PCIe slots you plan to use are directly connected to the processor. You can run a PCIe slot wired to the chipset, but you will risk running into bottlenecks.
A sure way to identify your physical PCIe x16 slot connection would be to identify the lanes allocated to it, as 16 or 8 lanes will directly link with the processor. The topmost PCIe-Slot is almost always connected to the CPU, but do consult your Motherboard manual to make sure this is correct.
Now, your CPU supports PCI-E 5.0 (backwards compatibility with earlier PCI-E versions), and it has total of 28 PCI-E lanes, while 24 of them are usable.
Additionally, from X670E chipset, CPU communicates with MoBo via 12 PCI-E 4.0 lanes.
CPU specs: https://www.amd.com/en/product/12741
I am sure your reply comment next about PCI-E 5.0 relates to that. if you could supply the bigger picture for this specific mobo and for this specific M.2 drive. eg what is the bandwidth limit for this M.2 drive, and the bandwidth limits for each of the M2 slots.
I think your next comment is implying the other 4 are all the same?
with the earlier question, to know what the bandwidth of my drive is, and then the bandwidth of PCe 4.0x4
While numbers are big, this is only numbers and you can see them in synthetic benchmarks. Only real world use where PCI-E 3.0 or 4.0 or 5.0 difference is shown, is when you move LARGE files from one drive to another. Then, the copy/paste takes different time. Otherwise, you can't tell a diff. Heck i even couldn't tell a diff when i moved from 2.5" SATA SSD to PCI-E 3.0 x4 drive, whereby speed diff is 600 MB/s vs 3300 MB/s.
I know this could be expensive, but have you or Tom's hardware tried testing hardware without fans, in different usage scenarios to see if and where the machine becomes permanently kaput?
Fanless PC is very much a thing. Also, many reviews are done on the open bench, where there are no additional case fans (or enclosed PC case) in that matter.
However, do note that building a fanless PC is far harder than building a mini-ITX build, since you are restricted severely by the component TDP. E.g when fanless heatsink is rated to cool 95W CPU and 250W GPU, you can not slap in 120W CPU or 300W GPU. With high end, high heat output hardware, fanless PC isn't cost effective since the heatsinks would be massive in size. Thus, high-end hardware needs active cooling, both to reduce the PC dimensions, weight and also cost.
Now, your CPU supports PCI-E 5.0 (backwards compatibility with earlier PCI-E versions), and it has total of 28 PCI-E lanes, while 24 of them are usable.
Additionally, from X670E chipset, CPU communicates with MoBo via 12 PCI-E 4.0 lanes.
CPU specs: https://www.amd.com/en/product/12741
now the manual p16 also says:
PCI_E1 supports up to PCIe 5.0 x 16
PCI_E2 supports up to PCIe 5.0 x 8, where the expander converts to M.2, does that expander just convert to one M.2 slot, nothing else? huge gizmo to support tiny gizmo?
PCI_E3 supports up to PCIe 5.0 x 4 at the cost of USBC socket labelled (2) on document p23
that PCI_E3 5.0 x 4, is that double the speed of PCI-e 4.0 x 4 = M2_4?
I think you said in the other topic that each new PCI-e version is double the earlier version? wouldnt that PCI_E3 then be 15.75 GB/s?
can this be converted to a 3rd 15.75 GB/s M.2? maybe in the future?
or maybe some kind of PCI-e card with USB4 sockets?
my 2010 mobo got USB3 via a PCI card, possibly PCI-e card. the mobo itself is just USB2.
I eventually found why that card was problematic, I had forgotten to attach the power!
there were no instructions, and I hadnt noticed the power socket originally!
only problem with that USB3 card, is Ubuntu Linux 8.x cannot see drives attached to it, but Linux Mint can, but Linux Mint doesnt have some of the useful progs that the Ubuntu 8.x CD has preinstalled.
read speed similar to write speed I think because it is solid state, I think with magnetic and optical drives, write speeds are much slower than read speeds?
that is an efficient utilisation of the facility! if the socket were lower than the disk, that would be a bottleneck, and if a lot higher, that would be wasted bandwidth! but I suppose something elsewhere on the bus could utilise the wasted bandwidth?
I think I will install it on M2_3 then as the earlier numbers seem to be more important, and I want to leave M2_4 for a 22110
While numbers are big, this is only numbers and you can see them in synthetic benchmarks. Only real world use where PCI-E 3.0 or 4.0 or 5.0 difference is shown, is when you move LARGE files from one drive to another. Then, the copy/paste takes different time. Otherwise, you can't tell a diff. Heck i even couldn't tell a diff when i moved from 2.5" SATA SSD to PCI-E 3.0 x4 drive, whereby speed diff is 600 MB/s vs 3300 MB/s.
where speed makes a difference is for sector by sector backups of entire drives, or for reinstating a drive from a backup, and also for virus scanning a drive.
I accept USAFret's idea of background backing up and differential backups which also is more efficient, where it doesnt add noticeable time, and similarly background virus scanning.
with my 2010 mobo, I think 250Gig can take 8 hours, but with the 2023 HP laptop USB3 one copy I did was much faster, possibly 2 or 3 hours.
for this kind of thing, faster drives could save hours of time!
But of course the source drive can be the bottleneck, where if that is slow, eg a WD Blue says 6GB/s SATA, that could be a bottleneck, depends on how the read speed of the source drive compares with the write speed of the destination drive.
6GB/s would be 41 seconds for 250Gig, absolutely no way that would happen, I am sure it will take hours.
the thing is I dont want to do backups to a permanently connected drive, I dont know if you can power off these M.2 drives?
with SATA drives, has anyone created a SATA cable with power switch?
I suppose with extenders I could disconnect the extenders, I dont like disconnecting at the equipment sockets, only at extenders, as these things eventually wear out and you have to junk the equipment. I have bought dozens of extenders for all my equipment, HDMI extenders, USB3 extenders, USB4 extenders for video, mini USB extender for my camera,
I prefer my backup drives to be powered down except when either backing up or restoring a backup, in order to reduce wear and tear. but I might try USAFret's idea for the system drive, where it ongoingly creates coherent backups.
I bought a stainless steel Kenwood microwave oven many years ago, and kept the mains on all the time, and maybe a month after the warranty ran out, it went kaput. I suspected it had some kind of timer based self destruct. All my earlier microwaves also only lasted 1 or 2 years, a DeLonghi ran out before the warranty and I got a cash refund on the warranty!
so when I replaced the Kenwood with another very similar Kenwood, I only powered on when using the microwave, and it lasted many years!
could be 7 years. the turntable finally began malfunctioning in recent weeks, where I now have a brand new Kenwood which does air frying also, no idea what air frying is! it also does 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100% power, where the previous one only did I think 30%, 50% and 100%. with this one I no longer slide things off the turntable, as I think that wears out the bearings. I now lift and place objects, sliding is prohibited!
from that experience, I prefer to switch equipment off at the mains unless its in use. I think with some medicines also, the medicine self destructs after the expiry date, no idea how they do that.
I think you are interpolating then, where you have spaced the 2 RPMs and the 2 voltages by the same number of steps, which is a different way of thinking the interpolation. there are several ways to think of interpolation, the way I use is via "similar" triangles on the graph:
y
^
|12 D
|
| / |
|
|V C
| / |
| |
| / |
| |
| / |
| |
| /
|3.4 A----------B-----E
|
|
|
|
|
|
|
|
O------------------------------->x
600 R 2000
ABC and AED are "similar" triangles (ie same 3 angles)
thus CB/AB = DE/AE, because this is the scaling factor,
ie
(V-3.4)/(R-600) = (12-3.4)/(2000-600)
what you are doing is subdividing AE and DE into the same number of steps, which will accomplish the same effect, which is a theorem. voltage_step/rpm_step will be the same scaling factor as above. to show this, if you slide the diagonal line segment down, you get a further similar triangle from the horizontal and vertical steps.
Fanless PC is very much a thing. Also, many reviews are done on the open bench, where there are no additional case fans (or enclosed PC case) in that matter.
However, do note that building a fanless PC is far harder than building a mini-ITX build, since you are restricted severely by the component TDP. E.g when fanless heatsink is rated to cool 95W CPU and 250W GPU, you can not slap in 120W CPU or 300W GPU. With high end, high heat output hardware, fanless PC isn't cost effective since the heatsinks would be massive in size. Thus, high-end hardware needs active cooling, both to reduce the PC dimensions, weight and also cost.
The M2_1 is on the CPU side and thus fastest. I'm not familiar with your motherboard but on most motherboards using a NVMe drive in this slot will downgrade the PCIe x16 GPU slot from 16x to 8x.
By leaving the M2_1 slot open, you free up all the lanes for the GPU.
A choice must be made whether it's more important to have the fastest SSD performance or to leave all lanes for the GPU. Default choice is to use the NVMe drive with the OS in that slot.
Next generation of intel chipsets will no longer have this problem.
the booting was noisy, but it was in a different part of the sound spectrum from magnetic hard drives, it was a kind of sporadic grinding noise nearer to the spectrum of voice, and thus not so irritating.
once booted, say you were working on a text file eg a program or a picture with Deluxe Paint, then it was total silence.
you say the floppy sound was constant, that is a disingenuous comment, it was only constant whilst loading a program, or a file, or booting, and that sound was sporadic not constant, AND usually no more than maybe a minute.
a PC fan or magnetic drive with some machines, that is constant sound, where the entire session is continual sound,
sound from loading data or programs is not "constant", eg optical drives often make a lot of noise when loading or writing, but when not loading, silence.
the "whacking" is not a proper complaint as that is you making the noise of your own volition, not the machine!
you shouldnt whack the machine, as that isnt going to help in any way!
each Amiga I bought was about £395, even though bought maybe 4 years apart.
did you use a real Amiga then?
the real problem with floppies wasnt the noise, but was the slowness, and one would mitigate variously, eg to use compressed .lha files, and to do work in the ram disk, just copying back to the floppy at the last moment.
for people who dont know the Amiga, the "guru meditations" were error codes when the machine crash, which you had to decipher from the manuals, ie you had to meditate on the numbers!
if you often worked on the same stuff, you started to know where the crashes would be, and learnt how to avoid them.
anyway you are exaggerating the noise, the main work or game session would be silence, other than say audio for a game.
you just cant compare the noise of an Amiga, where the main session is total silence other than audio, with various PCs where its a continual sound which gradually melts your brain!
you say the floppy sound was constant, that is a disingenuous comment, it was only constant whilst loading a program, or a file, or booting, and that sound was sporadic not constant, AND usually no more than maybe a minute.
The drive motor moved once every second or so unless a diskette was in the drive. With a floppy in the drive, it would either noisily read from it, or sit in complete silence once it had nothing to read.
So there was a way to shut up the floppy drive, yes but left to it's own devices it would never shut up.
I know Amigas. The 1200 was a great machine. But let's not gloss over details.
read speed similar to write speed I think because it is solid state, I think with magnetic and optical drives, write speeds are much slower than read speeds?
E.g M2_3 - PCI-E 4.0 x4 = 7,87 GB/s full bandwidth is allocated to the slot, regardless if there is a drive in it or not.
So, even when you have a drive in it that uses most of the bandwidth, but leaves something unallocated, the unallocated bandwidth remains as unused overhead (aka wasted bandwidth).
In backup situation to external drive, entire backup takes as long as how fast is the slowest component in the chain.
E.g: 990 Pro - USB 3.0 - 2.5" SATA SSD.
Speeds: 36 Gbps - 5 Gbps - 6 Gbps.
In this case, the slowest component is USB 3.0. With this, it doesn't matter how fast the source drive is, be it PCI-E 4.0 x4 or 5.0 x4, since you'll still be limited by USB 3.0 5 Gbps speeds. And USB 3.0 is always a bit slower than SATA3.
But of course the source drive can be the bottleneck, where if that is slow, eg a WD Blue says 6GB/s SATA, that could be a bottleneck, depends on how the read speed of the source drive compares with the write speed of the destination drive.
Western Digital Caviar Blue HDD, will NEVER do 6 Gbps. For SATA 3, 6 Gbps = 600 MB/s, and is the maximum bandwidth of SATA3. Drive itself is FAR slower.
I used to have WD Blue 1TB [WD10EZEX] HDDs, and now looking up my old benchmark results, it did: read 112 MB/s and write 102 MB/s. So, essentially 1/6 of what SATA3 is capable of.
In comparison, my 2.5" SATA SSD, Samsung 870 Evo 2TB, does: read 453 MB/s, write 398 MB/s. About 4x times better than my HDD was capable of, while still leaving headroom of what SATA3 is capable of: 600 MB/s.
My current OS drive, Samsung 970 Evo Plus 2TB, which is PCI-E 3.0 x4 drive, does: read 3350 MB/s, write 3375 MB/s, while the maximum bandwidth of PCI-E 3.0 x4 is 3940 MB/s. So, quite close, leaving ~600 MB/s (0,49 GB/s) as headroom.
You should be able to put the drives into "sleep" mode within Win power plan settings, just like you can do with HDDs. Though, i haven't personally tested it.
But you can not completely power them off when system is on and you're booted into OS.
Even if there is one, it's a bad idea. You have no way of knowing if the system accesses the drive and if you kill the power via manual switch, you can end up with data corruption or even drive failure. Hence why there is "Shut Down" command with PCs, rather than you just flipping the power switch when you're done with PC.
I bought a stainless steel Kenwood microwave oven many years ago, and kept the mains on all the time, and maybe a month after the warranty ran out, it went kaput. I suspected it had some kind of timer based self destruct. All my earlier microwaves also only lasted 1 or 2 years, a DeLonghi ran out before the warranty and I got a cash refund on the warranty!
so when I replaced the Kenwood with another very similar Kenwood, I only powered on when using the microwave, and it lasted many years!
could be 7 years. the turntable finally began malfunctioning in recent weeks, where I now have a brand new Kenwood which does air frying also, no idea what air frying is! it also does 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100% power, where the previous one only did I think 30%, 50% and 100%. with this one I no longer slide things off the turntable, as I think that wears out the bearings. I now lift and place objects, sliding is prohibited!
With home appliances, i prefer to buy industrial/restaurant grade hardware, rather than consumer grade, since those are built far better and are far more reliable.
E.g the microwave oven i have, which also has conventional (air) cooking and also grill functions, has lasted me for ~9 years now, while being powered on at all times. Then again, i bought high-end, restaurant grade microwave oven, from well respected German company in the field: ProfiCook.
The system (MoBo) doesn't know the RPM range of your fans, so, it only can show you voltage. And you have to calculate the RPM at any given voltage by yourself. Or set the voltage and look what RPM feedback fan gives to the MoBo.
Worst that can happen: CPU overheats, thermal throttles (you'll loose CPU compute power) and when temps rise even higher, CPU WILL kill the power to the PC. Whereby whatever you were doing, is lost. And that can easily lead to data corruption. Maybe even storage drive failure.
None of the smart phones are comparable to the compute power of the desktop PC. Hence why they don't need active cooling. IF the smart phone would have the same level of compute power as the desktop PC, it will be so hot that you can't even hold it in hand, let alone use it.
The 2010 PC BIOS shows only one option of Hard Disk perhaps because you only have one HDD(SSD) connected to it? Since BIOS doesn't show you those devices that can be potentially connected to, and booted off from. Instead, it shows you the currently connected devices.
Your new UEFI will identify all drives by name and also gives you option among those.
In boot order, i have plethora of devices. This list is to rearrange the boot order.
Now, when i select one of them, e.g "Hard Disk", my UEFI displays and identifies all drives that i have connected to my PC, whereby i can select which drive to boot off from.
Bootable drives in my system are 970 Evo Plus and 960 Evo. Other two are data drives without OS.
CPU and MoBo are designed to match each other, where there are no PCI-E 4.0/5.0 lane sharing between PCI-E and M.2 slots. It's one of the pros of high-end MoBo you have. Lesser (cheaper) MoBos can have the lane sharing issue but for the most part, GPU and one M.2 drive can live freely in the system, without conflicting each other bandwidth wise.
You should be able to put the drives into "sleep" mode within Win power plan settings, just like you can do with HDDs. Though, i haven't personally tested it.
There may be a simpler menu to deal with this, but one way i know works is to go to Control Panel - System and Security - Power Options, then select whichever plan you are using and edit it. Then select "Change advanced power settings. This finally opens the Power Options menu (part of the old control panel) where you can select after how many minutes of inactivity a hard disk should be turned off.
Even though it's not possible (i think) to manage power off time for drives individualy, the timers for each drive are separate.
By default this value is quite low, 10 minutes maybe? It causes annoying pauses when using file explorer if it has to wait for a drive to "wake up".
I've set it to 60 minutes. Your mileage may vary.
Indeed. The power supply is fine on it's own and even the CPU will probably be fine, but good case ventilation is easy to do and doesn't require noisy fans, or many of them. Just enough will do. Maybe one exhaust fan is all it takes.
Larger fans move more air so they don't need to spin at high revolutions.
Noctua and beQuiet! sell very silent fans. Scythe may be another manufacturer but i have no personal experience.
On a personal note, i have a PSU fan, 2 case fans and a CPU cooler fan running right now and i can sort of hear something so it's not completely silent. It's barely perceptible though.
Not quite. Also, you're forgetting the static pressure part. Sure, 200mm fan may have high CFM but it will have poor static pressure due to low RPM and blade design. So, other than it looking nice and somewhat spinning, there isn't much use for fans bigger than 165mm, like 200/230mm etc.
A video to watch, namely the part starting at 3:10 (but you can watch the whole video if you like);
On a personal note, i have a PSU fan, 2 case fans and a CPU cooler fan running right now and i can sort of hear something so it's not completely silent. It's barely perceptible though.
It all comes down to airflow vs noise. Sure, you can run your fans at below audible level but if fans hardly move any air, what use is to use case fans at all?
What i have, are:
2x 140mm front intake
1x 120mm bottom intake
3x 140mm top exhaust
1x 140mm rear exhaust
and
2x 120mm fans on CPU cooler (also acting as exhaust)
Of course, GPU too has 2x fans but those doesn't spin when GPU is below 60C. And my PSU also has a fan, but PSU airflow is completely separate from the rest of the PC case and PSU fan doesn't spin when load on PSU is below ~40% (that would make ~260W on my 650W unit). Did measure my PC max power draw during bench and got 212W, so, my PSU fan never spins either.
I keep my case fans on barely audible level, so that i know they are there and working. And with so many case fans, to maintain the proper airflow, i don't have to have my fans spinning any faster than that. Now, if i would have far less fans, e.g only 2x, then for proper airflow, both fans must spin far faster. Which also, in turn, increases the audible noise a lot.
Facebook
Twitter
Instagram