8GB in BIOS, 3.13GB useable?
Hey folks!
I'm running CentOS (x86_64) with an old windows laptop and can only use 3.13GB RAM, despite the BIOS seeing 8GB. Code:
[root@highhigh ~]# uname -a First off, please note that I'm aware the onboard graphics is most likely nabbing around 512MB!! :) Unable to flash the bios, the bios updates are only available as .exe files from Lenovo. Using "sudo lshw" we can see the memory: Code:
[root@highhigh ~]# sudo lshw We can see here that only some of the memory is currently useable: Code:
[root@highhigh ~]# free -m The board is PAE compatible, from the looks of below: Code:
[root@highhigh ~]# grep -i pae /proc/cpuinfo |
Can you use top on console?
It might be nessecary to flash the Bios. |
Quote:
My original post was using stats from htop to easily find the viewable RAM size. :( Quote:
Thanks for the reply, Keruskerfuerst. Any other suggestions? :) |
.exe files are often just containers designed to unzip into a specific Windows directory, and their contents accessible using your favorite distro's unzip extractor.
|
8GB in BIOS, 3.13GB useable?
Update through windows PE
|
A couple of thoughts:
* You should only update your BIOS if you're sure that will fix the problem - it's not clear to me that doing that will fix it. * If you're using a 64-bit version of CentOS and therefore your machine must be a 64-bit machine, if it's working, then; that can't be the problem that you're machine is only a 32-bit machine. If you were using a 32-bit system, then it will only be able to use 4GB of RAM and not 8GB of RAM. * On my CentOS 7.4 system (that has 8GB of RAM), 7.7GB is usable and not the whole 8GB of RAM - because some of it is reserved by the hardware for it's own use and therefore is not available to the OS. * It seems strange to me given the above that your machine's BIOS see's the 8GB of RAM, but the OS only see's 3.13GB of RAM. So, a couple of suggestions: Are you sure that the OS is only seeing 3.13GB of RAM? Have you had a look at another utility to make sure that IS the case? Have you ran memtest on your machine to see if there ain't any problems with the installed RAM, that maybe stopping the OS seeing the rest of the RAM installed? Bear in mind that it's quite normal that at least some of the RAM will be unavailable to the OS for it's use and therefore, the OS will not be able to address the whole 8GB of RAM. Also bear in mind that exactly how much is available to the OS, will most probably vary depending on your hardware. |
Quote:
Quote:
Quote:
Quote:
Quote:
Appreciate the input and time taken to give help on all the above posts. Thank you! |
Quote:
I'll try some memtests before proceeding.. |
Quote:
Code:
[root@highhigh ~]# memtest-setup |
Try running it off a live CD/USB instead. As, it can't test memory in use by the OS (or memory the OS does not have access to) while CentOS is running.
https://ask.fedoraproject.org/en/que...-support-uefi/ |
Thanks for the advice.
As the bios likely needed flashing and I also struggled to format a decent memory tester onto a USB from my Mac I've taken a strange route but, we're getting there.. - installed some light MSDOS OS, just to test and see if i could get some memory checks / bios flash running. ..no luck - downloaded windows 8.1 (as this version matches the Lenovo drivers on their support website). - attempted to install windows 8.1 using three USB's as HDD's. Who knew, it doesn't like that! - attempted to use an external drive, needed to be GPT format.. had to remove all my wonderful content to format the drive :( - found out then that windows wont accept external drives as main HDD, did I format for nothing? - Not to be defeated. I pulled apart the external drive, removed the disk and frame. - found an adapter, mounted external disk inside the faulty laptop. - booted (after many bios hurdles and more reboots than you could count) and installed 8.1. when loading 8.1, device info says something like: RAM: 8GB (3.24GB usable) ..argh! - transferred the bios flash to this new setup. - flashed bios. - rebooted. - checked 'MSCONFIG' > advanced options > maximum memory. - rebooted. - unable to use more than 3.24GB. - downloaded MemTest86+. - flashed a USB into a bootable memory tester. - running MemTest, computer just turned itself off, I assume due to heat? no errors were found YET, but ill place it on some books and try again. I did take it apart a few days back and give the heatsink and board a very good clean, as well as poke a few larger exit vents for that hot air.. it should be fine on some books, we hope! I'll check back with updates, thanks for your help so far. MemTest (like everything else) see's the RAM but the OS's cant. There is no memory restriction options in the BIOS either. Remember I have swapped the sticks about: If i remove one stick of RAM it sees 3GB, i swap for the other and it sees 3GB. I use either on the other RAM slot individually, same problem.. SO, it seems one stick is not at fault, nor is the individual slot at fault. WHY OH WHY?! :( :( :( :( |
Quote:
That is NOT normal. If that's true, then you have a hardware problem that cannot be fixed with software. I'd say based on that your RAM is faulty or a over-heating problem. Try determine which one it is. If your machine has not just "turned itself off" before running memtest, then I'd say it's faulty RAM that's the issue. |
memtest86 v7.4 (not memtest86+) burned to CD WFM on Gigabyte 1151/Kaby Lake (DDR4).
|
https://i.imgur.com/BwqlyrJ.jpg
running MemTest86+ quite successfully now. I can see the two RAM sticks at 4096 MB each at the bottom yet at the top it states: Memory: 3508M. Any ideas? I'm assuming if this is not normal maybe PAE is not being invoked properly? Unsure, haven't used MemTest before.. |
Quote:
|
Stick one:
https://i.imgur.com/IOErEZw.jpg Stick two: https://i.imgur.com/OXZDzzY.jpg I have an idea, I'll remove one stick and run the RAM test and paste the results. Then, I'll swap and do the same. :) |
Looks like it's the motherboard, ill order a new one online.
Thanks for all your input guys, greatly appreciated! |
Quote:
|
Quote:
Whats your thoughts on this: - MoBo see's both sticks of RAM. - only able to use one stick of RAM. - removing one stick, computer boots. - removing another stick, computer boots. - swapping either stick onto the other RAM slot ..computer doesn't boot. In fact, what i found was when i used the other RAM slot, the screen backlight didn't even come on! We know that the onboard GPU uses a portion of RAM as it's memory (explaining why 500MB was missing on all tests). Whats the thoughts here, I'm assuming this could be a dropped laptop at some point and have a lil damage on the board. Assuming if this was a CPU issue it's not likely that one slot wouldn't be working.. Share your thoughts, very interested. |
Based on everything you've said and described;
I'd say that it's either once again faulty RAM or a faulty memory controller. It's most certainly a hardware issue IMHO. |
Swapping either RAM stick to the identical slot on the board uses that individual stick of RAM to it's expected capacity.
Conclusion: not the RAM. Quote:
Quote:
|
some mobos require ram sticks to be two of a pair.
i once had this; i bought one extra ram stick, and found out it didn't fit with the other. only solution was to sell them, and get a matching pair. also some mobos are very picky about how you put the ram in. say you have four slots, but only two sticks, you must put them in slot 1 & 3, no other way. there might be the tiniest amount of dirt or oxidisation on the stick or in the plug. use some canned air, maybe even a wipe with cloth or gentle brush (but one that doesn't lose hairs). |
Quote:
Try ondoho's suggestions, and if you still have the same problem; Replace the RAM in question with known GOOD RAM that you have ***verified*** as being ***100% compatible*** with your board, and if you still have the same problem; then my bet would be the memory controller that is faulty. Look at/download the manual FOR the board in question and this should tell you EXACTLY which RAM modules ARE compatible with it. |
Quote:
Quote:
Quote:
It does look near perfect now and no good results after that. Possibly something underneath, I'll see if i can unmount it without melting solder etc :D Thanks for the suggestions! |
Quote:
Neither stick works in slot B. :( I have tested using either stick in slot B (with nothing in slot A) and the machine wont boot, the screen doesn't even come on. The graphics are onboard graphics that use a percentage of RAM, it makes sense if no RAM were available this would fail steal it's needed allocation and therefore to power the GPU at all. The above is why i'm 100% sure it's not the sticks, both passed memtest with flying colours when tested individually on Slot A. Quote:
|
Once again:
Quote:
Is the RAM slot in question loose in any way? Does it move even slightly when you touch it - the slot itself I mean - not the RAM module itself. |
Quote:
Quote:
|
Well then, here are your possibilities;
* The RAM module(s) themselves - maybe unlikely given what we know to date. * The RAM slot itself - maybe, but I have my doubts about that. * The memory controller itself - most likely to my way of thinking, once again based on what we know to date. The only other suggestion I can currently think of would be to try replacing the processor with a known good one; this way if it IS the memory controller, that should fix it and you will more likely than not know in that case, that it is indeed the memory controller at fault. Good luck with it. |
I don't see anywhere that you did a BIOS reset before or after flashing. How much trouble is it to remove the motherboard's battery or find a reset jumper? I wouldn't spend any money on anything without trying a BIOS reset first.
|
Quote:
Thanks for the advice |
Moderator response
Moved: This thread is more suitable in <Linux - Hardware> and has been moved accordingly to help your thread/question get the exposure it deserves.
|
there is a linux kernel option of whether to support 4g or 64g of memory please insure your CentOS kernel hsa the 64G set before wondering about a 4G limit on some systems you can read this in: /boot/config # this one, kernel options /boot/System.map # not this one on some kernels the config is also stored inside the kernel and can be retrieved with a utility that slips my mind # CONFIG_NOHIGHMEM is not set # CONFIG_HIGHMEM4G is not set CONFIG_HIGHMEM64G=y |
Quote:
Technically speaking, 32bit Windows is as capable of utilizing up to 64GB of the 36bit PAE addressing space as 32bit Linux is, but for Microsoft deliberately laming Win7 (and presumably all subsequent versions to keep the FICTION intact) in 2009 in order to force the gullible consumers to buy their new 64bit OS, that they had just spent mega-bucks developing. See the following links for further discussion: https://www.raymond.cc/blog/make-win...an-4gb-memory/ https://superuser.com/questions/5227...b-of-ram#95309 https://www.makeuseof.com/tag/unlock...ows-pae-patch/ |
Quote:
I appreciate that you've gone through the effort to try and help on this post but in my previous posts I've stated that: - the system is 64bit, including the windows installation. - the CentOS installation was also 64bit. - Ubuntu installs were 64bit. Windows see's closer to 3.5GB RAM, due to the onboard graphics stealing some! As mentioned, I've tested these sticks on different slots and it appears one of the board slots is not working. As above, appreciate your input. I really wish this was the case, it would be a much easier / cheaper fix :) |
All times are GMT -5. The time now is 09:29 AM. |