Tweet
Hello,
I wanted to upgrade an old Samsung R540 with 8 GB of RAM yesterday. Should be a five minute job, right? Open the maintenance panel, swap the modules, close it up again. Well...
Turns out I've been running into quite a bit of trouble instead, an quite unexpectedly so. Despite buying a matching pair (continuous serial numbers) of OK-quality and JEDEC-compliant Hynix sticks, they refuse to work in the machine and all attempts at troubleshooting have failed. And because of the type of errors I've been encountering, I don't think the new RAM is defective.
Specs:
NP-R540-JT04DE
i5-480M, HM55, Mobility HD 5470
2x ADATA 2 GB PC3-10600S DDR3-1333 (1.5 V) -> 2x 4 GB Hynix HMT451S6MFR8C-H9 PC3-10600S DDR3-1333 (1.5 V)
BIOS 03KP.M008.20100927.LDG (latest)
The system has been and remains to be rock-stable with the ADATAs. However, from previous experience I know that the UEFI seems to be somewhat buggy: I never could get it to fire up reliably or at all when attempting to boot from an UEFI-installed version of Win7-64 or various flavours of Linux. And running Memtest 7.5 (i.e. UEFI) results in an immediate freeze just like certain older UEFI-equipped machines I've read about elsewhere and in the PassMark forum. At least with the ADATAs, MT 4.3.7 runs fine, however: two sessions á 3.5 hours without any errors in addition to a LinX stresstest under Windows. Based on that, I'd rule out defective RAM slots or memory controller.
Intel specifies the CPU for a maximum of 2x 4 GB @ DDR3-800 or DDR3-1066, and there's a section about upgrading the memory in the R540 user manual, so I'd assume swapping out the factory 2x 2 GB sticks for 8 GB should in theory be supported. I probably should have been concerned by the fact that my internet searches failed to come up with syslog dumps or any other kind of proof that somebody, somewhere had managed to successfully upgrade this type of laptop to the maximum of 8 GB. I did come across two failed attempts resp. troubleshooting requests, but figured using brand-name parts within JEDEC specs should have me covered.
As it stands, the Hynix sticks fail in the exact same manner. No matter the combination (2x 4 GB, 4 GB alone in either slot, 1x 2 GB + 1x 4 GB mixed), neither Memtest 4.3.7 nor 7.5 can even complete the first pass. The machine either reboots or freezes after two seconds at most. While Win7-64 fails to start with 8 GB installed, the SPD readout in MT 7.5 seems normal and Live Lubuntu 16.04.1 x64 did boot to the desktop for what it is worth.
The UEFI is completely barren and basically doesn't have any options other than turning the touchpad and wired LAN chip on or off...not that it made any difference. Installed memory is identified with the correct capacity, however. Samsung never bothered to release any BIOS updates, so the current version is still as shipped from the factory in late 2010.
Any ideas? Is the UEFI so buggy that it simply cannot handle 4 GB modules despite claims to the contrary? I've experienced incompatibility issues with OEM and off-brand vendors memory in the past, but never had either variant of Memtest behave in such a manner.
Kind Regards
I wanted to upgrade an old Samsung R540 with 8 GB of RAM yesterday. Should be a five minute job, right? Open the maintenance panel, swap the modules, close it up again. Well...
Turns out I've been running into quite a bit of trouble instead, an quite unexpectedly so. Despite buying a matching pair (continuous serial numbers) of OK-quality and JEDEC-compliant Hynix sticks, they refuse to work in the machine and all attempts at troubleshooting have failed. And because of the type of errors I've been encountering, I don't think the new RAM is defective.
Specs:
NP-R540-JT04DE
i5-480M, HM55, Mobility HD 5470
2x ADATA 2 GB PC3-10600S DDR3-1333 (1.5 V) -> 2x 4 GB Hynix HMT451S6MFR8C-H9 PC3-10600S DDR3-1333 (1.5 V)
BIOS 03KP.M008.20100927.LDG (latest)
The system has been and remains to be rock-stable with the ADATAs. However, from previous experience I know that the UEFI seems to be somewhat buggy: I never could get it to fire up reliably or at all when attempting to boot from an UEFI-installed version of Win7-64 or various flavours of Linux. And running Memtest 7.5 (i.e. UEFI) results in an immediate freeze just like certain older UEFI-equipped machines I've read about elsewhere and in the PassMark forum. At least with the ADATAs, MT 4.3.7 runs fine, however: two sessions á 3.5 hours without any errors in addition to a LinX stresstest under Windows. Based on that, I'd rule out defective RAM slots or memory controller.
Intel specifies the CPU for a maximum of 2x 4 GB @ DDR3-800 or DDR3-1066, and there's a section about upgrading the memory in the R540 user manual, so I'd assume swapping out the factory 2x 2 GB sticks for 8 GB should in theory be supported. I probably should have been concerned by the fact that my internet searches failed to come up with syslog dumps or any other kind of proof that somebody, somewhere had managed to successfully upgrade this type of laptop to the maximum of 8 GB. I did come across two failed attempts resp. troubleshooting requests, but figured using brand-name parts within JEDEC specs should have me covered.
As it stands, the Hynix sticks fail in the exact same manner. No matter the combination (2x 4 GB, 4 GB alone in either slot, 1x 2 GB + 1x 4 GB mixed), neither Memtest 4.3.7 nor 7.5 can even complete the first pass. The machine either reboots or freezes after two seconds at most. While Win7-64 fails to start with 8 GB installed, the SPD readout in MT 7.5 seems normal and Live Lubuntu 16.04.1 x64 did boot to the desktop for what it is worth.
The UEFI is completely barren and basically doesn't have any options other than turning the touchpad and wired LAN chip on or off...not that it made any difference. Installed memory is identified with the correct capacity, however. Samsung never bothered to release any BIOS updates, so the current version is still as shipped from the factory in late 2010.
Any ideas? Is the UEFI so buggy that it simply cannot handle 4 GB modules despite claims to the contrary? I've experienced incompatibility issues with OEM and off-brand vendors memory in the past, but never had either variant of Memtest behave in such a manner.
Kind Regards
Comment