1 |
On 10/08/18 02:00, Mick wrote: |
2 |
> On Thursday, 9 August 2018 17:32:33 BST Alan Grimes wrote: |
3 |
>> [resend, list was down...] |
4 |
>> |
5 |
>> I've been meditating on the memory gremlin on my system... |
6 |
>> |
7 |
>> The ram is Corsair, 3000mhz. (never had any problem with their sticks in |
8 |
>> any system ever.) |
9 |
>> |
10 |
>> Motherboard is an early release mini-ATX B350 board from Asus... |
11 |
>> |
12 |
>> Chip is a R7 1800X |
13 |
>> |
14 |
>> The pattern is: all cells test good on memcheck but occasionally there |
15 |
>> is a bit error somewhere. I think it is a signaling issue between the |
16 |
>> ram module and the memory interface in the cpu. |
17 |
>> |
18 |
>> After meditating on it, I don't think there's anything I can do about it |
19 |
>> given the STUPID settings the BIOS goes to... The problem with the BIOS |
20 |
>> is that it considers only what the RAM tells it, it does not take into |
21 |
>> account that the CPU is rated at 2667mhz... Well there's the answer, |
22 |
>> this is AMD's first product with DDR4 support, and it's not super |
23 |
>> awesome so simply acknowledging the limitation there, and setting the |
24 |
>> memory interface to 2666 (which is what the BIOS offers), it won't be |
25 |
>> super fast but it damn well should work. =| |
26 |
> Keep an eye on MoBo firmware updates, Asus are usually OK in providing updates |
27 |
> to stabilise their chipsets, as long as the bugs are fixable in software. |
28 |
> |
29 |
> Also, if the BIOS offers DRAM timing settings increase the latency a notch and |
30 |
> see if that helps. |
31 |
> |
32 |
If you can locate it to a location range, you can use a kernel argument |
33 |
to exclude that area of memory. The hard part is to map the range. Had |
34 |
one system running that way for years. |
35 |
|
36 |
BillK |