[HELP] DDR3 2133MHz stable timings

[gohelljustme]

MBD: ASRock > FM2A68M-HD+
CPU: AMD A10-7800 @ 3.5GHz
GPU: AMD Radeon R7 Series (Integrated)
RAM: Corsair Dominator Platinum CMD8GX3M2A2400C11

QVL: https://www.asrock.com/mb/AMD/FM2A68M-HD+/#Memory

Hi! Firstly, I have to say that this RAM was a gift, so I checked it (just in case) with MemTest86 on default BIOS settings with 3 passes and no errors. I am looking for a job right now, so I don't have money for big changes in my setup or to buy new PC.

I don't know much about memory overclocking and I have totally zero experience in it, so I am looking for help with getting 2133MHz stable timings on this setup if it is possible. Of course I am not waiting for ready-steady-go memory timings, because it's impossible, but I want to know how to start overclocking safely with brain to get what I want.

Few problems(?) I have noticed when I were in BIOS:
1. BIOS DRAM Voltage is from 1.27V to 1.70V and I can adjust it by 6.625V, when Voltage for 2400MHz is 1.65 (That means I can set 1.63 or 1.70).
2. AMD APU A10-7800 Max Memory Frequency is 2133MHz (that's why I want to get 2133 instead of 2400).


If you will need additional informations - let me know!

Kind regards,
Michael

 

[Darkbreeze]

First of all, ALL memory configurations need to be ready-steady-go, ALL the time, because if they are not you are looking at the serious potential for micro-errors corrupting EVERY single thing that gets read and re-written. That means your entire OS and any files you work on, change or edit.

Memory errors, even micro-errors that do not cause any noticeable problems like blue screens or freezing, are no joke. Testing is required, and should be done, on ANY memory configuration beyond the basic SPD JEDEC default configuration because anything beyond that, whether tested by the DRAM manufacturer or not, whether verified or not by them, is still technically an overclock and may still result in issues.

And, the maximum memory speed for YOUR motherboard is 2400mhz, not 2133mhz. However, it would likely take a considerable amount of fiddling and testing, to get them to run at that speed and those timings. And it also may not. It might like them right from the start.

So for those modules, I would probably start by simply enabling the XMP or AMP profile in the bios, saving settings and then running Memtest86 for four passes to determine if they are stable at those settings. If not, then you'll likely need to manually configure the memory at whatever speed you wish, using the profile timings of 11-13-13-31 and the tested voltage of 1.65v. You might actually be able to run them at a lower voltage if you choose 2133mhz since 1.65v is what they used when testing at 2400mhz. All of this is outlined on the product page you linked to.

 

[gohelljustme]

Thanks for fast answer. I will let you know about results later.

 

[gohelljustme]

First of all, ALL memory configurations need to be ready-steady-go, ALL the time, because if they are not you are looking at the serious potential for micro-errors corrupting EVERY single thing that gets read and re-written. That means your entire OS and any files you work on, change or edit.

Memory errors, even micro-errors that do not cause any noticeable problems like blue screens or freezing, are no joke. Testing is required, and should be done, on ANY memory configuration beyond the basic SPD JEDEC default configuration because anything beyond that, whether tested by the DRAM manufacturer or not, whether verified or not by them, is still technically an overclock and may still result in issues.

And, the maximum memory speed for YOUR motherboard is 2400mhz, not 2133mhz. However, it would likely take a considerable amount of fiddling and testing, to get them to run at that speed and those timings. And it also may not. It might like them right from the start.

So for those modules, I would probably start by simply enabling the XMP or AMP profile in the bios, saving settings and then running Memtest86 for four passes to determine if they are stable at those settings. If not, then you'll likely need to manually configure the memory at whatever speed you wish, using the profile timings of 11-13-13-31 and the tested voltage of 1.65v. You might actually be able to run them at a lower voltage if you choose 2133mhz since 1.65v is what they used when testing at 2400mhz. All of this is outlined on the product page you linked to.

So, I did memory test after setting XMP 1.3 Profile in UEFI with result 4 passes without errors. What should I do now? Run stress test for X hours or leave it just like that?

#Edit
CPU-Z Memory tab. It looks like I got 2133MHz frequency like I wanted.

https://i.imgur.com/e2vWcYI.png

 

[Darkbreeze]

Yes, it does look that way. SO, if you passed four passes of Memtest86, the next thing to do would be to run Prime95 on a custom pattern.

I will just copy/pasta this from another member I'm currently helping in a similar situation.

Best method for testing memory is to first run four passes of Memtest86, all 11 tests. The paid version has 13 tests but the free version only has tests 1-10 and test 13. So run full passes of all 11 tests. Be sure to download the latest version of Memtest86. Memtest86+ has not been updated in MANY years. It is NO-WISE as good as regular Memtest86 from Passmark software.

If there are ANY errors, at all, then the memory configuration is not stable. Bumping the DRAM voltage up slightly may resolve that OR you may need to make adjustments to the primary timings. There are very few secondary or tertiary timings that should be altered. I can tell you about those if you are trying to tighten your memory timings.

After your memory will pass Memtest for 4 full passes, it is still not necessarily stable.

You should run Prime95 version 26.6 (And ONLY version 26.6) choosing the Custom test. Or you can use a newer version of Prime95 with the Custom test selected but you will need to make the following change.

If you wish to use a newer version than 26.6 make the following edit to the "local.txt" file located in the Prime95 folder.

CpuSupportsAVX=0 (instead of CpuSupportsAVX=1)

Make sure Blend is selected, first. Then, click on "Custom". Input a value of 512k in the minimum FFT size field. Leave the maximum FFT size field at 4096k. In the "Memory to use" field you should take a look at your current memory allocation in either HWinfo or system resource monitor. Whatever "free" memory is available, input approximately 75% of that amount. So if you currently have 12GB (approximately) free out of 16GB total, you should input 9216mb (Which is 75% of 12GB) Generally speaking I usually just input half of whatever the full amount of memory is, so 8192mb if there is 16384mb total.

Do not change the time to run each FFT size. Click run and run the Custom test for 8 hours. If it passed Memtest86 and it passes 8 hours of the Custom test, the memory is 100% stable, or as close to it as you are ever likely to get. If you get errors, (and you will want to run HWinfo alongside Prime95 so you can periodically monitor each thread as Prime will not stop running just because one worker drops out, so you need to watch HWinfo to see if there are any threads not showing 100% usage which means one of the workers errored and was dropped) then you need to either change the timings, change the DRAM voltage or change the DRAM termination voltage, which should be approximately half of the full DRAM voltage.

If you cannot pass Memtest86 with the memory at the default JEDEC SPD of 2133mhz with everything left on the auto/default configuration, then there is likely something physically wrong with the module AND you SHOULD run Memtest86 to test the memory at the default, non-XMP or custom profile settings BEFORE ever making any changes to the memory configuration so that you will know if the problem is a setting or is a physical problem with the memory. No memory should ever fail to pass Memtest86 when it is at the default configuration that the system sets it at when you start out or do a clear CMOS.

Running and passing 8 hours of Realbench, after verifying that the memory passes these other tests, would be highly recommended as well. Especially if you also have a CPU overclock configured, although, CPU stability if it is overclocked should be done BEFORE any memory configuration is done. Realbench WILL test with AVX instructions, so there is no need to worry about not using them in Prime. Unlike Prime95, Realbench will use realistic AVX workloads and issues with CPU or Memory stability should become obvious as the program will error out.

 

[gohelljustme]

Yes, it does look that way. SO, if you passed four passes of Memtest86, the next thing to do would be to run Prime95 on a custom pattern.

I will just copy/pasta this from another member I'm currently helping in a similar situation.

Best method for testing memory is to first run four passes of Memtest86, all 11 tests. The paid version has 13 tests but the free version only has tests 1-10 and test 13. So run full passes of all 11 tests. Be sure to download the latest version of Memtest86. Memtest86+ has not been updated in MANY years. It is NO-WISE as good as regular Memtest86 from Passmark software.

If there are ANY errors, at all, then the memory configuration is not stable. Bumping the DRAM voltage up slightly may resolve that OR you may need to make adjustments to the primary timings. There are very few secondary or tertiary timings that should be altered. I can tell you about those if you are trying to tighten your memory timings.

After your memory will pass Memtest for 4 full passes, it is still not necessarily stable.

You should run Prime95 version 26.6 (And ONLY version 26.6) choosing the Custom test. Or you can use a newer version of Prime95 with the Custom test selected but you will need to make the following change.

If you wish to use a newer version than 26.6 make the following edit to the "local.txt" file located in the Prime95 folder.

CpuSupportsAVX=0 (instead of CpuSupportsAVX=1)

Make sure Blend is selected, first. Then, click on "Custom". Input a value of 512k in the minimum FFT size field. Leave the maximum FFT size field at 4096k. In the "Memory to use" field you should take a look at your current memory allocation in either HWinfo or system resource monitor. Whatever "free" memory is available, input approximately 75% of that amount. So if you currently have 12GB (approximately) free out of 16GB total, you should input 9216mb (Which is 75% of 12GB) Generally speaking I usually just input half of whatever the full amount of memory is, so 8192mb if there is 16384mb total.

Do not change the time to run each FFT size. Click run and run the Custom test for 8 hours. If it passed Memtest86 and it passes 8 hours of the Custom test, the memory is 100% stable, or as close to it as you are ever likely to get. If you get errors, (and you will want to run HWinfo alongside Prime95 so you can periodically monitor each thread as Prime will not stop running just because one worker drops out, so you need to watch HWinfo to see if there are any threads not showing 100% usage which means one of the workers errored and was dropped) then you need to either change the timings, change the DRAM voltage or change the DRAM termination voltage, which should be approximately half of the full DRAM voltage.

If you cannot pass Memtest86 with the memory at the default JEDEC SPD of 2133mhz with everything left on the auto/default configuration, then there is likely something physically wrong with the module AND you SHOULD run Memtest86 to test the memory at the default, non-XMP or custom profile settings BEFORE ever making any changes to the memory configuration so that you will know if the problem is a setting or is a physical problem with the memory. No memory should ever fail to pass Memtest86 when it is at the default configuration that the system sets it at when you start out or do a clear CMOS.

Running and passing 8 hours of Realbench, after verifying that the memory passes these other tests, would be highly recommended as well. Especially if you also have a CPU overclock configured, although, CPU stability if it is overclocked should be done BEFORE any memory configuration is done. Realbench WILL test with AVX instructions, so there is no need to worry about not using them in Prime. Unlike Prime95, Realbench will use realistic AVX workloads and issues with CPU or Memory stability should become obvious as the program will error out.

Test has ended few minutes ago with result of 0 errors and 0 warnings on all workers and all workers had 100% uptime at 100% CPU usage.

I will let you know about Realbench stress test results in this post when it will comes to end.

 

[Darkbreeze]

Well, that looks great so far.

 

[gohelljustme]

My PC is right after Realbench stress test (8 Hours + Up to 8 GB) and everything is ok in this point (test passed). I will continue to looking at memory stability while playing in three different games with three different requirements in next three days. If anything will happened I will let you know in this thread. If there will be no errors/BSODs/something I will mark this topic as "Solved".

Thanks for your time!