Corsair vengeance LPX 2666 MHz dual channel memtest

Hello,

I am new to this forum, nice to meet you all.

I did a memtest86 earlier today and am not sure what all the figures are saying, I noticed under memtest the CAS latency is 19-15-15-31, is this normal for my RAM? How do I know if my system is already using all the 2666 MHz ?

I also only get 4 pass on memtest after 6 hours, is this speed normal ?

Image here

I am using dual channel 2x8 Gb corsair vengeance LPX 2666 MHz, placed under slot 1 and slot 3 (same colour, heard that ram should be placed to the same colour slot for dual channel), both are using XMP profiles (configured under bios).

Additional info =
Mobo : gigabyte Z170X G1 Gaming 3
Procie : intel i5 6600k (oc under bios to 4.2 Ghz)
GPU : Asus strix gtx 970

Appreciate all the advise and input given, thanks!

Are there multiple XMP profiles in bios for your ram? Perhaps its using the wrong one? What do the mem tabs in CPU-Z report?

Hi ninjacore, I have only 1 XMP profile under bios (Default).
Below is the screenshot from CPU-Z

Memory Tab :
Slot #1 :
Slot #3 :

Thanks

It looks like it is reporting accurately in CPU-Z (assuming the ram is rated at 2666/C16).

I would trust that one over what you're seeing in memtest. Possibly the version of memtest integrated into your motherboard is out of date and isn't accurately picking up the xmp profile?

The passes in Memtest86 take forever, so yeah four passes in six hours is normal.

And Memtest isn't the most accurate as far as timings go. Hell, it often reported the wrong CPU speeds as well. Don't worry about it.

For memory placement, make sure you're doing it according to the manufacturer's recommendation. Your mainboard manual will have the correct placement. Usually it is the same color slots, but it might be the 2nd and 4th slots for best performance/stability.

Memtest86/86+ has problems to read memory settings correctly. It also has problems to run in multi threading mode. Tests take so much time as it runs them on single CPU core. Probably multi-core mode won't work at all ( at least it's never working for me ).
Memtest is showing CL19-15-15-31 timings just because it's 1st option on the timing table in SPD ( one of the settings invisible to users ).

If you test memory then better make it in Windows. If you want to check stability in long tests then run HCI memtest, AIDA64 stability test ( memory+cache ) or Prime95 blend. There are other tests too but these 3 seem to find any errors faster and are using most of the RAM. I'm mainly using AIDA64 which is covering about 95-99% available RAM ( available so what's left after loading system files ).

Woomack said:

Memtest86/86+ has problems to read memory settings correctly. It also has problems to run in multi threading mode. Tests take so much time as it runs them on single CPU core. Probably multi-core mode won't work at all ( at least it's never working for me ).
Memtest is showing CL19-15-15-31 timings just because it's 1st option on the timing table in SPD ( one of the settings invisible to users ).

If you test memory then better make it in Windows. If you want to check stability in long tests then run HCI memtest, AIDA64 stability test ( memory+cache ) or Prime95 blend. There are other tests too but these 3 seem to find any errors faster and are using most of the RAM. I'm mainly using AIDA64 which is covering about 95-99% available RAM ( available so what's left after loading system files ).

Click to expand...

Woomack,

Isn't AIDA64 a weak memory / CPU test? It doesn't seem to put much strain on anything
and temps remain fairly low. I use Orthos (Prime 95 with a nice GUI) for two hours or so
and the difference is obvious and much more reassuring that everything is running correctly.

I ask because I only overclock to the limits of stock vcore and OEM cooler capabilities.

New AIDA64 stability test is great for everything but I'm using it mainly for RAM tests. It's using nearly all RAM and finds errors quick but for memory tests you don't have to pick any other tests than cache and memory. Then CPU won't heat up much while memory controller will be still loaded. You can check windows resources monitor. Even if you have 32GB+ RAM then it will use about 95-99%.
At least for me it works about as good as Prime95 blend with that one exception that CPU load/heat is lower ( it shows 100% load but generated heat is lower ). It still uses all CPU threads. I was also comparing some other tests like XTU, HCI memtest and some more but I'm using AIDA64 for many other things so I just rather stick to it than to 3-4 other applications which are giving me similar results.

Woomack said:

New AIDA64 stability test is great for everything but I'm using it mainly for RAM tests. It's using nearly all RAM and finds errors quick but for memory tests you don't have to pick any other tests than cache and memory. Then CPU won't heat up much while memory controller will be still loaded. You can check windows resources monitor. Even if you have 32GB+ RAM then it will use about 95-99%.
At least for me it works about as good as Prime95 blend with that one exception that CPU load/heat is lower ( it shows 100% load but generated heat is lower ). It still uses all CPU threads. I was also comparing some other tests like XTU, HCI memtest and some more but I'm using AIDA64 for many other things so I just rather stick to it than to 3-4 other applications which are giving me similar results.

Click to expand...

Womack,

Thank you! I understand, I was just not using what I have to best efficiency.
I rarely overclock my RAMs such as my new build because they are fast enough
by default, it's my CPU needs better monitoring so I'll make that adjustment
to my testing.

I have similar build for gaming. I mean 6600K+Z170, single 256GB SSD, GTX960, 550W Corsair PSU but I'm changing memory from time to time. CPU runs at 4.2GHz ~1.25V just because motherboard has no LLC options and during full load voltage drops to ~1.2V. At these settings it was working on stock cooler from haswell i3 without issues but later I changed cooling to water just because I had parts from other rig and wanted to have quiet cooling on graphics card. I was also reviewing Cryorig C7 cooler which is great replacement for stock Intel cooler. Temps were about 20*C lower and it's quite cheap. Size is about the same as stock cooler.
Right now I'm checking 2x8GB TridentZ on this rig. I'm not really using any stability tests ( only daily apps and games ) but I won't hide that I was testing 4-5 similar kits so I know what to expect even though this one is overclocking better.
If memory has some stability issues or is damaged in any way then in most cases won't even enter windows. Lower stability issues can be almost always corrected by timings or voltages.

Woomack said:

I have similar build for gaming. I mean 6600K+Z170, single 256GB SSD, GTX960, 550W Corsair PSU but I'm changing memory from time to time. CPU runs at 4.2GHz ~1.25V just because motherboard has no LLC options and during full load voltage drops to ~1.2V. At these settings it was working on stock cooler from haswell i3 without issues but later I changed cooling to water just because I had parts from other rig and wanted to have quiet cooling on graphics card. I was also reviewing Cryorig C7 cooler which is great replacement for stock Intel cooler. Temps were about 20*C lower and it's quite cheap. Size is about the same as stock cooler.
Right now I'm checking 2x8GB TridentZ on this rig. I'm not really using any stability tests ( only daily apps and games ) but I won't hide that I was testing 4-5 similar kits so I know what to expect even though this one is overclocking better.

Click to expand...

Same idea here, I bought an OEM non-K 6600 cooler from ATMINSIDE and I feel I can go higher on CPU
overclock like you have. I did set BIOS manually to 1.2 vcore and stock voltage for RAMs @ 1.25V.
This is aworking computer so I really don't have one to play with and crash all the time while testing .
However I WILL look at that cryorig_c7 cooler because I really don't want a brick hanging off my board
for a CPU cooler.

I reach about 60C - 61C on a two hour Orthos blend test on this OEM cooler @ 4000MHz.