Does permanent usage of memory XMP Profiles shorten the life of the system?

[NDRE28]

Hello! 

 

Is it OK to use memory with an XMP Profile loaded permanently? 

 

1. Will it shorten the life of the memory, CPU or motherboard?

 

2. Will it create system instability and overheat?

[Origami Cactus]

Yes

No

It shouldn't, but if your system is unstable lower the ram speed. It depends on your motherboard and platform thought.

[NDRE28]

10 hours ago, Origami Cactus said:

Yes

No

It shouldn't, but if your system is unstable lower the ram speed. It depends on your motherboard and platform thought.

My motherboard is Gigabyte Z390 Aorus Master. 

 

My CPU is Intel 9700K. 

 

My current memory is Kingston HyperX Fury DDR4-2666 CL16-18-18 @1.2V (it works at these speeds without using an XMP Profile, it’s Plug-N-Play). 

 

I am interested into buying a HyperX Predator DDR4-3200 CL16-18-18 @1.35V. 

[Origami Cactus]

20 minutes ago, NDRE said:

My motherboard is Gigabyte Z390 Aorus Master. 

 

My CPU is Intel 9700K. 

 

My current memory is Kingston HyperX Fury DDR4-2666 CL16-18-18 @1.2V (it works at these speeds without using an XMP Profile, it’s Plug-N-Play). 

 

I am interested into buying a HyperX Predator DDR4-3200 CL16-18-18 @1.35V. 

3200mhz is so slow, it will work fine, and wont affect anything.

If you were running 4266mhz ram i would be worried, but 3200mhz at 1.35mhz is totally fine. In fact, that is exactly the ram i am running now, same timing and everything.

[porina]

1, not in any meaningful way

2, maybe.

 

Things have got a lot better over the years, but compatibility is still not 100%. Some boards, some CPUs, some ram just wont get along. The faster the ram, the higher the chance it doesn't work. Last night I put in HyperX 4000 ram into a system with i3-7350k on Z710 chipset mobo. The ram has XMP profiles at 3600 and 4000, so I tried 3600 first. It booted, but running ram stability test it gave errors so that was out. I'll try different ram in it later. 

 

Don't forget that by having faster ram, you potentially allow the rest of the system to do more work. If you're close to the limit of stability already, it could push you over. As long as the system has adequate cooling, and you're not doing a crazy overclock of the CPU, it should be fine.

 

Also, unless you really need more ram, or want different looking ram, I wouldn't bother with an upgrade from 2666 to 3200. It will give a small performance boost here or there, but it isn't going to really change much.

[NDRE28]

12 hours ago, porina said:

1, not in any meaningful way

2, maybe.

 

Things have got a lot better over the years, but compatibility is still not 100%. Some boards, some CPUs, some ram just wont get along. The faster the ram, the higher the chance it doesn't work. Last night I put in HyperX 4000 ram into a system with i3-7350k on Z710 chipset mobo. The ram has XMP profiles at 3600 and 4000, so I tried 3600 first. It booted, but running ram stability test it gave errors so that was out. I'll try different ram in it later. 

 

Don't forget that by having faster ram, you potentially allow the rest of the system to do more work. If you're close to the limit of stability already, it could push you over. As long as the system has adequate cooling, and you're not doing a crazy overclock of the CPU, it should be fine.

 

Also, unless you really need more ram, or want different looking ram, I wouldn't bother with an upgrade from 2666 to 3200. It will give a small performance boost here or there, but it isn't going to really change much.

Going from DDR4-2667 CL16 to DDR4-3200 CL16 results in a theoretical speed improvement of 16%. 

Do you know how much that would be in percentages in “real world” applications or in benchmarks?

[porina]

34 minutes ago, NDRE said:

Going from DDR4-2667 CL16 to DDR4-3200 CL16 results in a theoretical speed improvement of 16%. 

Do you know how much that would be in percentages in “real world” applications or in benchmarks?

For gaming benchmarks you're going to have to look for other sources yourself. It isn't zero difference, but it wont be the 16% difference either. Expect it to be on the lower end.

 

For non-gaming benchmarks, most don't seem to have too big a factor in bandwidth. The only two I know of as good examples that do scale with ram are Aida64 Photoworxx and Prime95 large FFT.