pb xmp impossible 12900k drr5 6000 ?

remy.g158402dc

New member
Joined
Nov 7, 2024
Messages
14
Hello to you, I am sorry for all of us for what is happening to us with processors 13 and 14, so I changed to a 12th i912900k and I of course updated with the latest version of MAG motherboard bios Z790 TOMAHAWK max WIFI and I get a memory overclocking failed error message when I try to activate the xmp and run the ddr5 in 6000 is this normal in your opinion it comes from the processor which does not support xmp or it can come from a poorly configured bios option, I read that it was necessary to deactivate the secure boot to supposedly give authorization to the machine is overclocked, is this true in your opinion? thank you for your kind feedback.
 
Have you done a Clear CMOS? What is the exact RAM kit? It is not necessary to disable Secure Boot.

True 13th and 14th gen have a slightly improved IMC (integrated memory controller) compared to 12th gen, but also, there is a variation in IMC quality (within a certain range) for each CPU individually. Even taking this into account, DDR5-6000 should be possible with your CPU. Let's see what RAM kit model you have.
 
Have you done a Clear CMOS? What is the exact RAM kit? It is not necessary to disable Secure Boot.

True 13th and 14th gen have a slightly improved IMC (integrated memory controller) compared to 12th gen, but also, there is a variation in IMC quality (within a certain range) for each CPU individually. Even taking this into account, DDR5-6000 should be possible with your CPU. Let's see what RAM kit model you have.
First of all thank you very much for your help, here is the information you asked me

CORSAIR VENGEANCE RGB DDR5 RAM (4x16Go) 6400MHz CL32
Intel XMP Compatible iCUE
 
CORSAIR VENGEANCE RGB DDR5 RAM (4x16Go) 6400MHz CL32

4x 16 GB if you want 64 GB total is not ideal, always prefer 2x 32 GB. See RAM explained: Why two modules are better than four / single- vs. dual-rank / stability testing.

Any total RAM capacity you want, try to reach it with two modules only, especially with DDR5. First i'd make sure you actually need 64 GB total for your workloads, this would only be required for professional workloads. For daily use including gaming, you'd have more than enough RAM if you only use 2x 16 GB in slots A2 and B2, and it should work at XMP no problem. If you don't have any software that benefits from 64 GB, then this amount only makes things worse, not better. With four modules on a board with daisy-chained RAM slots, you worsen the electrical parameters, and of course you double the stress on the memory system.

This RAM configuration might've been just about ok on a "true" 13th/14th gen CPU (the -K models), but it might cause problems for a 12th gen CPU. The best thing to do then is to enable XMP, but also setting "DRAM Frequency" to something lower manually. You could try DDR5-5600 there, this should make it work. Or, if you realize that 32 GB total are enough, you would just use two modules and that should work at -6000, if not even -6400.
 
4x 16 GB if you want 64 GB total is not ideal, always prefer 2x 32 GB. See RAM explained: Why two modules are better than four / single- vs. dual-rank / stability testing.

Any total RAM capacity you want, try to reach it with two modules only, especially with DDR5. First i'd make sure you actually need 64 GB total for your workloads, this would only be required for professional workloads. For daily use including gaming, you'd have more than enough RAM if you only use 2x 16 GB in slots A2 and B2, and it should work at XMP no problem. If you don't have any software that benefits from 64 GB, then this amount only makes things worse, not better. With four modules on a board with daisy-chained RAM slots, you worsen the electrical parameters, and of course you double the stress on the memory system.

This RAM configuration might've been just about ok on a "true" 13th/14th gen CPU (the -K models), but it might cause problems for a 12th gen CPU. The best thing to do then is to enable XMP, but also setting "DRAM Frequency" to something lower manually. You could try DDR5-5600 there, this should make it work. Or, if you realize that 32 GB total are enough, you would just use two modules and that should work at -6000, if not even -6400.
I thank you and will come back very quickly to give you feedback based on your advice.
 
4x 16 GB if you want 64 GB total is not ideal, always prefer 2x 32 GB. See RAM explained: Why two modules are better than four / single- vs. dual-rank / stability testing.

Any total RAM capacity you want, try to reach it with two modules only, especially with DDR5. First i'd make sure you actually need 64 GB total for your workloads, this would only be required for professional workloads. For daily use including gaming, you'd have more than enough RAM if you only use 2x 16 GB in slots A2 and B2, and it should work at XMP no problem. If you don't have any software that benefits from 64 GB, then this amount only makes things worse, not better. With four modules on a board with daisy-chained RAM slots, you worsen the electrical parameters, and of course you double the stress on the memory system.

This RAM configuration might've been just about ok on a "true" 13th/14th gen CPU (the -K models), but it might cause problems for a 12th gen CPU. The best thing to do then is to enable XMP, but also setting "DRAM Frequency" to something lower manually. You could try DDR5-5600 there, this should make it work. Or, if you realize that 32 GB total are enough, you would just use two modules and that should work at -6000, if not even -6400.
Hello to you, I have come to give you feedback and have activated memory try it to select ddr5 5600 cl36 by activating xmp it calibrates to 6000 on profile 1 so I only activated the memory try it option which seems to be the option to select the operating mode to adapt the ddr, does this seem correct to you?

I also activated game boost and I wonder if this is normal or not because my cpu went from 3.20 ghz to 5 ghz should I deactivate this game boost option or is there a risk of damaging the cpu. Thank you in advance for your answers
 

Attachments

  • IMG_0353.jpg
    IMG_0353.jpg
    1 MB · Views: 28
  • IMG_0354.jpg
    IMG_0354.jpg
    1,006.9 KB · Views: 30
Hello to you, I have come to give you feedback and have activated memory try it to select ddr5 5600 cl36 by activating xmp it calibrates to 6000 on profile 1 so I only activated the memory try it option which seems to be the option to select the operating mode to adapt the ddr, does this seem correct to you?

There are two good methods, either should work:
1) XMP enabled, but setting "DRAM Frequency" to DDR5-5600 by hand too, only then pressing F10 to save and exit. It should use XMP timings, but at DDR5-5600.
2) XMP disabled, and "Memory Try It!" with the speed/timings you want to try.

I also activated game boost and I wonder if this is normal or not because my cpu went from 3.20 ghz to 5 ghz should I deactivate this game boost option or is there a risk of damaging the cpu. Thank you in advance for your answers

GameBoost is one of the worst functions in the entire BIOS on Intel-based boards. That's because it is a completely generalized, cookie-cutter auto-OC mechanism that tries to overclock any CPU using brute force, no matter how well-suited it is for it in terms of added voltage etc., or how it behaves individually (each CPU is different). This function always does more harm than good to the way the CPU is running.

Do not worry about the speed shown in the BIOS, the 3.2 GHz shown in the BIOS are only the CPU's base frequency. However, this base frequency is irrelevant, as it will clock much higher under load in Windows, and much lower when there is no load (like 0.8 GHz, for power saving). So pay no mind to the CPU speed shown in the BIOS on the top left, you will hardly ever see that speed in Windows.

To properly optimize the way the CPU is running, you should go completely in the other direction. By undervolting a bit to what your CPU actually requires for full stability, you can even gain some performance, completely without overclocking. That's because the high-end CPU (and GPU) models are all pushed too far from factory. Intel, AMD and NVIDIA, they only want to win benchmarks in the launch reviews, they don't care about power draw or efficiency. But if you reduce the voltage a bit to what your specific CPU actually needs, then it has a lower power draw, and all of a sudden, it can use that to boost higher and increase their perforance within the available power budget (or temperature budget). Not to mention it tends to run cooler and better than ever before.

I explain how that works in my Guide: How to set good power limits in the BIOS and reduce the CPU power draw.
 
There are two good methods, either should work:
1) XMP enabled, but setting "DRAM Frequency" to DDR5-5600 by hand too, only then pressing F10 to save and exit. It should use XMP timings, but at DDR5-5600.
2) XMP disabled, and "Memory Try It!" with the speed/timings you want to try.



GameBoost is one of the worst functions in the entire BIOS on Intel-based boards. That's because it is a completely generalized, cookie-cutter auto-OC mechanism that tries to overclock any CPU using brute force, no matter how well-suited it is for it in terms of added voltage etc., or how it behaves individually (each CPU is different). This function always does more harm than good to the way the CPU is running.

Do not worry about the speed shown in the BIOS, the 3.2 GHz shown in the BIOS are only the CPU's base frequency. However, this base frequency is irrelevant, as it will clock much higher under load in Windows, and much lower when there is no load (like 0.8 GHz, for power saving). So pay no mind to the CPU speed shown in the BIOS on the top left, you will hardly ever see that speed in Windows.

To properly optimize the way the CPU is running, you should go completely in the other direction. By undervolting a bit to what your CPU actually requires for full stability, you can even gain some performance, completely without overclocking. That's because the high-end CPU (and GPU) models are all pushed too far from factory. Intel, AMD and NVIDIA, they only want to win benchmarks in the launch reviews, they don't care about power draw or efficiency. But if you reduce the voltage a bit to what your specific CPU actually needs, then it has a lower power draw, and all of a sudden, it can use that to boost higher and increase their perforance within the available power budget (or temperature budget). Not to mention it tends to run cooler and better than ever before.

I explain how that works in my Guide: How to set good power limits in the BIOS and reduce the CPU power draw.
ok so if my cpu from 3.20 went to 5.00 ghz that's no problem but the gameboost option is moderately suitable if I understand how to configure its processor.

Icône de validation par la communauté
 
GameBoost is really not good on Intel. If you go by my thread that i linked, you will realize how much better that method is.
BTW, if you let the CPU boost normally, it will go up to 5.2 GHz under appropriate load. Don't fixate on the meaningless number shown in the BIOS.
 
ok I am ready to follow you on the adventure. apart from that, when I have a little means I will change the ddr5 according to your advice for 2x32.
I will come back to you another time and thank you for everything also I wish you a good weekend and good continuation in your commitments

Icône de validation par la communauté

GameBoost is really not good on Intel. If you go by my thread that i linked, you will realize how much better that method is.
BTW, if you let the CPU boost normally, it will go up to 5.2 GHz under appropriate load. Don't fixate on the meaningless number shown in the BIOS.
 
Last edited:
GameBoost is really not good on Intel. If you go by my thread that i linked, you will realize how much better that method is.
BTW, if you let the CPU boost normally, it will go up to 5.2 GHz under appropriate load. Don't fixate on the meaningless number shown in the BIOS.

Hello again, I had a blue screen and I wanted to lower the RAM to 5400 but there are several CL38 CL36 profiles. my ram as a reminder

CORSAIR VENGEANCE RGB DDR5 RAM (4x16Go) 6400MHz CL32

the blue screen tells me a bugcheck error 1001 so I suppose that my ram is not stable at 5600 but how can I select a DDR5 profile without CL32
Do I have to remove two bars to make them work in 6000 or should I lower the frequency to 5400 for the moment I have configured it on the 5400 CL36 profile against the old 5600 CL36, is there a risk for the machine, thank you in advance for your help
 

Attachments

  • bugchjeck.png
    bugchjeck.png
    52.6 KB · Views: 19
  • IMG_0368.jpg
    IMG_0368.jpg
    756 KB · Views: 19
Last edited:
The CL numbers refer to the timings, the higher the numbers, the looser they are (good for stability), the lower the numbers, the tighter they are (more demanding). But i think the DDR speed will have the biggest effect. So, if you used DDR5-5600 CL36 before, you could try DDR5-5600 CL38. While it's listed below CL36 in the BIOS, it's actually slightly looser. But this would only make it fully stable if you already only had a very slight instability. So this alone might not be enough. So then you could also lower the speed, DDR5-5400 like you have now.

I recommend checking the RAM with a stress test like TestMem5 which i link in my RAM thread under 5). Don't wait for a bluescreen, this stress test will tell you much quicker if it's stable or not.
 
The CL numbers refer to the timings, the higher the numbers, the looser they are (good for stability), the lower the numbers, the tighter they are (more demanding). But i think the DDR speed will have the biggest effect. So, if you used DDR5-5600 CL36 before, you could try DDR5-5600 CL38. While it's listed below CL36 in the BIOS, it's actually slightly looser. But this would only make it fully stable if you already only had a very slight instability. So this alone might not be enough. So then you could also lower the speed, DDR5-5400 like you have now.

I recommend checking the RAM with a stress test like TestMem5 which i link in my RAM thread under 5). Don't wait for a bluescreen, this stress test will tell you much quicker if it's stable or not.
very good, I will now apply this parameter and come back to you very quickly, hoping to give you a most positive result and thank you once again for your skills and your details, always well explained. Have a good evening until then or have a good day when you read me
 
The CL numbers refer to the timings, the higher the numbers, the looser they are (good for stability), the lower the numbers, the tighter they are (more demanding). But i think the DDR speed will have the biggest effect. So, if you used DDR5-5600 CL36 before, you could try DDR5-5600 CL38. While it's listed below CL36 in the BIOS, it's actually slightly looser. But this would only make it fully stable if you already only had a very slight instability. So this alone might not be enough. So then you could also lower the speed, DDR5-5400 like you have now.

I recommend checking the RAM with a stress test like TestMem5 which i link in my RAM thread under 5). Don't wait for a bluescreen, this stress test will tell you much quicker if it's stable or not.
hello again, I have just installed memtest5 successfully, a lot of people talk about using the Extreme1@anta777 test but other test profiles are also available, so depending on you you should run this test to get a reliable result for how long ? (see screenshot)
 

Attachments

  • dfdfshhh.png
    dfdfshhh.png
    27.2 KB · Views: 17
You can use the profile you mentioned, it's good, but personally i always use the 1usmus_v3 one. I found it fast and reliable to detect RAM instability. You can try both and see if one can detect something the other can't.

You run the test until it finishes. It depends on your RAM capacity and your RAM speed how long it takes, and of course on the profile you use. The 1usmus_v3 should take anything from 15, 20 minutes up to 40, 50 minutes with a lot of RAM, i would guess.
 
You can use the profile you mentioned, it's good, but personally i always use the 1usmus_v3 one. I found it fast and reliable to detect RAM instability. You can try both and see if one can detect something the other can't.

You run the test until it finishes. It depends on your RAM capacity and your RAM speed how long it takes, and of course on the profile you use. The 1usmus_v3 should take anything from 15, 20 minutes up to 40, 50 minutes with a lot of RAM, i would guess.
here is the result no errors found
 
Good, that's what you want. If there are any further problems, then they should not come from the RAM.
yes I used it as1usmus_v3 one you suggested thank you again also I have a small question for my personal culture, would you know why there are several identical profiles on the selection of parameters 5800 CL38 5800 CL38 5800 CL35 5800 CL35 5800 CL35 Are two 5800 CL35s for example the same parameters, it's very confusing. Thank you again for your support, I hope it will help others who have the same problem.
 
Are two 5800 CL35s for example the same parameters, it's very confusing.

Probably using slightly different subtimings, but since it only shows the CL timing in the list (CAS Latency), it appears as though it was identical. Most likely, when you compared two such seemingly identical profiles under "Advanced DRAM Configuration" later (where all the timings are visible), you would find some differences. But this is something for more advanced users. If you found a good profile for your four modules that has decent performance and is 100% stable, that's the main thing.

Thank you again for your support, I hope it will help others who have the same problem.

Always happy to help.
 
Back
Top