Can the CL / Latency affect compatability?

[d3ton4tion]

I am building a new PC and I opted for an Asus B650 Proart Creator + Ryzen 7 7800X3D + 2x16 G.Skill Flare X5 DDDR5-6000 as main components. The board is however not recognizing the memory sticks. The Asus B650 lists the Flare X5s on the QVL list, however with a different latency value (The list has CL36, I accidentally bought CL32). Do latency values affect the compatability between RAM, board and CPU that much ? And does this mean I have to order the CL36 version now? Or is there a workaround? 

[Kisai]

4 minutes ago, d3ton4tion said:

I am building a new PC and I opted for an Asus B650 Proart Creator + Ryzen 7 7800X3D + 2x16 G.Skill Flare X5 DDDR5-6000 as main components. The board is however not recognizing the memory sticks. The Asus B650 lists the Flare X5s on the QVL list, however with a different latency value (The list has CL36, I accidentally bought CL32). Do latency values affect the compatability between RAM, board and CPU that much ? And does this mean I have to order the CL36 version now? Or is there a workaround? 

Not generally unless you're mixing pairs.

 

It is possible to buy RAM that explicitly doesn't work with a chipset because it's too much of a mismatch.

[Shimejii]

Each kit is different, you got one thats not validated. It still has a good chance to work, just enable XMP/EXPO and see if it works. Your kit should be fine as its within tolerance. IF you have any instability with the ram kit then id suggest getting a 6000mhz CL 30 kit on the QVL list for the mobo.

[Tetras]

7 minutes ago, d3ton4tion said:

The board is however not recognizing the memory sticks.

What do you mean by "not recognising"?

[RONOTHAN##]

2 minutes ago, d3ton4tion said:

Do latency values affect the compatability between RAM, board and CPU that much ?

Kinda. There are 4 main things that affect memory compatibility: the PCB, the memory IC, and the frequency it's rated at. The PCB is usually the same for any kit of similar rating, so that should be the same between those. The frequency is going to be the same, so that just leaves the IC. Motherboards need to have specific support for each memory IC in order to work properly, and the CL30/32 kits have a different I than the CL36 kits on average, with the 36 kits being Samsung based and the 30/32 kits being some variant of Hynix. 

 

However, because Hynix memory is currently some of the best performing stuff you can buy, it's the first thing motherboard manufacturers optimize for. Therefore the CL32 kit should actually be slightly more compatible than the CL36 kit. 

 

 

The reason your kit wasn't on there was because the QVL isn't exhaustive, so they just didn't test it. It doesn't mean that it doesn't work, more likely it wasn't on hand to test. 

[Somerandomtechyboi]

No

 

Either the board is being pissy or the rams are faulty and id usually lean towards the former as boards are the most failiure/error prone component in a pc right next to rams (particularly xmp or mismatch ic)

 

Thats either hynix m or a die and no sane board manufacturer would not support those ics since theyre among the best ddr5 ics (atleast hynix a die is, mdie is stuck to ~7600 max)

 

6000 shouldnt cause any issues with the cpu imc either, CL is a universal timing so not that either, could be subtimings but again no sane ram manufacturer would make a ram kit thatd only work on intel or vice versa

 

Its also an expo kit so even if the board has an older bios which tend to have issues with xmp that shouldnt be the issue either

 

So yeah i blame board unless the rams are proven faulty

[podkall]

4 minutes ago, Somerandomtechyboi said:

No

 

Either the board is being pissy or the rams are faulty and id usually lean towards the former as boards are the most failiure/error prone component in a pc right next to rams (particularly xmp or mismatch ic)

 

Thats either hynix m or a die and no sane board manufacturer would not support those ics since theyre among the best ddr5 ics (atleast hynix a die is, mdie is stuck to ~7600 max)

 

6000 shouldnt cause any issues with the cpu imc either, CL is a universal timing so not that either, could be subtimings but again no sane ram manufacturer would make a ram kit thatd only work on intel or vice versa

 

Its also an expo kit so even if the board has an older bios which tend to have issues with xmp that shouldnt be the issue either

 

So yeah i blame board unless the rams are proven faulty

don't forget blaming poor RAM inserts, ram training or whatever DDR5 does, etc. - before you blame the RAM

[d3ton4tion]

5 minutes ago, Somerandomtechyboi said:

No

 

Either the board is being pissy or the rams are faulty and id usually lean towards the former as boards are the most failiure/error prone component in a pc right next to rams (particularly xmp or mismatch ic)

 

Thats either hynix m or a die and no sane board manufacturer would not support those ics since theyre among the best ddr5 ics (atleast hynix a die is, mdie is stuck to ~7600 max)

 

6000 shouldnt cause any issues with the cpu imc either, CL is a universal timing so not that either, could be subtimings but again no sane ram manufacturer would make a ram kit thatd only work on intel or vice versa

 

Its also an expo kit so even if the board has an older bios which tend to have issues with xmp that shouldnt be the issue either

 

So yeah i blame board unless the rams are proven faulty

I just reseated the RAM three times and after the third trie it finally accepted it  so yea probably was the board/maybe something in the slots. 

[Somerandomtechyboi]

1 hour ago, d3ton4tion said:

I just reseated the RAM three times and after the third trie it finally accepted it  so yea probably was the board/maybe something in the slots. 

All thats left is to stability test the rams

 

I personally use prime95 largeffts but theres others like tm5 (anta777 profile), hci memtest, etc.

 

1 hour ago, podkall said:

don't forget blaming poor RAM inserts, ram training or whatever DDR5 does, etc. - before you blame the RAM

i mean the gold contacts on the ram can get dirty and cause it to not detect, a wipe down with some 99% IPA + tissue or even a wet tissue will do fine but im lazy and just use my fingers which also works but has a chance of cutting your finger and yea thats happened to me like 2x by now, apparently a fiberglass pcb has some somewhat sharp edges

 

And yep ddr5 ram training completely forgot about that

[podkall]

2 minutes ago, Somerandomtechyboi said:

i mean the gold contacts on the ram can get dirty and cause it to not detect, a wipe down with some 99% IPA + tissue or even a wet tissue will do fine but im lazy and just use my fingers which also works but has a chance of cutting your finger and yea thats happened to me like 2x by now, apparently a fiberglass pcb has some somewhat sharp edges

 

And yep ddr5 ram training completely forgot about that

I mean, RAM can work when not fully inserted, or it can not work, or it can cause issues while working most of the time...

 

[Somerandomtechyboi]

4 minutes ago, podkall said:

I mean, RAM can work when not fully inserted, or it can not work, or it can cause issues while working most of the time...

 

 

i actually have an instance of that happening to me, im lazy and take out only 1 side of the rams just so it doesnt detect but i can easily put the rams back in by pushing down, but for whatever reason this time where id usually expect it to get stuck on whatever postcode that signifies no ram detected (i think it was e0 or something for amibios) it instead proceeded to go past d5 code and got stuck at 13 code, d5 means the rams detect but seems like it stopped at 13 due to only being a half detect

 

Pushed the rams just abit more without fully inserting them and now you get these cursed photos

 

Now that i think of it im geniuinely curious what happens if i were to overclock the rams in this state considering ram is very trace length sensitive

 

Welp if i find some time ill definitely be having fun with this =)

[podkall]

1 minute ago, Somerandomtechyboi said:

 

 

i actually have an instance of that happening to me, im lazy and take out only 1 side of the rams just so it doesnt detect but i can easily put the rams back in by pushing down, but for whatever reason this time where id usually expect it to get stuck on whatever postcode that signifies no ram detected (i think it was e0 or something for amibios) it instead proceeded to go past d5 code and got stuck at 13 code, d5 means the rams detect but seems like it stopped at 13 due to only being a half detect

 

Pushed the rams just abit more without fully inserting them and now you get these cursed photos

 

Now that i think of it im geniuinely curious what happens if i were to overclock the rams in this state considering ram is very trace length sensitive

 

Welp if i find some time ill definitely be having fun with this =)

honestly the photo is not that cursed, the RAM pins can't be seen which means they are very much likely at least 70-80% in contact

[Somerandomtechyboi]

1 minute ago, podkall said:

honestly the photo is not that cursed, the RAM pins can't be seen which means they are very much likely at least 70-80% in contact

No theyre very clearly visible

 

Its that grey looking thing at the botton of the rams that you can see receeds (actually makes decent contact) as you look to the left of the pic

 

maybe i should have taken the photo from a diff angle

 

Still though imagine 3000+ ddr3 with this nonsense