Chloe's RAM OC thread for everyone

[tabascosauz]

Viritually all CPUs and mobos are able to run 3733/1866 and many can do 3800/1900.

I dunno if I'd say that with certainty here. The 4000-4400 QVLs for 4-layers are for single rank sticks. These Kingston sticks are dual rank (some 4-layer mATX QVLs drop all the way to 3733), and the B550M Elite isn't a particularly high end board even amongst 4-layer mATX boards - it shares a PCB with the DS3H, not the Pro or Pro-P.

But you are right in that there's only one way to find out

I'll have a nap soon and I'll let MT5 run for a while.... if this 3466 is stable, then it's time for a boot and for 3533..

And hell no I don't loose timings, no no and no.

How'd TM5 turn out?

It's not a matter of loose timings, CJR just doesn't scale tRCD and tRP, period. You can make the same early mistakes as I did and try fruitlessly to make 18-18 stable at 3600+, but it's the integrity of your own drive data and Windows on the line. There's a good reason why you see CJR XMP loosen from 16-18-18 @ 3200, to 16-19-19 @ 3600 and eventually to x-20-20 or x-22-22 @ 4000.

 

[Mr.Scott]

Remember back in the DDR1 days when some motherboards offered the enigmatic CAS 1.5 (and even 1!!!) options? Did anyone ever get those to work? Winbond BH-5 memories...
*shakes head in drunken stupor*

CAS1 never existed. 1.5 however...........

 

[Keullo-e]

How'd TM5 turn out?

Seems fine. Weird tho that now it takes for like a minute to boot yet still everything seems to be fully stable.

 

[Splinterdog]

I have four 8GB Corsair Vengeance LPX 2400 sticks and have never bothered to overclock them.
Would it be worth it and would I notice any performance difference?

 

[lexluthermiester]

Would it be worth it and would I notice any performance difference?

Not really. OCing RAM is more of a hobbyist effort. However, there's no harm in pushing RAM to it's limits as long as you don't go crazy on voltage.

EDIT:
Due to a complaint, I'd like to be clear this post is just my opinion. Such SHOULD be clearly obvious but...

 

[PaulieG]

Let's just say that if you're lucky to hit higher speeds without much trouble, you'll discover real fast that the only reason you can do that is because of Geardown

Helpful starting point would be to know what RAM you *really* have in there, kinda informs pretty much everything you'll be doing: http://www.softnology.biz/files.html

Good starter guide: https://github.com/integralfx/MemTestHelper/blob/master/DDR4 OC Guide.md#integrated-circuits-ics

Best to test with HCI Memtest, you'll get tired of booting to a USB stick real quick.

I prefer to test with the OCCT memory test for an hour or so for quick stability testing if I'm looking to stay in Windows.

 

[80251]

There are several kinds of ram. one is the animal and 2 is running into something and 3 is the ram you find in your computer or laptop.

Dodge makes Ram too! In 2500 and 1500 flavors!

I have four 8GB Corsair Vengeance LPX 2400 sticks and have never bothered to overclock them.
Would it be worth it and would I notice any performance difference?
View attachment 201039

Lexluthermeister is probably right because you wouldn't be able to push those sticks very far, especially in a 4x8GiB configuration. If you overclocked and tuned your memory you might see some small increase in minimum FPS in some games.

Here's a video showing some diff. in gaming perf. going from 2666Mhz to 3200Mhz, but not much if anything when going for 4133Mhz. to 4400Mhz.

 

[RJARRRPCGP]

I speak always in DDR

The best I got before Memtest 86 failed with at least 2 errors-per-mhz, was 390(195mhz) on an Asus A7N8X-E Deluxe in 2004, albeit it was PC2700 DDR SDRAM. (2.5-3-3-7) (512 MB)

Then for some reason, I browsed and it looked like another person reported a 195 Mhz wall, possibly even for PC3200 DDR1. Is that a common Asus A7N8X-family issue? IIRC, mine was a late-revision A7N8X.

I never tried to get into Windows at higher than 195 Mhz FSB.

 

[Keullo-e]

The best I got before Memtest 86 failed with at least 2 errors-per-mhz, was 390(195mhz) on an Asus A7N8X-E Deluxe in 2004, albeit it was PC2700 DDR SDRAM. (2.5-3-3-7) (512 MB)

Then for some reason, I browsed and it looked like another person reported a 195 Mhz wall, possibly even for PC3200 DDR1. Is that a common Asus A7N8X-family issue? IIRC, mine was a late-revision A7N8X.

I never tried to get into Windows at higher than 195 Mhz FSB.

Can't say, IIRC I've owned only one A7N8X(-X?) and by that time, I wasn't into retro overclocking that much as I didn't have any spare place to work with old stuff. These days I just put a testbed on my TV desk and use my TV as monitor when overclocking and benching.

My best DDR modules were those BH-5 HyperX ones I mentioned. They probably had even more potential, but that 3.3V was maximum I could get (and that was by soldering a wire from 3.3V line straight to the memory), the board was MSI K8N Neo2 Platinum (S939, NForce3 Ultra).

 

[RJARRRPCGP]

Seems fine. Weird tho that now it takes for like a minute to boot yet still everything seems to be fully stable.

Did you check the log for "corrected" WHEAs? OTOH, mine feels fast at XMP for Vengeance LPX DDR4 3000. (and 1500 Mhz Fclk and Uclk)

 

[Keullo-e]

Did you check the log for "corrected" WHEAs?

Works fine now and boots as fast as normally, funno what was causing that. And event viewer is something that I don't have a clue what to look there from.

 

[RJARRRPCGP]

And event viewer is something that I don't have a clue what to look there from.

Look for events related to "WHEA-Error-Logger" or similar.

 

[Keullo-e]

Look for events related to "WHEA-Error-Logger" or similar.

Nothing there since since the 3rd of May.


e: Seems stable

 

[Mr.Scott]

The best I got before Memtest 86 failed with at least 2 errors-per-mhz, was 390(195mhz) on an Asus A7N8X-E Deluxe in 2004, albeit it was PC2700 DDR SDRAM. (2.5-3-3-7) (512 MB)

Then for some reason, I browsed and it looked like another person reported a 195 Mhz wall, possibly even for PC3200 DDR1. Is that a common Asus A7N8X-family issue? IIRC, mine was a late-revision A7N8X.

I never tried to get into Windows at higher than 195 Mhz FSB.

Bios related.
There are modded bios's out there right now that make that board run like a NF7. Very impressive.

 

[delshay]

CAS1 never existed. 1.5 however...........

View attachment 200719

You have CL1.5..but is it "truly active".

AFAIK there is only one motherboard where it is truly working. You know it is working, because there is a tank slapper where you will not be able to clock that high. The motherboard in question is a "ASUS" motherboard.

Anyone that takes overclocking memory deadly seriously regardless if it's DDR1, DDR2, all the way to DDR4 & upcoming DDR5 should have AIDA64 Engineering software. This will show what is active at the memory controller.

CPUZ may report CL1.5, but is it truly active. But as I said, you will know it is working, as there is a huge massive cut in clock speed.

 

[Keullo-e]

You have CL1.5..but is it "truly active".

AFAIK there is only one motherboard where it is truly working. You know it is working, because there is a tank slapper where you will not be able to clock that high. The motherboard in question is a "ASUS" motherboard.

Anyone that takes overclocking memory deadly seriously regardless if it's DDR1, DDR2, all the way to DDR4 & upcoming DDR5 should have AIDA64 Engineering software. This will show what is active at the memory controller.

CPUZ may report CL1.5, but is it truly active. But as I said, you will know it is working, as there is a huge massive cut in clock speed.

IIRC DFI nF4 Lanparty boards worked with real CL1.5 if the RAM just was capable of it.

 

[freeagent]

You have CL1.5..but is it "truly active".

I can’t speak for him, but I can say he knows his hardware and would be the one guy I wouldn’t doubt..

 

[delshay]

IIRC DFI nF4 Lanparty boards worked with real CL1.5 if the RAM just was capable of it.

DFI Lanparty motherboard at first had fake CL1.5, ie it was never really working. There was an update to the BIOS to fix it, if I remember correctly, but I don't know if it was truly fixed.

To tell if CL1.5 is working, first push as far as you can at CL2, then switch to CL1.5. It should fail to boot at the same frequency. If it boots at the same frequency then CL1.5 is not working.

Here when switching from CL2 to CL1.5 I get hugh massive errors. Reducing the clock speed is the only way to get rid of all the errors. This is how I know it is working, but I also have Professional Hardware Diagnostic Card to confirm it is working.

EDIT: As already pointed out, I have only found it working on one ASUS motherboard. You be lucky if you can even pass 208Mhz when using CL1.5. Trust me it's a true tank slapper, massive cut in overclocking.

 

[Mr.Scott]

DFI Lanparty motherboard at first had fake CL1.5, ie it was never really working. There was an update to the BIOS to fix it, if I remember correctly, but I don't know if it was truly fixed.

To tell if CL1.5 is working, first push as far as you can at CL2, then switch to CL1.5. It should fail to boot at the same frequency. If it boots at the same frequency then CL1.5 is not working.

Here when switching from CL2 to CL1.5 I get hugh massive errors. Reducing the clock speed is the only way to get rid of all the errors. This is how I know it is working, but I also have Professional Hardware Diagnostic Card to confirm it is working.

EDIT: As already pointed out, I have only found it working on one ASUS motherboard. You be lucky if you can even pass 208Mhz when using CL1.5. Trust me it's a true tank slapper, massive cut in overclocking.

We have had this discussion before at other sites.
DFI LP Ultra D with modded bios. Yes, it will boot at CL2. No it won't boot at CL1.5 at the same frequency, but it's not that much lower. Difference for me is CL2 = 265 MHz and CL1.5 = 253 MHz.
SPI 32M and 1M times are faster at CL1.5 also. Granted, only slightly, but still faster.
On certain platforms, you can also access CL1.5 with software within windows.

I do not wish to have this discussion with you again. It never ends well.

 

[delshay]

We have had this discussion before at other sites.
DFI LP Ultra D with modded bios. Yes, it will boot at CL2. No it won't boot at CL1.5 at the same frequency, but it's not that much lower. Difference for me is CL2 = 265 FSB and CL1.5 = 253 FSB.
SPI 32M and 1M times are faster at CL1.5 also. Granted, only slightly, but still faster.
On certain platforms, you can also access CL1.5 with software within windows.

I do not wish to have this discussion with you again. It never ends well.

A difference of 12MHz between CL2 & CL1.5. ..I'm going to recheck that when I have time, because I think I was getting a significant bigger cut in frequency than what you are posting, granted my timings are tighter within the SPD.


EDIT EXAMPLE: For those user(s) that are interested in AIDA64 "Engineering" here is a screenshot of memory timings. What you enter into the BIOS or the SPD does not mean it is active at the controller. You can see this by comparing what is entered (left side SPD) & what is truly active (right side MC0, MC1).. ect

But even with this software I don't trust it, This is why I have a PHD, ie, Professional Hardware Diagnostic Card.

 

[lexluthermiester]

A difference of 12MHz between CL2 & CL1.5. ..I'm going to recheck that when I have time, because I think I was getting a significant bigger cut in frequency than what you are posting, granted my timings are tighter within the SPD.


EDIT EXAMPLE: For those user(s) that are interested in AIDA64 "Engineering" here is a screenshot of memory timings. What you enter into the BIOS or the SPD does not mean it is active at the controller. You can see this by comparing what is entered (left side SPD) & what is truly active (right side MC0, MC1).. ect

But even with this software I don't trust it, This is why I have a PHD, ie, Professional Hardware Diagnostic Card.

Hmm...

I do not wish to have this discussion with you again. It never ends well.

I'm inclined to agree.

@delshay
Let's let this go. Nothing personal.