Corsair VENGEANCE 2 x 32GB DDR4 - OC HELP

[noobzin]

The gear down option is grayed out ?

 

[tabascosauz]

Yea seriously , kind of thinking should have gone with an asus

The Strix is an alright board, for the right price. Would be my pick for a daily

I think MSI has a different location in the BIOS for Geardown. I can't check at the moment though - are you sure its GDM greyed out and not commandrate greyed out? It's the right design, if GDM is on then 1T/2T is irrelevant

Not impossible and you only have to do it once... if he had one lol.

I figure the design choice is because they figure it's a feature most people shouldn't need to use. Just my take on it. Honestly with modern hardware, you don't really need it.

3 failed attempts, on the 4th the board should post defaults. It just seems to take forever.....

The MSI board needs the I/O shield to come off to access pins/route a button, but can't do that if it's in a case, 2 of the screws are on the bottom of the board

On paper, auto recovery should work, but as usual with AMD firmware things don't quite work that way...

I've had all of the B550 ITX boards except ASRock's higher end option. All of them except the ASRock (reckon if i pushed it enough it would too lol) have frozen mid-try and not even tried to do a second let alone third attempt. The Aorus gives up approximately 30% of the time (and annoyingly requires draining the battery to completely clear CMOS) , and the Strix the worst offender upwards of 50% of the time. The Strix's unique twist is that it goes straight to BIOS recovery and demands a reflash...but clear CMOS fixes it

Funny enough it's the Impact that actually works better for some reason, but it's the one where auto recovery is irrelevant lol (POST code + full OC buttons)

 

[freeagent]

I just use the reset switch to clear the cmos. If I need to reset the machine I just flip the PSU switch.

Very helpful when dealing with memory

 

[Deleted member 185158]

The Strix is an alright board, for the right price. Would be my pick for a daily

I think MSI has a different location in the BIOS for Geardown. I can't check at the moment though - are you sure its GDM greyed out and not commandrate greyed out? It's the right design, if GDM is on then 1T/2T is irrelevant



The MSI board needs the I/O shield to come off to access pins/route a button, but can't do that if it's in a case, 2 of the screws are on the bottom of the board

On paper, auto recovery should work, but as usual with AMD firmware things don't quite work that way...

I've had all of the B550 ITX boards except ASRock's higher end option. All of them except the ASRock (reckon if i pushed it enough it would too lol) have frozen mid-try and not even tried to do a second let alone third attempt. The Aorus gives up approximately 30% of the time (and annoyingly requires draining the battery to completely clear CMOS) , and the Strix the worst offender upwards of 50% of the time. The Strix's unique twist is that it goes straight to BIOS recovery and demands a reflash...but clear CMOS fixes it

Funny enough it's the Impact that actually works better for some reason, but it's the one where auto recovery is irrelevant lol (POST code + full OC buttons)

Man that'd strange to hear those strix boards don't recover well. I've never had a problem, but have B400 series Strix boards and B660 ddr5 boards. They all seem to work ok on post recovery. But I also try and take advantage of the extensive memory training offered in these boards as well.

Cpu overclocks aside from memory overclocks, yeah they don't exactly recover well from that. Not many boards do imo.

 

[Mussels]

HI guys,
looking for help to OC frequency and lower some latency. Corsair VENGEANCE 2 x 32GB DDR4 3200 B die 16 GB Micron. anything after 3400 it wont boot. Right now im running xmp profile 16-20-38 1.35 v @3400 mhz. would like to go to 3600-3800 and lower latency which right now is 63.



Ryzen 5 5600x 4.65ghz 1.25v
MSI B550I GAMING EDGE
Corsair VENGEANCE 2 x 32GB DDR4 @3400 mhz CMH64GX4M2E3200C16

sasdsad

Image sasdsad hosted in Freeimage.host

freeimage.host

SoC voltage to 1.15V, DDR voltage to 1.4V
use slots 2 and 4

Hynix memory will let you raise clocks, but wont let you lower timings
samsung memory will let you raise clocks and alter timings, but needs to be kept under 40c

Reading other posts i'd say enable XMP and use all your stock timings and try 3600, if that works with the voltages i suggested then and only then, try tweaking TRFC - mine defaulted to over 1000, but 704 is stable and cut a lot off aida64's latency results

 

[noobzin]

1st successful boot

I lost track now , what do I do ? Looks like I’m hitting 64.1 latency

 

[tabascosauz]

Man that'd strange to hear those strix boards don't recover well. I've never had a problem, but have B400 series Strix boards and B660 ddr5 boards. They all seem to work ok on post recovery. But I also try and take advantage of the extensive memory training offered in these boards as well.

Cpu overclocks aside from memory overclocks, yeah they don't exactly recover well from that. Not many boards do imo.

To be sure, you're completely right. Nothing beats a physical cmos button. Board acting up? BAM and off to the races again. Even if it's a power button repurposed as one (I just hang it out the back of the case if Ill be working on mem for a few days)

That MSI board is the most unique in the world for all the wrong reasons (it also has a nonstandard cooler backplate). Put a standard backplate on and it'll start breaking the flat caps on the back

1st successful boot

VDDG and VDDP fixed, good stuff. CCD doesnt always have to be as high, but set as you see fit.

VSOC too low for what VDDG is. Keep VSOC 0.05 above the VDDGs. Since you're still figuring things out, stay at 1.1V or above for VSOC - if you don't need all of it shave it off later when things stable.

methinks it's just a bad bin of Rev.B, might just need to bump VDIMM a bit from what you expect. 1.35V normal for 3600CL16 but that's from the perspective of a decent average bin. Happens all the time with all Corsair 3200CL16, give em the best ICs (B-die, Rev.E, Rev.B) and Corsair finds the worst samples possible to put in the kit.

tRFC you're already on the tight end for Rev.B. Higher latency can come from instability, you haven't memtested this profile yet.

@The King works on Rev.B a lot.

 

[Deleted member 185158]

To be sure, you're completely right. Nothing beats a physical cmos button. Board acting up? BAM and off to the races again. Even if it's a power button repurposed as one (I just hang it out the back of the case if Ill be working on mem for a few days)

That MSI board is the most unique in the world for all the wrong reasons (it also has a nonstandard cooler backplate). Put a standard backplate on and it'll start breaking the flat caps on the back

Working s939 right now. One of the worst things to have to go through is a cmos reset when running high bus clocks. Gotta work back up to the OC most of the time. Benching, not a daily setup though. DFI LanParty SLI-DR. And I've had to reset it several times in the past few days.

MSI has come quite a long way. Wouldn't even think about using one of their boards for the above adventure.

 

[noobzin]

To be sure, you're completely right. Nothing beats a physical cmos button. Board acting up? BAM and off to the races again. Even if it's a power button repurposed as one (I just hang it out the back of the case if Ill be working on mem for a few days)

That MSI board is the most unique in the world for all the wrong reasons (it also has a nonstandard cooler backplate). Put a standard backplate on and it'll start breaking the flat caps on the back



VDDG and VDDP fixed, good stuff. CCD doesnt always have to be as high, but set as you see fit.

VSOC too low for what VDDG is. Keep VSOC 0.05 above the VDDGs. Since you're still figuring things out, stay at 1.1V or above for VSOC - if you don't need all of it shave it off later when things stable.

methinks it's just a bad bin of Rev.B, might just need to bump VDIMM a bit from what you expect. 1.35V normal for 3600CL16 but that's from the perspective of a decent average bin. Happens all the time with all Corsair 3200CL16, give em the best ICs (B-die, Rev.E, Rev.B) and Corsair finds the worst samples possible to put in the kit.

tRFC you're already on the tight end for Rev.B. Higher latency can come from instability, you haven't memtested this profile yet.

@The King works on Rev.B a lot.

View attachment 259312

Ok done that, running some tests
Would like to see lower latency for sure

 

[freeagent]

Man that'd strange to hear those strix boards don't recover well. I've never had a problem, but have B400 series Strix boards and B660 ddr5 boards.

I got a Strix -F first, and I loved that board! There was a Strix XE on heavy discount locally so I snagged it and gave the -F to my sons build. I like the XE more.. honestly the best boards I have used. My -F doesn't really like the Adatas that I have in it right now. They love my B-Die though.. smooth sailing.

 

[ir_cow]

Congrats! I wasn't a believer....

 

[noobzin]

Thank you all !
I’m done for the day , great improvements !
thanks

 

[Deleted member 185158]

I got a Strix -F first, and I loved that board! There was a Strix XE on heavy discount locally so I snagged it and gave the -F to my sons build. I like the XE more.. honestly the best boards I have used. My -F doesn't really like the Adatas that I have in it right now. They love my B-Die though.. smooth sailing.

My B450-I ROG board clocks memory like an Apex. I'll just leave that alone there...I.

And that's something very noticeable for ddr5 lga 1700 boards. 2 or 4 dimm slots. My B660-I has absolutely no issues posting 6400mhz. The B660-G shits at 6000mhz though.

Thank you all !
I’m done for the day , great improvements !
thanks

I see them primaries went from 16-20-20 to 18-22-22. That 2T command rate can sometimes do wonders for stability with minimal penalty.

Good work.

 

[noobzin]

Dimm hitting 48c while gaming
Normal ? Crazy?

 

[tabascosauz]

Dimm hitting 48c while gaming
Normal ? Crazy?

What GPU?

Not even close to crazy, pretty normal for any 200W+ GPU in a regular tower case with 1.4V+ VDIMM on ram. If you don't have crazy abundant airflow, only use a normal tower cooler / watercooling, without a RAM cooler or any other fan pointed directly at the RAM.

Generally it's only Samsung 8Gb B-die that cares a lot about the temp, because once it hits 40-50C (depending on how tight you run/high VDIMM) it'll just destabilize by itself. Micron Rev.B is also kinda temp sensitive but should not be to that degree.

If you find you're failing memtests past a certain temp, crank your case airflow up, or stand up a small 80/92mm fan on the back of the video card pointing at the RAM.

 

[noobzin]

6900 xt gpu by powercolor
Ok thank you. It was pretty good airflow
3 x120 noctua fans on front intake
One whatever bottom of gpu
1 noctua 140 rear exhaust
And 2x120 aio h100i on top exhaust
Temos are 34-36 idle
58-60 full

 

[tabascosauz]

6900 xt gpu by powercolor
Ok thank you. It was pretty good airflow
3 x120 noctua fans on front intake
One whatever bottom of gpu
1 noctua 140 rear exhaust
And 2x120 aio h100i on top exhaust
Temos are 34-36 idle
58-60 full

60 is a little too warm - Rev.B is not THAT temp sensitive, but doesn't mean it isn't entirely. Point a spare fan at it. But I can see why, dual rank on a 2DIMM board, no space in between.

You'd be surprised what "good" airflow looks like for higher VDIMM ram. Just having a lot of intake fans in the front doesn't count for shit, not unless you run them at 1500rpm+ all day long. For my B-die at 1.55V a small 92mm ram fan meant the difference between 38C and 50C. I run a fierce push-pull C14S with two 140mm industrial noctuas, but I can now only do about 43C on the sticks under load no thanks to the 3070 Ti (similar power draw to 6900XT, albeit I lopped off about 50W).

The flow-through cutout on the back of these GPUs nowadays is great for the GPUs themselves, but an absolute disaster for the ram.

At one point I tried replacing my C14S with a 120mm tower cooler, even trying a Corsair ram cooler to compensate. 1.5V just wasn't gonna work. You're not quite that high on VDIMM yet, but imo it's not really voltage that makes the biggest difference - air 6900XT, dual rank, sticks crammed together in a 2DIMM board, AIO/custom loop contributing no airflow, and not enough case airflow are the big culprits.

 

[The King]

Sorry googled the wrong motherboard.

Dropping your subtimings will improve performance.

If your running tFAW 16 RRDS should be 4.

RRDS 4
RRDL 6
WTRS 4
WTRL 12
tWR 16
tRTP 8

Can try these settings and see if there is any improvement.

 

[noobzin]

60 is a little too warm - Rev.B is not THAT temp sensitive, but doesn't mean it isn't entirely. Point a spare fan at it. But I can see why, dual rank on a 2DIMM board, no space in between.

You'd be surprised what "good" airflow looks like for higher VDIMM ram. Just having a lot of intake fans in the front doesn't count for shit, not unless you run them at 1500rpm+ all day long. For my B-die at 1.55V a small 92mm ram fan meant the difference between 38C and 50C. I run a fierce push-pull C14S with two 140mm industrial noctuas, but I can now only do about 43C on the sticks under load no thanks to the 3070 Ti (similar power draw to 6900XT, albeit I lopped off about 50W).

The flow-through cutout on the back of these GPUs nowadays is great for the GPUs themselves, but an absolute disaster for the ram.

At one point I tried replacing my C14S with a 120mm tower cooler, even trying a Corsair ram cooler to compensate. 1.5V just wasn't gonna work. You're not quite that high on VDIMM yet, but imo it's not really voltage that makes the biggest difference - air 6900XT, dual rank, sticks crammed together in a 2DIMM board, AIO/custom loop contributing no airflow, and not enough case airflow are the big culprits.

I live in fresno ca


Also space wise this is where the pc end up . Below the desk but I did move away from the wall bit and put two wood blocks in the bottom and it has leg stands ^^. I meant cpu temos 34-36 idle full load 58-60. DIMM after the oc work yesterday was 40-45 idle , 45-49 c gaming

 

[dgianstefani]

Don't go above 1.4v.

 

[tabascosauz]

I live in fresno ca

You have A/C? Judging from the 5600X idle temps your ambient is similar to my upstairs 5700G PC in a Lone L5 case. Gets real warm up there, 30c+ even with central AC.

If you want to get a different kit, something like a 2x8GB 4000CL19 kit from Patriot or Crucial's 4000CL18 is probably a sidegrade at best (if you're lucky and get like 3800-4000CL14 out of it), or an actual downgrade. I have both kits and they're not what I'd describe as amazing OCers. You're throwing away your dual rank advantage (which counts for up to a few hundred MT/s worth). Since you're already at 3600, may as well just keep working on what you have.

Your load temps are also going to be a problem for any B-die kit.

 

[noobzin]

You have A/C? Judging from the 5600X idle temps your ambient is similar to my upstairs 5700G PC in a Lone L5 case. Gets real warm up there, 30c+ even with central AC.

If you want to get a different kit, something like a 2x8GB 4000CL19 kit from Patriot or Crucial's 4000CL18 is probably a sidegrade at best (if you're lucky and get like 3800-4000CL14 out of it), or an actual downgrade. I have both kits and they're not what I'd describe as amazing OCers. You're throwing away your dual rank advantage (which counts for up to a few hundred MT/s worth). Since you're already at 3600, may as well just keep working on what you have.

gotcha. yes ac runs all the time between 75-79 F. The 3600 settings passed all the mem tests but it was doing weird things while gaming , battlefield2042 and the temp was a bit high so i went back xmp default @ 3400 and stop messing for a bit

I tan the Aida benchmark with the ram setting like we worked on yesterday and cpu oc to 3.65 and during the test the pc restarted by itself so idk

 

[tabascosauz]

gotcha. yes ac runs all the time between 75-79 F. The 3600 settings passed all the mem tests but it was doing weird things while gaming , battlefield2042 and the temp was a bit high so i went back xmp default @ 3400 and stop messing for a bit

I tan the Aida benchmark with the ram setting like we worked on yesterday and cpu oc to 3.65 and during the test the pc restarted by itself so idk

Let's just simplify things, if you are going to memtest properly, use TM5.

Memory Testing with TestMem5 TM5 with custom configs

Hello everybody I am just making a very light tutorial with a collection of custom config files and a DOWNLOAD LINK for TM5 v0.12 anta777 absolut config Official Intel DDR4 24/7 Memory Stability Thread None of the work is mine but it seems like a pretty good and fast testing app Update...

www.overclock.net

Running two different untested OCs (CPU and ram) at the same time and trying to figure out what went wrong is a pointless exercise.

 

[noobzin]

update.
ran a few tests @ 3600 again today, everything went fine it seems.

 

[tabascosauz]

update.
ran a few tests @ 3600 again today, everything went fine it seems.

Your CLDOs are all out of wack again. Looking decent though.

As a side note, for TM5 we don't use serj config, it's the default and it's not very good. I was under the impression that the overclock.net zip already included the better config (anta777 extreme) as default, but it seems like it doesn't on yours? Try deleting cfg.link in the /bin folder to see if the config changes.

Anyways, bunch of the .cfg files should be in the zip. anta777 extreme1 and 1usmusv3 are good. Just delete existing MT.cfg and cfg.link, and rename the desired config to MT.cfg.