intel 13/14th gen fast RAM experiences: >7,000 MT/s is XMP stable!

[Dr. Dro]

I think my board was bad. I bought an open box Apex Encore at microcenter and have some 8000 2x24 patriot kit coming in the mail. This new board ran the current 7200 2x16 memory fine. I could only clock the memory to 7400 with 7200xmp set. We'll see what happens with the new kit.
Board says MC score is 80 but that doesn't mean anything. I'll run aida64 here after a while and see what it says.

I only wish I could get an Apex Encore... they just never showed up here in Brazil. Actually, none of the refresh Z790s made it yet. I don't think any importer even bothered.

I'm looking at a Z690 Aorus Tachyon that showed up here recently... the deal it being Z690 as well, i'm unsure at how much better it'd be against my MSI Ace. If my brother goes ahead with his idea of building a 14700K system, I think i'll give him my motherboard as he doesn't overclock and intends to get a lot of RAM and buy it for myself... if it hasn't sold by then. Think the shop has just one.

 

[ir_cow]

Board says MC score is 80 but that doesn't mean anything. I'll run aida64 here after a while and see what it says.

Yeah MC score doesn't mean much of anything from what I can tell. Mine 71 and I can do 8400 (stupid voltages) with 2 DIMMs. 8600 with 1 and 9000 into windows (just not stable for that).

But don't hype yourself up too much. Silicon lottery is real. 7600-8600 is the range mostly.

The two 14900K I bought last month was worse. User (confidential) bought the better one from me.

I only wish I could get an Apex Encore... they just never showed up here in Brazil. Actually, none of the refresh Z790s made it yet. I don't think any importer even bothered.

I'm looking at a Z690 Aorus Tachyon that showed up here recently... the deal it being Z690 as well.

Mine (Z690 Tachyon) maxed out at 7800. 8000 was bootable, mostly. Just never stable. Stupid voltages too. I liked it a lot, but the Z790 took so long to come I just got a Apex instead.

 

[fusseli]

I think my board was bad. I bought an open box Apex Encore at microcenter and have some 8000 2x24 patriot kit coming in the mail. This new board ran the current 7200 2x16 memory fine. I could only clock the memory to 7400 with 7200xmp set. We'll see what happens with the new kit.
Board says MC score is 80 but that doesn't mean anything. I'll run aida64 here after a while and see what it says.

Cool, keep us updated! I'm doing very minor tweaks one at a time and running them with stress tests overnight. Nothing major to report other than underlying stability at 7800c36 XMP is real, and that most of the new Gigabyte "easy tweaking features" do make aida64 scores go up, but aren't stable

Also there are new RAM overclocking features that sounds efficiency oriented, Dynamic Memory Boost and Real-Time Memory Frequency Overlocking. It sounds like the second of those just throttles between JEDEC and XMP... no thanks.

Here's with the enhancements disabled in advanced memory settings, and no other change besides tREFI to 65532. For some reason tRFC is grayed/locked out at 624. We'll see how this does overnight!

 

[Upgrayedd]

This patriot 2x24 8000 kit had 2 errors trying the 7800xmp profile last night. Ran the 7600 profile no problems. I'll prob just stick with 7600 with this setup. 8000 C38 XMP is a no go.

If I had a timing template that I could copy and input values into that calculated possible correct timings for a particular speed that would be sweet but there's like 87bajillion timing values you can edit on this board and it's just too much. I noticed other 8000kits are cL40 not cl38.

 

[ir_cow]

This patriot 2x24 8000 kit had 2 errors trying the 7800xmp profile last night. Ran the 7600 profile no problems. I'll prob just stick with 7600 with this setup. 8000 C38 XMP is a no go.

If I had a timing template that I could copy and input values into that calculated possible correct timings for a particular speed that would be sweet but there's like 87bajillion timing values you can edit on this board and it's just too much. I noticed other 8000kits are cL40 not cl38.

Have you tried manually entering the SA, Tx and VDD2 voltages? Pretty much don't need to adjust anything besides enable XMP and go. No need to change any sub-timings that isn't in the SPD data on the ASUS Apex for 8000. Of course you can for more performance adjustments.

Could be your CPU IMC limitation here as well , but you won't know until you try the manual adjustments for the memory controller. 7600 to 8600 is the range. Unfortunately that is the lottery aspect.

 

[fusseli]

I had no luck getting the boosted tREFI stable. Nothing above the XMP 7605 seems to be stable. Are there other timing dependencies that need to be changed too when changing it?

 

[Upgrayedd]

Have you tried manually entering the SA, Tx and VDD2 voltages? Pretty much don't need to adjust anything besides enable XMP and go. No need to change any sub-timings that isn't in the SPD data on the ASUS Apex for 8000. Of course you can for more performance adjustments.

Could be your CPU IMC limitation here as well , but you won't know until you try the manual adjustments for the memory controller. 7600 to 8600 is the range. Unfortunately that is the lottery aspect.

I really don't know what those voltages are and what would be a safe long term voltage.

I had it passing memtest at 7600xmp but as I took the pc over to my nephews he loaded up metro exodus and hard froze twice in a row.. the game didn't run but 30 seconds in the game world.. really getting tired of this machine..I bought an i9 a few days ago when i bought the apex encore incase I was still having problems after buying the apex thinking it could be the cpu so I'll get both so I don't have to come back again right away... I might crack it open and try it.. that or my contact frame isn't right and it's putting too much pressure still...
I set the XMP the 7800 profile timings and manually set the speed to 7600 and it didn't crash anymore.
Or is there a looser profile other than the default XMP1? cause I have XMP2 and XMP TWEAKED but the descriptors for these settings are pretty bad so it's not quite clear on what does what.

 

[fusseli]

I hate to say it but maybe the contact frame damaged the mobo

 

[ir_cow]

XMP2 takes the XMP data direct. XMP1 alters some of the sub-timings for you. Same with XMP Tweak. I use XMP2 for reviews as its what the SPD values really are, not what ASUS deems "better" for performance.

Asus is tricky withe XMP stuff. Probably shouldn't even be called the same thing to avoid confusion from the actual XMP profiles.

As for voltages, I don't remember if it's this thread or the other one you started first I wrote the voltages in for ya already.

With an ASUS Apex MB and 8000 binned memory, it sure isn't those two giving you problems. You could just have really shitty CPU IMC. You'll just have to spend the time to figure it out on your own. No one can tell you what it "suppose" to be. CPU lottery is real.

 

[Toothless]

I hate to say it but maybe the contact frame damaged the mobo

Having seen in person and done an install myself, the contact frames will not damage a board unless you decide to sneeze all over everything or had way too much alcohol. They're VERY easy to install and the 14900k desktop with the frame on that my wife uses has zero issues. Actually that rig does everything and more perfectly.

 

[ir_cow]

@Toothless agreed. Hard to mess up unless you bent some pins prior to install.

 

[Upgrayedd]

XMP2 takes the XMP data direct. XMP1 alters some of the sub-timings for you. Same with XMP Tweak. I use XMP2 for reviews as its what the SPD values really are, not what ASUS deems "better" for performance.

Asus is tricky withe XMP stuff. Probably shouldn't even be called the same thing to avoid confusion from the actual XMP profiles.

As for voltages, I don't remember if it's this thread or the other one you started first I wrote the voltages in for ya already.

With an ASUS Apex MB and 8000 binned memory, it sure isn't those two giving you problems. You could just have really shitty CPU IMC. You'll just have to spend the time to figure it out on your own. No one can tell you what it "suppose" to be. CPU lottery is real.

I'm going back over there tomorrow to get the PC back.
I'll try xmp2 I've only been trying xmp1 pretty sure. I'll go back and look through my thread and see what you have written about voltages. Are you suggesting they might need to be raised or just hard set?

Having seen in person and done an install myself, the contact frames will not damage a board unless you decide to sneeze all over everything or had way too much alcohol. They're VERY easy to install and the 14900k desktop with the frame on that my wife uses has zero issues. Actually that rig does everything and more perfectly.

How tight do you put it? Till the screws stop? I mounted everything until it stopped, x pattern on the frame and I tightened the noctua d15 all the way.

 

[Toothless]

How tight do you put it? Till the screws stop? I mounted everything until it stopped, x pattern on the frame and I tightened the noctua d15 all the way.

Just be reasonable with the tightness. I did it enough where it's not gonna come lose (insert "that ain't coming out" dad joke) but not enough to give the board issues. I learned from the best!

 

[fusseli]

Okay the gloves are off. I’ve added an absolutely necessary Noctua fan to my ram

 

[ir_cow]

I love it! I got a big 120 nuctua fan at the testbench and 90 on my work computer sitting on the video card pointing at it. I like to keep the temps within reason.

 

[fusseli]

Oddly the little 40mm fan helps significantly (it's silent at full speed so I'm running it at full speed), running testmem5 was getting me to ~60C and now I just did a run and hit 50C on one stick.... but 63C on the other... how is that possible??? lol

Nevermind it appears to have been a fluke; here's ~15 minutes into an Extreme run on testmem5, both sticks are sitting at ~50C. YES ladies and gentlemen, you are witnessing a ~10C drop in DDR5 7800 ram temps from adding a 40mm x 25mm Noctua fan using an Asus ROG fan bracket

 

[ir_cow]

Probably all the air is going down the middle. So front of one where and the ICs are while the back of the others PCB.

@fusseli . Just curious what those SA, TX and VDD2 voltages are that gigabyte set for you (assuming not manually entered). Can you share a full Hwinfo screenshot? They arent all in the same place is why.

 

[nguyen]

7200MT XMP just works, and using Buildzoid timings (which HUB showed in their video) and I spent less than 2mins to get the most out of my RAM kit

I use Phantek T30 on the top as intake to blow air over the DIMM, PMIC temp only goes to 42C in game

 

[fusseli]

Probably all the air is going down the middle. So front of one where and the ICs are while the back of the others PCB.

@fusseli . Just curious what those SA, TX and VDD2 voltages are that gigabyte set for you (assuming not manually entered). Can you share a full Hwinfo screenshot? They arent all in the same place is why.

Here are all voltages, I'm not sure what you mean, and yes this is all stock/auto using "spec enhance" (the only one that's stable with this 7800 mt/s) in Gigabyte bios. I'm secretly hoping they drop a bios update at some point, there are some glitchy things here and there with all of the restarts and settings combinations I have tried. I've done a bios factory reset and started fresh to get away from unseen settings walks that seem to creep up from toggling various things around of auto/enabled/disabled (wtf is auto doing anyway?). Earlier today I had bios literally freeze qhile adjusting q-fan settings for this new fan

 

[ir_cow]

@fusseli Thanks. I was just curious. Looks like that motherboard doesn't report VDD2, but often its the same as the VDDQ_TX. So my guess your at SA 1.3, TX 1.4, VDD2 1.4. It's close to what I need for 8400 (1.375, 1.4, 1.4), but I also have a 2-slot motherboard. Nice thing about those is less voltage needed on those rails I noticed.

 

[fusseli]

@fusseli Thanks. I was just curious. Looks like that motherboard doesn't report VDD2, but often its the same as the VDDQ_TX. So my guess your at SA 1.3, TX 1.4, VDD2 1.4. It's close to what I need for 8400 (1.375, 1.4, 1.4), but I also have a 2-slot motherboard. Nice thing about those is less voltage needed on those rails I noticed.

Interesting on the 8400 note. I have not tried to up my speed yet but once I'm locked in with good latency and stability I will try for 8000 or more just to see if it's possible.

What seemed hopeless is finally getting somewhere again, as evidenced just now! I am *finally* getting signs of stability with tweaked XMP 7800 CL36. Thanks for @nguyen on the mention to Bullzoid, I can't find anything over 7200 to directly use but I looked for the trends in how he went from tweaked 6000 to tweaked 7200 for some clues and it seems to have worked. The trend that remains fixed for me is that I can't touch tREFI... I don't even let my eyes settle on it in bios anymore, it stays at 7605 or I crash/error on stress testing.

It's possible the RAM cooling is helping. I don't break ~50 C now and I am gaming stable for hours on end and don't break 45 C that way. Also my cpu biscuits score changed itself, it was 92.270 and now it's 92.418 .... whatever that means

Aaaaaannddd the above wasn't OCCT stable. So I backed off the timings more and to get stable again it looks pretty similar to before. This is slow going:

 

[Upgrayedd]

Just be reasonable with the tightness. I did it enough where it's not gonna come lose (insert "that ain't coming out" dad joke) but not enough to give the board issues. I learned from the best!

The pressure definitely matters here. I've tried so many reseats with this frame and it's just so sensitive. I took the frame off and put the stock ILM back on. 7600 passed XMP2. 7800 had 1 error on pass 3/4. Didn't even try 8000 C38. I did some reading last night and read the TG frame is very finicky and the thermalright frame might be better. I read that frames and high end asus boards don't get along with high frequency ram. Also read that a convex shaped air cooler could have problems with frames?
There should definitely be some rubber feet or something under the corners of this TG frame that give the feeling of resistance so you know you don't need to go anymore. I don't have a tiny little torque driver to get the .6Nm tightness like they want. Why would they rely on anyone to have a torque driver?
After one particular baby finger tightening and a failed memtest I went to take the cooler off and check the tightness of the frame screws and notice that one screw was very very loose and took a few free spins to get any resistance, the other 3 were fine and didn't take any resistance free spins.
I'll take a look at some voltages and if nothing helps I might crack the i9 box and try for 8000 C38 xmp2.

 

[ShrimpBrime]

The pressure definitely matters here. I've tried so many reseats with this frame and it's just so sensitive. I took the frame off and put the stock ILM back on. 7600 passed XMP2. 7800 had 1 error on pass 3/4. Didn't even try 8000 C38. I did some reading last night and read the TG frame is very finicky and the thermalright frame might be better. I read that frames and high end asus boards don't get along with high frequency ram. Also read that a convex shaped air cooler could have problems with frames?
There should definitely be some rubber feet or something under the corners of this TG frame that give the feeling of resistance so you know you don't need to go anymore. I don't have a tiny little torque driver to get the .6Nm tightness like they want. Why would they rely on anyone to have a torque driver?
After one particular baby finger tightening and a failed memtest I went to take the cooler off and check the tightness of the frame screws and notice that one screw was very very loose and took a few free spins to get any resistance, the other 3 were fine and didn't take any resistance free spins.
I'll take a look at some voltages and if nothing helps I might crack the i9 box and try for 8000 C38 xmp2.

You've reseated the cooler several times. Did you pull the cpu and clean the contacts and reseat that yet?

Mount the cooler normal. You only need a few inch pounds per screw to have a good clamping force.

For example, a 0.138" screw would only need about 2.14 inch pounds per each of 4 screws to get 125 pounds of clamping force. For only 2 screws about 4.2 inch pounds to achieve the same clamping force. ( or there abouts depending on washers and so forth.) I use these formulas for Peltier clamping force for best thermal transfer.

Need to turn on memory training. Most of them on auto is disabled. I'd definitely start with early memory training on enabled to start with.

 

[ir_cow]

I didn't waste money on a $60 Thermal grizzly frame. Paid $7 for the Thermalright one. Bought 4 in total (yes all are being used). All the same. X pattern until finger tight. I've probably done it 30-40 times now at least. I'm Nov it was about 10 trying different 14900Ks and motherboards.

Never once did "overtightened" happen. The same can't be said for waterblocks. Those monoblocks that screw in from the back give me trouble all the time.

Unrelated to the contact frame, I personally wouldn't go past those voltages you have already for long term use. Truthfully, no one knows how long it will last. Could be 1 year or 10 years. Some of those crazy XoC guys go with 1.4+ on the SA all day long, but they aren't planning on keeping it past the next socket either.

I've noticed what I call a roll over voltage (aka playing dead). For example if I put in 1.4v SA after a few of hours I'll start to get errors no matter if the freq is 6000 or 8400. Same goes with the other two. 1.45 and it will get mad at me eventually. Combine all three and it's like 30 minutes.

For most of the 12th gen I thought it was the MB and memory giving me grief. Nope, just the CPU.

I think the all these CPUs just like the IMC have different limits. no one can agree what's "safe" or produces errors.

 

[Upgrayedd]

You've reseated the cooler several times. Did you pull the cpu and clean the contacts and reseat that yet?

Mount the cooler normal. You only need a few inch pounds per screw to have a good clamping force.

For example, a 0.138" screw would only need about 2.14 inch pounds per each of 4 screws to get 125 pounds of clamping force. For only 2 screws about 4.2 inch pounds to achieve the same clamping force. ( or there abouts depending on washers and so forth.) I use these formulas for Peltier clamping force for best thermal transfer.

Need to turn on memory training. Most of them on auto is disabled. I'd definitely start with early memory training on enabled to start with.

Nah the cpu pads and mobo pins look fine. I just followed the noctua instructions. Go until it stops, isn't that what the pressure springs are for? I've never had a problem with cooler mounting pressure but I did back off like a full turn or two from stop.
I'm fairly certain training is on. I'll get some stats here a in a bit for you.