Asus Z490 stuff

[Braegnok]

BIOS 1003 is solid on Maximus Extreme XII board.



Loading my old 5.3 all core profile and running Prime95 now,..

[geneo]

Ditto for Hero

[morph.]

BIOS 1003 is solid on Maximus Extreme XII board.



Loading my old 5.3 all core profile and running Prime95 now,..

Still, on 901 and running 4 sticks @4133, I think there is still an improvement for you there are some values of mine I could probably tighten too but haven't bothered yet.

[kumiko]

i upgrade from 1002 to this new 1003, and when is finish to reebot, i stuck with white light VGA and error 79, i shutdown and on againd and start as usually, is any problem with this??

[Braegnok]

Still, on 901 and running 4 sticks @4133, I think there is still an improvement for you there are some values of mine I could probably tighten too but haven't bothered yet.

This board will take any value I put in on any timing and not crash at re-start, it will boot Windows every time, or go to F1,.. it's crazy.

Lowering tRFC from 367 only reduces Read/Write/Copy & raises my Latency, basically for 180ns x 4200 = 756000 divided by 2000 = [378] is what it should be running at, but it seems to do better at 367,.. and for 140ns my tRFC would be [294] and I can run at 294 stable but unfortunately Latency drops & Read/Write/Copy also drop, so seems 367 is my sweet spot with this memory kit.

Lowering tWTR L/S I leave both in Auto while adjusting tWRRD sg/dg, and can go 26/22 and 6/2 stable but reduces Copy and raises Latency, so settled on 29/25 & 8/4.
CAS Write Latency (tCWL) is strange on this board it should be same as Primary (tCL) unless that is odd #, like 17 than rule is change 1-lower to 16. But like I said this board is strange and my tCL is 16 so setting tCWL on 16 will cause issues while adjusting other values, so I leave it in Auto while adjusting than set it when finished. And this crazy board does not like an even #,.. in Auto it will run 15 till I get really tight on other settings than change to 13 but generally it likes 15, so I settle for that, as I have tried 16, 15, 14, 13, and 15 works best, vary odd indeed.

Lowering tCKE seems to have no effect on anything, I have tried 8, 6, 4, and stable at all, and Read, Write, Copy, and Latency stay the same, so I settle at 6, as I think running a lower # 4 might put more stress on memory controller for no gain.

[Ezveedub]

Anyone run into a debug code 32 boot loop on these Asus mobos? Seem I find scant info on this and my Hero Xii is now stuck with this issue and won't post. Just CPU & DRAM Q-LED come on on start up and then starts again with debug LED showing 32 before looping. Tried reflashing two different bios in recovery, pulling CMOS battery, swapping DDR, pulling CPU and reinstalling, same issue, stuck at code 32 rinse and repeat. Pulling out all memory results in memory debug Q-LED on and long beeping. Also tried different memory. Had 2x16gb, tried 4x8gb and 2x8Gb same results. I'm thinking this mobo crapped out. Gonna try another mobo here soon to check, but weird to just up and do this from clear CMOS only reset.

Resolved this issue after a few days of back and forth testing, including a trying a Gaming Strix-E, another 10900K & MSI Tomahawk mobos. Found my reset button on the case it was installed in was incorrectly wired to be closed circuit on at all times and this caused the Asus mobos to stop at 32 or 3A on initial post from a cold boot when the bios is reset and loop the initial posting over and over. The original Aorus z390 Master monk in this case seems to ignore this and worked fine.


Sent from my iPhone using Tapatalk Pro

[morph.]

This board will take any value I put in on any timing and not crash at re-start, it will boot Windows every time, or go to F1,.. it's crazy.

Lowering tRFC from 367 only reduces Read/Write/Copy & raises Latency, Default is 734 so I settled on 1/2 that and called it good, as I have tried 300 up to 400.

Lowering tWTR L/S I leave both in Auto while adjusting tWRRD sg/dg, and can go 26/22 and 6/2 stable but reduces Copy and raises Latency, so settled on 29/25 & 8/4.
CAS Write Latency (tCWL) is strange on this board it should be same as Primary (tCL) unless that is odd #, like 17 than rule is change 1-lower to 16. But like I said this board is strange and my tCL is 16 so setting tCWL on 16 will cause issues while adjusting other values, so I leave it in Auto while adjusting than set it when finished. And this crazy board does not like an even #,.. in Auto it will run 15 till I get really tight on other settings than change to 13 but generally it likes 15, so I settle for that, as I have tried 16, 15, 14, 13, and 15 works best, vary odd indeed.

Lowering tCKE seems to have no effect on anything, I have tried 8, 6, 4, and stable at all, and Read, Write, Copy, and Latency stay the same, so I settle at 6, as I think running a lower # 4 might put more stress on memory controller for no gain.

Your latency is okay but could be lower IMHO did you tweak TXP/PPD? IMHO your bandwidth is lacking for 4200...

At 4133 with the new 1003 bios, this is what I get... For some reason, I couldn't post with tRDRD_SG 6 so had to settle for 7.

Will run an overnight stability test if it's okay ill proceed to lower tRAS to 34 and test again and if stable again I can start to dial down some voltages.

[martine-dee]

So, my new settings are in. Nothing spectacular, and I override the CPU settings with XTU anyway.

@BloodBaron ^ up there. I compute the DRAM voltage by stretching the range 2133 MHz - 4000 MHz linearly to the 1.2 - 1.5 Volts. So, X MHz would mean:

1.2 + (X - 2133) / (4000 - 2133) * (1.5-1.2)

Volts.

@Braegnok touche, that is impressive for the kit. Not sure if rasising DRAM RAS# to CAS# Delay from 15 to 17 (vs. my setup) did something there. I might get that AIDA64 thing at some point, to compare the latency, and play better with the RAM settings. Tho, it looks like a subscription software model, which I don't endorse or support for plainly checking RAM latency every now and then. I am not that much of a large customer.

[Braegnok]

So, my new settings are in. Nothing spectacular, and I override the CPU settings with XTU anyway.

@BloodBaron ^ up there. I compute the DRAM voltage by stretching the range 2133 MHz - 4000 MHz linearly to the 1.2 - 1.5 Volts. So, X MHz would mean:

1.2 + (X - 2133) / (4000 - 2133) * (1.5-1.2)

Volts.

@Braegnok touche, that is impressive for the kit. Not sure if rasising DRAM RAS# to CAS# Delay from 15 to 17 (vs. my setup) did something there. I might get that AIDA64 thing at some point, to compare the latency, and play better with the RAM settings. Tho, it looks like a subscription software model, which I don't endorse or support for plainly checking RAM latency every now and then. I am not that much of a large customer.

I generally run in default settings with memory @ 3600MHz tuned instead of using XMP.

Time Spy run @ 4200MHz. http://www.3dmark.com/spy/16964375

You will enjoy tuning your memory and checking results with AIDA64, it is a nice way to get a bit more performance, confirming results.

I'm testing another 3600MHz G.SKILL kit, F4-3600C14D-32GTRS it's a 2x16GB kit, I disabled DIMM0 A, B and will see how 2x16 bars work with the two unused DIMM slots disabled.

[morph.]

I generally run in default settings with memory @ 3600MHz tuned instead of using XMP.

Blue Room run @ default settings. http://www.3dmark.com/vrpbr/96478

You will enjoy tuning your memory and checking results with AIDA64, it is a nice way to get a bit more performance, confirming results.

I'm testing another 3600MHz G.SKILL kit, F4-3600C14D-32GTRS it's a 2x16GB kit, I disabled DIMM0 A, B and will see how 2x16 bars work with the two unused DIMM slots disabled.

Funny enough I just ordered a new set of 3600 14-15-15-35 gskill as well and it's 2x16 too (F4-3600C14D-32GTRS). Hoping to dial it up to around 4200-4400 with decent timings...