©ASUSTEK COMPUTER INC. ALL RIGHTS RESERVED.

Shamino · ‎05-21-2020

Ataemonus · ‎07-09-2020

I flashed BIOS version 0704 for the Hero. Quite a few bugs in it, very obvious ones at that.
First, the APM settings I set to Auto, they revert to whatever random setting they want so I have to set them again.
Second, instead of RAM compatibility, it's now not even working at SPD, which would be Auto or Manual, cycles forever until in POSTs again in safe mode. That is with Crucial Ballistix RGB 3200C16, one kit of 2x32GB sticks. This is quite silly, I mean....I have to enable XMP for the system to boot. SPD is 2666 with some very loose timings, so ...
I tried Auto, I tried manual, I was afraid that would be that until I can reflash BIOS, but it worked on XMP, at least that.
But I want my SPD back, it's like 40C in the room where the PC is, already too hot, I don't want the 1.35V through the RAM in these conditions until I have my AC back.

Also, one bug that has been in the ASUS BIOSes for ... EVER.
Setting the onboard LEDs to All On in ON mode, POST code to Auto, and OFF state to Stealth Mode, sets Stealth mode for ON state and that's that. Every single ASUS board I tried does that, I mean, I tried dozens of different boards. WTF ?
I have to post, then go and redo those settings again, although they were confirmed and shown in the change list before exiting the BIOS previously. That is ridiculous.

GemueseMonster · ‎07-10-2020

Might be a stupid question, but has IO Latency Offset effect reversed from previous generation boards?

I remember lower offset resulting in lower RTL/IO, but with the M12A its higher offset = lower RTL/IO

GemueseMonster · ‎07-14-2020

GemueseMonster wrote:
Might be a stupid question, but has IO Latency Offset effect reversed from previous generation boards?

I remember lower offset resulting in lower RTL/IO, but with the M12A its higher offset = lower RTL/IO

nvm, seems fixed in 0704.

but now IOLs dont train with IO Latency Offset above 4500MT/s... whaaat? :eek:

Apothysis · ‎07-14-2020

So there's a couple of timings exposed in MemTweakIt that show huge potential in regards to mem oc, tXP and PPD.

Tested in AIDA:
tXP 7 + PPD 1 (Default) = 37.4 ns
tXP 4 + PPD 1 = 36.3 ns
tXP 4 + PPD 0 = 35.0 ns

There's a thread on reddit about this showing gains for everyone who tests it. However, these timings are not exposed in the UEFI meaning you have to set them with memtweakit every single reboot - a huge pain obviously. Could we have these timings made available on a UEFI-level? Or is there another way we could make the settings stick?

Thanks

GemueseMonster · ‎07-15-2020

Apothysis wrote:
So there's a couple of timings exposed in MemTweakIt that show huge potential in regards to mem oc, tXP and PPD.

Tested in AIDA:
tXP 7 + PPD 1 (Default) = 37.4 ns
tXP 4 + PPD 1 = 36.3 ns
tXP 4 + PPD 0 = 35.0 ns

There's a thread on reddit about this showing gains for everyone who tests it. However, these timings are not exposed in the UEFI meaning you have to set them with memtweakit every single reboot - a huge pain obviously. Could we have these timings made available on a UEFI-level? Or is there another way we could make the settings stick?

Thanks

which version of memtweakit and what board?

I haven't found a working version for the M12A yet

bscool · ‎07-15-2020

GemueseMonster wrote:
which version of memtweakit and what board?

I haven't found a working version for the M12A yet

MemTweakIt 2.02.48 works on the Apex and came on the included CD.

I tested and didn't see a difference in latency changing tXP but the PPD from 1 to 0 did reduce latency.

sladenlol · ‎07-16-2020

Hey guys,

Ive Upgraded from 7700k to 10900k and now my gpu score dropped from 30700 in firestrike to 29500.
Everything is the Same As before, i checked gpu it runs 16x
No heat issues either.
Is this to be expected cause the Ring Bus Has more latency cause of more cores?
Ive read something like that and thats why Many firestrike benchers use 7700k-8700k.

Betroz · ‎07-16-2020

sladenlol wrote:
Is this to be expected cause the Ring Bus Has more latency cause of more cores?
Ive read something like that and thats why Many firestrike benchers use 7700k-8700k.

No, the 10900K is not slower than a 7700K. Maybe you should reformat your C: drive and perform a clean Windows install with all the newest drivers.

Edit : One thing you can try first, is to remove all the CPU power limits in BIOS.
Edit 2 : Make sure your RAM is not running at 2133 Mhz speed in BIOS. If it does, that will ruin your performance.

sladenlol · ‎07-16-2020

Betroz wrote:
No, the 10900K is not slower than a 7700K. Maybe you should reformat your C: drive and perform a clean Windows install with all the newest drivers.

Edit : One thing you can try first, is to remove all the CPU power limits in BIOS.
Edit 2 : Make sure your RAM is not running at 2133 Mhz speed in BIOS. If it does, that will ruin your performance.

Im Not talking About cpu performance, im talking About Ring Bus and PCIe performance.
My cpu performance is fine, but i have less gpu score in firestrike, cause of more Ring Bus latency i guess.
I have all Power Limits disabled and Run 5,3GHz allcore with 4200 cl15 ram

Betroz · ‎07-16-2020

sladenlol wrote:
Im Not talking About cpu performance, im talking About Ring Bus and PCIe performance.
My cpu performance is fine, but i have less gpu score in firestrike, cause of more Ring Bus latency i guess.
I have all Power Limits disabled and Run 5,3GHz allcore with 4200 cl15 ram

Latency with Z490/10900K is about ~2ns worse compared to Z390/9900K with the same memory settings, but that is not the cause of your problem. What is, I don't know.

Edit : **latency with AIDA64 memory test**

Asus Z490 stuff