Hello, curious if updating to this bios might fix my board power cycling 3 times before posting. Also the PowerNow option is not working on the official release, the voltage doesnt go down when idling. Pretty much sold on the latest bios, so I’m keen to try out options.
@barisuraz well that sucks. DRAM problem? also have an issue with voltage reading in the bios vs software. my core voltage in bios is 1.270 while in software (hwmonitor/hwinfo) is 1.250. there is also some kind of offset with DDR voltage and SoC(sadly i dont remember the exact figures). does your mobo have consistent readings as with software?
@barisuraz @banjanjan - sounds like DRAM Training issue, can you guys disable dram training and test if same?
OR Fail Count set to 3, can be set to 1 - I checked, and default is 3 - it’s in CBS, maybe @ket left suppressed though, I can’t tell for sure but it looked like he enabled visibility.
<Request> Flagship X470 Gaming 7 Wifi Motherboard BIOS Fix
Yeah I believe it is related to the DRAM as it only happens when the ram is OC’d over 2400mhz. I will try those but I don’t really know how to disable DRAM training.
I looked around a little for DRAM Traning, but couldn’t find it, maybe @ket can say more on that since he’s much more familiar with the BIOS settings that are and are not visible.
Fail_CNT should be visible for you guys I think, in CBS, if not it will need un-suppressed still because I can see it there. Default setting is 3, and I’ve helped users before on other boards change this to 1, sounded just like what you guys are mentioning.
If you change that to 1 and it only reboots once, then you know it’s failing with something in your current memory settings, hence the 3 reboots
@banjanjan Thats a DRAM related issue your memory is unstable. If you are running stock frequencies try swapping the DIMM slots your memory is in. @Lost_N_BIOS As far as I have seen there is no DRAM training option in the GT7 firmware - anywhere.
@ket thank you for the key. I’m running 2933 for a 2666 rated kit. But Even at stock settings, the issue still persists. It does run 2933 on windows though. I’ll try swapping slots.
@ket I’m running 2933 on a 3000 rated kit and it’s perfectly stable in Windows but it cycles 3 times before booting with any settings other than the default 2400. The issue still happens when I change slots.
Thanks @ket - I assumed it was a memory issue, and know the Fail_CNT change to one from three will prove to them this is the issue when reboots change to one instead of three, some setting is unstable.
Stock settings do not always = should be / will be stable 
Also, stock settings for your board do not equal stock settings for the memory, the SPD profiles only contain so many settings, and that’s less than BIOS contains.
Your board/BIOS choose 20-30+ settings that aren’t part of the memory’s programmed settings, all selected automatically based on whatever the engineers programmed them to use at set speeds.
When Fail_CNT happens, whatever sub-setting triggers that failure is tried x amount of times (Whatever the setting is set to) and then when it fails, some other board selected higher value is applied instead, then it boots properly.
@Lost_N_BIOS thats the problem Biostar completely ballsed up what the board defaults to wether you load the XMP profile or not, put bluntly, the Biostar timings are simply fucked. Their not much better when you load an XMP profile either Biostar can’t even get their code working properly to set the timings in the SPD (thats Serial Presence Detect for the less tech savvy) data. My particular test kit uses an XMP profile with 15-15-15-15-35-50 primary timings, which being Zen with Gear Down enabled should be 16-15-15-15-35-50 yet the GT7 sets 16-15-15-22-36-75… I could probably go on a 10 page rant right about now being former R&D for Mushkin Memory but it wouldn’t be anything we both haven’t already ranted about in PMs 
Although stock timings, should always, without exception, be stable. If they are not its either due to a bad stick of RAM, or poor memory compatibility. When memory kits are binned the timings and frequency they are binned at are tested at a 2T Command Rate (Gear Down: Disabled) 1T Command Rate (Gear Down: Enabled) is not guaranteed but will work in 99.9% of cases. This is why it is an utter laughing stock that disabling Gear Down at a frequency above 2400MHz causes a no POST on the GT7. Biostar couldn’t have messed up their memory compatibility more if they tried.
@barisuraz Yep thats just crappy memory compatibility on Biostars part. When Biostar can’t even get compatibility right for G.Skill memory of all things you know they have just done a terrible, terrible job on memory compatibility.
@banjanjan Ok let us know how things go knowing the model number for your memory would help me suggest some timings that might play nicer for you.
For me I had no chance to get my ram working properly in bank a. Only way to get it running is bank b.
@ket swapping slots doesn’t fix it. I’m using this cheapo ddr4 from Team Group because of availability. Model: TEAM ELITE PLUS DDR4 2666( TPD44G2666HC19BK) Black/gold. The default timings are already crappy as they are, Tcas for 2933 is 20 since Biostar automatically(idk why) sets it to that value if I tried to put, say 18. timings are 20-19-19-40. Wish I had hands on a better kit, since I am not yet confident in tweaking so much in mem settings.
I hear that @ket - I was more speaking about how some boards/brands set tighter sub-timings than others, and often none of those whether loose or tight take into account what’s in the SPD and some of the available/auto set timings may not even be included in all SPD/XMP’s.
You are certain far more knowlegable than I am about this subject, but that’s always been how I’ve understood and looked at this in general regarding auto being OK on one board/brand but possibly not on another at seemingly same settings (or auto/XMP)
And yes, as we’ve discussed, I think they’ve messed it up plenty for memory compatibility. All the complaints and issues I’ve seen mentioned tell me as much without even having to use the board.
@Screemer @banjanjan @barisuraz - does setting FAIL_CNT in CBS to 1 instead of 3, change reboot behavior to one reboot before starting instead of 3?
Thanks @barisuraz - Then this setting needs to be unsuppressed with FF’s at the Question ID, or change the Not 17 02 maybe? @ket probably knows best which way to do it in this BIOS
It’s in both A5E369C8-ABF9-4B43-B212-FF1BFD35666D_CbsSetupDxeZP
and 3E7788CA-2BFC-4FCE-8122-5176CA492D9F_CbsSetupDxeRV
Example
QuestionId: 0x40 equals value 0x1 {12 86 40 00 01 00}
Change to
QuestionId: 0x40 equals value 0x1 {12 86 40 00 FF 00}
Or, change the not 17 02 to False 47 02 or True 46 02 (Unsure which works when question ID asked preceding)
I always simply FF that question ID value, negating the suppress if original intended target answer
@ket can you unsupress everything in CBS and PBS? I mean this custom bios is obviously aimed at users with at least at a decent level of knowledge about computers so I don’t think there will be a problem.
[quote="banjanjan, post:113, topic:32253"] TPD44G2666HC19BK [/quote]
From what I know of that kit it is rated for 19-19-19-42 @ 1.2v and Hynix ICs. I'd try 16-16-16-36 @ 1.35v, test with HCI memtest for at least 3 hours for memory errors and if it passes try stepping up to 2666 then re-test.
[quote="barisuraz, post:117, topic:32253"] @ket can you unsupress everything in CBS and PBS? I mean this custom bios is obviously aimed at users with at least at a decent level of knowledge about computers so I don't think there will be a problem. [/quote]
Most suppressed options are just debug things, duplicates or things that are not helpful to overclockers at all theres only a couple options I could add that might be useful.
[quote="Lost_N_BIOS, post:116, topic:32253"] Thanks @barisuraz - Then this setting needs to be unsuppressed with FF's at the Question ID, or change the Not 17 02 maybe? @ket probably knows best which way to do it in this BIOS
It's in both A5E369C8-ABF9-4B43-B212-FF1BFD35666D_CbsSetupDxeZP
and 3E7788CA-2BFC-4FCE-8122-5176CA492D9F_CbsSetupDxeRV
Example
QuestionId: 0x40 equals value 0x1 {12 86 40 00 01 00}
Change to
QuestionId: 0x40 equals value 0x1 {12 86 40 00 FF 00}
Or, change the not 17 02 to False 47 02 or True 46 02 (Unsure which works when question ID asked preceding)
I always simply FF that question ID value, negating the suppress if original intended target answer [/quote]
No need to bother with any of that, simply find the Hex code for the option you want to expose and move it outside of the suppress constraints.
[quote="Lost_N_BIOS, post:114, topic:32253"] I hear that @ket - I was more speaking about how some boards/brands set tighter sub-timings than others, and often none of those whether loose or tight take into account what's in the SPD and some of the available/auto set timings may not even be included in all SPD/XMP's.
You are certain far more knowlegable than I am about this subject, but that's always been how I've understood and looked at this in general regarding auto being OK on one board/brand but possibly not on another at seemingly same settings (or auto/XMP)
And yes, as we've discussed, I think they've messed it up plenty for memory compatibility. All the complaints and issues I've seen mentioned tell me as much without even having to use the board.
@Screemer @banjanjan @barisuraz - does setting FAIL_CNT in CBS to 1 instead of 3, change reboot behavior to one reboot before starting instead of 3? [/quote]
Some boards can vary with timings they use but typically now you will find from board to board timings are set pretty much the same with maybe just an odd difference here and there in the sub-timings which are often tweaked by the manufacturer for better performance at a certain frequency or to maximise frequency and / or compatibility only if something has been extremely poorly coded / engineered will timings be drastically different and of course when an XMP profile is loaded those timings in the XMP should be being applied by the board - no exceptions ever. Biostar have just completely failed here and must sort it out.
@Lost_N_BIOS.
Actually never tried it. My cl16 dual ranked b-die ram is working fine in bank b with 3200mhz and tight subtiming.
just tried the BIOS mod! what can I say! FANTASTIC! Biostar should seriously think to hire you guys! Only 2 notes:
1) slow post time (cold boot) for my hynix AFR ram@2933 Mhz (savage4 pro) used ryzen mem calculator without problems;
2) DDR4 CAD settings in my case didn’t worked (with CAD memory settings no boot). Left to auto and it works!
keep up the good work guys!