@cicoria 1. Thats normal for the GT7 I noticed it too with the vanilla firmware however Biostar insist “We did not notice any significant difference when comparing boot times”. That is despite me saying I have five other X370/470 boards that do not exhibit that long boot time. 2. That sounds like normal behaviour if you set ohms too high or too low. I’ve tested that feature on the GT7 I have and it worked properly however that setting is very subjective and will vary depending on what ICs the memory you have uses. Try downloading Ryzen Timing Checker, input the ohms values it reports manually and see if the board boots. If you could disable the Gear Down option at or aove 2667MHz memory frequency to see if the board will POST that would be very helpful in determining just how broken the gear down option is and how many people it is affecting.
@ket
1) yes boot up times are very similar with the official biostar firmware. The slow part comes out with ram @2933 standard/2100 system boot up super fast!
2) my ohms are set up according to what ryzen mem calculator say to me.
3) I tried Gear down "disabled" with my ram @2933 (Savage 4 pro black hynix AFR) and booted up with no problems! cpu-z says (correctly?) command rate 1t
@ket Tried to expose Fail_CNT but couldn’t get it to show up no matter what. Can you give me a hand? I also tried to just change the count from 3 to 1 in hex but I don’t think I was successful as it still cycles 3 times before posting. Task manager shows a 4sec improvement in the Last BIOS Time. Not sure if it’s related to me messing around with the Fail_CNT.
Things that can still be updated:
I know I have newer versions for all PSP and ASM1543x Firmwares, collected from other OEMs. And I haven’t even done much research lately, I’m certain there could be even newer than what I have.
Biostar_X370GT7__Z_Extract.rar (3.35 MB)
@barisuraz
Go to GUIDs 3E7788CA-2BFC-4FCE-8122-5176CA492D9F and A5E369C8-ABF9-4B43-B212-FF1BFD35666D, extract the PE32 images.
Convert PE32 images with Universal IFR extractor, look for "DRAM Timing Configuration" in both files
Copy the Hex code, search for it in HxD
Once found delete the Hex, then enter it after 29 02
Repeat the process for both files, save the changes
Replace the PE32 images with the edited ones
All done
@lardkag - are you checking the modified BIOS, or stock? All that info looks like stock results?
This is all things I already updated
LAN -
EFI Realtek UNDI (940F3713-D7A2-48E0-A77A-F5DAFC87E308)- 2.040 >> Updated to >> 2.045 (4/26/18) << *** Date corrected ***
OROM Realtek Boot Agent GE (940F3713-D7A2-48E0-A77A-F5DAFC87E307) - 2.64 >> Updated to >> 2.66 (5/26/17)
CPU uCode -
CPUID 00800F82 >> 08008206 (2-12-2018) >> Updated to >> to 0800820B (6-20-2018)
EFI AMD GOP Driver - 2.2.0.0.0_signed (2/6/2018) >>> Updated to >>> 2.4.0.0.0_signed (5/9/18)
EFI AMD GOP Driver - 1.59.0.0.0 (7/25/2015) >>> Updated to >>> 1.66.0.15.50_Signed (4/23/18)
OROM VBIOS Raven - 016.002.000.006.000000 (3/13/18) >>> Updated to >>> 016.002.000.007.000000 (6/8/18)
OROM VBIOS Carrizo - 015.048.000.070.007160 (9/13/16) >>> Updated to >>> 015.048.000.070.007160 = Null/Latest Already
EFI AMD GOP Driver - 2.2.0.0.0_signed = 12FA6BCD-E5C0-4E61-8BC6-3876EC6C2083 = 72KB
EFI AMD GOP Driver - 1.59.0.0.0 = DFD5316F-E5C0-4E61-8BC6-3876EC6C2083 = 96KB
OROM VBIOS Raven - 016.002.000.006.000000 = 13F67334-DFB1-490D-A444-D7173A71C3D6 = 53KB
OROM VBIOS Carrizo - 015.048.000.070.007160 = F1004391-BE0F-40F0-94A6-7C45154C9114 = 64KB
EFI AMD RAIDXpert2-Fxx - 9.2.0-00041 (03/19/2018) >> Updated to 9.2.0-00070 (05/14/2018)
OROM AMD RAIDXpert2-Fxx - 8.1.0-00046 (06/08/2017) >> Updated to 8.1.0-00064 (06/08/2017)
OROM AMD RAIDXpert2-Fxx - 8.1.0-00046 (06/08/2017) >> Updated to 8.1.0-00064 (06/08/2017) (Yes, these are two different, but very similar oroms)
Someone send me final BIOS file you guys are using now and I can expose the Fail_CNT if you don’t end up getting it. I have so many versions here, I’m not sure which or if I have the final one ket posted for you all to use.
@ket your above instructions lost me and I mod BIOS 
DRAM Timing Configuration is at top, without any suppress near it, did you mean for them to find this, then locate the Fail_CNT underneath it and move that outside suppress?
That too is confusing to me, sorry Wouldn’t it be move outside of “end if” Very confusing thinking about how to do that, maybe because I have not used that method?
Suppress If {0A 82}
QuestionId: 0x40 equals value 0x1 {12 86 40 00 01 00}
Not {17 02}
End {29 02}
Numeric: Fail_CNT, VarStoreInfo (VarOffset/VarName): 0xC2, VarStore: 0x5000, QuestionId: 0x5D, Size: 1, Min: 0x0, Max 0xA, Step: 0x0 {07 91 5F 01 60 01 5D 00 00 50 C2 00 10 20 00 0A 00}
Default: DefaultId: 0x0, Value (8 bit): 0x3 {5B 06 00 00 00 03}
End {29 02}
End If {29 02}
Change to??
Suppress If {0A 82}
QuestionId: 0x40 equals value 0x1 {12 86 40 00 01 00}
Not {17 02}
End {29 02}
End If {29 02}
Numeric: Fail_CNT, VarStoreInfo (VarOffset/VarName): 0xC2, VarStore: 0x5000, QuestionId: 0x5D, Size: 1, Min: 0x0, Max 0xA, Step: 0x0 {07 91 5F 01 60 01 5D 00 00 50 C2 00 10 20 00 0A 00}
Default: DefaultId: 0x0, Value (8 bit): 0x3 {5B 06 00 00 00 03}
End {29 02}
If not, how? Much easier this way I think, but maybe it’s not ideal? And, what does the Not 17 02 do/mean? That may defeat my usual FF edit below, not sure?
Suppress If {0A 82}
QuestionId: 0x40 equals value 0x1 {12 86 40 00 FF 00}
Not {17 02}
End {29 02}
Numeric: Fail_CNT, VarStoreInfo (VarOffset/VarName): 0xC2, VarStore: 0x5000, QuestionId: 0x5D, Size: 1, Min: 0x0, Max 0xA, Step: 0x0 {07 91 5F 01 60 01 5D 00 00 50 C2 00 10 20 00 0A 00}
Default: DefaultId: 0x0, Value (8 bit): 0x3 {5B 06 00 00 00 03}
End {29 02}
End If {29 02}
Ahh! Looking again, further up @ket , I see what you mean I think! Does unsuppressing here, defeat all the other individual suppresses below for each setting?
Suppress If {0A 82}
64 Bit Unsigned Int: 0x1 {45 0A 01 00 00 00 00 00 00 00}
Ref: DRAM Timing Configuration, VarStoreInfo (VarOffset/VarName): 0xFFFF, VarStore: 0x0, QuestionId: 0x38, FormId: 0x7034 {0F 0F B7 00 B7 00 38 00 00 00 FF FF 00 34 70}
Ref: DRAM Controller Configuration, VarStoreInfo (VarOffset/VarName): 0xFFFF, VarStore: 0x0, QuestionId: 0x39, FormId: 0x7035 {0F 0F B8 00 B8 00 39 00 00 00 FF FF 00 35 70}
Ref: CAD Bus Configuration, VarStoreInfo (VarOffset/VarName): 0xFFFF, VarStore: 0x0, QuestionId: 0x3A, FormId: 0x7036 {0F 0F B9 00 B9 00 3A 00 00 00 FF FF 00 36 70}
End If {29 02}
Change to >>
Suppress If {0A 82}
64 Bit Unsigned Int: 0x1 {45 0A 01 00 00 00 00 00 00 00}
Ref: DRAM Controller Configuration, VarStoreInfo (VarOffset/VarName): 0xFFFF, VarStore: 0x0, QuestionId: 0x39, FormId: 0x7035 {0F 0F B8 00 B8 00 39 00 00 00 FF FF 00 35 70}
Ref: CAD Bus Configuration, VarStoreInfo (VarOffset/VarName): 0xFFFF, VarStore: 0x0, QuestionId: 0x3A, FormId: 0x7036 {0F 0F B9 00 B9 00 3A 00 00 00 FF FF 00 36 70}
End If {29 02}
Ref: DRAM Timing Configuration, VarStoreInfo (VarOffset/VarName): 0xFFFF, VarStore: 0x0, QuestionId: 0x38, FormId: 0x7034 {0F 0F B7 00 B7 00 38 00 00 00 FF FF 00 34 70}
@ket thanks for those timings. they worked though i only did memtest for 100% coverage since i need to do something else. trC automatically sets itself to 69 and wont go lower though. either way i’m satisfied with this degree of control. this bios is more accurate with vcore settings than biostar’s latest offering. although SoC voltage still has some kind of overhead reading. it also does 2 cycles from 3 cycles till post.
@Lost_N_BIOS Heres a quick example of what the code would look like when correctly modified and the modded file extracted with Universal IFR. In a nutshell, moving the primary reference outside the suppress constraints activates all the options for it no need to expose things one by one 
Also, the latest firmware image is the one available for download.
0x9877 Suppress If {0A 82}
0x9879 64 Bit Unsigned Int: 0x1 {45 0A 01 00 00 00 00 00 00 00}
0x9883 Ref: DRAM Controller Configuration, VarStoreInfo (VarOffset/VarName): 0xFFFF, VarStore: 0x0, QuestionId: 0x39, FormId: 0x7035 {0F 0F B8 00 B8 00 39 00 00 00 FF FF 00 35 70}
0x9892 End If {29 02}
0x9894 Ref: DRAM Timing Configuration, VarStoreInfo (VarOffset/VarName): 0xFFFF, VarStore: 0x0, QuestionId: 0x38, FormId: 0x7034 {0F 0F B7 00 B7 00 38 00 00 00 FF FF 00 34 70}
0x98A3 Ref: CAD Bus Configuration, VarStoreInfo (VarOffset/VarName): 0xFFFF, VarStore: 0x0, QuestionId: 0x3A, FormId: 0x7036 {0F 0F B9 00 B9 00 3A 00 00 00 FF FF 00 36 70}
0x98B2 Ref: Data Bus Configuration, VarStoreInfo (VarOffset/VarName): 0xFFFF, VarStore: 0x0, QuestionId: 0x3B, FormId: 0x7037 {0F 0F BA 00 BA 00 3B 00 00 00 FF FF 00 37 70}
0x98C1 Ref: Common RAS, VarStoreInfo (VarOffset/VarName): 0xFFFF, VarStore: 0x0, QuestionId: 0x3C, FormId: 0x7038 {0F 0F BB 00 BB 00 3C 00 00 00 FF FF 00 38 70}
0x98D0 Ref: Security, VarStoreInfo (VarOffset/VarName): 0xFFFF, VarStore: 0x0, QuestionId: 0x3D, FormId: 0x7039 {0F 0F BC 00 BC 00 3D 00 00 00 FF FF 00 39 70}
0x98DF End Form {29 02}
@banjanjan No problem. I’d say its definitely worth testing that memory @ 2933 as well even Hynix kits tend to accomplish that without too much difficulty, 3200 might be a bit of a push, but still in the realms of possibility.
i finally got the fan issue fixed. i flashed the non modded bios to rom2 and wrote down the values it defaults to for the quiet preset.
afterwards i set those values in the modded bios and finally it worked. i would have never found the right setting by manually trying them.
now i can use your very very useful mod bios!
thx again for your all the work you put into this.
and something everyone fiddling with ram controller and timings should be aware off: https://www.overclock.net/forum/11-amd-m…ml#post27705714
Updated firmware is now live
Thanks @ket that example is same as one I mentioned on my final edit! Sorry for the initial confusion, I missed that top-most DRAM Timing Configuration entry, until I took a third look
That’s too funny, I also need the password now 
Never mind, dug it up, will send your way @barisuraz
@ket - is this new update exposing the Fail_CNT we’ve been discussing? If not, I’ll send out a tester to the few that have been discussing this with me past few days.
Sad to report that Fail_CNT doesn’t effect anything related to boot time or power cycling before POST.
Maybe something else is causing that for you then @barisuraz ? @ket why is that setting in two GUID’s? Can you enable one instead of the other, maybe they differ?
Fail_CNT is the number of times the board will attempt to POST before reverting to default settings. Options are at 2 GUIDs likely where one firmware image is used for several different boards can’t think of any other reason why its being done like that. Your problem definitely sounds like unstable settings or just poor memory compatibility bar.
I always wondered why there is two CBS option GUID’s anyway, guess I never thought of these being like Dell BIOS and used on different models
I never looked through them carefully, just assumed you could maybe use one or the other.
Just look at the GT8 firmware, its 97% identical to the GT7 