BIOS updates for AM4 motherboards

Alright, flashed 1.40, no BCLK adjustment in the factory condition. No Sub-Timings adjustments either, just the Processor OnDie Terminationn (ProcODT).
Though, lots of small visual tweaks that I'm liking, some that I'm not totally thrilled over:
Liking the display of timings when things are running Auto.
The little red "pencil" (barely looks like one) next to the input boxes is kinda annoying lol
There's the "Memory Try It !" option, which I think might've been in 1.3 as well. If not, it just is a number of preset memory speeds and timings. Whether or not they include different SUB timings as well is anyones guess.
CPU and Mainboard temps are now under the CPU/RAM speed instead of next to the Game Boost window. In their stead is the A-XMP button with profiles. A nice touch, though I kinda miss the easy to see (big font/in your face) temp readings heh Oh well.
Microcode reads as 800111C, sadly I was too big of an idiot to think to look at what my v1.10 showed :| lol
Aww they removed the really low DDR4 speeds lol I was kinda hoping to leverage those for uber low timings! :shame: 1866 is now the lowest.
CnQ option now available in the CPU Features section, as well as SMT option, Spread Spectrum and IOMMU.
The Settings area is still pretty sparse given what is actually all available in the BIOS hidden in other inaccessible areas.
Lastly I notice CPU-NB voltage defaults to 0.904V now instead of 0.954V
​
So with the summary concluded and nothing interesting for me to play with, time to set defaults and apply my modded BIOS.
Back in a flash! Get it?! Flash?? :hilarious: Cuz... BIOSes ne.. need to be flashed.... *sigh* :shifty: lol


EDIT: Preliminary results are in.
A 1700X boots into the BIOS, its speed reads 3.40Ghz, this we all know.
An MSI Titanium v1.40 BIOS lacks BCLK option, this anyone else can easily confirm.
My modded v1.40 BIOS has a BCLK option, this I can easily prove.
Whether or not that BCLK option actually does anything, that is what we want to know....

So does it? Looks to be the case thus far! :smug:
After setting "CPU Base Clock (MHz)" to 101, saving a rebooting...
CPU Speed reported by BIOS: 3.43GHz (y)
BCLK Ahoy!.png


Though, a bit of a lawl moment (cuz of Engrish) when trying to apply my pre-1.40 OC Profile:
Got some defauls HOLES.png


No biggie really. The most annoying "loss" here for me is the fan profiles I created, but nothing I'll lose sleep over. Off to Windows now :)


EDIT2: Proof is in the puddin'
We BClockin.png


And for anyone curious, my (our?) Titanium default speed on Base Clock is 99.8MHz, so this is in fact a a 1.1MHz increase. So YAY IT WORKS! :)
UPDATE: I couldn't wait... Looks like it's just that semi-faux BCLK option. It maxes at 103MHz lol Oh well. That'll at least be good for DDR4-3020 at 2933, or DDR4-3296 at 3200. Guess these Titaniums may lack the hardware for any higher, after all. That or, and fingers crossed, they're just limiting the max for now to play it safe (since the PCIe at 3.0 link starts to flake out around 107MHz)​

HOWEVER, the real test is yet to come... Will I be singing the same AGESA 1.0.0.4a Blues (I couldn't wait... again. I'm indeed singing the "No-3200 Blues" :( At least I'll have BClk to tide me over), or will my system still rock the DDR4-3200 speeds w/o issue? Looks like AIDA doesn't know what Motherboard I have anymore though. Not entirely sure if that's down to my modded BIOS or not. CPU-Z reads it fine. Nevertheless, I'll flash back to 1.40 unmodded tomorrow, dump it like I did with 1.10, and then mod that instead of the one I downloaded.

Tune in next week tomorrow night!
Same Bat time! (if not earlier)
Same Bat channel!
 
Last edited:
ok so I have some what successfully updated my UEFI. everything appears ok and i was able to dial in my OC with ease but there is a issue, not long after getting everything up and running I restarted to get back and change some fan setting in the BIOS and I have been getting this ever since

20170415_020316[1].jpg


every time I reset or power up I get pulled in to the UEFI to this screen, I'm not sure what it is going on about ROM Image?, I have done some reading and it seems to be a issue with using XMP speeds, anyone have any ideas let me know pls



EDIT: Never mind my over clock is totaly not working all though it show it was working in the BIOS :\
going to look in to this more might just go back to F5d as it worked
 

Attachments

  • 20170415_014124[1].jpg
    20170415_014124[1].jpg
    129.4 KB · Views: 48
Last edited:
Had to roll back to F5d, could not get a overclock to work with XMP enabled even thought the XMP profile seems to be working as in i had 1600MHz DDR running with F5 but any over clock i set would fail and go to the default clock, will have to look in to this further but for now I'm going to use what works
 
Got the 1.40 BIOS for MSI Titanium. So far so good.

Was able to set A-XMP to Profile 2 on the Flare X, which loaded as 3200 speed 14 latency. No problems (previously enabling XMP w/ my old Intel RAM would cause system to fail POST).

Also, boot time was reduced from 45 seconds to 27 seconds, which is still somewhat longer than I'd prefer but acceptable.
 
Alright so a slight update with the Titanium on my Modded v1.40 BIOS, not the full update like I wanted to make...
As I mentioned in earlier post, the limit is 103MHz which is a bummer and 3200 no longer works either on my TridentZ CL15 kit. Now I haven't tried much to get 3200 working, but I'll throw timings, voltage and monkey with the ProcODT.
So in addition to that two things. First, a BCLK of 103 doesn't seem to work. Whether that is due to 2933 timings being tighter and my RAM just not jiving with that, or due to it being at the ragged edge of what the crappy BClk hardware is capable of (hence why 103 is the limit?). It'll POST fine, and work in the BIOS, but the Storage controller doesn't quite like it, and crashes completely before Windows has a chance to load beyond the BootLoader which you enter the Troubleshooter, or right when Windows logo shows and the system restarts. 102 works fine so far, and as such is yielding DDR4-3000, but memory performance is quite a ways down from 3200.

As it stands right now, it's looking more likely that I'll be going back to 1.10, unless I can managed to somehow rip out this AGESA and replace it with whatever is in 1.10... However, I suspect that's beyond my skillset, but we'll see. The memory latency at 3000 is 75.8ns vs my 3200's 76.9ns (same timings), and that could completely be down to the CPU clock being 90MHz higher. (LOL A Sempron 2600+ on an ASRock board with a GeForce6100 chipset and SINGLE CHANNEL DDR400 get 77ns)

EDIT: Before-bed update... Flashed back to 1.10 just to see if Microcode would go back. I can confirm it DOES flash the old Microcode on our Titaniums :)
More interestingly... I flashed a more heavily 'unlocked' (nothing really special) modded 1.40 and let Auto timings take over this time which meant 16-15-15-35... and 3200 worked! Though a huge eyebrow raiser is my L3 speeds >_> (and no, I forgot to flash factory 1.4 and then dump it, so motherboard unfortunately still reads Unknown)
cachemem bios 1.40.png


I mean sure the latency is down, as advertised, but my god that L3 is like 300GB/s lower!! Whatever they've done, is very "Yikes" worthy, as they've mucked something up...

EDIT 2: Seems to just have been a fluke with that boot. I restarted and while I was doing so put BCLK up to 102 again to see if it'd be OK at 3200, which it is, and reran tests. L3 is back at it's 374GB/s.
 
Last edited:
starmax said:
I installed this today and got 2666 working in my G.Skill Ripjaw 3200 V series. Nothing higher would post. I was previously stuck at 2400.
Click to expand...
bobzdar said:
New asus prime b350m-a bios (0604) with updated agesa out:

http://dlcdnet.asus.com/pub/ASUS/mb/SocketAM4/PRIME_B350M-A/PRIME-B350M-A-ASUS-0604.zip

I'll try it tonight and see if I can get some more speed out of my ram.
Click to expand...

I installed tonight and now am at 2666 on the ram, so steady improvements. Launch bios was 2133, then a couple of days later a new bios got it to 2400 with looser timings, then 2400 cas15, now 2666cas15. No post at 2933cas15. I haven't tried 2933 at looser timings (and probably won't bother) but so far it looks to be rounding into shape.
 
gigaxtreme1 said:
Wasn't there something about timings defaulting to round numbers for stability?
Click to expand...
Wouldn't, in this case, there need to be decimal values to achieve "round numbers"? Either way, I get what you're referring to, and I really really wish I knew what the nonsense is behind these systems not keeping Odd tCL numbers. Apparently they work on AMD's officially supported speeds only (as per the Gigabyte Guide's last page), so 2133, 2400, 2666 all will function at 15-15-15-35 (for example). Whereas 2933 and 3200 both will actually be running at 16-15-15-35. It's actually bugging the hell out of me that it does that. heh Mainly because I bought a 15-15-15-35 DDR4-3200 kit, and it seems silly having 16-15-15-35 :p Running CL14 at 3200 worked fine... until this new and shitty AGESA, which doesn't seem to like CL14, but works fine at CL16. I mean CAS is an important timing, so having to skip every other number kinda sucks when you're a low-level enthusiast who is out to squeeze every last ounce of performance. :D
*clears throat*
Join me, [H]ard brothers and sisters, we must Ryze up, and demand AMD free our Odd CAS timings!! lol (Yes, it was bad, and, yes, I feel bad. Don't worry! ;))
 
I don't know how the microcode works in any way., but hopefully they get a handle on all these switches that are employed by it in the next update.I am going to buy a Ryzen system once the bugs are ironed out or maybe the next chipset version, Like an X390 or somesuch.
 
Formula.350 said:
Wouldn't, in this case, there need to be decimal values to achieve "round numbers"? Either way, I get what you're referring to, and I really really wish I knew what the nonsense is behind these systems not keeping Odd tCL numbers. Apparently they work on AMD's officially supported speeds only (as per the Gigabyte Guide's last page), so 2133, 2400, 2666 all will function at 15-15-15-35 (for example). Whereas 2933 and 3200 both will actually be running at 16-15-15-35. It's actually bugging the hell out of me that it does that. heh Mainly because I bought a 15-15-15-35 DDR4-3200 kit, and it seems silly having 16-15-15-35 :p Running CL14 at 3200 worked fine... until this new and shitty AGESA, which doesn't seem to like CL14, but works fine at CL16. I mean CAS is an important timing, so having to skip every other number kinda sucks when you're a low-level enthusiast who is out to squeeze every last ounce of performance. :D
*clears throat*
Join me, [H]ard brothers and sisters, we must Ryze up, and demand AMD free our Odd CAS timings!! lol (Yes, it was bad, and, yes, I feel bad. Don't worry! ;))
Click to expand...

I've never been able to run my RAM at 3200 C14 due to cold boot issues. Quite happy with 3200 C16 to be honest. The new AGESA code significantly reduced latency anyway so performance @ C16 is probably better now than it would have been at C14 with the old BIOS/AGESA.
 
gigaxtreme1 said:
I don't know how the microcode works in any way., but hopefully they get a handle on all these switches that are employed by it in the next update.I am going to buy a Ryzen system once the bugs are ironed out or maybe the next chipset version, Like an X390 or somesuch.
Click to expand...

I mean, there aren't really any 'bugs'. Yeah, people aren't hitting extremely high o/c's on the ram (limit seems to be around 3400 at tighter timings now, which is pretty darn fast), but official speed is 2133 and pretty much everybody is well beyond that at this point. I haven't seen any actual bugs (like devices not working, or instability at stock speeds or anything like that), so you can sit on the sidelines waiting for 2-300mhz more ram speed or enjoy the system now. Or spend a little extra for ram on the QVL list that runs at rated speed and not worry about it.

The only reason I see for waiting is if you need 64GB of ram at high speeds for some very specific use case (not even sure what that would be), or if you're waiting for an intel response.
 
On a Ryzen build I just tossed together, the first thing I did once posted was upgrade the bios to F3 on the Gigabyte K7. Then I set the ram to the XMP profile, which did 3000 (or 2933 I guess), and 15-17-17-35. Corsair 16GB kit, 3000 15-17-17-35.
No issues.
 
bobzdar said:
I mean, there aren't really any 'bugs'. Yeah, people aren't hitting extremely high o/c's on the ram (limit seems to be around 3400 at tighter timings now, which is pretty darn fast), but official speed is 2133 and pretty much everybody is well beyond that at this point. I haven't seen any actual bugs (like devices not working, or instability at stock speeds or anything like that), so you can sit on the sidelines waiting for 2-300mhz more ram speed or enjoy the system now. Or spend a little extra for ram on the QVL list that runs at rated speed and not worry about it.

The only reason I see for waiting is if you need 64GB of ram at high speeds for some very specific use case (not even sure what that would be), or if you're waiting for an intel response.
Click to expand...
I believe AMD's official supported max speed is DDR4-2667, if not, I know it's 2400.

dave343 said:
On a Ryzen build I just tossed together, the first thing I did once posted was upgrade the bios to F3 on the Gigabyte K7. Then I set the ram to the XMP profile, which did 3000 (or 2933 I guess), and 15-17-17-35. Corsair 16GB kit, 3000 15-17-17-35.
No issues.
Click to expand...
For the record, it'll be running at 16-17-17-35, not 15 (as I mentioned above). Odd-numbered I know work on 2133, pretty sure I tested them working at 2400, and I think 2667 but that's just me speculating based on MSI new BIOS option with built-in profiles.
 
STILL stuck at 2133 on my x370 Titanium after the recent 1.4 BIOS with Dominator 2400 10-12-12-28. Nothing will work. Neither xmp nor manual voltage and timing adjustments. Somehow during all these flashes my command rate became set at 2t as well, and there isn't even an option in the bios to change it wtf?
 
Earlier than expected.

ASUS PRIME B350-PLUS BIOS 0609
Improve memory stability

I'm at work, can't try it here.
 
Asrock just released this for my killer ac/sli
2.30 4/26/2017 Windows 6.57MB Update Agesa Version string "SummitPI-AM4 1.0.0.4a"

Wish they had better descriptions
because their older one said
2.00 3/31/2017 BIOS 5.84MB 1.Update AGESA to 1.0.0.4a
2.Precision Boost - Adjusts clock speeds in 25MHz increments in BIOS setup

So I have zero idea what that bios update will do
 
I got the same thing a couple times on my Gigabyte GA-AB350-Gaming 3. I just reflashed the BIOS every time it happened. I was testing the limits of my OC with various voltages and just figured that it was a result of the testing. It never happened before F6 and I am using the XMP profile for Flare X DDR4-3200. It's not recurred since I settled on an OC and voltage.

Kato1144 said:
ok so I have some what successfully updated my UEFI. everything appears ok and i was able to dial in my OC with ease but there is a issue, not long after getting everything up and running I restarted to get back and change some fan setting in the BIOS and I have been getting this ever since

View attachment 22008

every time I reset or power up I get pulled in to the UEFI to this screen, I'm not sure what it is going on about ROM Image?, I have done some reading and it seems to be a issue with using XMP speeds, anyone have any ideas let me know pls



EDIT: Never mind my over clock is totaly not working all though it show it was working in the BIOS :\
going to look in to this more might just go back to F5d as it worked
Click to expand...
 
bobzdar said:
I mean, there aren't really any 'bugs'. Yeah, people aren't hitting extremely high o/c's on the ram (limit seems to be around 3400 at tighter timings now, which is pretty darn fast), but official speed is 2133 and pretty much everybody is well beyond that at this point. I haven't seen any actual bugs (like devices not working, or instability at stock speeds or anything like that), so you can sit on the sidelines waiting for 2-300mhz more ram speed or enjoy the system now. Or spend a little extra for ram on the QVL list that runs at rated speed and not worry about it.

The only reason I see for waiting is if you need 64GB of ram at high speeds for some very specific use case (not even sure what that would be), or if you're waiting for an intel response.
Click to expand...

How about 32GB on 4 dimms????? 16 GB dimms are all dual rank. I can't go above 2400mhz on 4 eight GB dimms that are Samsung B-die single rank.
 
os2wiz said:
How about 32GB on 4 dimms????? 16 GB dimms are all dual rank. I can't go above 2400mhz on 4 eight GB dimms that are Samsung B-die single rank.
Click to expand...

What about it? Jedec max is 2133 for ddr4, so you're technically overclocked above that which is why it's an xmp(intel) or docp(amd) profile. I think most stuff is working and has been working at 2133, which is what is non-oc spec. That was kind of my point. Yeah, you can't push the memory to high speeds in a lot of cases, but that doesn't make it a buggy platform as technically you're overclocking...

I get it, it's not as mature as Intel, but I certainly never expected it to be. That doesn't make it buggy, I haven't seen any realy bugs that are platform wide - yes some motherboards have/had issues, but that's not a platform issue as it's limited to specific models, that's a motherboard issue.
 
bobzdar said:
What about it? Jedec max is 2133 for ddr4, so you're technically overclocked above that which is why it's an xmp(intel) or docp(amd) profile. I think most stuff is working and has been working at 2133, which is what is non-oc spec. That was kind of my point. Yeah, you can't push the memory to high speeds in a lot of cases, but that doesn't make it a buggy platform as technically you're overclocking...

I get it, it's not as mature as Intel, but I certainly never expected it to be. That doesn't make it buggy, I haven't seen any realy bugs that are platform wide - yes some motherboards have/had issues, but that's not a platform issue as it's limited to specific models, that's a motherboard issue.
Click to expand...

Don't be a putz. Who here is not overclocked? I had ddr3 2400 running on my Crosshair V with Piledriver, that was overclocked. I don't need a bible thumper to preach to me. It is NOT an unreasonable expectation to run higher than 2400mhz when this memory situation sorts itself out. There are some people with 4 dimms running 2933mhz and even 3200mhz if you bother to read all the forums say on overclock.net They are not the majority, but my ram is specked the same as theirs , just do not know all the tricks of the trade that they are employing.If they are using b clock that is a no go on MSI Titanium.
 
MSI just released their 1.5 official bios for the X370 Titanium board. It is supoposed to improve memory compatibility. I have not tested it on 4 dimms yet and I have been running at 3200mhz memory speed for the last 6 weeks since I installed Ryzen certified b-die memory from G. Skill Flare X. But as i mentioned with beta 1.52 a few days ago,now the official1.5 update has allowed a 14.5 % improvement in fps. This confirmed on cinebench 15.038 open GL test. Now where is the troll Juanrga, who has refused to address this improvement which now puts Ryzen fps performance startlingly close to an I7 7700k?
 
Gideon said:
Ahh damn I thought it was a tile memory game :mad:
Click to expand...

True enough ,but it has to translate into fps improvements of some kind in the vast majority of games that employ open gl support. I am NOT a big gaming freak. Don't dare tell me it means nothing. You test it out . I reported the information that does NOT obligate yo go out and buy a host of games I really have no interest in. My gaming interest is limited to strategy games.
 
Last edited:
I updated to MSI 1.5 BIOS (from 1.4) and there is literally no performance difference, at least in Unigine Superposition. I got practically the same score (actually one point less).
 
  • Like
Reactions: xorbe
like this
cybereality said:
I updated to MSI 1.5 BIOS (from 1.4) and there is literally no performance difference, at least in Unigine Superposition. I got practically the same score (actually one point less).
Click to expand...
that bench has very little impact with CPUs. So it isn't a good indicator of CPUs at all.
 
Gigabyte x370 gaming k5
Ryzen 5 1600 (USA/Malaysia)
F1 Firmware (initial): can set multiplier up to 37-ish without adjusting voltages, ram maxes at 24 multi/2600.
F2 Firmware (AGESA 1004a, misc): any CPU multi above 32 reqs more volts (VDD), but mem multi can go up to 26.66 stable now, 2800.
 
os2wiz said:
Stellmeit means nothing more thanananomaly?An anomaly that consistently happens in cinebench???
Click to expand...
some times what you type makes no sense. So I will state this and maybe it will clear up whatever you were trying to convey to me.

The Superposition bench has little to no impact on the CPU so it would not likely show positive or negative results of a bios update.

Cinebench is a much better indicator of CPU performance and likely will show CPU increases as a result of a bios update if it does in fact increase CPU performance.
 
I also tried 3DMark Fire Strike Ultra. It went from 18,224 to 18,938 in physics test, about a 4% increase.

However, the old score was from when I first built the computer and I have done a some other changes and tweaks since then (new RAM, Windows Update, video drivers, AMD chipset drivers, Ryzen Balanced Profile, etc.) so it could be anything.
 
You must log in or register to reply here.
Back
Top