My RAM is now clocked to 3466 stable at CL18 with no more effort put in than enabling XMP and setting it from 3600 to 3466. Everything seems stable so far and has passed all my stress tests etc so after a few more days of testing I will get into the overclocking side of things and see how far I can push the CPU on this BIOS as well as my RAM. 

Any Taichi owners wondering if the Beta is any good, it seems to be the most stable version I have tested, even allowing higher RAM speeds than previous versions on my Corsair Vengeance LPX DDR4 3600 kit.

RAM: CMKGX4M2B3600C18R (R denotes Red for the color, it is identical to modules of the same number ending in other letters)

GPU: Gigabyte GTX 960 G1 Gaming 2gb

CPU: Ryzen 1600x

That should be all the hardware that can make a difference to stability. My drives are all SATA and none of my M.2 slots are populated presently. 

I will update once I have done more testing and overclocking  

My idle temps in OS are sub 30c so it isn't anything to be concerned about. The UEFI usually shows higher than idle temps, typically in my case around 50c which I would consider normal due to the load the UEFI puts on the CPU. 65.5c however can't be accurate. 

u dont have issues with ram latency/slowpoke bandwith?

mine fatal1ty doesnt fare well with that 1.0.0.2 agesa (4.63)

latency around 75, bandwith in 40GB ranges at 3400mhz hmm

so im stuck with 4.61 (68ns with ~53gb read/write)

Cuba1hr, sadly that seems to be the trend with AM4, it is extremely dependent on hardware configurations. Something as simple as different RAM chips can make a huge difference. My kit is Samsung B die. If my tests follow the same pattern as Kerberos then I may not be as lucky as it seems, those scores are worse than my kit ran at 2800 on the launch BIOS my board came with.

I have only updated my BIOS about 12 hours ago, much testing still to do. I can say that it is more stable (without needing to tweak the UEFI first) than the 3.0 version I was running. I needed to alter quite a few settings with that to get my RAM stable at 3200 and the rest of my system bug free. 

4.64 appears to contain the newest AGESA available for the Taichi. I decided to give it a go and see where I land with my config. Still have a lot of fine tuning to do and then I will get onto benchmarks etc to compare to my previous settings/BIOS version. If I find any interesting workarounds/fixes I will post them here. 

Sadly it does look like AMD have left us in the cold with the 4xx chipsets being an improvement over 3xx and no real improvements to be had with newer UEFI/AGESA versions. I guess they need to sell the new boards/chipsets. I hope I am wrong but so far, I have not seen any evidence to prove me wrong. Some reviews even show first gen Ryzen performing better and achieving higher RAM frequencies on the X470 chipset. It does at least seem RAM dependent but even if it is only a few cases, it is a little disheartening. 

I will have more for you all in the next few days. So far, I am not dissapointed but that all depends on my benchmark results. Out of the box stability at the cost of performance is not worth it, I would rather have to tweak things a little and have better performance. 

With beta BIOS names I do know that if a version is L1.43 then it is based on P1.40. Sometimes an earlier branch was found to be more compatible with intended changes. 

Again, this is just my own speculation based on some rather limited understanding of software development but it does seem to fit the BIOS naming. 

As for overclocking, so far as all the reviews seem to be discovering, overclocking the new 2000 series CPUs is pointless. AMD's boost clock system works incredibly well and allows for much higher frequencies on 1 - 3 cores than you can achieve across all cores. This means that for gaming, you are actually far better off leaving things stock as you will get better per core performance with the boost.

If you are running highly threaded productivity tasks then it is more worth overclocking but even then, the gains are not all that great, particularly at the cost of higher thermals and power consumption. In short, the new Ryzen+ CPUs are very good right out of the box, great for more casual users but a little disappointing for those of us that like to fine tune things. That said, you can still tweak all the different performance states and get more out of the system that way.....

The beta includes an AGESA update. Ryzen first gen has been both a train-wreck and a resounding success. It does seem that a lot of the issues faced with Ryzen 1k are hardware rooted (chipset, IMC etc) so many of the fixes in the 400 series boards will likely never come to 300 series boards, at least, they won't work the same way. While the AGESA is meant to be common across all AM4 platforms it can't correct hardware level issues. 

All of us on Ryzen first gen are in the same boat. The issues are not specific to ASRock, from what I have seen all manufacturers have struggled with getting AM4 stable, some with better and some with worse success. Worse still is that which boards work best change from one BIOS update to the next. This is why I recommend that if you find a board and BIOS that works as you need, stick with it. I can only hope that we at least see a stable BIOS for every 300 series board, even if there are still quirks (RAM speeds etc).

I haven't got much time today for more testing but I will be going through 4.64 over the next few weeks.

1. RAM is no more or less stable (with my kit) than 3.0 I was using before the update.

2. RAM latency and R/W speeds are more or less identical to what I achieved with 3.0

3. RAM speeds and timings achievable with the Beta are the same as 3.0 (3333 max stable @CL18)

4. Benchmarks show absolutely no change

I was stable at 3466 on my RAM for about a week then began experiencing the occasional blue screen. With some looser timings I can stop those but at CL18 my timings are not great to begin with so I just dropped to 3333 and set my timings to CL16. 

Bugs?

CPU temps read stupidly high in the UEFI. I am sure there are more but that is the only one that stands out as a potential issue (for some users). A lot of people get scared when they see high temps in the UEFI so this is something that needs addressing IMO. 

Still fine tuning and tweaking. I have yet to do any CPU overclocking, honestly I generally run stock on this system with my poultry GTX 960 there really isn't any point in overclocking anyway.    

I am not familiar with 3.0, in shop they installed 3.20 I think when they tested board with my 1700X and my Corsair - they have said it was stable  on 2933 (tested for 1-2 hours I guess) and then shipped the components to me. I had troubles with RAM on that version. Soon there was 4.40 bios and after so many tries on that one it was stable for earlier mentioned 13 days (windows update restarted my computer then).

I cannot reproduce that situation with 4.64 or 4.70 and cannot go back with 2700X. Now the PC is running for 2 days 15hours, raised DRAMV for 0.01V more (1.38V, shown 1.40 in picture), tomorrow morning is the critical period, if the patient survives early morning, it will live:)...but I doubt, every time it's the third day BSOD, just after problem with NTP update...I think. I have seen that in my event viewer before "unexpected restart" few times, don't know if it is correlated, but error more than 90% of time is MEMORY_MANAGEMENT and few IRQL_LESS_OR_NOT_EQUAL.

Here are problematic  MIN & MAX values in CPUID HWMonitor and HWInfo on BIOS 4.64:

https://imgur.com/a/fRHCu2B" rel="nofollow - https://imgur.com/a/fRHCu2B

@Cuba1hr

Have you tried running the RAM at 2133 or 2400 MHz with all other settings on auto? If so, does your PC crash at these RAM speeds also?

When running the RAM at 2800 or 2933, have you tried changing the command rate to 2T and disabling Geardown mode?

http://www.w3.org/2000/svg" rel="nofollow - http://www.w3.org/2000/svg" >

http://www.w3.org/2000/svg" rel="nofollow - http://www.w3.org/2000/svg" >