new beta bios for X399 is up on ASRock site
Printed From: ASRock.com
Category: Technical Support
Forum Name: AMD Motherboards
Forum Description: Question about ASRock AMD motherboards
URL: https://forum.asrock.com/forum_posts.asp?TID=8670
Printed Date: 22 Dec 2024 at 11:44am
Software Version: Web Wiz Forums 12.04 - http://www.webwizforums.com
Topic: new beta bios for X399 is up on ASRock site
Posted By: lowdog
Subject: new beta bios for X399 is up on ASRock site
Date Posted: 12 May 2018 at 3:23pm
|
Beta bios L1.91E based on Agesa 1.0.0.6 for X399 is up on ASRock site. Testing atm so will see how it goes. Straight up XMP 3200Mhz is still a fail, it needs to be tweaked. With 4 x 16GB b-die modules the bios setting for ram and RttWr specifically need to be set to 3 to gain stability at speeds over 2933Mhz. If RttWr is not tweaked to 3 it defaults to auto which is disabled and the system is flakey at ram speeds over 2933MHz Below settings are proving stable so far but still early days.  ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Replies:
Posted By: ssateneth
Date Posted: 12 May 2018 at 3:59pm
|
LINK!!!!
http://asrock.pc.cdn.bitgravity.com/BIOS/TR4/X399%20Taichi%28L1.91E%29ROM.zip" rel="nofollow - http://asrock.pc.cdn.bitgravity.com/BIOS/TR4/X399%20Taichi(L1.91E)ROM.zip ------------- MB: X399 Taichi, UEFI L3.32 VGA: EVGA 1080 Ti FTW3 PSU: Seasonic PRIME 1300 W PLATINUM CPU: AMD Ryzen Threadripper 2950X MEM: 4x16GB, 3600MHz G.Skill Trident Z F4-3600C17Q-64GTZKW |
Posted By: lowdog
Date Posted: 12 May 2018 at 4:01pm
|
Fat Pro http://www.asrock.com/mb/AMD/Fatal1ty%20X399%20Professional%20Gaming/index.asp#BIOS" rel="nofollow - http://www.asrock.com/mb/AMD/Fatal1ty%20X399%20Professional%20Gaming/index.asp#BIOS ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: TANWare
Date Posted: 12 May 2018 at 6:15pm
|
Your screen shot is unreadable as it is small only. ------------- x399 Taichi, 1950x, 32GB 3466 DDR4 2x500GB 960EVO,Asus 1080TI http://www.pcambrosia.com/images/PassmarkEVO960_e.jpg" rel="nofollow - Passmark |
Posted By: JayB33
Date Posted: 12 May 2018 at 6:21pm
|
The forum's auto resizing it. Right click - view image or open image in new tab and use the magnifying glass. Glad to see there's a new BIOS.
|
Posted By: lowdog
Date Posted: 12 May 2018 at 6:22pm
|
Click link then resize. https://imgur.com/zhh8Nwq" rel="nofollow - https://imgur.com/zhh8Nwq Or as JayB33 above mentioned, right click image in first post and open in new tab or ask corona etc. ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: TANWare
Date Posted: 12 May 2018 at 7:01pm
Thanks, with IE I do not get a new tab option. I as well seem to have rttwr as disabled but I could not find that setting in bios.
 ------------- x399 Taichi, 1950x, 32GB 3466 DDR4 2x500GB 960EVO,Asus 1080TI http://www.pcambrosia.com/images/PassmarkEVO960_e.jpg" rel="nofollow - Passmark |
Posted By: TANWare
Date Posted: 12 May 2018 at 7:32pm
|
Only issue so far is the fan tuning somehow disables controlling fan1 speed. I use this jumper for the Enermax pump and it just ran a real slow speed. Went to UEFI defaults and then just manually se up fan speeds and all is fine again. Just though I would throw that out there is others are having problems. ------------- x399 Taichi, 1950x, 32GB 3466 DDR4 2x500GB 960EVO,Asus 1080TI http://www.pcambrosia.com/images/PassmarkEVO960_e.jpg" rel="nofollow - Passmark |
Posted By: Meluto
Date Posted: 12 May 2018 at 10:56pm
|
" rel="nofollow - It seems stable, my discoveries so far: 1.- the dram fsb is a little elevated in comparison with 2.0, from 1399 to 1407 mhz (2 x DDR4 3200 (PC4 25600) Desktop Memory Model F4-3200C16Q-32GTZR that STILL CANNOT GO OVER 2800 mhz) 2.- the core speed of the 1950x cpu oc'ed to 3900 mhz is also a little higher, from 3899 - 4002 mhz to 3900 to 3925 mhz) 3.- Dare I to say the temperature reported seems a little lower?, 1- 2 C° degrees, might be a placebo effect though XD 4.- Fans, usbs, sdd's, dvd rom, all are recognized and work well. 5.- Booted into W10 x64 without any problems My 64 gb of 3200 mhz rated memory (see up for details) will not boot beyond 2800, but at that speed is rock solid stable, paired with the cpu oc at 3900 marks >3240 points in cinebench and can run prime95 large fft endlessly, temps keep around 60 C° using ekbw 3 fans radiator kit with, of course, using XSPC Raystorm Neo Waterblock and metal tim (ambient temp is high, around 26-27 C°) At the end of the day, this new bios might not be very useful if the XMP mem settings are not better implemented, but thanks anyway for the effort to the Asrock team ------------- Taichi x399 MB 64GB RAM G.Skill - 2x F4-3200C16Q-32GTZR kits Threadripper 1950x EKWB water loop, XSPC CPU block 550MX 1Tb ssd & 12Tb hdd 2x EVGA SC BE 1080 ti SB AE-5 sound Phantek Enthoo Primo case |
Posted By: MNMadman
Date Posted: 13 May 2018 at 1:34am
|
Works great for me, though P2.00 was working fine as well. Wouldn't allow me to use BIOS Flashback to install it though, for whatever reason. Flashed from BIOS fine. Had no problems with XMP timings on my G.Skill F4-3200C14Q-32GTZRX RAM with P2.00. Still have no problems on L1.91E. Currently running the 3232 Fast preset timings on Ryzen DRAM Calculator 1.1.0 beta 2, was stable on P2.00 and seems stable on L1.91E. Will be verifying that with my usual stress tests. CPU is 1950X @ 4.141GHz using 1.36875v (41 x 101, fixed overclock, fixed voltage) 16c/16t (SMT disabled), RAM at 3232MHz, SOC at 1.05000v fixed, CPU and SOC LLC at Level 2. Using Socket interleave but will be testing Die and Channel as well. |
Posted By: lowdog
Date Posted: 13 May 2018 at 5:16am
|
@ TANWare your modules are more than likely 8Gb single rank so I think if you use the dram calculator you will see that RttWr is disabled at default and doesn't need to be set. My modules are 16GB dual rank so they will require different values under the Data Bus Configuration settings in bios. ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: TANWare
Date Posted: 13 May 2018 at 8:09am
|
Ok then, RttWr does not even show then. ------------- x399 Taichi, 1950x, 32GB 3466 DDR4 2x500GB 960EVO,Asus 1080TI http://www.pcambrosia.com/images/PassmarkEVO960_e.jpg" rel="nofollow - Passmark |
Posted By: MNMadman
Date Posted: 13 May 2018 at 9:27am
Well, it turns out L1.91E isn't as stable as P2.00 for me. I tested P2.00 to over 16,000% in RAM Test with no errors, but L1.91E had two errors in over 19,000%. That's at the Dram Calculator's 3232 Fast preset.
It's there, deeper in the DRAM Timings, as are the other settings in the Termination block. I just set it again this morning for the beta BIOS. Edit: It's in OC Tweaker > DRAM Timing Configuration > Data Bus Configuration |
TANWare wrote:Posted By: lowdog
Date Posted: 13 May 2018 at 9:30am
" rel="nofollow -
What program is "Ram Test" ??? Does it run on Win 10 ??? I just use HCI Memtest ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: MNMadman
Date Posted: 13 May 2018 at 9:40am
http://www.overclock.net/forum/18051-memory/1644432-great-new-memory-stability-tester-ram-test.html#post26511421" rel="nofollow - Link |
Posted By: SoniC
Date Posted: 13 May 2018 at 9:06pm
|
Hell yes! Now I am on 4x16GB @3200 CL14! Do NOT, I repeat --- do NOT touch the "memory clear" option in UEFI (One of the recommendations in Ryzen Calculator). This has lead to boot loops and I thought cause of too high RAM clocks which was not the case.  ------------- -=SoniC=- TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280, AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU |
Posted By: SoniC
Date Posted: 13 May 2018 at 11:12pm
Ten euros for a small piece of software? No, thank you. Maybe if someone is testing RAM stability on a regular basis... but for such a small piece of software it is a hefty sum. ------------- -=SoniC=- TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280, AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU |
Posted By: MNMadman
Date Posted: 13 May 2018 at 11:47pm
" rel="nofollow -
As always, it's up to you. I consider system stability to most definitely be worth that paltry sum, especially since I'll be using it for years to come. Of course, I've also purchased AIDA64 Extreme, the entire FutureMark (now UL) suite, and other software. I also build a new liquid-cooled system every one to to years -- the longest I've had a system since I started building them is 26 months. So yeah, 10 euros is nothing. ----- As to your other post, I haven't seen any consequence to enabling Memory Clear. I consider clearing the memory after training a good thing. Others have it disabled and it works fine. So like pretty much every setting in the BIOS, there is no universal "do it" or "don't do it". |
Posted By: SoniC
Date Posted: 13 May 2018 at 11:48pm
|
There are a lot of free alternatives (especially to this memtest). Slower, yes, but I am not in a hurry. ------------- -=SoniC=- TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280, AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU |
Posted By: MNMadman
Date Posted: 13 May 2018 at 11:53pm
As I said, up to the user whether or not it's worth it. I also do a lot more stress testing to make sure overclocks are stable than most people do. I don't ever just use one program to test anything. |
Posted By: lowdog
Date Posted: 14 May 2018 at 1:45pm
|
Ok got 4 x 16GB @ 3333MHz seemingly stable - more testing to come yet but raising ProcODT to 68 from stock 60 helps heaps!!! Prevents the system from sh*tting itself on re boot and boot up so far. EDIT; nah not stable  ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: SoniC
Date Posted: 14 May 2018 at 11:32pm
|
Great to hear lowdog! I will just experiment with improving the timings, 3200 is enough for me for now ;-) ------------- -=SoniC=- TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280, AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU |
Posted By: kgonepostl
Date Posted: 16 May 2018 at 12:30am
|
" rel="nofollow - Anybody able to boot into windows 7 with the new bios? |
Posted By: TANWare
Date Posted: 16 May 2018 at 5:05am
|
I went back to 2.0 as the beta seemed less stable overclocked. I too though on my 3466 ram set the ProcODT to 68.6. Before at default of 60 3200 MHz was rock solid but at 3466 MHz the ram could get flakey and the wireless seemed to always end up with an error. Now over 24 hours and no issues. Always got the IRQ before error before. ------------- x399 Taichi, 1950x, 32GB 3466 DDR4 2x500GB 960EVO,Asus 1080TI http://www.pcambrosia.com/images/PassmarkEVO960_e.jpg" rel="nofollow - Passmark |
Posted By: lowdog
Date Posted: 16 May 2018 at 7:15am
|
" rel="nofollow - [URL=][/URL]Notes from AMD on Agesa 1.0.0.6; we began beta testing a new AGESA (v1.0.0.6) that is largely focused on aiding the stability of overclocked DRAM (>DDR4-2667). We are now at the point where that testing can begin transitioning into release candidate and/or production BIOSes for you to download. Depending on the QA/testing practices of your motherboard vendor, full BIOSes based on this code could be available for your motherboard starting in mid to late June. If you?�re the kind of user that just needs (or loves!) virtualization every day, then AGESA 1.0.0.6-based firmware will be a blessing for you thanks to fresh support for PCI Express® Access Control Services (ACS). ACS primarily enables support for manual assignment of PCIe® graphics cards within logical containers called ?�IOMMU groups.??The hardware resources of an IOMMU group can then be dedicated to a virtual machine. This capability is especially useful for users that want 3D-accelerated graphics inside a virtual machine. With ACS support, it is possible to split a 2-GPU system such that a host Linux® OS and a Windows VM both have a dedicated graphics cards. The virtual machine can access all the capabilities of the dedicated GPU, and run games inside the virtual machine at near-native performance. This is certainly a complicated setup for most users, but I have no doubt that there will be a whole lot of you enthusiastically nodding at this news. We?�re grateful for your feedback and your patience, and we hope the new support for ACS serves you well. What's Next For Memory AGESA 1.0.0.6 officially adds 26 new parameters that can improve the compatibility and reliability of DRAM, especially for memory that does not follow the industry-standard JEDEC specifications (e.g. faster than 2667, manual overclocking, or XMP2 profiles). I'm seeing NONE of these fixes in beta bios L1.91E based on agesa 1.0.0.6 from ASRock!!! This beta bios is no more stable than previous bios 2.00 based on Agesa 1.0.0.4. XMP profiles in ASRock bios 2.00 based on Agesa 1.0.0.4 were unstable and XMP profiles in bios L1.91E based on Agesa 1.0.0.6 are just as unstable! So looks like Asrock is NOT making the effort to incorporate those fixes and added parameters into the new bios based on Agesa 1.0.0.6 AMD put them there for a reason so why do we not get the benefit of them in our bios??? ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: MNMadman
Date Posted: 16 May 2018 at 9:06am
|
Calm down man. This is a beta BIOS, and likely won't be the last one before a full release comes with the 1.0.0.6 AGESA code. For all we know, those 26 RAM parameters aren't exposed in the BIOS -- they may be acting in the background and something nobody will ever see, no matter who makes the board. |
Posted By: SoniC
Date Posted: 17 May 2018 at 12:37am
|
Those params are not visible for us, was the same with AGESA 1.0.0.6 for AM4 -- they (AMD) have introduced over 50+ params but visible were only 10 or 15. BTW I don't quite agree with you lowdog. There are DRASTIC improvements in the memory compatibility I am seeing. I could never, I repeat - NEVER reach 3200 stable, even with Ryzen calc. Now I do without breaking a sweat (with some tuning based on the calc). And now -- it's the fourth day without a reboot with some (heavy) gaming, browsing, encoding etc. and 100% stable. ------------- -=SoniC=- TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280, AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU |
Posted By: lowdog
Date Posted: 17 May 2018 at 4:46am
Perhaps I'm a harsh judge and your right SoniC, I need to give it more time and see how it goes.  But.....The only time I seem stable @ 3200Mhz is if I leave the timings at what XMP sets even though I tweak ProcODT to 68.8 and RttWr to 80. then whenever I run RAM Test it's always error free even after cold boots, awake from sleep or restars....If I set XMP and then tweaked timing as per dram calculator it would run RAM Test error free but then after a cold boot or wake from sleep or entering the bios and re starting, any thing random like that and RAM Test would throw errors yet with XMP timings there has been no such errors. So it seems like the timings dram calcularor were saying to set were not the best. ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: orionbg
Date Posted: 17 May 2018 at 12:33pm
Sooo let me see if I understood you correctly: You are saying that at XMP profile settings (the ones the RAM manufacturer have set for that specific RAM) everything runs properly, but when you try to change those to some other values that runs the RAM out-of-specs, and you got from a third party software that has nothing to do with either AMD, AsRock or the RAM manufacturer, you are experiencing issues... Interesting... What was that you were complaining for again???  ------------- Asrock X399 Taichi, AMD Threadripper 1950X, 32GB G.SKILL TridentZ RGB 3200MHz, 256GB Samsung PM961 nvme, 512GB Samcung 850 EVO, 512GB Intel 5400S, GTX1080Ti FE, Lian-Li PC-O11WGX case, Custom W.Loop |
Posted By: lowdog
Date Posted: 17 May 2018 at 12:54pm
|
Still intermittent errors in RAM Test after some reboots and just had a bluescreen crash so even XMP isn't stable @ 3200MHz on latest beta bios. What can seem perfectly stable and 1000% error free in HCI Memtest @ 3200Mhz both straight XMP or tweaked timings can become unstable after a reboot, it's hit and miss but there is something going on at reboot, cold boot or boot to bios - reboot that is changing something on a bios level and causing instability. ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: wardog
Date Posted: 18 May 2018 at 11:01am
" rel="nofollow -
You DO have Mem Training set to Disabled, right? That is the ONLY thing that would at boot. Well......... No remnants of Ryzen Master Utility hanging around? .
|
Posted By: wardog
Date Posted: 18 May 2018 at 11:03am
|
Too, remember. Every new AGESA release is a ground up rewrite regarding BIOSes. New releases are not something I would consider mature enough to reliably OC on.
|
Posted By: lowdog
Date Posted: 18 May 2018 at 11:23am
|
mem training has been auto and training stopped when I upped ProcODT from 60ohm to 68.8ohm , I'll try with it disabled but don't know if it will make any difference. ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: SoniC
Date Posted: 19 May 2018 at 7:59pm
|
New Ryzen DRAM Calculator 1.2.0 beta2 is out> http://www.overclock.net/forum/13-amd-general/1640919-ryzen-dram-calculator-1-1-0-beta-2-overclocking-dram-am4.html" rel="nofollow - http://www.overclock.net/forum/13-amd-general/1640919-ryzen-dram-calculator-1-1-0-beta-2-overclocking-dram-am4.html ------------- -=SoniC=- TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280, AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU |
Posted By: SoniC
Date Posted: 21 May 2018 at 4:35am
|
I've been playing a little with tightening the timings and here is my result. 3200 v1 was the initial set (safe), 3200 v2 is with tighter ones (fast). Plus I have raised the ProcODT to 68.6 Ohm which helped with the stability or RAM. Also raised the Voltage of RAM from 1.35 to 1.36. SOC is on 1.1V. MEMORY CLEAR (contrary to what the Ryzen DRAM Calculator recommends) is set to AUTO. I have left memory training untouched... with ProcODT 68.6 it boots superfast anyways. 3200 CL14 v1>  3200 CL14 v2>  3200 v2 timings:  To sum things up: FINALLY, a golden BIOS/UEFI for me! THANK YOU AMD and THANK YOU ASROCK.
------------- -=SoniC=- TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280, AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU |
Posted By: lowdog
Date Posted: 21 May 2018 at 4:55am
Yes 68.8 ProcODT definitely helps with booting and stability with dual rank 16GB modules. And agreed new Agesa has helped greatly with stability with high speed ram @ 3200Mhz especially a 4 x 16GB config  ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: MNMadman
Date Posted: 28 May 2018 at 9:55am
|
Went back to P2.00 BIOS. L1.91E added nothing for me, and made my RAM settings less stable. Even the Safe settings stopped being stable. ------------- Heatripper Threadkiller // 1950X @ 4.141GHz 16C/16T // Taichi // 32GB 3232 RAM // // Titan Xp @ 2075MHz/12,000MHz // 2TB+512GB NVMe SSDs // Custom Liquid Cooling Loop // |
Posted By: SoniC
Date Posted: 29 May 2018 at 1:58am
|
" rel="nofollow - MNMadman - sorry to hear... E1.91 is a blessing for me... Running now 4x 16GB DR 3200 CL14 with super-tight timings, 1000000% stable (bought the software you recommended at the end lol... too much hassle with HCI ;-) ). I thought about OCing it to 3400 CL16 maybe... but is it worth it? I already achieved ~97000 MB/s read write and copy in AIDA64. ------------- -=SoniC=- TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280, AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU |
Posted By: lowdog
Date Posted: 29 May 2018 at 6:51am
|
Heard the best way to test memory stability is from a cold boot run memtest to 200% if no errors then shut down comp and do another cold boot (turn power off at the wall I assume) then rum mentest to 200% again, do this procedure 3 or 4 times and chuck in a few war reboots as well and if there is no errors in memtest then should be good. Lots are saying with Ryzen that they can run memtest (HCI Mentest or Ramtest) to over 1000% without error but after a reboot or cold boot then memory testing will fail and give errors ....so worth a try. @ Sonic.....good luck with 3400, I reckon it's a fat chance of getting that speed with DR 16GB sticks especially when using 64GB.....I can run 3333 with 4 x 16GB but it will always give an error with ramtest, usually before it hits 100% testing and I haven't been able to stabilize that speed yet. ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: SoniC
Date Posted: 29 May 2018 at 9:42pm
|
" rel="nofollow - Thanks lowdog, however I really ask myself if it would be worth the hassle... To be frank I was totally fed up with the results of my experiments (till E1.91 and the big success with 3200).... So this time I am asking myself if it would provide any real benefits compared to the current setup and throughput. 3200 CL14 tight timings -- ca 97000 MB/s in read write and copy AIDA64 benchmark vs 3400 (probably) CL16 with looser timings? I mean the infinite fabric would probably run faster but if it would be really noticeable? Anyone have experience in this area? ------------- -=SoniC=- TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280, AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU |
Posted By: lowdog
Date Posted: 30 May 2018 at 4:52am
|
Here you go, 4 x 8GB stick of Flare X; 3333MHZ Safe V1 timings  3466MHz V1 Fast timings  3600MHz V2 Safe timings  ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: lowdog
Date Posted: 30 May 2018 at 4:55am
And here is 4 x 16GB @ 3333Mhz V1 Safe timings ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: ssateneth
Date Posted: 01 Jun 2018 at 4:04am
|
" rel="nofollow - Anyone know how to fix the bug where my clock is ticking faster with a threadripper overclock on 1.91E? I don't use sleep mode and it's affecting my benchmark scores. ------------- MB: X399 Taichi, UEFI L3.32 VGA: EVGA 1080 Ti FTW3 PSU: Seasonic PRIME 1300 W PLATINUM CPU: AMD Ryzen Threadripper 2950X MEM: 4x16GB, 3600MHz G.Skill Trident Z F4-3600C17Q-64GTZKW |
Posted By: Xaltar
Date Posted: 01 Jun 2018 at 5:32am
|
It will probably be sorted with the release version. Bugs like this are fairly common with beta releases. As it is something that only happens when overclocked it probably slipped through the less stringent in house beta testing phase. Thanks for posting the info, I will make sure ASRock see it. ------------- 
|
Posted By: ssateneth
Date Posted: 01 Jun 2018 at 6:58pm
|
thanks @xaltar. Yeah it only appears if the CPU is overclocked past the point of XFR being turned off. I noticed something was up when my AIDA memory and cache scores went down the further I increased core speed then turns out my clock was ticking about 20% faster, making up for the 20% loss in benchmark scores (latency scores were not affected, probably because it's measured some other way) ------------- MB: X399 Taichi, UEFI L3.32 VGA: EVGA 1080 Ti FTW3 PSU: Seasonic PRIME 1300 W PLATINUM CPU: AMD Ryzen Threadripper 2950X MEM: 4x16GB, 3600MHz G.Skill Trident Z F4-3600C17Q-64GTZKW |
Posted By: lowdog
Date Posted: 06 Jun 2018 at 12:03pm
|
P-State overclocking completely broken on this bios as it is on bios 2.0 as well!!! Core clock drops but volts stay fixed and won't down volt!!!!! Will Asrock ever fix P-State overclocking......PLEASE FIX IT!!  ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: MNMadman
Date Posted: 08 Jun 2018 at 9:13am
I didn't have any problems with P-State overclocking on either BIOS version. CPU down-clocks and down-volts as expected. No need to get anything fixed. Haven't done P-State since I switched back to P2.00, but I'll test it again tonight just to be sure. ------------- Heatripper Threadkiller // 1950X @ 4.141GHz 16C/16T // Taichi // 32GB 3232 RAM // // Titan Xp @ 2075MHz/12,000MHz // 2TB+512GB NVMe SSDs // Custom Liquid Cooling Loop // |
Posted By: lowdog
Date Posted: 08 Jun 2018 at 9:16am
Well it doesn't work for me on either 1920X or 1900X ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: MNMadman
Date Posted: 08 Jun 2018 at 9:38am
Just confirmed again that it works. VID goes down to 0.887v and Vcore drops down to 1.112v. This is in the CPU [#0] Node #1: AMD Ryzen Threadripper 1950X: Enhanced section of HWiNFO64. CPU clock drops to 2000MHz minimum but mostly stays at 2200MHz until load is applied. In the ASRock X399 Taichi (Nuvoton NCT6779D) section, Vcore drops down to 1.104v minimum but mostly stays at 1.120v until load is applied. Not sure why yours wouldn't be working, unless you have a BIOS setting wrong.
------------- Heatripper Threadkiller // 1950X @ 4.141GHz 16C/16T // Taichi // 32GB 3232 RAM // // Titan Xp @ 2075MHz/12,000MHz // 2TB+512GB NVMe SSDs // Custom Liquid Cooling Loop // |
Posted By: lowdog
Date Posted: 09 Jun 2018 at 6:44am
" rel="nofollow -
Well, what are your bios settings? Wonder if it's only 1950X that P-state overclocking works properly with??? as both my 1900X and 1920X exhibit the same symptoms of only down clocking frequency but not down volting, voltage stays locked. I'm just setting vcore to offset and changing only P-State 0 frequency value and that's it.....same as I did on X370 Fat pro and it worked with my 1800X on that board so...??.???? ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: MNMadman
Date Posted: 09 Jun 2018 at 12:14pm
" rel="nofollow -
The relevant settings I have are: Vcore set to Offset at 0.22500 Core Performance Boost: Auto (don't know whether or not this matters) Global C-State Control: Enabled P-State 0 set to Custom (a0) and the rest at Auto ------------- Heatripper Threadkiller // 1950X @ 4.141GHz 16C/16T // Taichi // 32GB 3232 RAM // // Titan Xp @ 2075MHz/12,000MHz // 2TB+512GB NVMe SSDs // Custom Liquid Cooling Loop // |
Posted By: lowdog
Date Posted: 09 Jun 2018 at 12:16pm
" rel="nofollow -
Exactly as I have mine set yet no voltage down volting on 1900X or 1920X...just stay static! ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: orionbg
Date Posted: 11 Jun 2018 at 6:05pm
|
" rel="nofollow - My 1950X also downclocks without any issues and the vcore also goes down to 0.887v Have been behaving like that with both BIOS 2.00 and the Beta one. By the way, when you are experiencing that issue, that does the Dr. LED shows? "AA" or "40"
------------- Asrock X399 Taichi, AMD Threadripper 1950X, 32GB G.SKILL TridentZ RGB 3200MHz, 256GB Samsung PM961 nvme, 512GB Samcung 850 EVO, 512GB Intel 5400S, GTX1080Ti FE, Lian-Li PC-O11WGX case, Custom W.Loop |
Posted By: lowdog
Date Posted: 12 Jun 2018 at 6:11am
" rel="nofollow -
Yeah it does show AA and 40 often. ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: lowdog
Date Posted: 12 Jun 2018 at 6:30am
|
" rel="nofollow - I have all 3 of the M.2 slots occupied, not sure if that would make a difference with P-State overclocking???.....also if I overclock BLK to 101 I loose one of the PCI-E M.2 drives, it's failed to be recognised. Otherwise I can only ascertain that the failure of the vcore to down volt with P-State overclocking with my setup is due to using either a 1900X or 1920X......perhaps only a 1950X works properly and Asrock would need to look into this to confirm and rectify this behaviour if it is the case. ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: lowdog
Date Posted: 12 Jun 2018 at 8:08am
|
Just pulled all 3 x M.2 drives and still no down volting with P-State overclocking. ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: orionbg
Date Posted: 13 Jun 2018 at 2:22pm
For me, when it says AA, everything is ok but some times after a long idle session (left the PC for an hour or more) it changes to 40 and then the CPU is locket at 4GHz It is very random.... The interesting thing is that I have disabled all sleep related functions and the only thing that I've allowed it for the monitor to turn off after 15 minutes... ------------- Asrock X399 Taichi, AMD Threadripper 1950X, 32GB G.SKILL TridentZ RGB 3200MHz, 256GB Samsung PM961 nvme, 512GB Samcung 850 EVO, 512GB Intel 5400S, GTX1080Ti FE, Lian-Li PC-O11WGX case, Custom W.Loop |
Posted By: ssateneth
Date Posted: 17 Jun 2018 at 9:23am
|
P-State does not seem to work correctly for me either, but this is on 1.92 (I stopped using 1.91E because of the faster clock ticking issue). I only set core speed and voltage through pstate, leave all over core freq/voltage settings on main voltage/freq page on auto. It does seem to respect the pstate 0 (the highest speeds) but seems to ignore other pstates set or the BIOS setting doesnt match actual setting. It seems to be 4.2ghz > 3.6ghz > 3.36ghz > 2.88ghz when I want 4.2 > 4 > 3.6 > 2.8 > 1.6. voltages are seemingly ignored too. it only does 1.5 > 1.15 instead of 1.5 > 1.35 > 1.15 > 0.95 > 0.8. Too bad "zenstates" program only works with asus boards, no reason it shouldnt work with any x399 board... otherwise it might fix p state problem. ------------- MB: X399 Taichi, UEFI L3.32 VGA: EVGA 1080 Ti FTW3 PSU: Seasonic PRIME 1300 W PLATINUM CPU: AMD Ryzen Threadripper 2950X MEM: 4x16GB, 3600MHz G.Skill Trident Z F4-3600C17Q-64GTZKW |
Posted By: orionbg
Date Posted: 19 Jun 2018 at 7:02pm
|
" rel="nofollow - Version 1.92 ???? Where is this from? ------------- Asrock X399 Taichi, AMD Threadripper 1950X, 32GB G.SKILL TridentZ RGB 3200MHz, 256GB Samsung PM961 nvme, 512GB Samcung 850 EVO, 512GB Intel 5400S, GTX1080Ti FE, Lian-Li PC-O11WGX case, Custom W.Loop |
Posted By: SoniC
Date Posted: 20 Jun 2018 at 1:21am
Old Beta BIOS from the 2.00 times... the beta included the XMP 5-beep fix which 2.00 did not. Might be that these were two branches --> 1.92 beta parallel to 2.00 stable. E1.91 is definitely a newer branch than both of them. ------------- -=SoniC=- TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280, AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU |
Posted By: lowdog
Date Posted: 20 Jun 2018 at 5:56am
Been over a month now since 1.91E beta and still no update either beta or official to address broken P-State overclocking where vcore does NOT down volt when cpu goes idle......happens with both 1920X and 1900X......Asrock are soooooo slow!  ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: Xaltar
Date Posted: 20 Jun 2018 at 6:07am
|
My guess would be that Computex had our guys and gals at ASRock pretty busy getting products ready for reveal etc. Just a guess but I have noticed a slowdown in updates around this time of year. With all the new products soon to drop, hopefully they pick up the pace again. ASRock is a very small group of people compared to their competitors, it's one of the reasons we see such competitive price/value with them. The down side is that things can slow to a crawl when they are busy with new launches/conferences etc. Personally, I prefer the value prospect over fast updates. It may be frustrating at times but generally, updates aren't needed all that often. Ryzen/Threadripper have been the exception to that rule, still so many issues and gen 2 has already dropped.... -------------
|
Posted By: ajc9988
Date Posted: 25 Jun 2018 at 11:35am
|
http://imgur.com/cj7c9Ts" rel="nofollow - http://imgur.com/cj7c9Ts This is the reason I stopped using the beta bios. Same error for each thread. |
Posted By: wardog
Date Posted: 25 Jun 2018 at 1:06pm
hahahaha 8 Year Old software!
|
Posted By: ajc9988
Date Posted: 28 Jun 2018 at 10:02am
Doesn't matter if its old if it works. This often identifies instabilities very quickly. I then follow it up with HCI memtest and with Google Stressapp Test (GSAT). Also, if there is a thread error occurring that didn't occur on any prior BIOS, then there is a problem, is there not? |
Posted By: SoniC
Date Posted: 29 Jun 2018 at 12:15am
No one sane will support an 8+ years old software piece.. especially if the error doesn't necessarily mean instability but can be a problem with W10 & it's updates. I was on 2.00, I was on 1.92 and I am now on 1.91 and for me it is by far the best BIOS/UEFI till now. It has its quirks but stability is definitely not a problem for me (and I am talking about a system that runs 24x7).
------------- -=SoniC=- TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280, AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU |
Posted By: MNMadman
Date Posted: 29 Jun 2018 at 11:15am
|
The first time I tried L1.91E it wouldn't overclock as well as P2.00 did. Even memory stability was affected. I retried L1.91E recently and any overclock at all caused perceptible slowdowns even in basic Windows tasks with nothing else open. Worked great at Auto settings though. Switched back to P2.00 and everything returned to normal. ------------- Heatripper Threadkiller // 1950X @ 4.141GHz 16C/16T // Taichi // 32GB 3232 RAM // // Titan Xp @ 2075MHz/12,000MHz // 2TB+512GB NVMe SSDs // Custom Liquid Cooling Loop // |
Posted By: ajc9988
Date Posted: 29 Jun 2018 at 10:34pm
You do NOT listen well. With version 2.00 on Windows 10 Enterprise build 1803, with that software, I DO NOT GET THAT ERROR. The only difference between getting that error and not getting it was changing the bios AND it only occurred when I did an all core overclock, rather than using stock clocks. What does that mean? It means the beta is doing something incorrect on thread management with an all core overclock. So, you evidently haven't dug in to see what problems, like this one, can be found. To be clear, my 3600 MHz memory overclock did not cause this error and could pass just fine without the all core overclock (slightly tweaked from timings used before this BIOS, but part of the game). That is the only time it appeared. Since it worked before on Windows 10 and with all the updates, but stopped and kicked errors with the beta BIOS and only when overclocking the core speeds on the chip, tell me again how it is the software and Windows 10? Also, if you want an idea of my performance, check this handle (ajc9988) at HWBot to get an idea of what I can do with this board and my hardware. Edit: I didn't give all that info to you prior to this, but did to Asrock when I sent in the comment. But instead of instantly attacking someone, you might try thinking that they have some knowledge and that there is a basis for the statement. You look sheepish with all the information presented. Also, what I said here matches MNMadman's experience with auto vs overclock reactions. He likely was getting thread handler errors or similar types of errors. |
Posted By: SoniC
Date Posted: 30 Jun 2018 at 5:05am
|
I am not sheepish, but overclocking it in a different way. That is why my system is stable. ------------- -=SoniC=- TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280, AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU |
Posted By: ajc9988
Date Posted: 30 Jun 2018 at 10:03am
That is fine, and I can understand that. But an all core overclock is very common. That is why when I saw thread handling errors, I immediately stopped, troubleshot what caused the problem, and identified it as the overclocking on the cores, specifically keeping bclk at 100, then changing the multiplier for all cores. If it did it for one program that ran fine on the prior BIOS, it really is something for them to be aware of, especially approaching about 1 month before TR2. I'm hoping we see a new beta BIOS in the next week or so, that way to give feedback over the next month if we find anything. I do apologize for the more gruff response this morning. Guess the heat is getting to me (still no excuse). But I also hope I gave more information to show it is a legit concern and why. ------------- // 1950X @ 4.05GHz @ 1.2875V // Asrock X399 Taichi // G.Skill 4133 Trident Z @ 3600 CL14 17 17 17 34 48 4x8GB // // EVGA 980 Ti Classy // Custom Liquid Cooling Loop // |
Posted By: lowdog
Date Posted: 30 Jun 2018 at 10:21am
" rel="nofollow -
Just out of curiosity, did you try giving the all core overclock that was stable on bios 2.00 a slight vcore bump ^ when trying bios 1.91E?, perhaps that particular bios (1.91E) need more vcore than the previous bios to be stable. ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: lowdog
Date Posted: 30 Jun 2018 at 10:24am
|
Also most other vendors have Agesa 1.0.0.6 bioses as official out already for weeks now! Asrock is still on the first beta that has been out for 1 months so far without any update.....pretty sloooooow moving if you ask me.  ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: ajc9988
Date Posted: 30 Jun 2018 at 10:26am
To be honest, I cannot remember. After it failed, I remember going back to the 2.00 version. But, because we were headed into summer, around that time I started playing with what voltages worked with which multipliers for full stable in the hotter months so I cannot remember at all. I can probably go back and test in the next couple days. That is a great point. I do think I did try a higher voltage, but since I do not have it logged or clearly remember doing it, I cannot say for sure. Edit: Also, if my voltage looks low in my signature, it is because I paid the premium for the top binning from Silicon Lottery. This chip can do 4.1 on all cores at 1.35V. Also, the beta BIOS has been out since May 11th, which means we are fast approaching two months. ------------- // 1950X @ 4.05GHz @ 1.2875V // Asrock X399 Taichi // G.Skill 4133 Trident Z @ 3600 CL14 17 17 17 34 48 4x8GB // // EVGA 980 Ti Classy // Custom Liquid Cooling Loop // |
Posted By: lowdog
Date Posted: 30 Jun 2018 at 11:05am
|
2 months, yep!, your correct....makes their effort seem even more pathetic. I'll remember Asrocks lame bios support and how bugs over a year old haven't even been addressed let alone fixed.....cough X370 Taichi/Fatality Pro Gaming....bloody disgraceful!! Same is happening with X399.....P-State overclocking with 1900X and 1920X doesn't even work, clock gen non functional, XMP 3200 with Samy b-die still unstable....the list goes on and never gets fixed. ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: lowdog
Date Posted: 30 Jun 2018 at 11:12am
|
" rel="nofollow - If the error with bios 1.91E was happening at stock it would be a worry but if it's only when overclocked it's more than likely a voltage matter? ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: ajc9988
Date Posted: 30 Jun 2018 at 11:45am
I hear ya. I'm pretty sure voltage increase was the first thing I tried after finding it because it makes sense to try it first thing. I just cannot remember with certainty, which is why this weekend I'll try it with 1.35V@3.95GHz. Considering that is stable for 4.1GHz, it takes away all doubt doing it that way. If the problem is gone, then I'll lower voltage until the error occurs, but if it doesn't appear at all, I'll assume it got corrupted last time (which I also think I tried reflashing it to rule that out also, but, as I said, I was playing with a lot during that period, then trying to get in last minute benches before it got much hotter here). Either way, if still there, screenshots will follow. ------------- // 1950X @ 4.05GHz @ 1.2875V // Asrock X399 Taichi // G.Skill 4133 Trident Z @ 3600 CL14 17 17 17 34 48 4x8GB // // EVGA 980 Ti Classy // Custom Liquid Cooling Loop // |
Posted By: antorsae
Date Posted: 30 Jun 2018 at 4:16pm
| I totally agree Asrock support for addressing BIOS issues is unacceptable. I still recommend TR to friends but I will not recommend Asrock MB compared to the competition. |
Posted By: lowdog
Date Posted: 01 Jul 2018 at 9:16am
Exactly!....Asrock are just shooting themselves in the foot, they need to get some good bios programmers on board or else they will keep loosing customers because they can't sort out bios issues that other vendors don't have of at least fix. Honestly it was a first time Asrock buy for me when AM4 and TR4 came out, bought both the Fatality Pro Gaming boards and a X370 Taichi and I will not buy another Asrock motherboard again because their bios is nothing short of sub standard to put it simply. ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: ajc9988
Date Posted: 01 Jul 2018 at 11:23pm
First time Asrock owner, too. But, from the threads, etc., WE HAD LESS BIOS ISSUES THAN ASUS DID. They crank it out regularly, but have a lot of broken stuff in theirs. On offerings from MSI and Giga at the time of release, there really wasn't a good reason to get either of those. With that said, I would like a more open BIOS like ASUS offers. And, if ASUS didn't try to bundle the 10Gbps card with the MB (added $100 to the price and all of my network is setup for 1Gbps, meaning it would go to waste), then they would have been in consideration (although I'm happy I didn't have to go through their BIOS firmware again, was a Maximus VIII Extreme owner). Now, I do agree on getting a BIOS programmer that does what is needed. I don't know if Prema has worked much with AMD, but his work turning crappy Clevo stock BIOS into a working, thriving masterpiece is something I greatly respect. And he is starting to do desktop BIOS, or so I've heard. Forgot his actual name atm, but that is someone that knows how to take input from alpha and beta samples to good overclockers and tweak it to improve behavior and performance. I know Chew was giving feedback for awhile as well, don't know if that is still happening. But, I also read about all the others being broken on AGESA 1.0.0.6 back in May, don't know if it was ironed out or not by now. So let's keep with facts in calling for action. I got this board to get away from ASUS firmware nightmare at times, and on board breakdown, and because giga was sending more volts and blowing up AM4, along with other recent issues historically, and the breakdown on the MSI board. Now, if EVGA did an X399 Dark board, I think I would jump ship (maybe wait until 7nm so that I might get PCIe 4 or 5 next year, but still). ------------- // 1950X @ 4.05GHz @ 1.2875V // Asrock X399 Taichi // G.Skill 4133 Trident Z @ 3600 CL14 17 17 17 34 48 4x8GB // // EVGA 980 Ti Classy // Custom Liquid Cooling Loop // |
Posted By: lowdog
Date Posted: 02 Jul 2018 at 4:51am
|
" rel="nofollow - @ ajc9988 is you memory actually stable at 3600MHz??? have you run HCImemtest + ramtest with numerous warm/cold boots in between to test if booting doesn't cause instability???? No matter what I have tried with my 2 sets Samsung b-dies - 4 x 8GB Flare X 3200MHz and 4 x 16GB Trident Z 3200MHZ I can only have 100% stability @ 3066MHz. The 4 x 8GB combo will run 3200 - 3333 - 3466 - 3600 and show apparent stability @ 3200 - 3333 - 3466 going to 1000% HCIMemtest and ramtest as well but sooner or later after either a warm reboot or a cold reboot either one of those ram test programs will throw an error....so what previously had seemed stable now throws an error - WTF!.....where as the 3066MHz never fails. Same with the 4 x 16GB....perfectly stable @ 3066MHz.....appears stable @ 3200MHZ but eventually after a few reboots an error gets thrown in a ram test program. Got me stumped, no matter what I try it's always the same scenario with both sets of ram and I've tried with 2 x cpu a 1900X and a 1920X and results were the same. How can it run both HCIMemtest and ramtest stable then some how get screwed up after a reboot. ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: ajc9988
Date Posted: 02 Jul 2018 at 5:34am
|
I wrote a long explanation, but yes, it is stable. Between typing and submitting, connection reset and what I typed was lost, but I'm not gonna retype the long explanation right now. Yes I used HCI and GSAT, along with testram5, warm and cold boots. Over 1.075V on SoC and I lose all stability at higher speeds on the ram. 1.0375-1.075 is the sweet spot. I run 1.0375V, but if it ever fails to train, just increase slightly up to 1.075, then change it back down and it goes, no errors. Sorry it isn't the depth on my termination block settings, etc. that got lost for the moment due to reset in connection. ------------- // 1950X @ 4.05GHz @ 1.2875V // Asrock X399 Taichi // G.Skill 4133 Trident Z @ 3600 CL14 17 17 17 34 48 4x8GB // // EVGA 980 Ti Classy // Custom Liquid Cooling Loop // |
Posted By: ajc9988
Date Posted: 03 Jul 2018 at 11:03am
|
" rel="nofollow - Did anyone else see the Level1Tech review of the Asrock X399m Taichi? Even though the mATX board, he mentioned getting a beta BIOS of version 2.35 which fixed the P-state overclocking, etc. Really wish Asrock would either put it out or put out like a 1.92E or something like that soon. ------------- // 1950X @ 4.05GHz @ 1.2875V // Asrock X399 Taichi // G.Skill 4133 Trident Z @ 3600 CL14 17 17 17 34 48 4x8GB // // EVGA 980 Ti Classy // Custom Liquid Cooling Loop // |
Posted By: lowdog
Date Posted: 03 Jul 2018 at 11:11am
got link to review? Been waiting for best part of a year for FUKKKKKIN P-State overclocking to be fixed on X399 Fatality Pro Gaming with 1900X and 1920X FFS!!! ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: antorsae
Date Posted: 03 Jul 2018 at 3:15pm
https://level1techs.com/video/asrock-x399-m-taichi-review-linux-test
|
Posted By: lowdog
Date Posted: 03 Jul 2018 at 3:25pm
|
Thanks ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: TANWare
Date Posted: 03 Jul 2018 at 5:50pm
|
My SOC runs higher here. Even setting to 1.0375 in bios gives 1.045 in HWInfo and with level 5 at load 1.072. With level 1 under load it is 1.088. Before today it was set to auto. ------------- x399 Taichi, 1950x, 32GB 3466 DDR4 2x500GB 960EVO,Asus 1080TI http://www.pcambrosia.com/images/PassmarkEVO960_e.jpg" rel="nofollow - Passmark |
Posted By: MNMadman
Date Posted: 03 Jul 2018 at 11:31pm
Mine is similar. Set 1.00000 and it shows up as 1.02500 in HWiNFO64 (LLC 2). Doesn't go up under load though. Auto sets it to 1.10000 which is too high. ------------- Heatripper Threadkiller // 1950X @ 4.141GHz 16C/16T // Taichi // 32GB 3232 RAM // // Titan Xp @ 2075MHz/12,000MHz // 2TB+512GB NVMe SSDs // Custom Liquid Cooling Loop // |
Posted By: lowdog
Date Posted: 04 Jul 2018 at 6:09am
OMG, sent Asrock TSD an email stating Pstate overclocking was broken on X399 Fat Pro Gaming with 1920X and 1900X and the vcore would not down volt when idle.....they replied saying it worked and the vcore did down volt when idle and gave an example with screen shots of CPU-Z and a 1920X......at STOCK!!!!!!!!!!!! FFS!!!!!!!!!!!!  I had no alternative but to reply giving a detailed explanation on how to do a Pstate overclock and how the same procedure worked flawlessly on my X370 fat Pro Gaming....lets see how that goes, but I don't have high hopes for a fix.  ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: ajc9988
Date Posted: 04 Jul 2018 at 10:30pm
I get similar. With Level 3, in HWInfo, the V set for it under CPU #0 is at 1.0375, like set, with the VDDCR_SOC under Asrock X399 reads 1.064 to 1.072. So it looks like 25mV to 40mV is what we are looking at for a difference between what we set and what we get. Yours seems a bit higher than ours, but you are also using level 5. But, that has been a critique of Ryzen boards that seems now true for TR boards, which is that they may have got voltage to core right, but that less attention was paid to getting other voltages done correctly (potentially). As seen here ( https://youtu.be/iQB7ELwBbVI" rel="nofollow - https://youtu.be/iQB7ELwBbVI at around 3:10, going through MSI, Asus, and GB b350 boards) there can be a difference between what is set, what is read with a digital multi-meter, and what is read with HWInfo. GB board had the worst difference between software and multi-meter with around a 200-230mV difference. Asus had 40-100mV (mostly around 40mV) and MSI had 40-80mV, about (mostly also around 40mV). Unfortunately, they did not do testing on Asrock boards. If we were to estimate performance roughly in line with Asus and MSI, which GB has fried some of the Ryzen chips, so we will go with those other boards, even thought they are b350, our seeing 1.064 to 1.088 would still fall under the 1.15V SOC guidelines. Even at 1.075V, it may be above 1.15V, but is still clearly less than 1.2V. So this is something to keep in mind with the SoC Viktage with our chips, and is, in part, why I selected that range (aside from that range working and above that range receiving increasing instability). Now, there is another possibility, which is that with chew helping out with this board, the 25-40mV difference between set and read in software is the 40mV difference that was seen on MSI and Asus boards and the software is reading correctly the voltage that is going into the SoC. That would be awesome. But, I have not broken out my multi-meter to confirm (need to find it, might be something to check when I service my cooling loop soon). But, more than likely, it is like what I explained in the paragraph before this one. Anyone care to test? ------------- // 1950X @ 4.05GHz @ 1.2875V // Asrock X399 Taichi // G.Skill 4133 Trident Z @ 3600 CL14 17 17 17 34 48 4x8GB // // EVGA 980 Ti Classy // Custom Liquid Cooling Loop // |
Posted By: datonyb
Date Posted: 05 Jul 2018 at 1:24am
|
" rel="nofollow - having some experiance with multi meters im more concerned about everyones meters being variable e.g. cost and ability of meter and then calibration of the meter and design tolerances of the meter reading i mean has anyone even nulled the leads before taking readings ? ------------- [url=https://valid.x86.fr/jpg250]  [/url]3800X, powercolor reddevil vega64, gskill tridentz3866, taichix370, evga750watt gold |
Posted By: lowdog
Date Posted: 05 Jul 2018 at 5:02am
|
On my X370 fat Pro Gaming I tested vcore with DMM and level 5/auto LLC. SO if I set 1.35V then from memory if vcore under load when running Prime95 was 1.325V in HWiNFO64 the DMM reading was actually .025 mv lower than what was showing in software, specifically HWiNFO64 in this case. So in summary with vcore real volts at core when checked with DMM were lower by around .02 - .03 mv than what you'll see in software. So I would hesitate a guess that if your setting 1.4V for an overclock with level 5/auto LLC you would be well under the AMD safe limit for vcore. Heck do the math, even setting 1.45V for core with LLC 5/auto would give a lower vcore than AMD max safe volts of 1.425 when loaded. But personally I draw the line at 1.4V or lower for my max. Haven't tested with X399 but probable similar to X370. Think chew said SOC showed pretty similar characteristics to vcore but may have been a little tighter.....can't remember for sure....but no reason to get to paranoid about it. Setting 1.1V SOC I couldn't imagine being any sort of issue. I just keep mine set at 1.05V with LLC 5/auto with my ram @ 3066MHZ with fast timings which betters 3200MHz XMP for latency as I can't get 3200MHz completely stable with 4 x 16GB DR b-dies. ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: SoniC
Date Posted: 06 Jul 2018 at 6:36pm
|
" rel="nofollow - So when can we expect a BIOS with AGESA 1.0.7.1 for X399 boards? ASUS is like... miles ahead of us :-/ ------------- -=SoniC=- TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280, AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU |
Posted By: lowdog
Date Posted: 06 Jul 2018 at 7:36pm
|
" rel="nofollow - latest is 1.0.0.6 isn't it? ------------- X399 Fat Pro Gaming bios 3.10 - TR4 1900X - 64GB G-SKILL TridentZ F4-3200C14Q-64GTZ @ 3133MHz - Vega 64 AIO with EK block - WC Custom loop - 1500W Silverstone PSU - yay |
Posted By: SoniC
Date Posted: 06 Jul 2018 at 9:10pm
Nope, 1.0.7.1 is the latest according to elmor in the OCUK thread for X399 ASUS boards. http://www.overclock.net/forum/11-amd-motherboards/1642825-asus-x399-rog-zenith-extreme-complaint-bug-tracking-thread-64.html It displays as AGESA 1.0.0.6 but in reality it is 1.0.7.1 (they probably forgot to update the version string). Zenith's Beta BIOS 1102 has some probs with XFR but else it was nice -- allowed for even tighter memory timings. Probably because of the XFR bug we will get an 1.0.7.2. I really do hope that ASRock will release a BIOS including it. ------------- -=SoniC=- TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280, AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU |
Posted By: ajc9988
Date Posted: 15 Jul 2018 at 8:23am
|
So, hopefully they push something soon. Since July 5th, we've seen X370, X470, Z370, and B350 boards get firmware updates. Also, hopefully they not only fix the P-state OCing, they change that entire side of things to STOP USING HEX VALUES. I do not know hex, really, so a lot of nice options I'd like to use go dormant. I got annoyed at Asus not having their firmware up to snuff on a lot of things on my M8E board, and the ZE board seems to be called perpetual beta board, but, come on. Still have no qualms having passed on Giga and MSI on release though. Also, even if it cannot be done this round, two cool features for BIOSes are a search function and the BIOS OCing and Stress testing on EVGA. That really is a nice feature. But, even if not done, Asrock hopefully monitors their forums and sees this and pushes at minimum another beta. ------------- // 1950X @ 4.05GHz @ 1.2875V // Asrock X399 Taichi // G.Skill 4133 Trident Z @ 3600 CL14 17 17 17 34 48 4x8GB // // EVGA 980 Ti Classy // Custom Liquid Cooling Loop // |
Posted By: SoniC
Date Posted: 16 Jul 2018 at 12:36am
|
" rel="nofollow - ajc9988 - you know that some of the features are a part of what AMD delivers? Like the P-State OC? And I wouldn't cross my fingers that they will change their minds and start to use decimals instead of hex. Probably they consider this a "pro" feature and if someone cannot work with hex then he would better leave it be (or search for hex tables or YT tutorials) I hope that the power phase design is strong enough to support TR2 and if so -- that ASRock will release an updated AGESA to support it. ------------- -=SoniC=- TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280, AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU |
Posted By: ajc9988
Date Posted: 16 Jul 2018 at 3:52am
Personally, I think of P-state OC as a nice addition, but not something I care about. I set the voltage to locked, not offset, and run. Still, would be nice if they would just do the translation for us. As to the power phase, you have 8 phase using 60A IR3555M Dr. MOS, so you can do a bit on it. It will support stock and maybe lower overclocks. How high the VRM can support for all core OC, we'll find out soon enough.... Edit: so, if you think 1.5V*(40A*8MOSFETS)=480W, then the question is what amps will be used to drive what voltages on the new chips. And I think the 250W TDP was confirmed. I know that a binned 1950X chip can do 4.05GHz staying under 300W, potentially as low as 250W, so, it is hard to say exactly how far you can go. Der8aur did a video posted today doing 3.8GHz all core on a 6 phase server board on an Epyc 7601. Considering this is 8-phase with these MOSFETS, we should see some decent numbers. ------------- // 1950X @ 4.05GHz @ 1.2875V // Asrock X399 Taichi // G.Skill 4133 Trident Z @ 3600 CL14 17 17 17 34 48 4x8GB // // EVGA 980 Ti Classy // Custom Liquid Cooling Loop // |
Posted By: ajc9988
Date Posted: 17 Jul 2018 at 10:54pm
|
It's CHRISTMAS! New BIOS from Asrock. You have to upgrade to 2.3, then upgrade to 3.0. http://www.asrock.com/mb/AMD/X399%20Taichi/index.asp#BIOS" rel="nofollow - http://www.asrock.com/mb/AMD/X399 Taichi/index.asp#BIOS Will flash shortly. ------------- // 1950X @ 4.05GHz @ 1.2875V // Asrock X399 Taichi // G.Skill 4133 Trident Z @ 3600 CL14 17 17 17 34 48 4x8GB // // EVGA 980 Ti Classy // Custom Liquid Cooling Loop // |
Posted By: SoniC
Date Posted: 18 Jul 2018 at 12:20am
|
" rel="nofollow - !!! Any tests/results already ? WTF!!! NOTHING for X399 FPG !!! ! ! ! ! ! I want my BIOS! I want to upgrade! Wanna test!! ASRooooooock! Gimme my X399 FPG 3.00 dose !
 ------------- -=SoniC=- TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280, AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU |
Posted By: datonyb
Date Posted: 18 Jul 2018 at 12:24am
|
but its indicating just support for tr2 maybe will have extras who knows ------------- [url=https://valid.x86.fr/jpg250] [/url]3800X, powercolor reddevil vega64, gskill tridentz3866, taichix370, evga750watt gold |