Taichi BIOS V3.10 Strange Happenings
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=6021
Printed Date: 18 May 2024 at 6:20pm
Software Version: Web Wiz Forums 12.04 - http://www.webwizforums.com
Topic: Taichi BIOS V3.10 Strange Happenings
Posted By: Ken429
Subject: Taichi BIOS V3.10 Strange Happenings
Date Posted: 07 Sep 2017 at 5:31pm
|
" rel="nofollow - Updated the BIOS on my Taichi from 3.0 to 3.1. IBT ran ~160.16 seconds on 3.0 and 163.09 seconds on 3.1. The max core temp increased from 65.5C to 67.5C. Cinebench cpu numbers went from 1718 on 3.0 to 1717 on 3.1 and the OpenGL numbers went from 95.35 on 3.0 to 94.66 on 3.1. I guess the differences could be attributed to needle quiver? But, I have run several passes on different days and get the same results - slightly lower CPU scores and higher CPU temperatures. However, using Pstates for the 3.9GHz OC under 3.1 the minimum idle Vcore drops to ~0.368V versus the 0.900V in Pstate2. The minimum cpu speed dropped to 1950MHz versus the 2200MHz that is in the Pstate2 settings. That is a step backward from 3.0 where the minimum Vcore and cpu speed were the same as specified in Pstate2. Also, the max Vcore under load is increased by 0.400V when the "Global C-State Control is set to Disabled. I don't think that happened under BIOS v3.00. I have to ask what benefit is there to V3.10 versus V3.0 on the Taichi M/B to offset the apparent "broken" Pstate2 settings. Should have left well enough alone?! Wasted several hours - back to V3.00? ------------- X470 Taichi, 3800X (BIOS Defaults), BIOS v3.60, Sandisk Extreme Pro 240GB, 2 crucial MX500 1TB, G.Skill 2X 2X8 3200-14, EVGA 1660Ti SC, EVGA 750 G3, W10 1903 |
Replies:
Posted By: Xaltar
Date Posted: 07 Sep 2017 at 5:46pm
|
From what I have seen, I didn't bother with the update on my Taichi. My thinking is "if it ain't broke". In all likelihood 3.1 is just a few tweaks and further expanded memory compatibility. I have not got an official breakdown or anything but my system is stable at rated speeds on 3.0 do for me at least, 3.1 is tempting fate for no reason. I may pull my 3600 RAM kit out of my intel setup and see if 3.1 can manage 3600 stable but short of that I am quite happy with 2933 on my 2800 kit @1.2v  ------------- 
|
Posted By: Ken429
Date Posted: 08 Sep 2017 at 5:09pm
|
I went back to 3.00. I set the Global C-State Control setting to Auto after making the Pstate changes. Much to my surprise the idle voltage dropped to ~0.368V when Idle! I changed the Global C-State Control to Disabled and the minimum Vcore dropped to 0.880V when idle. Close enough to the 0.900V in Pstate2 I guess! However unlike BIOS v3.10 the Vcore under load did not increase by 0.400V. I must have left the Global C-State Control set to Disabled when I was happy with the 3.00 BIOS. Another thing I noticed when running the AMD Ryzen Master program was the MHz on cores 5,6 and 7 dropped to 1950MHz. Cores 1,2,3,4 and 8 settle down to the 2200MHz specified in the Pstate2 settings. So, I guess V3.10 wasn't quite a screwed up (compared to V3.00) as I originally thought. But one other anomaly With V3.100 was core 8 always ran at 3900MHz when the AMD Ryzen Master program was running. Also, the Max CPU Tdie as reported by HWiNFOR64 dropped back to 63.5C when running IBT for 10 passes under V3.00 versus the 67.5C when running the same test under V3.10. ------------- X470 Taichi, 3800X (BIOS Defaults), BIOS v3.60, Sandisk Extreme Pro 240GB, 2 crucial MX500 1TB, G.Skill 2X 2X8 3200-14, EVGA 1660Ti SC, EVGA 750 G3, W10 1903 |
Posted By: datonyb
Date Posted: 08 Sep 2017 at 9:13pm
|
to add to confusion my taichi with the latest bios now decides to downvolt itself on idle even though its set at 1.3v in p states and all the c6 cool n quiet etc is all disabled unless theres a bug in cpuz mine can sit at 3900mhz and drop to 0.480 / 0.608 it should or did with last bios sit pretty much at 1.290 ish all the time ------------- [url=https://valid.x86.fr/jpg250]  [/url]3800X, powercolor reddevil vega64, gskill tridentz3866, taichix370, evga750watt gold |
Posted By: Xaltar
Date Posted: 08 Sep 2017 at 9:44pm
|
That is by design I am pretty sure. My Intel systems all do that too. I think a lot of people forget that before power saving CPUs would remain at full frequency 24/7 but even then, used less power when not under load. It takes a lot less power to maintain say 4ghz at idle than it does at load so the vcore will lower to further save power depending on the load on the CPU. CPUz has also had issues with AM4 vcore readings, when I did my Taichi review I was seeing 2.68v in CPUz. The latest update could just be more compatible with CPUz's monitoring back-end. Regardless, there is never anything wrong with seeing lower voltages unless you are unstable  -------------
|
Posted By: parsec
Date Posted: 08 Sep 2017 at 9:48pm
" rel="nofollow -
I assume your Windows Power Plan is still Ryzen Balanced, and the Minimum Processor State is still at 90%? There are two C State option settings in my different X370 board's UEFI, another is buried somewhere in the AMD CBS options. ------------- http://valid.x86.fr/48rujh" rel="nofollow"> 
|
datonyb wrote:Posted By: datonyb
Date Posted: 09 Sep 2017 at 12:55am
|
yeah im not complaining just something i noticed upto and including version 3 the cpuz would stay at 1.296-1.306 even at idle (i know the current would not be flowing on idle and obviously the temps would be lower) but now its dropping on idle to 0.480-0.600ish its not causing an issue and if i start cinebench or something else it ramps directly back upto 1.296 im pretty sure i switched off any bios based extras for example theres a c6 and a global c6 in different places of bios and yes windows is the ryzen power plan i was just forwarding on my observations to the group on a side not i did have a play with ram again this morning i can boot upto 3600 cl16 at 2t and 1.4 volts no cold start testing though yet ( and thats where my ram plays up at anything over stock settings) im just testing the old NEMESIS of mine which was/is the cl14 timings at the moment its stable in windows yet to give it the cold boot test (this time ive upped the ram volts to 1.395) something i didnt want to do before now ------------- [url=https://valid.x86.fr/jpg250] [/url]3800X, powercolor reddevil vega64, gskill tridentz3866, taichix370, evga750watt gold |
Posted By: Ken429
Date Posted: 14 Sep 2017 at 5:04pm
|
" rel="nofollow - Anybody with a Taichi X370 M/B try the latest BIOS (3.20) just released? Any idea on why it was released just 12 days after 3.10 was released? Did ASRock address some of the new bugs introduced in 3.10? I really don't want to spend a few hours updating the BIOS just to find out I took a few steps backward from 3.00?! ------------- X470 Taichi, 3800X (BIOS Defaults), BIOS v3.60, Sandisk Extreme Pro 240GB, 2 crucial MX500 1TB, G.Skill 2X 2X8 3200-14, EVGA 1660Ti SC, EVGA 750 G3, W10 1903 |
Posted By: Ken429
Date Posted: 15 Sep 2017 at 7:08pm
|
Looks like I'm talking to myself again. I'll give 3.20 a try on the next rainy day and see what happens. ------------- X470 Taichi, 3800X (BIOS Defaults), BIOS v3.60, Sandisk Extreme Pro 240GB, 2 crucial MX500 1TB, G.Skill 2X 2X8 3200-14, EVGA 1660Ti SC, EVGA 750 G3, W10 1903 |
Posted By: parsec
Date Posted: 15 Sep 2017 at 10:17pm
The description for the X370 Taichi UEFI 3.20 sums it up fairly well. DDR4 memory with a stock/JEDEC speed of 2666, will remain at 2666 instead of being down-clocked to 2400. Ryzen and now some new Intel processors have a maximum memory speed spec (NOT an OC) of 2666/2667. At least one memory manufacture is now producing DDR4 memory whose stock JEDEC (NOT XMP) speed is 2666. Prior to this for Ryzen, 2400 was considered the highest stock/JEDEC memory speed, for example the G.SKILL FlareX 3200 memory has a stock/JEDEC speed of 2400. Most of the earlier DDR4 memory designed for Intel has a stock/JEDEC speed of 2133. So it's a simple change that most likely won't affect you, so no need to update. What new bugs are in 3.10? It takes you a few hours to update the UEFI/BIOS? I assume that means restoring your non-default settings? EDIT: The drop in VCore below 0.400V with Cool n Quiet and C6 enabled, has been in my X370 Killer SLI/ac board's UEFI for as long as I can remember. Why does that surprise you? My Ryzen PC is perfectly stable when that happens. Also, the core speeds at 2200MHz and ~1950MHz at idle are also standard for my board with the CPU power saving options enabled, and the Windows Power Plan Minimum Processor state adjusted from the Ryzen Balanced 90%. ------------- http://valid.x86.fr/48rujh" rel="nofollow">
|
Posted By: Ken429
Date Posted: 16 Sep 2017 at 12:57am
|
Ok, so I exaggerated the time a little bit! This time when I update the BIOS from 3.00 to 3.20 I'll make sure I've documented exactly what V3.00 was doing - Max Core Temp when running IBT for 10 passes increased ~3C under 3.10, Core 8 never dropping to 2200 MHz when running the AMD Master Ryzen program under 3.10, the Vcore increasing by 0.40 (both Idle and under load) when Global C State was Enabled in 3.100 - it does not do that in 3.00. I agree that the Vcore dropping to 0.368V when idle is probably not an issue BUT Pstate2 was set to 0.900. Maybe I was wrong but under 3.00 the system only took ~10 seconds to do a cold boot to the first BIOS splash screen with memory set to XMP. Under 3.10 it took 20+ seconds. When I went back to 3.00 it took 20+ seconds to do the same thing!! Now When I go back to the default memory setting instead of the XMP setting it takes ~10 seconds. Maybe just maybe I thought I was running 3.00 with XMP set and was getting the ~10 second boot to BIOS splash screen and I was running with the default memory setting but I don't think so. That's why I said 2 hours to go back, I don't think I ever got there!! This thing has a mind of it's own. Anyway thanks for the response - other than the minor stuff above my system runs great and I'm generally very happy. ------------- X470 Taichi, 3800X (BIOS Defaults), BIOS v3.60, Sandisk Extreme Pro 240GB, 2 crucial MX500 1TB, G.Skill 2X 2X8 3200-14, EVGA 1660Ti SC, EVGA 750 G3, W10 1903 |
Posted By: datonyb
Date Posted: 16 Sep 2017 at 1:07am
|
a guy posted today that when he upgraded to bios 3.1 on taichi boot times went very long he reset back to v3 and then flashed v3.1 again from usb boot times are now identical personally on my taichi i think the boot times are swift with my flavour of pstate tuning (which is locked 3.9 and 1.3 volts) all c&q and c6 off version 3.1 has bought unexpected down volting according to cpuz to the half a volt area i dont even know why its downvolting now with same setting as v3 when it would stay rigid at 1.3 volts at all times ------------- [url=https://valid.x86.fr/jpg250] [/url]3800X, powercolor reddevil vega64, gskill tridentz3866, taichix370, evga750watt gold |
Posted By: Ken429
Date Posted: 17 Sep 2017 at 6:16pm
|
datonyb, parsec, You made me try BIOS v3.20. Like I said this thing has a mind of it's own. I Flashed the BIOS to 3.20 via the Internet and carefully changed the BIOS settings one at a time until I got to the "Final" Psate settings. Much to my surprise I'm back to where the system was under 3.00. It's running at 3.9GHz @ 1.35 Vcore with XMP settings turned on. The system boots to the first BIOS screen in ~10 seconds like it did under 3.00 before I flashed to 3.10!! I don't know what magic is occurring but after 3.10 I could not get 3.00 to boot to the first BIOS screen in less that 20 seconds with XMP enabled. datonyb, I had this system running 3.9 @ 1.30V and thought it was very stable until I met Handbrake and tried making an MP4 file from a Blue Ray disk that was located on an SSD drive! Something about the input to Handbrake from the SSD caused the system to hang until I increased the Vcore to 1.35V. ------------- X470 Taichi, 3800X (BIOS Defaults), BIOS v3.60, Sandisk Extreme Pro 240GB, 2 crucial MX500 1TB, G.Skill 2X 2X8 3200-14, EVGA 1660Ti SC, EVGA 750 G3, W10 1903 |
Posted By: datonyb
Date Posted: 17 Sep 2017 at 7:22pm
|
@ ken yes i think i might have been lucky with my cpu it does seem to be pretty good at low voltage (whether this is anything to do with the early rumours that the plain 1700 was binned with lower voltage stability in mind i know not) however it seems later cpu's (of which mine is not ) are getting reports of better stability with lower volts, of course this is never a sure bet as a 1600 r5 i was involved two weeks ago demands nothing less than 1.35v @ 3.8ghz or it plays up anyways im glad your finding latest bios ok ( i must confess for a real change this time i havnt tried 3.2 yet, im usually the first crash test dummy to try them) ------------- [url=https://valid.x86.fr/jpg250] [/url]3800X, powercolor reddevil vega64, gskill tridentz3866, taichix370, evga750watt gold |
Posted By: Ken429
Date Posted: 17 Sep 2017 at 8:02pm
|
" rel="nofollow - [URL=][/URL]parsec, Can you shed some light on the CPU temps reported by AMD. BIOS version 3.10 and 3.20 give you the option to use different temps to control the (in my case) the radiator fans. Back on 3.00 it seemed to monitor the Tctl as reported by HWiNFO64. At least the Customize fans settings I was using worked close to what I would have expected. Now, under BIOS V3.20 with the CPU Fan Temp Source set to Monitor CPU (as it was under 3.00) the fans never get to the max speed when running IBT. When I change the CPU Fan Temp Source to Monitor Tctr1 the Radiator fans jump to the max RPM almost immediately when running IBT using the settings I had in 3.00. As reported by HWiNFO64 when running IBT for 10 passes: CPU AMD Ryzen 1700X CPU (Tctl) 45.5C to 84C CPU (Tdle) 25.5C to 64C ASRock X370 Taichi CPU 28.5C to 47.5C It appears BIOS Monitor Tctr1 relates to the HWiNFo CPU (Tctl) and the BIOS Monitor CPU relates to the HWiNFO ASRock CPU? The most reasonable one appears to be the HWiNFO CPU (Tdle) but I don't think that is one of the temps monitored by the BIOS. I can't believe the 28.5C to 47.5C can be the CPU temp? ------------- X470 Taichi, 3800X (BIOS Defaults), BIOS v3.60, Sandisk Extreme Pro 240GB, 2 crucial MX500 1TB, G.Skill 2X 2X8 3200-14, EVGA 1660Ti SC, EVGA 750 G3, W10 1903 |
Posted By: nangu
Date Posted: 18 Sep 2017 at 3:29am
|
Ken, Tdie is the real measured junction temp inside the AMD CPU. The Tctl temp reading is Tdie + 20° offset, as AMD describes. It was controversial why AMD added the 20°c offset to "X" versions of Ryzen CPUs. The 1700 non "X" as example, don't have this offset so Tdie = Tctl. AMD said they added the offset in order to allow better fan curve management, Personally I find it confusing and no real value to consumer. Anyway, I recommend to use the Tctl in your case to drive fans, taking care the temp reading is 20°c higher than the real CPU temp, so adjust your curve accordingly. The CPU reading relates to the thermal diode on the CPU socket, so it's a Socket temperature and not the reeal CPU temperature. ------------- R7 1700 @3.90 1.25v - GSkill TridentZ 3200c16 Hynix MFR @2933 14-16-16-32 - Fatal1ty Gaming X370 K4 - Gigabyte GTX 1070 G1 - WD 256 Black M2 Nvme as Windows 10 boot drive - EVGA Gold 650W |
Posted By: Ken429
Date Posted: 18 Sep 2017 at 4:10am
|
Thanks for the response, Looks like Tctl (in HWiNFO language) = Tctr1 in (ASRock language)! Anyway, setting the CPU Fan1 Temp Source to Monitor Tctr1 works as advertised. Funny though there is no mention of Tctr1 in the Fan tuning section of the BIOS but it looks like it is using the Tctl or Tctr1 for the Fan1 graph but call it Monitor CPU! ------------- X470 Taichi, 3800X (BIOS Defaults), BIOS v3.60, Sandisk Extreme Pro 240GB, 2 crucial MX500 1TB, G.Skill 2X 2X8 3200-14, EVGA 1660Ti SC, EVGA 750 G3, W10 1903 |
Posted By: parsec
Date Posted: 18 Sep 2017 at 9:13am
" rel="nofollow -
Sorry, had to work today, started this post earlier, and only finished it now. Tdie and Tctrl are AMD terms for the Ryzen CPU temperature. Tctrl = Tdie + 20° C. Or put another way, Tctrl has a +20° C offset from Tdie. You can see that perfectly in the HWiNFO readings. The ASRock "CPU Temperature" seems to be similar to the AMD CPU socket temperature, used on earlier AMD platforms. That's the best I can tell you about it. It changes very little and has a small range of temperatures. This temperature was used in all the early ASRock Ryzen boards. Recently it was changed to Tctrl, and now with the latest Ryzen UEFI updates, we can choose between the two. The ASRock CPU Temperature is the default setting. So which one is correct? AMD said Tctrl is used to control the CPU cooler's speed on some Ryzen processors. Not clear at all, is it? Is that the stock AMD Ryzen CPU coolers, or after market coolers? If it only applies to some Ryzen processors, then which ones? We need to ask AMD which of these two temperatures is correct, since I have never seen that identified in a Ryzen processor review. I have a 1700X too, and would like to believe that Tdie is the correct Ryzen processor temperature. Do I know with certainty it is? No. A post above said it is, great, but just show me where AMD said that. Why ASRock chose Tctrl as an option for the fan speed monitoring temperature, instead of Tdie, I don't know. The ASRock utilities like A-Tuning and F-Stream use the ASRock CPU Temperature for fan speed control, and in the latest Ryzen board UEFI updates, that is what "CPU Temperature" is. Just before your board's 3.10, ASRock changed from the ASRock CPU Temperature, to Tctrl. ASRock Ryzen users have complained about the use of both the ASRock CPU Temperature, and Tctrl. Users that don't like the constant fan speed changes don't like Tctrl. Users that care about the actual CPU temperature being too high don't like the ASRock CPU Temperature. Using Tctrl can be fine if you configure a custom fan speed curve to match your fans and your tolerance for fan noise. The built in fan speed profiles will only satisfy you by pure chance, since all PC fans are different in their characteristics. If you don't see the Tctrl option, which is not in the FANTastic Tuning screen in the UEFI, you need to check the individual fan speed configuration options in HW Monitoring screen. We now have "CPU Temperature", Tctrl, and Mother board temperature as the three options. At least that is what I have in UEFI version 3.20 for the X370 Killer SLI/ac board. I doubt that the X370 Taichi version 3.10 and 3.20 is any different. ------------- http://valid.x86.fr/48rujh" rel="nofollow">
|
Posted By: Ken429
Date Posted: 18 Sep 2017 at 6:30pm
|
parsec, Thanks for the information, what got me going was the change from 3.00 to 3.20. ASRock decided to change to the Temp source for CPU. Good thing I got EK-Vardar F2 140 (1600) fans on the radiator. Even at full speed when using Tctr1 as the source in the BIOS they are not too noisy. However they do get to the max speed almost immediately when running IBT but then that's probably OK. I did try the "Standard" setting in the BIOS and the fans ran slower and quieter when running IBT but the Temp according to what HWiNFO reported was ~3C higher. I've played with the settings in the BIOS based on the higher temps being used as the source and got the fan speed back to something less than full speed when running IBT as they were under 3.00 which had little or no effect on the HWiNFO reported temps. ------------- X470 Taichi, 3800X (BIOS Defaults), BIOS v3.60, Sandisk Extreme Pro 240GB, 2 crucial MX500 1TB, G.Skill 2X 2X8 3200-14, EVGA 1660Ti SC, EVGA 750 G3, W10 1903 |
Posted By: nangu
Date Posted: 19 Sep 2017 at 12:25am
" rel="nofollow -
Here is where AMD said that: https://community.amd.com/community/gaming/blog/2017/03/13/amd-ryzen-community-update Cheers. ------------- R7 1700 @3.90 1.25v - GSkill TridentZ 3200c16 Hynix MFR @2933 14-16-16-32 - Fatal1ty Gaming X370 K4 - Gigabyte GTX 1070 G1 - WD 256 Black M2 Nvme as Windows 10 boot drive - EVGA Gold 650W |
Posted By: Ken429
Date Posted: 19 Sep 2017 at 4:13am
|
" rel="nofollow - [URL=][/URL]I see what you mean, it's clear as mud. What is the actual temperature of my 1700X processor when running IBT for ten passes? I would like to believe it's 64C but based on my 4790K experience the 84C seems more like what I would expect. Oh well, whatever the CPU temperature is the 1700X/Taichi beats the heck out of my 4790K's when running Handbrake which is what I built the thing for in the first place. But the more I play with it the better I like it as an all around system. Maybe I had better sell my Intel stock and buy some AMD? ------------- X470 Taichi, 3800X (BIOS Defaults), BIOS v3.60, Sandisk Extreme Pro 240GB, 2 crucial MX500 1TB, G.Skill 2X 2X8 3200-14, EVGA 1660Ti SC, EVGA 750 G3, W10 1903 |