Page 20 of 46 FirstFirst ... 10 18 19 20 21 22 30 ... LastLast
Results 191 to 200 of 456

Thread: Latest C7H BIOS

  1. #191
    ROG Enthusiast Array
    Join Date
    Oct 2013
    Reputation
    10
    Posts
    33

    Here my UV profile which i like a lot. GPU running at 0.762V @ 1645/6014 (1080Ti Aorus Xtreme)
    CPU 3600 under load @ 0.89V stable and RAM 3800/1900 @ 1.36V (with optimized subtimings)
    Run stabilitytests for ~7 hours with this profile.
    ProcODT currently at 30.

    Click image for larger version. 

Name:	3800 Ultra Low Spannung Stabitest.png 
Views:	0 
Size:	412.1 KB 
ID:	81296


    Click image for larger version. 

Name:	3800 1.36V Stable Settings.jpg 
Views:	0 
Size:	534.7 KB 
ID:	81297


    Of course you ask yourself why is this guy running his 3600 with only 3.4GHz?
    Because i love to undervolt and this CPU with such a low clock and fast RAM is faster than what i need.
    I would go even lower, but my MB wont allow me^^
    I am playing at native 3840x2160 (mostly 3800x1800 pimped with reshade, to have DLSS Effect without a RTX card)
    F.e. look at Forza 4 Benchmark @ 1440p Ultra, even if i max OC my 1080Ti, it is too slow, even a 2080Ti wouldnt be enough.

    Edit: Asus wont allow me to upload the Forza results... (always failed the upload..)
    You can check it here, how i am faster than a oced 2700X with fast RAM etc with my very strong UV profile
    https://www.computerbase.de/forum/th...#post-22968580

    Edit 2:

    My CPU doesn t see 40°C during gaming for hours and everything is very silent, i don t even hear my PC sitting next to me on the table.
    (adjusted fan speeds, not with those from the screenshot)
    With PBO enabled and one of my faster RAM profiles i get ~200 AVG FPS in Forza 4 CPU Render, and ~160 FPS mins, but i dont see, why i should do this, because with nerarly ZERO Power Consumption and ~idle temps (UNDER LOAD), my 3600 is fast enough for every game i play
    Go and undervolt your systems guys, no need to run your components @ stock most of the time, so go and analyze what you really need.


    Edit 3:

    I am a lil bit pissed because of the BIOS situation, everyone is getting fresh BIOS (Asrock/GB etc), it would be nice if someone else (not only Shamino) would post Beta Bios here. BTW for my second system here i bought a ASUS C6H for 96€ in a shop in germany, hope that was not a mistake... So come on ASUS, we stick with you, even if we dont get new BIOS the fastest, but come on dont treat us like stepchildren, if you know what i mean.

    Greetings @ all form Gerrmany
    Last edited by Duvar; 08-07-2019 at 12:50 PM.

  2. #192
    ROG Enthusiast Array KeithMyers PC Specs
    KeithMyers PC Specs
    MotherboardCrosshair VII Hero
    ProcessorRyzen 7 3900X
    Memory (part number)F4-3200D-16GTZ
    Graphics Card #1GTX 1080 Hybrid
    Graphics Card #2RTX 2080 Hybrid
    Graphics Card #3RTX 2080 Hybrid
    Storage #1Samsung 960 EVO M.2 256GB
    CPU CoolerXSPC 360 V3 radiator Aqua Computer Kryos Cuplex Next
    CaseThermaltake Level 20 XT
    Power SupplyCorsair AX1200
    OS Ubuntu 18.04.3 LTS
    Network RouterASUS RT-AC68R

    Join Date
    Jun 2014
    Reputation
    10
    Posts
    45

    Question Huh ? ?

    Quote Originally Posted by xeizo View Post
    That's because IF doesn't adjusts itself to Fclk/Mem automatically over 3533. Even when set at 1:1 it doesn't work. IF setting is sort of hidden away under OC/PBO at the bottom of the second bios tab and a couple submenus deep, if you set it manually at 1800 latency will be good at 3600 too.
    Huh? ? Do you mean that the FCLK setting directly below the memory clock setting does not apply on the main Extreme Tweaker page? I set the FCLK manually for half the memory clock setting. 3533 memory clock setting gets 1767Mhz FCLK setting and 3600Mhz memory clock setting gets 1800Mhz FCLK setting. I've never even opened the OC/PBO option since I manually all-core overclock.

  3. #193
    ROG Enthusiast Array KeithMyers PC Specs
    KeithMyers PC Specs
    MotherboardCrosshair VII Hero
    ProcessorRyzen 7 3900X
    Memory (part number)F4-3200D-16GTZ
    Graphics Card #1GTX 1080 Hybrid
    Graphics Card #2RTX 2080 Hybrid
    Graphics Card #3RTX 2080 Hybrid
    Storage #1Samsung 960 EVO M.2 256GB
    CPU CoolerXSPC 360 V3 radiator Aqua Computer Kryos Cuplex Next
    CaseThermaltake Level 20 XT
    Power SupplyCorsair AX1200
    OS Ubuntu 18.04.3 LTS
    Network RouterASUS RT-AC68R

    Join Date
    Jun 2014
    Reputation
    10
    Posts
    45

    Question

    Quote Originally Posted by Othello View Post
    There was essentially no load on the CPU when taking those measurements. What I found interesting was difference between the measurement and reported values. When I monitored the CPU voltage during a Blender run I noticed a fair amount droop. The settings below seems to have reduced the amount of droop by about 50%.

    My CPU:
    LLC = 4 and 7, yes there are 2 LLC levels now, not sure why.
    Wish someone would explain this too. Bewildering that both CPU and SoC get two LLC setting automatically as soon as you move off Auto. And the relationship between the top LLC setting is tied to the second LLC setting too. I wonder if the two settings are tied to the two cpu chiplets.

    Do Ryzen 3700X users get two LLC setting? That would disprove my dual chiplet theory.

  4. #194
    ROG Enthusiast Array MattTheTech PC Specs
    MattTheTech PC Specs
    Laptop (Model)Nothing Interesting
    MotherboardASUS ROG Crosshair VII Hero
    ProcessorRyzen 9 3900x
    Memory (part number)3200C14D-16GFX (32GB Total)
    Graphics Card #1EVGA GTX 1080ti FTW3 Hybrid
    Sound CardHT Omega eCLARO
    Monitor3 Wall Mounted-1x Samsung 55" 4K TV (Runs 1440p 120hz), 2x LG 32" 1080p
    Storage #1Samsung 970 Evo
    Storage #23x Samsung 850 Evo 512GB in Raid 0 For 1.5TB Array Game Drive + ZFS 30TB NAS/Server For Storage
    CPU CoolerEKWB Dual Custom Water Loop
    CaseModded Rosewill Evo Armor 2 (I know all reviews hated it, but after my Mods its been a GREAT CASE)
    Power SupplyCorsair RM1000i
    Keyboard Corsair RGB k70 Lux V2
    Mouse Logitech G700s
    Headset Oculus Rift
    Mouse Pad Razor Vespula
    Headset/Speakers 7.1 Custom Built Surround Sound System
    OS Win 10 Pro 64bit
    Network RouterASUS RT-AC1900p
    Accessory #1 10Gb Nic
    MattTheTech's Avatar
    Join Date
    May 2015
    Reputation
    10
    Posts
    48

    Quote Originally Posted by gupsterg View Post
    I have launch C7H board. It is one of the bugged ones where chipset voltage in SW monitoring and Probeit point is ~50mV higher than reality/set value. So do be aware they can have issues.

    Secondly the Probeit points ever so slightly seem to be affected by load line effect, not as much as C6H. Why I say this is I used 3x differing 2700X with the C7H and several differing RAM kits, always Probeit point would be ~0.05V higher for DRAM than set point. Now with R5 3600 same board, same RAM, on Probeit point for DRAM the difference between set value and DMM read is ~0.03V.

    The other factor when comparing software reads and dmm is polling interval of SW. HWINFO default polling rate is 2000ms, lower it and it will better reflect reality. Next you wanna look at average value rather than min/max/current. The averages are more in line with DMM, this was also the case with C6H.

    The C6H Super IO chip was configured with worse granularity, so in SW monitoring averages were best to ref. The C7H has better granularity, see the OP of OCN C7H thread for PDF by Elmor, is also in the C7H Essential thread here on ROG. The C7H also has a "OP-AMP" (see red arrow below, AS324M-E1). From comments of Buildzoids, it supposed aid accuracy on readings from CPU.

    Attachment 81295

    When ASUS WMI was implemented on C7H we gained the (VRM) readings which were more accurate than the non (VRM), this was down to an EC FW update, the C6H never got that, as it is suspectable to bricking when EC FW updates are done.
    I really appreciate the detailed insight.

    I too have the same batch of boards you have. Bought it on Launch day, and it also shows the Chipset being .05v higher. Seems to have caried over to the 3900x as well.

  5. #195
    ROG Enthusiast Array MattTheTech PC Specs
    MattTheTech PC Specs
    Laptop (Model)Nothing Interesting
    MotherboardASUS ROG Crosshair VII Hero
    ProcessorRyzen 9 3900x
    Memory (part number)3200C14D-16GFX (32GB Total)
    Graphics Card #1EVGA GTX 1080ti FTW3 Hybrid
    Sound CardHT Omega eCLARO
    Monitor3 Wall Mounted-1x Samsung 55" 4K TV (Runs 1440p 120hz), 2x LG 32" 1080p
    Storage #1Samsung 970 Evo
    Storage #23x Samsung 850 Evo 512GB in Raid 0 For 1.5TB Array Game Drive + ZFS 30TB NAS/Server For Storage
    CPU CoolerEKWB Dual Custom Water Loop
    CaseModded Rosewill Evo Armor 2 (I know all reviews hated it, but after my Mods its been a GREAT CASE)
    Power SupplyCorsair RM1000i
    Keyboard Corsair RGB k70 Lux V2
    Mouse Logitech G700s
    Headset Oculus Rift
    Mouse Pad Razor Vespula
    Headset/Speakers 7.1 Custom Built Surround Sound System
    OS Win 10 Pro 64bit
    Network RouterASUS RT-AC1900p
    Accessory #1 10Gb Nic
    MattTheTech's Avatar
    Join Date
    May 2015
    Reputation
    10
    Posts
    48

    Quote Originally Posted by KeithMyers View Post
    Huh? ? Do you mean that the FCLK setting directly below the memory clock setting does not apply on the main Extreme Tweaker page? I set the FCLK manually for half the memory clock setting. 3533 memory clock setting gets 1767Mhz FCLK setting and 3600Mhz memory clock setting gets 1800Mhz FCLK setting. I've never even opened the OC/PBO option since I manually all-core overclock.
    I am curious about this as well. After I read your comment I went directly to that setting in the Advanced/AMD OC Settings and found every single place where there is a FClk setting, there is 3 of them, 2 in OC, and 1 in CBS that I could find, and set them all to 3600, went back into windows and tested my latency with AIDA64 and had the same exact result.

    Is there no good way to tell if the FClk Setting actually took and which one actually does it. I also set the Ratio 1 to 1 manually with that setting to to be certain and same exact result. I was hoping I would shed a few MS. At 3600Mhz CL14 I am at 67ms.

  6. #196
    ROG Enthusiast Array MattTheTech PC Specs
    MattTheTech PC Specs
    Laptop (Model)Nothing Interesting
    MotherboardASUS ROG Crosshair VII Hero
    ProcessorRyzen 9 3900x
    Memory (part number)3200C14D-16GFX (32GB Total)
    Graphics Card #1EVGA GTX 1080ti FTW3 Hybrid
    Sound CardHT Omega eCLARO
    Monitor3 Wall Mounted-1x Samsung 55" 4K TV (Runs 1440p 120hz), 2x LG 32" 1080p
    Storage #1Samsung 970 Evo
    Storage #23x Samsung 850 Evo 512GB in Raid 0 For 1.5TB Array Game Drive + ZFS 30TB NAS/Server For Storage
    CPU CoolerEKWB Dual Custom Water Loop
    CaseModded Rosewill Evo Armor 2 (I know all reviews hated it, but after my Mods its been a GREAT CASE)
    Power SupplyCorsair RM1000i
    Keyboard Corsair RGB k70 Lux V2
    Mouse Logitech G700s
    Headset Oculus Rift
    Mouse Pad Razor Vespula
    Headset/Speakers 7.1 Custom Built Surround Sound System
    OS Win 10 Pro 64bit
    Network RouterASUS RT-AC1900p
    Accessory #1 10Gb Nic
    MattTheTech's Avatar
    Join Date
    May 2015
    Reputation
    10
    Posts
    48

    Quote Originally Posted by KeithMyers View Post
    Wish someone would explain this too. Bewildering that both CPU and SoC get two LLC setting automatically as soon as you move off Auto. And the relationship between the top LLC setting is tied to the second LLC setting too. I wonder if the two settings are tied to the two cpu chiplets.

    Do Ryzen 3700X users get two LLC setting? That would disprove my dual chiplet theory.
    And again, I second Keith's Question.

  7. #197
    ROG Enthusiast Array KeithMyers PC Specs
    KeithMyers PC Specs
    MotherboardCrosshair VII Hero
    ProcessorRyzen 7 3900X
    Memory (part number)F4-3200D-16GTZ
    Graphics Card #1GTX 1080 Hybrid
    Graphics Card #2RTX 2080 Hybrid
    Graphics Card #3RTX 2080 Hybrid
    Storage #1Samsung 960 EVO M.2 256GB
    CPU CoolerXSPC 360 V3 radiator Aqua Computer Kryos Cuplex Next
    CaseThermaltake Level 20 XT
    Power SupplyCorsair AX1200
    OS Ubuntu 18.04.3 LTS
    Network RouterASUS RT-AC68R

    Join Date
    Jun 2014
    Reputation
    10
    Posts
    45

    Cool My 3900X latencies in Linux

    Quote Originally Posted by MattTheTech View Post
    I am curious about this as well. After I read your comment I went directly to that setting in the Advanced/AMD OC Settings and found every single place where there is a FClk setting, there is 3 of them, 2 in OC, and 1 in CBS that I could find, and set them all to 3600, went back into windows and tested my latency with AIDA64 and had the same exact result.

    Is there no good way to tell if the FClk Setting actually took and which one actually does it. I also set the Ratio 1 to 1 manually with that setting to to be certain and same exact result. I was hoping I would shed a few MS. At 3600Mhz CL14 I am at 67ms.
    I can't ever compare my latencies to any of the posts since I don't do Windows and don't have any of the tools that everyone uses. I can see latencies with Intel MLC and Geekbench4.

    keith@Serenity:~/Downloads/Utils/MLC$ ./mlc
    Intel(R) Memory Latency Checker - v3.6
    Measuring idle latencies (in ns)...
    Numa node
    Numa node 0
    0 53.4

    Measuring cache-to-cache transfer latency (in ns)...
    Local Socket L2->L2 HIT latency 15.3
    Local Socket L2->L2 HITM latency 24.3

    https://browser.geekbench.com/v4/cpu/14148236

  8. #198
    ROG Guru: Orange Belt Array xeizo PC Specs
    xeizo PC Specs
    Laptop (Model)HP i7-7500U 16GB 256GB NVMe
    MotherboardAsus Maximus VIII Hero
    Processor6700K
    Memory (part number)32GB HyperX/CorsairLP @ 3200MHz
    Graphics Card #1MSI 1080Ti @ Water
    Sound CardSteinberg UR22
    MonitorAOC 1440p 27" S-PLS
    Storage #1256GB Samsung 950 Pro NVMe
    Storage #22xSSD:s, 4xHDD:s
    CPU CoolerAntec 620 AIO push-pull 2xNidec Gentle Typhoon 1450rpm
    CaseCorsair Carbide 200R
    Power SupplyCooler Master V550S
    Keyboard Logitech G710+ MX Brown (50$ on sale)
    Mouse Steelseries Sensei RAW
    Headset Sennheiser HD600
    Mouse Pad Steelseries, hard
    Headset/Speakers Swissonic ASM5 active studio monitors (thomann.de)
    OS Windows10 preview build 16237
    Network RouterAsus RT-N66U

    Join Date
    Apr 2016
    Reputation
    38
    Posts
    351

    Quote Originally Posted by MattTheTech View Post
    I am curious about this as well. After I read your comment I went directly to that setting in the Advanced/AMD OC Settings and found every single place where there is a FClk setting, there is 3 of them, 2 in OC, and 1 in CBS that I could find, and set them all to 3600, went back into windows and tested my latency with AIDA64 and had the same exact result.

    Is there no good way to tell if the FClk Setting actually took and which one actually does it. I also set the Ratio 1 to 1 manually with that setting to to be certain and same exact result. I was hoping I would shed a few MS. At 3600Mhz CL14 I am at 67ms.
    Here it is, this setting sometimes gets itself "wrong" and you get high latency, but if you set it manually like I did it alwyas stays the same and always low lattency. Just setting Fclk isn't enough. Just wondering why this is hidden away like this.

    Click image for larger version. 

Name:	4.png 
Views:	0 
Size:	933.0 KB 
ID:	81310

    I discovered it by accident when it stayed at 1900 even after a bios reset and I got abnormal good results.

  9. #199
    ROG Guru: Brown Belt Array gupsterg PC Specs
    gupsterg PC Specs
    MotherboardAsus Maximus VII Ranger
    Processori5 4690K
    Memory (part number)HyperX Savage 2400MHz 16GB
    Graphics Card #1Sapphire R9 Fury X (1145/545 Custom ROM)
    MonitorAsus MG279Q
    Storage #1Samsung 840 Evo 250GB
    Storage #2HGST 2TB
    CPU CoolerThermalRight Archon SB-E X2
    CaseSilverstone Temjin 06
    Power SupplyCooler Master V850
    Headset/Speakers HyperX Cloud
    OS Win 7 Pro x64 / Win 10 Pro x64
    gupsterg's Avatar
    Join Date
    Feb 2015
    Reputation
    103
    Posts
    1,017

    Quote Originally Posted by KeithMyers View Post
    Huh? ? Do you mean that the FCLK setting directly below the memory clock setting does not apply on the main Extreme Tweaker page? I set the FCLK manually for half the memory clock setting. 3533 memory clock setting gets 1767Mhz FCLK setting and 3600Mhz memory clock setting gets 1800Mhz FCLK setting. I've never even opened the OC/PBO option since I manually all-core overclock.

    Quote Originally Posted by xeizo View Post
    Here it is, this setting sometimes gets itself "wrong" and you get high latency, but if you set it manually like I did it alwyas stays the same and always low lattency. Just setting Fclk isn't enough. Just wondering why this is hidden away like this.

    Click image for larger version. 

Name:	4.png 
Views:	0 
Size:	933.0 KB 
ID:	81310

    I discovered it by accident when it stayed at 1900 even after a bios reset and I got abnormal good results.
    I've never had to use the AMD menu setting in your screenie on UEFI 2406/2501/0068/2602. I use the Memory Frequency & FCLK Frequency on Extreme Tweaker page TBH. Be aware you can capture UEFI screen shot to USB stick by pressing F12.
    Intel Defector AMD Rebel

    TR 1950X - Custom WC - Asus Zenith Extreme Alpha - G.Skill Trident Z 3400MHz C14 QC - RX VEGA 64
    R5 3600 - Custom WC - Asus Crosshair VII Hero Wifi - G.Skill Ripjaws V 3800MHz C16 - Sapphire HD 5850 Toxic

    24/7 OC: i5 4690K @ 4.9GHz CPU@1.255v 4.4GHz Cache@1.10v - Archon SB-E X2 - Asus Maximus VII Ranger
    Sapphire Fury X (1145/545 ~17.7K GS 3DM FS)

  10. #200
    ROG Enthusiast Array MattTheTech PC Specs
    MattTheTech PC Specs
    Laptop (Model)Nothing Interesting
    MotherboardASUS ROG Crosshair VII Hero
    ProcessorRyzen 9 3900x
    Memory (part number)3200C14D-16GFX (32GB Total)
    Graphics Card #1EVGA GTX 1080ti FTW3 Hybrid
    Sound CardHT Omega eCLARO
    Monitor3 Wall Mounted-1x Samsung 55" 4K TV (Runs 1440p 120hz), 2x LG 32" 1080p
    Storage #1Samsung 970 Evo
    Storage #23x Samsung 850 Evo 512GB in Raid 0 For 1.5TB Array Game Drive + ZFS 30TB NAS/Server For Storage
    CPU CoolerEKWB Dual Custom Water Loop
    CaseModded Rosewill Evo Armor 2 (I know all reviews hated it, but after my Mods its been a GREAT CASE)
    Power SupplyCorsair RM1000i
    Keyboard Corsair RGB k70 Lux V2
    Mouse Logitech G700s
    Headset Oculus Rift
    Mouse Pad Razor Vespula
    Headset/Speakers 7.1 Custom Built Surround Sound System
    OS Win 10 Pro 64bit
    Network RouterASUS RT-AC1900p
    Accessory #1 10Gb Nic
    MattTheTech's Avatar
    Join Date
    May 2015
    Reputation
    10
    Posts
    48

    Quote Originally Posted by KeithMyers View Post
    I can't ever compare my latencies to any of the posts since I don't do Windows and don't have any of the tools that everyone uses. I can see latencies with Intel MLC and Geekbench4.

    keith@Serenity:~/Downloads/Utils/MLC$ ./mlc
    Intel(R) Memory Latency Checker - v3.6
    Measuring idle latencies (in ns)...
    Numa node
    Numa node 0
    0 53.4

    Measuring cache-to-cache transfer latency (in ns)...
    Local Socket L2->L2 HIT latency 15.3
    Local Socket L2->L2 HITM latency 24.3

    https://browser.geekbench.com/v4/cpu/14148236
    Wow, so it appears I am doing something wrong.

    I wanted to give you a basis of comparison, since Geekbench is on Windows, I figured I would run it. I see you are on version 4.3.4, I don't have that version in my archive, but I do have 4.3.3, so I figured that would be close enough.

    I don't know for sure if scores across the Windows and Linux versions are supposed to be comparable, but I think they are, I have to look that up. But man... I am clocked at 4225Mhz all core right now, Its what I have been running while trying to get my memory dialed in before going back to my Per CCX Overclock. But you are clocked at 4.15Ghz, so I thought for sure our scores would be pretty damn close to each other, considering you are on Linux and we all know the Linux scheduler just handles Ryzen (and indeed all processors) better than Windows. Not to mention there is less junk Running in the background. Keep in mind I optimize Windows pretty damn heavily to not have any spying junk or services, tasks, or programs that I don't need running in the background, so I have much closer to the "fresh" desktop environment as possible, and when I got blown away by you, I went into my services and task manager and turned off EVERYTHING, including HWinfo, I literally had nothing but Explorer, and Geekbench running when I did the test I am about to post.

    And particularly for Integer based tests you fricking MURDERED my score. I remained pretty competitive throughout the other types of tests, and even beat you barely on memory tests and crypto tests, but that was by less than 2 percent on memory and barely over a half percent on Crypto, so margin of error with both. But then we get to the Integer tests, and you just plain and simply WHOOPED my A$$ by 29% (These are the Multicore Scores I am citing here, with single core you still bested me in Integer by 12.5%)!!! That is CRAZY!

    So I don't know if Linux is just that much more optimized for Integer based computing, or if I am just doing something horribly wrong with my setup, considering we have damn near identical hardware. You can dig further into the details of each score by adding a ".gb4" at the end of the online scores to get a better picture if anyone is curious. But now I am wondering if I should Boot up Ubuntu and see if I see a change in scores...

    Anyways, here is my Result:
    https://browser.geekbench.com/v4/cpu/14210825

    and if you want to see the broken down score, where you have to sift through text but can get an idea of how I scored on each test then you can go to:
    https://browser.geekbench.com/v4/cpu/14210825.gb4

    Any insight into this would be awesome!

    And Keith, You have an awesome setup man! I hope this Geekbench score helps create a bit of an idea of where you system sits. Hopefully more people will post their Geekbench 4 scores, I am just wondering if someone can confirm if the Linux vs Windows scores are supposed to be comparable.
    Last edited by MattTheTech; 08-08-2019 at 08:29 AM.

Page 20 of 46 FirstFirst ... 10 18 19 20 21 22 30 ... LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •