Sounds to me as if there appears some type of a conflict after enabling it in BIOS, thus the recommendation to keep it disabled at all times and increasing the performance in CPU bound games.
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Ryzen "2" ?
- Thread starter beany_bot
- Start date
Sounds to me as if there appears some type of a conflict after enabling it in BIOS, thus the recommendation to keep it disabled at all times and increasing the performance in CPU bound games.
Most modern mobo's bios's don't give you an option to enable or disable HPET. It's simply enabled by default anyway, because it's up to the OS to make the call to use it or not. If you alter the boot script of the OS to not use HPET, that's all it means, the OS won't use it. It still dosn't alter the fact that the bios has it enabled.
It's nice to see a followup, but HPET isn't a level playing field since nowadays all modern operating systems just default to TSC for Intel and I think APERF for AMD since AMD has some issues with TSC accuracy. There are a lot of other timers Windows will resort to before ending up on HPET, which comes with a performance impact for both platforms (previously more so on AMD systems, but after meltdown the tables have turned).
HPET seems to have been affected by the meltdown+spectre fixes on the Intel side since applications need to switch to kernel mode to read HPET.
@kitfit1 I'm running custom bios with HPET disabled same as spectrum for south bridge. And few other tweeks that are not possible to change in Normal Asus Bioses.
YES i have benefit's from not having HPET AT ALL latency goes down.
Also there is software script to reduce Windows latency. My windows and system in general is tweeked to the maxxxxx
C7H is stable on my C6H settings. Fan and sensors are FIXED !!!!! Finally
YES i have benefit's from not having HPET AT ALL latency goes down.
Also there is software script to reduce Windows latency. My windows and system in general is tweeked to the maxxxxx
C7H is stable on my C6H settings. Fan and sensors are FIXED !!!!! Finally
Found it for ya. Thats how You cut latency from 1500us to 50us :]
https://forums.guru3d.com/threads/windows-timer-resolution-tool-in-form-of-system-service.376458/
I'm always after max max tune up options
https://forums.guru3d.com/threads/windows-timer-resolution-tool-in-form-of-system-service.376458/
I'm always after max max tune up options
I get similar LatencyMon readings without doing any "tweak" apart from disabling Global C-States in the UEFI.
And doesn't that tool just make the Windows timers less accurate? Not sure it's going to help that much, TSC and APERF, which is what Windows is using should have no overhead.
And doesn't that tool just make the Windows timers less accurate? Not sure it's going to help that much, TSC and APERF, which is what Windows is using should have no overhead.
cant remember if that runs good on windows 10 try DPC latency checker I'm on 53-500 there before i was at 500-1500.
Dont have moded bios for C7H yet thats coming at some point
I just like to dig deepest i can in system. Think I got every possible registry tweek done on this system. Slammed new board no reinstall ect scores only 5 in cb15 less than C6H :]
Will wait with full windows reinstall and tweeking till 2700x
BTW. My pc got all C states enabled downlocking downvolting coreparking.
Last edited:
Permabanned
- Joined
- 23 Apr 2014
- Posts
- 23,552
- Location
- Hertfordshire
Trying it now with -0.1 and PE level 3. All core clock goes slightly higher(4050 ish) but max boost slightly lower (4350).
All core ends up about 1.2v and single core boost does hit 1.45, think that's fine on these processors.
Seems these can be pushed quite hard with bclk increase, bit more offset voltage and PE, 4.5+ boost clocks, not sure its worth it though.
Just trying to get my memory stable at 3500MHz at the moment, boots into windows fine using The Stilts 3466 C15 timings but IBT fails after a few loops. So it's close, time to play with the SOC voltage
Got me a few extra points in CB ~1830
Got me a few extra points in CB ~1830
Permabanned
- Joined
- 23 Apr 2014
- Posts
- 23,552
- Location
- Hertfordshire
Yeah, didnt pass burn test with -0.1 or -0.9ish.
was that under water?
Dry Ice.
Please don't misunderstand me. I am saying the testing was a level playing field. They both had a brick the same size tied to their ankle. Just one dealt with it better. The results obviously weren't right.
I am very happy to see Intels results restored tbh, I was doubtful of the new chips besting the 8xxx stuff in most games.
@Zeed disabling C-States, core parking and all of the power saving features will do a lot more for your latency than changing timer accuracy, if anything changing timer accuracy will make programs like LatencyMon a lot less accurate and might be the reason why you're seeing 'lower' latency numbers in it. Also don't use DPC latency checker for Windows 7 and up, it's not going to be accurate, LatencyMon is better, but I think you can affect its accuracy if you change timer resolution (which I presume is what that application is doing).
Ryzen is really good with power management, disabling C-states, using a fixed frequency, using the high performance power plan will only result in a minimal increase in power used.
@Scougar Oh, I agree, but Anadtech should have known better than to force HPET for all of their testing since it's non stock behaviour.
But on the other hand they did discover something neat, don't think that many people realized that HPET incurred a performance penalty after the spectre+meltdown patches, albeit if you think about it then it's obviously something that was going to be affected by the OS and microcode patches since HPET is a system call. HPET being higher resolution on Intel probably adds to the bigger performance impact the Intel platforms had.
On another note, AMD seems to have some issues with TSC accuracy according to Agner Fog. Seems the way to mitigate that is to run Ryzens at a fixed frequency.
Ryzen is really good with power management, disabling C-states, using a fixed frequency, using the high performance power plan will only result in a minimal increase in power used.
@Scougar Oh, I agree, but Anadtech should have known better than to force HPET for all of their testing since it's non stock behaviour.
But on the other hand they did discover something neat, don't think that many people realized that HPET incurred a performance penalty after the spectre+meltdown patches, albeit if you think about it then it's obviously something that was going to be affected by the OS and microcode patches since HPET is a system call. HPET being higher resolution on Intel probably adds to the bigger performance impact the Intel platforms had.
On another note, AMD seems to have some issues with TSC accuracy according to Agner Fog. Seems the way to mitigate that is to run Ryzens at a fixed frequency.
i wonder how he got the ram that fast. mine is slower on dice than air on the ram.
I have 3600Mhz ram (bought it September 2016) , and worked without issue with the 1800X.
The problems with Ryzen start when you try to overclock ram above their factory rated speeds.
I bet if anyone has 4000-4500mhz module, and plug it on the Ryzen 2 with X470 board, it will work also.
Not With High cpu OC usually I cant run my cpu OC AND memory at rated 3733. Can run Stock cpu and 3733 on C7H could not even boot 3733 up on C6H !!!!