2017 Hardware Thread

solarstarshines said:

Yes Sir

Well, I think it is a given that Ryzen can't run on an Intel Platform, so they would have to be talking about hamstringing the 'competition'.

If, and I emphasize if, AMD is telling reviewers to do that, I have a few observations.

1. They are absolutely stupid. Word getting out of them doing this would ruin AMD's reputation, not to speak of the reviewer's reputation.

2. Again, if AMD is telling reviewers to do that, it is very unethical and borders on defrauding the public. It would also show how high they regard their customers.

3. It would be a tacit admission that they cannot compete with Intel.

I would really like to know what @kipper has to say about this. Does it sound like something AMD would do and is this type of request common among reviewers?

I would like to know which reviewer that came from. I honestly find it hard to believe that AMD would be that stupid.

I've always sort of liked Paul. I did notice he sort of breezed over that last part.

For the HPET portion, it can be a good idea using intel too on some setups, but can be deadly for those that use overclocking software instead of BIOS.
I recommend one of you OC UEFI/BIOS Gurus to take a look at the complete thread.
I don't quite understand it yet, but I would like to here your recommendation, I think I have HPET also on in BIOS and that could be my stability problem with some things, like using REALBENCH that screws up my system.:

Enabling HPET in BIOS is just half way of enabling HPET, it needs to be enabled in OS too, and in a way that it's the only timer used.

By default windows uses combination of TSC+ACPI timers, not matter if HPET is enabled in BIOS.
TSC+LAPICs Low performance (slow timers + syncing)
LAPICs low performance (slow timer - no syncing)
TSC+HPET medium performance (slow and fast timer + syncing)
HPET high performance (fast timer - no syncing)
HPET + platformclock=true will give you best timer resolution, frame rate and lowest DPC latency.
You can test timer ratio and QueryPerformanceFrequency with WinTimerTester 1.1 http://www.mediafire...xzo9n84d8lze9nb
The higher the QueryPerformanceFrequency is the better is performance. You only get high frequency with HPET. The other timers will give you significantly less frequency. Also note that if your ratio is not 1.0000 you are off set (or you have wrongly OC'ed), enable HPET and you should be without sync problems.
If you ever want to go back to default timers admin cmd:
bcdedit /deletevalue useplatformclock
Varying depending on setup, one should get increase up to +30 FPS and from the between.Online games is a good example of boost from HPET.

TSC+LAPICs Low performance (slow timers + syncing) = 2.76MHzLAPICs low performance (slow timer - no syncing) = 3.5Mhz
TSC+HPET medium performance (slow and fast timer + syncing) = 3.8Mhz
HPET high performance (fast timer - no syncing) = 14.3MHz
Run the WinTimerTester 1.1 to see your QueryPerformanceFrequency

Then try with HPET, you'll be amazed.

Because if you enable HPET in BIOS you will mix it with other timers and CPU needs to sync between them. You need to force HPET only and reboot. Mixing HPET with other timers is not good, but using HPET only will give you best performance. People don't know about that Windows needs to be forced to use HPET only - that's why the bios setting alone can undergrade their performance.
bcdedit /set useplatformclock true (then reboot) enable HPET
bcdedit /deletevalue useplatformclock (then reboot) disable HPET

TWEAK: Enable HPET (in BIOS and OS) for better performance and FPS - OS Customization, Tips and Tweaks - Neowin Forums
Also it could explain why I have problems with intel(R) Integrated Clock Controller Service when I downclock from a previous OC setting, like in my post #308 above.

Also:

Hardware timer info

TSC RegisterSome Intel and AMD processors contain a TSC register that is a 64-bit register that increases at a high rate, typically equal to the processor clock. The value of this counter can be read through the RDTSC or RDTSCP machine instructions, providing very low access time and computational cost in the order of tens or hundreds of machine cycles, depending upon the processor.
Although the TSC register seems like an ideal time stamp mechanism, here are circumstances in which it can't function reliably for timekeeping purposes:

• Not all processors have TSC registers, so using the TSC register in software directly creates a portability problem. (Windows will select an alternative time source for QPC in this case, which avoids the portability problem.)
• Some processors can vary the frequency of the TSC clock or stop the advancement of the TSC register, which makes the TSC unsuitable for timing purposes on these processors. These processors are said to have non-invariant TSC registers. (Windows will automatically detect this, and select an alternative time source for QPC).
• On multi-processor or multi-core systems, some processors and systems are unable to synchronize the clocks on each core to the same value. (Windows will automatically detect this, and select an alternative time source for QPC).
• On some large multi-processor systems, you might not be able to synchronize the processor clocks to the same value even if the processor has an invariant TSC. (Windows will automatically detect this, and select an alternative time source for QPC).
• Some processors will execute instructions out of order. This can result in incorrect cycle counts when RDTSC is used to time instruction sequences because the RDTSC instruction might be executed at a different time than specified in your program. The RDTSCP instruction has been introduced on some processors in response to this problem.

Like other timers, the TSC is based on a crystal oscillator whose exact frequency is not known in advance and that has a frequency offset error. Thus before it can be used, it must be calibrated using another timing reference.
During system initialization, Windows checks if the TSC is suitable for timing purposes and performs the necessary frequency calibration and core synchronization.
PM ClockThe ACPI timer, also known as the PM clock, was added to the system architecture to provide reliable time stamps independently of the processors speed. Because this was the single goal of this timer, it provides a time stamp in a single clock cycle, but it doesn't provide any other functionality.
HPET TimerThe High Precision Event Timer (HPET) was developed jointly by Intel and Microsoft to meet the timing requirements of multimedia and other time-sensitive applications. HPET support has been in Windows since Windows Vista, and Windows 7 and Windows 8 Hardware Logo certification requires HPET support in the hardware platform.

– – – – Begin Email Conversation 2 – – – –
From JS:
I like your article on this very much, but you neglected to mention the aspect of overclocking and it’s effects on timing.
For example, if I overclock my FSB via SetFSB with the default timer, WinTimerTester shows that the longer I run it, the more out of sync QueryPerformanceCounter and GetTickCount get.
This causes severe problems in some games, notably “Battlefield: Bad Company 2”. I’m only able to play it for a few minutes before the game gets horribly out of sync and is unplayable.
If I set useplatformclock to true, I don’t have this problem. I get a near perfect 1.000 ratio in WinTimerTester while overclocked, and those games don’t get out of sync anymore.
More info here, Must-Read: SetFSB O/C Game Problems FIXED | NotebookReview
Video – Crysis 3 – 30% Framerate and Performance Boost, and FPS Cap Removal [120Hz.Net – HyperMatrix]
– – – – End Email Conversation 2 – – – –

Timer Tweaks Benchmarked - TweakHound

Cliff S said:

For the HPET portion, it can be a good idea using intel too on some setups, but can be deadly for those that use overclocking software instead of BIOS....[snip]

I can't make heads or tails as to how this works, don't work, or performs so it's all mumble jumbo to me. That said, I've had it enabled in the BIOS cause I don't know. Did it when I built the system last year. Anyway I thought this might be a better link since it's much recent and was active as late as last week. Hope you can figure it out and enlighten us

• Windows 10 - disable HPET before install and enjoy low latencies!

sygnus21 said:

I can't make heads or tails as to how this works, don't work, or performs so it's all mumble jumbo to me. That said, I've had it enabled in the BIOS cause I don't know. Did it when I built the system last year. Anyway I thought this might be a better link since it's much recent and was active as late as last week. Hope you can figure it out and enlighten us

• Windows 10 - disable HPET before install and enjoy low latencies!

I have been reading around on it this morning, and basically, it helps for some, and makes things worst for others, and again doesn't do anything to a 3rd group. It depends on what you use your PC for, if and how you overclock(BIOS vs software OCing), and how high & stable the OC is.

Also what I call "settings drift", which is any setting in electronic equipment will drift after a while, like you TV/monitor calibration, your BIOS settings, and it seems for overclocking specially so, can show the set values, but should be checked and redone after a while for stability, as they can drift, meaning, for example
: I had VCore set to 1.45 and when it was set to Override, showed 1.450 initially in monitoring software, but after a month of using it, later in the monitoring software it showed the actual value to be 1.443, that is drift(it can also go in another direction(higher)).

So I'm guessing that the timers can also drift, but it's only noticeable on machines set to run on the edge.

Yeah, I kind of get some of it and understand timings, but in the end still can't figure if enabling/disabling nets any real performance to be of any worries. And, I'm still not sure if Windows 10 itself needs this done in OS as well, or if Windows 10 does it automatically. A lot of the talk on the OS side of enabling mainly centers around Windows 7, though there is some 8 & 10 talk. In the end, I can't tell if there's a performance boost or not. It's all maybe; it depends; probably.