FPS drawdowns in densely populated areas (The Witcher 3: Wild Hunt)

VOVAN WOLFovich
Can't it be like that, do you understand? Not in our reality that the fucking 1700 is a throat and does not pull the Witcher 3 in Novigrad. Are you out of your mind at all? Because of you, my paranoia began, I went to YouTube and 100% of vidos from 1600-1700 in The Witcher 3 were not lower than 90-100 + fps with a norm card. In Novigrad and so on. It just doesn't happen, you know? Why are you writing all this? Wake up.
What kind of death was he overclocked there? (FX) Up to 4.2 this is enough, almost for a runoff. With cooling from a turret for 2 pipes and a 90mm vent. I've had FX since 12 or 13 years old. It should be driven on the tire, not on the multiplier. And yes, you can use it without overclocking. But this does not negate the fact that even he will pull The Witcher 3 and he will not show such drawdowns. You can write anything, but this is not normal behavior and the problem is not that 1700 is WEAK and does not pull the game.
Well, about 228fps even in the voice with you. Ozu overclocking always gives an increase in FPS, and not as you write, just latency will drop a chunk, especially if you leave the frequency drain. In general, you write as much as possible that is not related to our reality.

Neon Abyss
Neon Abyss wrote:
1.39 is a gimmick, right?
So
Neon Abyss wrote:
It turns out 1.39 at 3000MHz isn't it too much?
In XMP 1.35v. I tried the same manually, it does not work stably (either a blue screen or crashes from games)
Neon Abyss wrote:
Did you twist the secondary or what?
Resale trogal.Poka not like this:
Spoiler
Neon Abyss wrote
a 3000 defoltnoy naprugi 99% zavedetsya
I defaulted 2133Mgts and 1.2v

vnlmeo
vnlmeo wrote:
1) This function should at a low priority
I have in the middle:
Spoiler
In Boklere :

But Ass Grove:

vnlmeo wrote:
the game does not lag in densely populated areas and everything goes + - smooth
Continuous lags. Not in the processor. Your card for the Witcher ultra settings is 3 bottom, only for medium. Perhaps you will set some settings high. I set high on 7950 in FHD
I even managed to shoot a video on it:
Spoiler

VOVAN WOLFovich
Spoiler
VOVAN WOLFovich wrote:
You monitored with minimum, average and maximum. I guess not, which means AVERAGE is indicated. It is necessary to analyze these 30 frames to argue that yes, it really was not.
I do not see the need for this. I'm mainly interested in the minimum observable FPS that shows the fraps / bandicam counter and which is always running as usual. if you monitor min max fps, then I admit that fps can sag somewhere for some reason when loading or something like that, but this is not a factor affecting gameplay. or the maximum FPS, which will jump under the over 90 for half a second. if we say so, the average FPS was something about 45-50, but in particular in novegrad, closer to the port, it strove to 30 and was not stable. I also did not monitor what the GPU is doing there all the time, but there was no serious emphasis on it. The only thing was when I gathered a couple of packs of wolves, the fps dropped to 20-25 fps on the GPU because of the haerfox. in general, you can tell a lot about fx, but this does not apply to the topic. the problem of fx-sa is low performance per core at the n-th frequency (it is comparable to the epic Intel core and fen 2 processors, while u5-7 are faster at the same frequency per stream, plus they have a number of problems). Kuruuzens are devoid of this problem and are comparable to normal Inel processors of their time. for example p1700 is approximately equal in benchmarks and 7 5960x at the same frequency. Perhaps the curuzen have some specific problems, like fx, in which they can begin to seriously lose in the performance of some tasks (but to be honest, I did not really monitor and was interested. In games sharpened for a multi-thread, the p 1700 works mostly not bad. Kuruuzens are devoid of this problem and are comparable to normal Inel processors of their time. for example p1700 is approximately equal in benchmarks and 7 5960x at the same frequency. Perhaps the curuzen have some specific problems, like fx, in which they can begin to seriously lose in the performance of some tasks (but to be honest, I did not really monitor and was interested. In games sharpened for a multi-thread, the p 1700 works mostly not bad. Kuruuzens are devoid of this problem and are comparable to normal Inel processors of their time. for example p1700 is approximately equal in benchmarks and 7 5960x at the same frequency. Perhaps the curuzen have some specific problems, like fx, in which they can begin to seriously lose in the performance of some tasks (but to be honest, I did not really monitor and was interested. In games sharpened for a multi-thread, the p 1700 works mostly not bad.
As for 4-core processors, the Witcher 3 no longer belongs to the era of 4-thread processors, so it is natural to expect that all stocks and 3 and 5 will not work normally in games because the norm is 8 threads. At the same time, the Witcher 3 is not processor-dependent with respect to the games of its time, both on the ratio of GPU requirements to a CPU and to a CPU with a conditional infinite video card (some kind of weak at a low resolution).

Neon Abyss
"I also have a slightly different risen" .... Ahahah, as if we are talking about some kind of console.
That's right ... and, there is no difference, all the risers are as one, regardless of the architecture, tdp, the number of cores and threads, the supported memory frequency, the chipset on which it runs, they are so risen ... just a bit different ... 🤣
vnlmeo
It would be nice to describe the entire configuration down to the power supply unit.
Regarding the voltage, it is better not to exceed 1.35v, on cheap motherboards, without heatsinks for VRM and proper cooling in general.
If there were any manipulations in the BIOS, then I would throw it off.
It would also not hurt to check the firmware for the motherboard, chipset and update.

IngwardIn
Hi.
1) As for overclocking the RAM, I already figured it out and decided to stop at 3000MHz at 1.2V for now.
2) A little later, I will throw off the link to the video test in the game before overclocking the RAM, so that you can more clearly see the whole picture of the PC in the game. Together with the test, I will throw off the complete PC configuration, if I don't forget.
3) Resetting the BIOS does not help, I tried it. "The number of characters on the screen" was and is at a minimum.

vnlmeo Changed the
message, read it from the phone and got a little confused.
You can also try to look into the paging file, if there are two disks, then put the paging not on the system disk. You can also try to fix the size of the paging file 8000-8000.

I do not advise listening to the mother's overlovers, who assure that during overclocking nothing can be disabled, it all depends on what you overclock, on what and through what.
The fact that you approach this business with pedantry is absolutely correct, but most likely overclocking to 3000 requires a change in the primary timings, the stability of the RAM can be checked by OCCT or Aida64.
Again, it would not hurt to check the firmware for the motherboard, chipset and update.
In general, if nothing helps, I would, in principle, run the components in the OCCT tests.

vnlmeo
To be honest, after watching the video, I did not notice colossal drawdowns in the settlement and more growth outside.
There were drawdowns, but in my opinion, these drawdowns more rested on the GPU, in those moments when Geralt was running on water, among the grass ...
In any case, overclocking RAM to 3200 is needed and should provide an increase, in some projects + -10-15 fps ...

IngwardIn
You watched the wrong video, since I have not yet uploaded the video to YouTube and did not upload it)
I thought about overclocking the RAM to 3200, but did not dare to increase it yet, because I need to get better acquainted with such a topic as timings and their calculation. I know that there is a special calculator for calculating timings, but since my mother is on the A320 chipset, I cannot select the desired chipset in the special program and calculate the timing data (

The Witcher 3 test on my PC (trash for quality)

vnlmeo
you so did not write your configuration.

In my opinion it is not normal that the load on the GPU processor drops to 0%. Together with it and the consumption of the card, respectively, there is a reason to sin on the power supply. Most likely, it simply cannot provide the required power for the video card.
During the entire test, the consumption was unstable and below 80%, which means that the card, in principle, does not receive proper power. From that and the FPS is low!

Either the power supply unit does not correspond to the configuration, or there is a malfunction. Run the POWER stress test with auto instructions at OCCT. If the PC goes out, then there is no need to go to the fortune-teller.
Well, as the most obvious option, what if you have it - check the back of the wire, whether it sits tightly in the power supply unit itself, if there is no BOOK DEEPER!
And check the pins / wires on the video card itself.

IngwardIn
IngwardIn wrote:
In my opinion it is not normal that the load on the GPU processor drops to 0%.
For the RX5xx series, this is the norm. I have the same hat on the RX588, except for games where you can turn on MSAAx8

vnlmeo
Well, show the frequency of your processor in the overlay. Maybe it works for you at 1550 MHz. Show it on all cores. And shoot a video. preferably using the driver with the keys Ctrl + Shift + E

Denis Kyokushin
Consumption of 0-30% would be the norm if there were no drops, freezes and low fps.
In my opinion, continuous consumption of 70-100% could still be called the norm.
In any case, this does not negate the fact that it will not be superfluous to conduct a number of stress tests.
And I certainly don't know the full potential of the 1700x, but the load on it is suspiciously low.
Still, it seems to me that the problem is that performance comes down to power.

Well, either at the extreme, precision boost is disabled.

vnlmeo
You might also think about a motherboard, but the A320 is not the best chipset, it hardly has a good power supply, and it may well overheat and not cope with the power supply. You can probably forget about overclocking the ozu on this chipset, if you manage to start it, then most likely with inadequate values.

IngwardIn
Maybe he turned on the energy saving

PSU: Aero Cool Eco-450W
Motherboard: Asrock A320M-DVS R4.0
RAM: AMD r744g2606u1s 4GB 2 dies
Video card: RX560 4GB
Processor: Ryzen 7 Pro 1700X
HDD WD-blue
IngwardIn
Here is the PC configuration.
A little later I will conduct tests and describe everything as it went.

Denis Kyokushin
If you are talking about a diet plan, then I have a balanced diet. I tried to change it to high priority, but it did not give much result.

IngwardIn
precision boost is probably off, so you need to take a look.

vnlmeo
Show the core frequencies on the video

I tried to find this precision boost of yours, but nothing worked. Apparently my mother doesn't support this feature.

vnlmeo
There is an AXMP profile in the BIOS, turn it on in position 2, this will increase the frequencies of the RAM, or it is simple where the overclocking of the RAM there will be a line that changes the values ​​of timings and frequencies and set the frequency a little higher and at this frequency the minimum timings for me, for example, work stably at 3466mhz, although the original maximum frequency was 3200mhz