light flickering sometimes in port royale

[Sno0py]

out of 20 benchmarks i had a lot of flickering on my rtx2080 and an i9900k. This bench flickers like mad for me is this my pc? im bringing it back anyway but i want to say two things.

1. this software has been unreliable and i wont ever be using it again. Timespy flickering and other applications running fine lead me to beleive it is this "benchmark"

2. This is the 5th return i have had for different reasons (really bad luck)

[Sno0py]

reasons for the other returns, just so you dont think it was because of me thinking it was a gpu error, it was PSU boot looping on a really bad OEM (powerspec). another was a possible video error but random dhcp watchdog violations, the last one was a psu fan that literally buzzed and shaked and sqealed even when walking past it.

This, however, gave me a lot of time to test the 2080.

[Sno0py]

I really want to see if anyone gets the same results if they run the bench a few times. i showed a friend the bench twice and it happened the second time, then i reroduced it. The reason i say this is becasue

THIS SOFTWARE IS NOT A BENCHMARK BY DEFINITION

if, after i return this ENTIRE pc, i can reproduce this issue, this software is broken and not a benchmark. i would advise anybody not to use it to test their GPU. No Beef with the Devs it just is a straight up waste of money with the way drivers are being rolled out and how this software has been behaving.

A benchmark is a standard that other similar items can be compared to — that way everything is measured against a common standard. Benchmark originally referred to a mark made on a wall or building by a surveyor for use as a reference point.

[Seamus]

Sigh. Update your damn drivers. That's what's causing the black flicker in time spy. Nvidia addressed this days ago.

As for light flicker in port royal, that's usually the symptom of an unstable overclock in my experience.

3dmark is fantastic, you just don't seem to understand how to use it.

[Sno0py]

i do understand how to use it, and i do not think you understand what i am saying here. Everything here is stock. 3Dmark as a company is is not my concern. This product, however, is NOT a benchmark if it was made for ray tracing and it glitches out on every singe RTX card. It is inconsistant with its own test.

What i am saying is literally any other program is more of a benchmark because this code is not stable with the newest driver nvidia released, on multiple benchmarks, and in different ways. You seem to be thinking i am talking about this code as a game that has something that can be patched out and we can wait for. I dont use this GPU to game only. the primary purpose of this program is to test.

And yeah i also told someone the timespy flicker is a driver issue to calm them down... i know already... The issue here is. How is this program benchmarking anything? Other DirectX12 applications do not have the flicker. It is just a source of anxiety for RTX owners.

Why would i advise anybody to test their hardware with this software that causes stress, a forum search, and cannot even finish the pass without it behaving how it is. you literally needed a hotfix driver to bench for a week.

Originally posted by Seamus:

Sigh. Update your damn drivers. That's what's causing the black flicker in time spy. Nvidia addressed this days ago.

As for light flicker in port royal, that's usually the symptom of an unstable overclock in my experience.

3dmark is fantastic, you just don't seem to understand how to use it.

[UL_Jarnis]

Originally posted by Sno0py:

out of 20 benchmarks i had a lot of flickering on my rtx2080 and an i9900k. This bench flickers like mad for me is this my pc? im bringing it back anyway but i want to say two things.

1. this software has been unreliable and i wont ever be using it again. Timespy flickering and other applications running fine lead me to beleive it is this "benchmark"

Time Spy Flickering is NVIDIA driver bug. Do not blame us for it.

NVIDIA released a hotfix driver to fix it.

You would know this had you read this forum at all before starting to go on a rampage.

[Sno0py]

Originally posted by UL_Jarnis:

Originally posted by Sno0py:

out of 20 benchmarks i had a lot of flickering on my rtx2080 and an i9900k. This bench flickers like mad for me is this my pc? im bringing it back anyway but i want to say two things.

1. this software has been unreliable and i wont ever be using it again. Timespy flickering and other applications running fine lead me to beleive it is this "benchmark"

Time Spy Flickering is NVIDIA driver bug. Do not blame us for it.

NVIDIA released a hotfix driver to fix it.

You would know this had you read this forum at all before starting to go on a rampage.

We have spoke before Jarnis, you should remember me that time when you totally did not bother to get past the demo for timespy and antagonized the customers. You apparently do not read the forums. i have dealt with you enough and kept my calm. This is another rude post from UL_Jarnis here.

I did not blame you, i am saying your software does not work for modern stuff at the moment and people SHOULD NOT USE IT FOR THAT PURPOSE.

I did not have a problem with 3Dmark, i am trying to get people to have a working computer. Now you have given me a reason to have a problem with the staff at 3Dmark. even if it was not this time the other thing was not acceptable, writing the Timespy flicker off as a "non issue" and saying it was your customers hardware.

It is not a benchmark by definition for a lot of reasons. please read my previous post. This software has caused co workers who use the RTX, friends, and many people who use this to bench i know personally A LOT of anxiety and stress. It would be a benchmark by definition if it was not made for RTX, but PR is a ray tracing bench, and an unreliable one. (wether it is your fault, nvidias fault, or all RTX cards fault) If Ray tracing is supposed to freak out occasionally due to card clock rates than your software is not a benchmark because it is not a uniform test. What im saying is it should not be called a benchmark at the moment, or used for that purpose. maybe if you dont care if your video card is failing it can work as a bench for thermals or crashes. you literally need a forum post to know that default benchmark for this program flickers on almost every single RTX card today.



Also run the stress test, wait till the second pass, and compare it to the first one. Use the stress test and wait for the second pass. You will see objects out of place and stuff like that. That i will blame on you guys.

[UL_Jarnis]

It works for modern stuff just fine. Port Royal is not an "unreliable bench". It does not "freak out" on normally working hardware and drivers. It was extensively tested for launch and the drivers that existed at that time. If you are seeing some kind of issue, the simple checklist is;

1. Am I overclocking? If yes, try without overclock

2. Did I update drivers recently? Try rolling back to an older driver to see if the problem goes away.

We are just as disappointed as you for any driver related problems. In the past, NVIDIA has been very reliable on drivers, but apparently the latest WHQL ones have some issues. Do not judge hardware stability on just one driver version.

If you want to complain about drivers, NVIDIA has specific forum thread about it.

https://forums.geforce.com/default/board/33/geforce-drivers/

[Sno0py]

Originally posted by UL_Jarnis:

It works for modern stuff just fine. Port Royal is not an "unreliable bench". It does not "freak out" on normally working hardware and drivers. It was extensively tested for launch and the drivers that existed at that time. If you are seeing some kind of issue, the simple checklist is;

1. Am I overclocking? If yes, try without overclock

2. Did I update drivers recently? Try rolling back to an older driver to see if the problem goes away.

We are just as disappointed as you for any driver related problems. In the past, NVIDIA has been very reliable on drivers, but apparently the latest WHQL ones have some issues. Do not judge hardware stability on just one driver version.

If you want to complain about drivers, NVIDIA has specific forum thread about it.

https://forums.geforce.com/default/board/33/geforce-drivers/

i dont think you know what i am trying to do or say here man. im not trying to compain about drivers. i am trying to inform people that i have seen this software act unreliably when looking for "normal 3D ray tracing" or whatever you want to call that ON MULTIPLE MACHINES. I also said i will have a new machine i will be testing tomorrow. if it happens again, on completely new hardware. when i get the time to test it ill be sure to let you know. i test everything stock, the OEM i am testing now is IBUYPOWER. i have been working on this all day and thought about how much stress it is causing so i want to inform people to not get stressed out and why.

I understand you are defending your software. I am trying to say it may not be a good tool to test anything with. Why would i roll back drivers that work with ray tracing and DX12 in other applications better than before? do you not see the problem is not with you or nvidia i am trying to help the average user. you are literally playing devils advocate and blindly defending this product. There is better software to try ray tracing out on and see if it is working properly with this driver.

Edit :
You wrote

It does not "freak out" on normally working hardware and drivers. It was extensively tested for launch and the drivers that existed at that time. If you are seeing some kind of issue, the simple checklist is.

the keywords are "that existed at that time". There are reasons to test the new, overall more stable drivers for the applications i use. your software cannot test those drivers so what are you arguing about?

[UL_Jarnis]

I'm arguing that we have no control over drivers. I also don't think NVIDIA breaks drivers on purpose, but if there is a driver issue, you should report it on their forums, not come here to complain that our benchmark is somehow bad (when it has not changed in any way recently).

[Sno0py]

i took a video of the second flicker instance but it is tied to social media right now (it is 4am and i am tired, ill get to it if someone wants it)

Please try these two things

1. Use the stress test and wait for the second pass. You should see objects out of place (the thing to the left of the gold statue when the stress test starts its second pass, and a robot is still near a doorway near the beginning, cuts between frames seem to also change in reference to the first pass.
2. Run in benchmark mode multiple times and see if you can produce a lighting flicker, 10-15 should be enough.

[Sno0py]

Originally posted by UL_Jarnis:

I'm arguing that we have no control over drivers. I also don't think NVIDIA breaks drivers on purpose, but if there is a driver issue, you should report it on their forums, not come here to complain that our benchmark is somehow bad (when it has not changed in any way recently).

dude the problem with the second pass thing is deffinatly 3DMark software in stress test mode.

I hope i am wrong about this trust me.

edit: why would i go to nvidia forums to find benchmark results. i am trying to test something on the new drivers with your software i payed for, using it for what it is intended for, in the forums for it.

***Also if the newest drivers are considered the default for the product... this is not a working benchmark as of now. you keep telling me it has not changed recently and i know that. I am simply using the word benchmark as it is defined.

I want to see repros, and see if anyone experienced the same thing. why else would i post the OP.

[Seamus]

Originally posted by Sno0py:

i took a video of the second flicker instance but it is tied to social media right now (it is 4am and i am tired, ill get to it if someone wants it)

Please try these two things

1. Use the stress test and wait for the second pass. You should see objects out of place (the thing to the left of the gold statue when the stress test starts its second pass, and a robot is still near a doorway near the beginning, cuts between frames seem to also change in reference to the first pass.
2. Run in benchmark mode multiple times and see if you can produce a lighting flicker, 10-15 should be enough.

1: This is entirely false if there's nothing wrong with your computer.
2: I spent four hours running port royal runs on my 2080 last week. I never once got the lighting flicker unless my overclock on my gpu was too high.

Everything you're complaining about is user error.

[Sno0py]

Originally posted by Seamus:

Originally posted by Sno0py:

i took a video of the second flicker instance but it is tied to social media right now (it is 4am and i am tired, ill get to it if someone wants it)

Please try these two things

1. Use the stress test and wait for the second pass. You should see objects out of place (the thing to the left of the gold statue when the stress test starts its second pass, and a robot is still near a doorway near the beginning, cuts between frames seem to also change in reference to the first pass.
2. Run in benchmark mode multiple times and see if you can produce a lighting flicker, 10-15 should be enough.

1: This is entirely false if there's nothing wrong with your computer.
2: I spent four hours running port royal runs on my 2080 last week. I never once got the lighting flicker unless my overclock on my gpu was too high.

Everything you're complaining about is user error.

Thanks for the info that is all i wanted. More results, what driver if you do not mind?

to answer your questions

1. That is what i am trying to find out, and the use of a benchmark.
2. When you say "last week" you did not mention what driver you had installed so this information's usefulness is limited, thanks anyway and please update.

To respond to your statement, I am asking someone to do something on a benchmark to see if it produces the same results i noticed.
here is the definition again-
A benchmark is a standard that other similar items can be compared to — that way everything is measured against a common standard. Benchmark originally referred to a mark made on a wall or building by a surveyor for use as a reference point.

[Seamus]

Originally posted by Sno0py:

Originally posted by Seamus:

1: This is entirely false if there's nothing wrong with your computer.
2: I spent four hours running port royal runs on my 2080 last week. I never once got the lighting flicker unless my overclock on my gpu was too high.

Everything you're complaining about is user error.

Thanks for the info that is all i wanted. More results, what driver if you do not mind?

to answer your questions

1. That is what i am trying to find out, and the use of a benchmark.
2. When you say "last week" you did not mention what driver you had installed so this information's usefulness is limited, thanks anyway and please update.

To respond to your statement, I am asking someone to do something on a benchmark to see if it produces the same results i noticed.
here is the definition again-
A benchmark is a standard that other similar items can be compared to — that way everything is measured against a common standard. Benchmark originally referred to a mark made on a wall or building by a surveyor for use as a reference point.

If you spent a few minutes reading the forum instead of writing posts complaining and attacking the devs, you'd know that the black flicker in time spy is caused by 430.39 and is fixed by the 430.53 hotfix.

I was using 39 for my port royal runs, which ran just fine, and am now on 53 to deal with the black flicker in time spy.

Seriously. User error.

And stop throwing out the dictionary definition of benchmark. No one cares.

[Sno0py]

Originally posted by Seamus:

Originally posted by Sno0py:

Thanks for the info that is all i wanted. More results, what driver if you do not mind?

to answer your questions

1. That is what i am trying to find out, and the use of a benchmark.
2. When you say "last week" you did not mention what driver you had installed so this information's usefulness is limited, thanks anyway and please update.

To respond to your statement, I am asking someone to do something on a benchmark to see if it produces the same results i noticed.
here is the definition again-
A benchmark is a standard that other similar items can be compared to — that way everything is measured against a common standard. Benchmark originally referred to a mark made on a wall or building by a surveyor for use as a reference point.

If you spent a few minutes reading the forum instead of writing posts complaining and attacking the devs, you'd know that the black flicker in time spy is caused by 430.39 and is fixed by the 430.53 hotfix.

I was using 39 for my port royal runs, which ran just fine, and am now on 53 to deal with the black flicker in time spy.

Seriously. User error.

And stop throwing out the dictionary definition of benchmark. No one cares.

Have you read this post once? i was involved with the the entire Timespy flicker issue and read the nvidia post, and you were a toxic element in every single post i saw literally making bets on who certain customers will blame. Please go elsewhere you are not helpful, nor do you understand the fact that i am trying to test Port Royal, not Timespy, on this driver in particular.

Shhhhh already. You really dont understand what a benchmark is, or that i am trying to test this benchmark on this driver, in particular, for my own reasons, as previously mentioned. I brought up Timespy to make a reference to the DX12 code having a flicker issue and it POSSIBLY being related. (not to mention just the fact that there was flickering on the driver in general) Go light a flame war somewhere else man. i want info, not your crap. Like really look at your posts why are you even here? to be a mean person to people having hardware issues? and yes i have read a LOT of your posts on A LOT on these threads because you seem to really care about antagonizing upset and frustrated customers with broken hardware. RTX hardware that is VERY expensive.