About to upgrade to an RTX 3080 & Ryzen9 5900x. Running 1440p, non-VR. Anyone else here have a similar setup that can recommend solid graphics settings?

[Tree_Beard]

I'm not playing very competitively, so ideally I'd like the settings balance to lean more toward quality than performance, within reason of course. 

My current rig is an i7-7700k and a GTX 1080. I get decent performance right now (again, non-VR), but hoped to get some community input on what settings others are running with the gear mentioned in the title. Screenshots would be great!

Thanks all. 

[xoxen]

Hi Tree_Beard,

I have a similar system like the one you want to go for. The CPU is sightly smaller but this should give you an idea of what to achieve:

Except of SSAA, SSLR and SSAO I put everything on max settings at 1440p resolution and achieve 75 FPS in the F-16 Mission King of the hills. Simply start the mission and do not change the view. If you want me to test other settings please let me know.

Cheers,

Xoxen

Edited November 3, 2021 by xoxen

[Tree_Beard]

Thanks, this is very helpful. 

[av8orDave]

I agree with the above post. My setup is similar (5950X, RTX 3080) and I run the same settings with about the same results as above. Now, if you run VR, all bets are off.

[Thinder]

On 11/3/2021 at 8:02 PM, Tree_Beard said:

I'm not playing very competitively, so ideally I'd like the settings balance to lean more toward quality than performance, within reason of course. 

My current rig is an i7-7700k and a GTX 1080. I get decent performance right now (again, non-VR), but hoped to get some community input on what settings others are running with the gear mentioned in the title. Screenshots would be great!

Thanks all. 

According to the title of your topic, you're looking at RTX 3080 & Ryzen9 5900x, this would take care of one bottlenecks you would have by upgrading your GPU only but from my PoV, it is also an overkill when it comes to the CPU.

If you run Bottleneck Calculator you'll figure that a 5600X would do you just fine, the Ryzen9 5900x will give you little gain when it comes to FPS especially if you don't take care of your RAM.

Bottleneck Calculator. Ryzen 5600X GTX 3080

Better upgrading to a Cl14 4 X 1 rank 3200MHz B.Die kit which will give you an extra 6%+ (6.04% in my case) headroom at 4K (where the CPU would throttle down under load because of controller limitation).

Controllers are limited to a maximum number of ranks (4) and frequency (3200MHz), but the Ryzen 5 5600X (and all new AMD GPUs) take advantage of lower latency a lot more than relying on higher frequencies, so there is no point in fitting 3600MHz if your sticks are 2 X ranks Cl16, your CPU will still see the same limits under load.

I recommand a B.die kit, this material is a lot more stable and the chips made of it allows for a much tighter timings than non B.Die RAM, this garanty that your RAM will be capable to run under the optimum settings which Cl16 doesn't allow for.

B-Die Finder

Of course there is a premium coming with this sort of RAM quality, but here are the advantages: No need to O.C your CPU, I gained 6.04% in CPU speed and 1.33% in GPU speed due to the fact that there is no longer any bottleneck in those channels (RAM-to-CPU or CPU-toGPU) and your system will keep running at a higher percentage of load, which is what you want with a ga,e like DCS,

6.04% is more than double what you can expect by O.Cing your CPU with a water cooling kit in average, during those tests, my CPU never went over 76°C and I only have a cheap Artic Freezer 7X air cooler.

Without even taking this into account, pitted with a 3080, my 5600X has only an average bottleneck percentage: 3.47% (GPU bottleneck), meaning it easily can cope with load even with a Ti since the high average is 10% at 4K and the bootleneck source is the GPU, that's before you use boost as well.

You don't need more than 32 GB of RAM. Your RAM will reprocess data a lot faster than 64GB of Cl16.

Your 5600X will cost you <> £197.00+ less than a Ryzen 9 5900X which will still be limited under load since they all share the same controller, from my PoV, it makes no sense to fit a 5900X and keep the RAM bootleneck with it, better spend the difference in a higher quality RAM kit and solve the issue from day one, you won't have any bottleneck at all, not with the RAM, not with the GPU at 4K.

Hope this help.

Edited November 5, 2021 by Thinder

[BitMaster]

I run a true 3200-14-14-14-34-17 kit, 4x16GB, and it is outperformed under full stress from the same kit at 3600-16-16-16-36-1T.  There is no drop in performance anywhere, neither my fps go down, my latency is ~99% the same where as Aida64 tells me I improved by a few nanoseconds, bandwidth is a little higher as expected. 

Bottom line, I cannot see any proof that 3200/14 is always better than 3600/16 nor does my CPU go nuts and cuts down on performance.

The good thing is, I saved quite a bit of cash getting the 3200/14 and oc them to 3600/16. They are basically the same kit. My other true 3600/16 kit ( only 32GB, 4x8) has 3200/14 as one of it's profiles... OK, I need to up the Volts to 1.4v but that is well within DDR4 specs for permanent usage.

[Thinder]

FACT i : You use a B.die kit which you can O.C and I have the proof and provided with the data, I would get similar results if I did O.C my kit at 3600MHz simply because those kits have the right range of timing to do so, looks like you missed some serious studies on the subject, in my case, I upgraded from a non-B.Die kit and I explained clearly what the difference was.

More to it, Intel technicians are using exactly the same G.skill kit than mine for O.C their RAM and I also already have been explaining the reaons why and how.

So, to your benefit I'll reiterate: Controllers have limitations due to number of ranks and lfrequency and for all Ryzen of this generation the recommanded frequency is 3200MHz, this is simply because they assume that the average user is not going to pay a premium for a B.Die kit and therefore woun't be able to provide the CPU with the right timings.

As a matter of fact, the day I tried to O.C my old RAM kit to 3600 MHz (Crucial 2x16GB DDR4-3200 Cl16), I lost about <> 1% in the same test at the same settings.

How to Overclock RAM.

B.Die have been used by previous generations of Intel and AMD players alike precisely for this reason, they both were relying more on higher frequencies for performances but this cannot be achieved with non-B.Die kit which is the reason for my system to gain 6.04% at 4K.

Please read carefully before replying. Cheers.

Oh, I forgot to mention, by O.Cing your RAM, you also lost your warranty, which is the reason why I didn't and am more than happy with the gain I got already.

MSI UK Tech Support Fri, 29 Jan at 11:50

 

Edited November 5, 2021 by Thinder

[BitMaster]

Honestly,

who cares about the Warranty with a RAM kit. How could they ever proof I ran them above 3200 ? There is no way to do that.

The only thing you have to obey with Ryzen is that you better don't force the IO-Die to go into 1/2 mode, somewhere between 3600-4000MHz that will likely happen.

If you stay below that crucial tipping point, say 3600 as a safe setting, you do not have to fear any downsides.

 

I am very sure Intel uses a whole bunch of kits to test and not only this ONE kit, be realistic.

[Thinder]

3 hours ago, BitMaster said:

I am very sure Intel uses a whole bunch of kits to test and not only this ONE kit, be realistic.

And yet they chose this one for their example, it's enough for me, plus I've been asking players in diverse MSI and AMD forums what they were using before contacting supports from MSI, AMD and G.skill.

Since he is planning on using a 3080 CPU, a 5600X with a Cl14 B.die RAM kit will be a very good match, even a Ti since the 3.7% bootleneck is a GPU bootleneck. No need for a 5900x.

This looks like the best solution to me, especially since I am planning myself to get a 3080Ti to play VR at 4K and I know that my CPU-RAM solution is up to the task.

Anyway, at the end of the day, it's up to every individual to make their own choice and we can only suggest solution but this one is tested and yourself know the benefit of a B.Die Cl14 kit, even if you chose to O.C yours, you wouldn't have been able to do it with a non-B.Die kit.

 

[Katj]

Honestly,
who cares about the Warranty with a RAM kit. How could they ever proof I ran them above 3200 ? There is no way to do that.
The only thing you have to obey with Ryzen is that you better don't force the IO-Die to go into 1/2 mode, somewhere between 3600-4000MHz that will likely happen.
If you stay below that crucial tipping point, say 3600 as a safe setting, you do not have to fear any downsides.
 
I am very sure Intel uses a whole bunch of kits to test and not only this ONE kit, be realistic.

I ran into some instability at a fabric clock of 1900 MHz. 1800 MHz is no problem, and I think most 5000-series CPUs can manage this.

Now I haven't even tried pushing the limits with the later BIOS versions that apparently are more stable. I might do that one of there days.

I went with a 5600x and 32 GB B-die ram, as I judged this to be the most bang for the buck. I don't think a 5900x will future proof a build much.

I fully expect to buy a new mid-range cpu and some ddr5 ram, a high-end GPU, and a new motherboard in a couple of years.

If I still want a top notch VR experience that is.

[Thinder]

1 hour ago, Katj said:

I ran into some instability at a fabric clock of 1900 MHz. 1800 MHz is no problem, and I think most 5000-series CPUs can manage this.

Now I haven't even tried pushing the limits with the later BIOS versions that apparently are more stable. I might do that one of there days.

I went with a 5600x and 32 GB B-die ram, as I judged this to be the most bang for the buck. I don't think a 5900x will future proof a build much.

I fully expect to buy a new mid-range cpu and some ddr5 ram, a high-end GPU, and a new motherboard in a couple of years.

If I still want a top notch VR experience that is.

The 3600X could already run at 3200MHz, but I noticed an increase in performances of 21% running the same test with the 5600X, it is a very good gaming processor, I think you won't regret it, if I were you, depending on the budget and availability I would go for a Ti, it would reduce the GPU bootleneck further and help with frame rate, it will do just fine until the next generation upgrade.

If not possible, a 3080 will be good anyway.

[Katj]

The 3600X could already run at 3200MHz, but I noticed an increase in performances of 21% running the same test with the 5600X, it is a very good gaming processor, I think you won't regret it, if I were you, depending on the budget and availability I would go for a Ti, it would reduce the GPU bootleneck further and help with frame rate, it will do just fine until the next generation upgrade.
If not possible, a 3080 will be good anyway.

I must have been unclear. I've had the 5600x and the 3080 for a year now. The Ti wasn't a thing then. Anyway, you're right, I don't regret it.

[Thinder]

39 minutes ago, Katj said:

 

I must have been unclear. I've had the 5600x and the 3080 for a year now. The Ti wasn't a thing then. Anyway, you're right, I don't regret it.

No, it's me, I misunderstood you. Sry...