AMD Ryzen For Simracing?

[Jacob Marshall]

Well, my reasoning was to get another couple of years out of AM4 as I don't see anything exciting sim-wise on the horizon to get me hyped for a whole new platform. (Ok, maybe AC2...)

I find Ryzen 7000 a little lacklustre, not far enough in front of the 3d to warrant a whole new platform. If i were moving to AM5 i would only be interested in the 7800X3D or whatever they call it. And as you say, the cost of MB's are a bit of a joke. Also, DDR5 needs time to more time to mature for better pricing.

I don't have all of the money together for a gpu yet, saving has become a bit of a slow process lately, but i have been watching the falling price of the 6800xt. If you can get a good deal, it is a lot of card compared to the 5700xt. If i'd had the money a couple of months ago i would have grabbed one but i think if i had the money today i would sit on my hands and wait.

The 7800 non xt Vs 6800xt should (hopefully) be in the same ballpark price and performance wise, maybe with a performance bump but definitely better power consumption and heat output (important with rising energy costs and for those two weeks of non air conditioned British summers...).
...

That's a really helpful reply man, cleared the air a lot for me and helped me decide.

Stick with AM4 for the next few years (same in that there's nothing exciting on the horizon), get all the value from that that I can and then decide on a GPU early next year.

The x3D is, I now believe, the better investment; it has longevity and sits on solid architecture, AM4. The thing with AM4 is, unless there's a bombshell development, 7xxxGPUs will work in a 4/5xx mother board.

x3D it is; cheers for the input, and I'll be sure to flame you, and hold you personally responsible for my decisions if I'm not happy with the CPU

 

[RasmusP]

I find Ryzen 7000 a little lacklustre, not far enough in front of the 3d to warrant a whole new platform. If i were moving to AM5 i would only be interested in the 7800X3D or whatever they call it. And as you say, the cost of MB's are a bit of a joke. Also, DDR5 needs time to more time to mature for better pricing.

99% agreed!
I was quite baffled both by the lack of good deals around black Friday but also amazed by the 3 things that did show good deals!
So I've bought:
- 7600x, 269€ vs. 329€
- Gigabyte B650 gaming x, 184€ vs. 249€
- 2x single stick Kingston 16GB 5600 CL36, 2x 83€ vs 107€

(which is 99% safely sk hynix die, the "new Samsung b-die" and might run up to 6400 cl32 with future CPUs and agesa/bios/chipset updates. Kingston has no deal with Samsung, apparently, and micron ddr5 can't run past 5200 yet.)

So while my excel sheet showed 644€ earlier in November for this upgrade, but with only 2x 8GB 5200 cl40 ram, which can't clock higher!, I now got 2x 16GB and way higher possible speeds for 619€!

Or if you take out one of the sticks:
536€

I paid 518€ for my 10600k + Z490 gaming x + 2x 8gb 3200 cl16 in May 2020 so the price was reasonably enough for me to go for it.

We did some benchmarking in AC and ACC here at RD with the 7700X and 5800x3D and they both basically double my fps, especially the 1% low!

Me and my 3080 are ready for SOL/pure, csp reflections, night racing etc!
With the i5, I always had annoying micro stutters here and there.
The jump to 5800x3D, Intel 12th/13th gen and now ryzen 7000 is insanely big compared to Intel 4th-11th gen!
And AMD used this stagnation to "rise" their Ryzens up to speed. And I love it

Currently waiting for my cooler adapter kit. Definitely gonna post some results over the next weeks!

 

[Bob Gnarley]

Awesome, have fun man!

 

[RasmusP]

Finally ready to race again
I only have the boxed cooler of a 5600 for now while waiting for some #6-32 screws to arrive to mount my Thermalright Macho.
So I've limited the power to 50W for now, which is why the results are a bit "underwhelming".
I'm very happy already though!

AC	1% low	avg
10600k	83	107,8
7600x	112	152
ACC	1% low	avg
10600k	60,8	69,7
7600x	77	85,6
rF2	1% low	avg
10600k	48	63
7600x	60	83
Witcher 3	1% low	avg
10600k	105,5	122,4
7600x	138,3	162,6

 

[Novopaine]

Finally ready to race again
...
I'm very happy already though!

Damn you! I've already convinced myself that I need to wait for the 3D part. Now you're making me second guess myself!

 

[RasmusP]

Raised the limit to 70W, put PBO to +200, curve optimizer -10 and raised the RAM speed to 6000, keeping the CL36 identical to the 5600 expo profile but then activated the 2 memory settings for automatically slightly better timings.

It's not much, but it's free performance for just slightly higher power consumption.
Smallest gain were the 1% lows in AC with +2.32%, while the biggest gain showed in the avg fps in ACC with +10.16%, which meant 94.3 fps instead of 85.6!

 

[Glaniu]

Coś jest nie tak :-(
Zmieniłem 5600x na 5800x3d i wyniki w ACC są gorsze na tych samych ustawieniach.

 

[RasmusP]

Coś jest nie tak :-(
Zmieniłem 5600x na 5800x3d i wyniki w ACC są gorsze na tych samych ustawieniach.

Hi, do you have some automatic translation running?
RD is an international, English forum but Google translator did a good job for your post

This has to be a graphics card limit if the fps are basically identical.
The 2.4 fps on average are normal fluctuations if you didn't do 10 test-runs with each cpu.

Go into your graphics settings in ACC and activate DLSS set to "Ultra Performance".
Looks super ugly, but takes a lot of the load away from the graphics card.
If your fps increase, you were limited by the graphics card and your 5600x was good enough for your settings.

If your fps stay identical with DLSS at ultra performance, there's some other issue going on!

 

[phaaser]

Only around 30% improvement from an old gen 10600 to a brand new 7600x!? From the reviews I expected much more! Did I miss something? Is the 13600k eventually the better option for ACC and rf2?

 

[Novopaine]

Only around 30% improvement from an old gen 10600 to a brand new 7600x!? From the reviews I expected much more! Did I miss something? Is the 13600k eventually the better option for ACC and rf2?

See the other thread.

https://www.racedepartment.com/threads/ryzen-am5-7000.250019/post-3618126

 

[RasmusP]

Only around 30% improvement from an old gen 10600 to a brand new 7600x!? From the reviews I expected much more! Did I miss something? Is the 13600k eventually the better option for ACC and rf2?

As you could see in the other thread, tuning the RAM, activating PBO and unlocking the power limit showed around 50% improvement.
"old" for the 10600k is a bit brutal. It was released in April/May 2020! But Intel accelerated the release of new gens and released 3 new gens within 2.5 years...
It's more like "one gen" old. 11th gen was a weak update that was slower in some games and 13th gen is more like what the 12th gen was supposed to be.

The Ryzen 7000 are a bit "lacking" compared to the Intel's 13th gen. That's a fact when averaging all available benchmarks.
However the RAM I got can run up to 6400 CL32 with future CPUs and I'll be able to sell my 7600X in 2 years and replace it with a 7800X3D or 9600X by simply updating the BIOS, loosen 2 screws on my cooler, throw in the new CPU, add a dot of thermal paste and tighten 2 screws.

An Intel upgrade will require selling the mobo too, re-installing Windows and all programs and spend 3 hours changing motherboards & re-doing cable management.

 

[hotak]

I just got a 5800x3d to replace my 3900x.
I hope it'll be worth it, I felt like I was CPU limited with my 3080 12gb running a Reverb G2: most sims need very low settings even when using opencomposite/openXR foveated rendering, and the framerate feels inconsistent in rFactor2.

Hopefully it won't be a downgrade for my other needs, since i'm running some very big neural networks on my GPU and I think tensorflow uses multicore quite well. Not sure if the cache can compensate for 50% less cores.

 

[Rebel_488]

Nah dog u gonna love it, a few guys on rf2 have gone the 5800x3d and the results are superb. Also in vr

 

[Spinelli]

7000x3d is destroying 13900Ks in leaked benchmarks. With the 13900KS probably only being 0-3 % faster in games than the 13900K, it looks like Intel will have no answer.

 

[hotak]

I got the 5800X3D.
What to say?
I'm just even more convinced that we absolutely need VR reviewers, and that most people doing reviews need to stop looking at 1440p/4k benchmarks to predict VR performance!

With a 3080 12Gb I can now run ALL of my games - excluding ACC - at 90FPS with increased details compared to the 3900X.
If I had to listen to reviewers, nobody would have ever predicted this, they're all like "VR = high resolution, you're clearly GPU limited". Fortunately I know enough about computers and graphics options, to be able to get my own conclusion, and I suspected that my problem was basically being "draw calls limited", wich is most likely true considering the FPS gains I got by upgrading the CPU.

If you have some doubts about what to upgrade first, I'll now strongly suggest to get a 5800X3D and only then think about updating the GPU: having consistent FPS is much more important than high FPS with some slowdowns, especially whan playing in VR.

 

[Spinelli]

If I had to listen to reviewers, nobody would have ever predicted this, they're all like "VR = high resolution, you're clearly GPU limited". Fortunately I know enough about computers and graphics options, to be able to get my own conclusion, and I suspected that my problem was basically being "draw calls limited", wich is most likely true considering the FPS gains I got by upgrading the CPU.

I agree 100%. I have seen so many times over 15 years of PC gaming, reviews, benchmarks, etc. where something seems to be GPU limited "on paper" but, even so, getting a new CPU does help. Or the other way around, you seem to be CPU limited yet upgrading the GPU still helped.

I remember a couple months ago in another thread someone asking about the same thing: if they should upgrade from their Ryzen 3900XT & Radeon 6900XT to a 5800X3D for very-GPU-demanding VR (Pimax 5K Plus). Some users such as myself advised them to definitely do so, there was at least one user who did the whole "VR = high resolution, you're clearly GPU limited" thing you mentioned and how the CPU upgrade would basically be a side-grade with negligable improvement. It's just not that simple. Yes, GPU will probably give the biggest increase in framerates, generally speaking, but you can still get respectable-to-large framerate improvements (especially with a beast like the 5800X3D) with CPU upgrades even when you're in situations which are very GPU-demanding.

 

[hotak]

I agree 100%. I have seen so many times over 15 years of PC gaming, reviews, benchmarks, etc. where something seems to be GPU limited "on paper" but, even so, getting a new CPU does help. Or the other way around, you seem to be CPU limited yet upgrading the GPU still helped.

I remember a couple months ago in another thread someone asking about the same thing: if they should upgrade from their Ryzen 3900XT & Radeon 6900XT to a 5800X3D for very-GPU-demanding VR (Pimax 5K Plus). Some users such as myself advised them to definitely do so, there was at least one user who did the whole "VR = high resolution, you're clearly GPU limited" thing you mentioned and how the CPU upgrade would basically be a side-grade with negligable improvement. It's just not that simple. Yes, GPU will probably give the biggest increase in framerates, generally speaking, but you can still get respectable-to-large framerate improvements (especially with a beast like the 5800X3D) with CPU upgrades even when you're in situations which are very GPU-demanding.

It's not jus this: when you are GPU limited, you can help with things like openXR and foveated rendering, or just decrease resolution/antialiasing.
when you are CPU limited, the only way is to reduce poligonal detail, and it won't help much anyway. Games like RBR have no way to help CPU. You end up using very low visual settings to avoid stuttering, and changing GPU, while increasing average FPS, won't help at all and won't allow to increase poly count, so you end up with overfiltered graphical settings that remind me of smartphone games. A good CPU coupled with a bad GPU will always bring more constant framerates and allow more balanced graphical settings (which is something reviewers never test, they always go for max quality, which is not what people actually play)

 

[RasmusP]

Nice to read that you got a nice improvement!

It's not jus this: when you are GPU limited, you can help with things like openXR and foveated rendering, or just decrease resolution/antialiasing.
when you are CPU limited, the only way is to reduce poligonal detail, and it won't help much anyway.

Yep..

You end up using very low visual settings to avoid stuttering

Indeed! If you're CPU limited and need to reduce graphics settings to get higher cpu-fps, you almost always end up by reducing the gpu load way more..
If your gpu is almost at its limit in 1080p but your CPU is struggling, you end up with so heavily reduced settings that you could play in 4k, lol.

something reviewers never test, they always go for max quality, which is not what people actually play

I agree but also don't agree. They do this to have the most drawcalls etc. possible.
Therefore they also test in 1080p or even 720p!

And gpu reviews are therefore most important in 4k to take the CPU out of the equation.

An important fact about reviews is that it's not about knowing how many fps you'd get with which hardware.
It's about showing relative performance.

So when upgrading, you note down your current fps with the settings you'd like to use or be happy with.
Then check how much percent better the new parts would be and then calculate what your new fps would be.

Depending on your bottlenecks, you have to lower or increase the resolution to get a cpu or gpu bottleneck and then do the comparison + calculation.


Btw, hardware unboxed is using medium settings for ACC in cpu reviews. They know it's "esports" so higher fps are needed.
No sense in knowing that a 13600k would achieve is 40 fps and 13900k 55 fps.

 

[Benzaah]

So I've just had a massive upgrade, ascended to a higher level and im feeling a bit underwhelmed/lost as i no longer have an upgrade path.

I was running a 3600xt with a 2060 super for the longest time. Im totally against Nvidias greed atm and swore i was not going to buy a 4080 or a 4070ti. Im an AMD fanboy so when a deal came on the 7900xt hellhound a week ago, I Finally bit the bullet and boy was it a great upgrade from the 2060. I sold my old gpu and a few other bits and had enough to upgrade my cpu which i was on the fence with.
Local stores had all the am4 on sale so i had the pick of a X3D or 5900x at 499au, 5800x at 429 or the 5700x at 299. I choose the X3D but didnt upgrade my cooler so still running a Deepcool GTE which is only a 4 pipe cooler. The cpu only hits 60 degrees playing games but will peack at around 82c benchmarking so i dont think its being throttled. It boosts slower single core than the 3600Xt i had(which admittedly was a golden chip).

I'm mainly playing AMS2 and I have TRIPLE 165hz 1440 lg panels I want to utilize. I thought I would be able to easily crank 165 frames with all settings Maxed but maybe my expectation were too high. I can get n165 with most settings on medium and Low SMAA, MSAA or FXAA settings but I think ill have to play around with the MSAA, SMAA and FXAA settings and figure out how much super sampling I can run.

Any thoughts on system tweaks or bench marks I can run to test performance. I have ACC and AMS2 to test. If there are any other games benchmarks I can run without buying the games, let me know.
Regards Benzaah.