Ryzen AM5 ; 7000

[Kek700]

I do not know if it is just what I see and so that makes you think what you notice is globally true.
This 8 months trying to sell stuff is very difficult, unless you sell something that people with money want. Trying to move down into the more affordable things, all you get is no interest or silly offers.
There is no shortage of money, perhaps it is just a particular section of society is really struggling, another section of society is having to better manage their finances, and then the rest are just getting very stupidly richer.
Who knows.

 

[Terry Rock]

How many of you guys on Ryzen use PBO, XML or another BIOS optimization feature during setup?
I'm wondering if those technologies are somehow influencing the swings some are seeing in frametime graphs during heavy gaming.
Do a few of you guys mind 'hard' selecting your RAM timings and voltages for a comparison test?
I am on a B450 platform but I have zero stuttering or drops in gameplay...and I am using an Intel ARC A770 card which is supposed to be a 'dog' compared to most modern cards.
I've selected my RAM to 3600MHz , 1T, CAS 16,19, 19, 38 (1.38 volts) for my inexpensive TimeTec 32GB ram kit.
The R7-5800X3D is being cooled by a Peerless Assassin 120 which never exceeds 89 degrees during hours of racing.
It typically hovers in the 65.6 degree range just browsing.
I'm wondering if the overclock is hitting a limit...throttling and then starting all over again for some.

 

[RasmusP]

Wouldn't that be a bit useless when comparing to AM5 (wrong thread)?
Unstable overclocks can of course lead to issues, but the difference between the architectures, windows patch version etc. probably is a lot bigger.

Maybe the Intel drivers really focus on stable frame times instead of peak fps.

 

[Terry Rock]

Wouldn't that be a bit useless when comparing to AM5 (wrong thread)?
Unstable overclocks can of course lead to issues, but the difference between the architectures, windows patch version etc. probably is a lot bigger.

Maybe the Intel drivers really focus on stable frame times instead of peak fps.

No!... not wrong thread.
I am trying to see if it is happening across all AMD platforms.
There are still guys on AM5 complaining about bad frametime in games.

 

[Durge Driven]

7800X3D - 20 negative - Expo - Thermals 80C
No need to push anymore for ISIMotor
There is nothing else I care about.

But I would still be waiting for a 9xxx3D / 800 chipset if I was buying this quarter.
Even 9700X is just as fast in games, better in everything else and same price.

 

[Kek700]

I had run my 7800x3d on its limit including ram. ( that is my version of maximum )
Eventually died on me , so I had to load the latest version of bios. Now run it standard, and will only tune it when and if necessary.
Funnily it now boots faultlessly and quickly..
never did before, took ages. I still get 144hz constantly with 5760 x 1080

 

[RasmusP]

No!... not wrong thread.
I am trying to see if it is happening across all AMD platforms.
There are still guys on AM5 complaining about bad frametime in games.

Aaahh okay,
well I had a 7600x for a year and now 7800X3D for 10 months.
Tuned timings that weren't really stable or super safe settings don't make any difference.
Always some frame time spikes in AC & ACC (didn't use anything else long enough).
Also tuned both CPUs to an unstable level (PBO with + Hz and negative curve optimizer).

No impact on the frame times. Also none of the throttling limits were reached (temps, power).

Funnily enough, although the rtss frame time line in the overlay is a flat line apart from the spikes every 30-90 seconds, it still sometimes looks like my gsync would be skipping a frame.
Exactly identical setup and settings in Win 10 give a smoother image, but strange gsync behavior.. (sometimes not engaging in borderless mode, sometimes it's flawless.)
In real fullscreen, it's smoother than win 11 (the mode with a short black screen when alt+tab out of the sim).

So my guess is that the people who complain either have issues with gsync/freesync or with the hybrid-fullscreen that win 10 and 11 use (different behavior in some situations though!).

 

[Kek700]

Interesting reading Rasmus account , pretty much coincide with my view, excluding the reasoning behind it, for me it was just an observation.
Since the new latest bios update that I was forced to use, ( i was on the very first bios implementation ) and now running the cpu and ram with no overlocking , just standard. I have not seen the stuttering that I once saw on a regular basis. But time will tell???

 

[Spinelli]

AMD Ryzen 5 7600X3D Processor - Exclusive to Microcenter Bundle (for now)

AMD has introduced the Ryzen 5 7600X3D, a 6-core, 12-thread processor based on the Zen 4 microarchitecture, enhanced with 3D V-Cache technology to boost gaming performance.

www.guru3d.com

Sweet CPU for those not looking to spend much, however, I do think it's price is a little too close to the 7800X3D's especially currently with the 7800X3D only $70 more.

The best price-to-performance combo, in my opinion, is to purchase all used parts along with a used 5800X3D. 5800X3D is truly a gaming beast. Unless you want to upgrade in the next 1 or 2 years, then I'd go with the 7000-series since it's motherboard will allow you to upgrade to a 9000 series in the future.

 

[Spinelli]

Tuned timings that weren't really stable or super safe settings don't make any difference.
Always some frame time spikes in AC & ACC (didn't use anything else long enough).
Also tuned both CPUs to an unstable level (PBO with + Hz and negative curve optimizer).

Unstable RAM timings and RAM/CPU/GPU overclocks can cause things like stutters, frame skips, etc. even if they don't actually decrease overall framerates. You want to make sure your settings are fully stable. Lower scores in stability tests can be an indication of instability even though it's "stable enough" to finish the test and not crash. I noticed this a lot when tuning my RAM.

Exactly identical setup and settings in Win 10 give a smoother image, but strange gsync behavior.. (sometimes not engaging in borderless mode, sometimes it's flawless.)
In real fullscreen, it's smoother than win 11 (the mode with a short black screen when alt+tab out of the sim).

So my guess is that the people who complain either have issues with gsync/freesync or with the hybrid-fullscreen that win 10 and 11 use (different behavior in some situations though!).

That is one of the reasons I refuse to move to Windows 11.

With Win 10, you can put the game in true full-screen mode by checking "disable fullscreen optimizations" on the program's EXE file. I do this with some of my games - especially the ones I take more seriously like sim racing. Apparently, doing this with Win 11 can actually mess things up because Win 11 works totally different in the "back end" so you don't want to do that with Win 11.

Also, with Win 10, I make sure to use "Ultimate Performance" power plan. It has the lowest DPC system latency of all power plans. Again, with Win 11 it's different. There's 2 different power type settings - "Power Plan" and "Power Mode". So you can set one to "Balanced" and another to "High Performance" at the same time...or something like that.

Also, with Win 11, with at least 1 of the power type settings, it's advised to use Balanced and not High Performance as the Balanced mode actually gives better performance. It's weird.

With Win 10, it's simple. For absolute best performance and/or lowest risk of issues, check "disable fullscreen optimizations" and use "Ultimate Performance" power plan (or at least "High Performance"). That's it. Done.

 

[RasmusP]

Unstable RAM timings and RAM/CPU/GPU overclocks can cause things like stutters, frame skips, etc. even if they don't actually decrease overall framerates. You want to make sure your settings are fully stable. Lower scores in stability tests can be an indication of instability even though it's "stable enough" to finish the test and not crash. I noticed this a lot when tuning my RAM.

Absolutely!
I didn't want to say, that unstable overclocks would be fine. They just didn't make any difference regarding microstutter for me.
ACC the fancy princess sim would crash with "fatal error" every now and then though.
It's quite ridiculous with ACC. All drivers in my team with a 3000,4000 or radeon 7000 GPU had to manually reduce the boost clocks from the stock config to get a fully stable ACC.



Good tips about Win10 & Win11! Completely agree.
I'm just not sure about the 2 different power things with Win 11. From my knowledge there's still just the normal power plans..

 

[Durge Driven]

What now Sinkholes ! lol

Gigabyte bios update for sinkclose is out for B650 and X570 .

W11 I set CPU and GPU power to default.
Never had a issue with using full screen windowed mode
Lemans Ultimate Ultra-High 24 hour race + Replay, when you can do that get back to me .. ( shrugs)

Fresh Windows 11 Pro
Stick your features Bill without them all my OSes run like trains.

Fresh Updates
Weird one, new Cumulative brings up AMD System driver I install manually.
It won't download just sits there " retry" lol
I do the performance fix preview, reboot and the AMD entry is gone.
However if I had clicked install all it would install.

Anyways pretty good install up to all apps
Check my control panel there are another 30+ standalone apps including stuff some would think " you can't use that standalone" well you can 98% of "stuff". I never reinstalled a sim app or software, just means you need a perfect registry import and saved C:\ files is all.

You should see some of the PC I have worked on for people.
I scroll this list on other towers my wrist gets tired ! lol

0.01 - 0.03 %

 

[Durge Driven]

That is all Bill I keep ( most you can't delete) but I never open or use this like ever !

OPEN SHELL just makes a laugh out of W11 design, concise, compact and to the point.

Another thing I like is how clean you can make Windows Explorer look.

 

[Durge Driven]

I do not know if it is just what I see and so that makes you think what you notice is globally true.
This 8 months trying to sell stuff is very difficult, unless you sell something that people with money want. Trying to move down into the more affordable things, all you get is no interest or silly offers.
There is no shortage of money, perhaps it is just a particular section of society is really struggling, another section of society is having to better manage their finances, and then the rest are just getting very stupidly richer.
Who knows.

 

[RasmusP]

Installed a fresh 23H2 on a free partition, updated to the magic new Scheduler KBxxxx whatever (current optional update).
Adjusted all drive letters to be identical to the 22H2 Windows.

Installed latest drivers, only launched Steam (on a separate drive, so it only needs to "repair the registry entries" to run without issues).
Went into ACC, 1024x786, Suzuka, 49 AI, got 160-170 fps.

Booted my 22H2 Win 11, identical ACC settings, 180-190 fps.

Gonna test my Win 10 tonight and also check if there's anything wrong with the 23H2 with the optional KB update..

Not a good first impression
I wouldn't know what might hamper the CPU performance, since clockspeeds etc. are the same and nothing was running in the background.
But I'll report back!

The slight freezes while navigating in Windows are still there, the microstutters in ACC are still there.

Gonna be interesting, once 24H2 rolls out for non-insiders, since it has a new kernel, so basically Windows 12.

 

[RasmusP]

For whatever reason, although I used symbolic links for all files apart from the controls.json, the graphics settings weren't identical.

So I've made new tests:
- Win 10 22H2 B19045.4780
- Win 11 22H2
- Win 11 23H2 KB5041587 (CPU Scheduler Update)

Windowed and fullscreen runs in 1024x784, but using a Replay with 40 cars, sitting inside the cockpit of a car almost at the last place.
Starting at the start/finish line of lap 1 and letting CapFrameX run for 100 seconds.

In the Windows "graphics settings":
- HAGS enabled,
- "Variable Refresh Rate" enabled
- Win 11: "Optimisations for windowed.." enabled
- Win 10: Performance set to Desktop applications

Here are the results:

Win 10 - windowed:

Win 10 - fullscreen:

Win 11 22H2 - windowed:

Win 11 22H2 - fullscreen:

Win 11 23H2 KB5041587 - windowed:

Win 11 23H2 KB5041587 - fullscreen:

Nice gains from the CPU scheduler update, but I'm quite surprised by the fps of my Win 10 installation! All Windows' have the same applications running in the background (basically none, nothing above 0.4% in Taskmanager).

 

[Spinelli]

For whatever reason, although I used symbolic links for all files apart from the controls.json, the graphics settings weren't identical.

So I've made new tests:
- Win 10 22H2 B19045.4780
- Win 11 22H2
- Win 11 23H2 KB5041587 (CPU Scheduler Update)

Windowed and fullscreen runs in 1024x784, but using a Replay with 40 cars, sitting inside the cockpit of a car almost at the last place.
Starting at the start/finish line of lap 1 and letting CapFrameX run for 100 seconds.

In the Windows "graphics settings":
- HAGS enabled,
- "Variable Refresh Rate" enabled
- Win 11: "Optimisations for windowed.." enabled
- Win 10: Performance set to Desktop applications

Here are the results:

Win 10 - windowed:
View attachment 782346

Win 10 - fullscreen:
View attachment 782347

Win 11 22H2 - windowed:
View attachment 782348

Win 11 22H2 - fullscreen:
View attachment 782349

Win 11 23H2 KB5041587 - windowed:
View attachment 782350

Win 11 23H2 KB5041587 - fullscreen:
View attachment 782351


Nice gains from the CPU scheduler update, but I'm quite surprised by the fps of my Win 10 installation! All Windows' have the same applications running in the background (basically none, nothing above 0.4% in Taskmanager).

So basically, screw Win 11.

What power plan was Win 10 set to?
What do you mean "performance set to desktop applications"?
These tests are using a 7800X3D (as in your sig), yes?

 

[RasmusP]

What power plan was Win 10 set to?

Ultimate Performance and CPU min/max both at 100%

What do you mean "performance set to desktop applications"?

Eh, this weird one:

These tests are using a 7800X3D (as in your sig), yes?

Yes. But my RAM is currently at 5400 with BuildZoids "Easy Timings for Hynix". Needed some absolute stability after ACC is throwing some "Fatal Errors" after the latest update..

So basically, screw Win 11.

Yes and no. I'm not sure why the performance is that bad in comparison. When I compared my 2020 Win 10 vs. the then fresh Win 11 22H2, the performance seemed to be identical.
I really like the multi-monitor behaviour of Win 11, but yeah, right now I'll race with Win 10.

 

[Durge Driven]

Ryzen 5 7600X3D im Test: AMD hat einen neuen Effizienzmeister!

Hier finden Sie Benchmarks in Spielen und Anwendungen zum Ryzen 5 7600X3D. Die Beobachtung wird jeweils durch einen Eintrag im CPU-Index abgeschlossen.

www.pcgameshardware.de

Buying that 7600X3D special I linked for anyone lucky to live USA could mean a step up in GPU if you take into consideration your overall package price, as well run on a cheap air cooler.

I know most will frown on 6 core for gaming now let alone futureproof.
I think this perfect for midrange GPU and 3440 resolution, well for sims anyway.
Thing is if you put the $180 saving all on CPU then it's chicken feed you would sell it S/H same price.
Later upgrade to 9xxxx at reduced price too, that's futureproof enough.

 

[Spinelli]

Yes. But my RAM is currently at 5400 with BuildZoids "Easy Timings for Hynix". Needed some absolute stability after ACC is throwing some "Fatal Errors" after the latest update..

The RAM probably won't make much difference - at least relative to other setups - since you're using an X3D chip so I wouldn't stress over RAM overclocking & timing-tuning much. Having said that, 5400 MHz is quite slow for DDR5 but I'm not sure how much difference even that slow speed makes with an X3D chip.

If you really want to check for RAM stability, I'd use Karhu for DDR5. I'd aim for 6 hrs without error - bare minimum - but ideally around 24 hrs. It's a one-time $10 price but widely regarded as one of the best (and fastest), if not the best, program for checking for RAM errors. Linpack Xtreme and Y-Cruncher w/ VT3 test may detect errors even quicker but those programs are not pure RAM stability tests, they also hammer the CPU. Heck, some PCs can't even run the Linpack X and Y-Cruncher tests at all because of the CPU heat regardless how stable their RAM is. Karhu and TestMem5 w/ various different TM5 profiles like 1Usmus v3, Anta Extreme, and Anta Absolut (possibly other profiles, I haven't kept up with TM5 in a year or 2) are the best pure RAM stability tests. I wouldn't even bother with other RAM testers like MemTestPro, MemTest64, MemTest 86, MemTest86+, etc. Unless there's been some significant updates to them recently, TM5 and especially Karhu, are the best at finding errors, and quickly (OCCT has gotten better recently but Karhu and then TM5 are still easily my top 2 and most others seem to agree).