• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Ryzen Owners Zen Garden

Why is ryzen 5800x3d considered to be so good? How do we know its not just a handful of selected games that see big gains in FPS?

I just watched this video here... I don't know... I got a ryzen 5600 at a 4.7 all core for $129 - and it matches and beats the 5800x3d in several games. At first I was skeptical this video was accurate, but then it showed games like Watch Dogs Legion and the 5800x3d smothering the non-x3d chips, which is accurate... but still that's only only one of a handful of games in reviews I have read... not sure if 5800x3d is worth $215 more... (I only looked at the 1440p benchmarks)

I have one just because I made an embarrassing error and killed my 5600X. Where it really shows it's advantage is down at 1080p, barely anything can touch it there where it's a CPU bound scenario, I game at 3440x1440 I am certainly not making use of that advantage that would be tangible other than having the extra threads over the 5600X

Edit: Now that CP2077 shot you just posted should be enough of an answer no? I mean it's right there with the top 5xxx's with a 4.4Ghz clock speed and half the cores.
 
Edit: Now that CP2077 shot you just posted should be enough of an answer no? I mean it's right there with the top 5xxx's with a 4.4Ghz clock speed and half the cores.

I was under the impression that games really don't use more than 6 cores anyway, so that's not a big deal really. Hardware Unboxed seems to say that, benches seem to prove it as well, since a 7900x matches a 7600x in basically every game
 
I was under the impression that games really don't use more than 6 cores anyway, so that's not a big deal really. Hardware Unboxed seems to say that, benches seem to prove it as well, since a 7900x matches a 7600x in basically every game
Yeah for most games that is entirely true, I would love to say "blame consoles" but that would be too easy. Devs have had years of multicore/thread CPUs they "could" make better use of...I guess my point was the pretty much locked down, no OC 5800X3D can stomp CPUs much more powerful in the arena it's meant for
 
@Space Lynx 5800X3D is an otherworldly beast for competitive FPS games.
 
The nice thing about X3D is how easy it is to cool. Just like a 5600X. Now compared to a tuned 5900X.. that means I don’t have to listen to fans ramp up and down, I can play in actual quietness. Outside of games.. well.. you can do better.
 
Yeah for most games that is entirely true, I would love to say "blame consoles" but that would be too easy. Devs have had years of multicore/thread CPUs they "could" make better use of...I guess my point was the pretty much locked down, no OC 5800X3D can stomp CPUs much more powerful in the arena it's meant for
We are not far off from decades now!!
 
Well, I have some benchmarks here on my new system. According to Cinebench R23, this new system is about half again as fast as my previous system.

Cinebench R23 Scores (Intel Core i7 8700K, all-core 4.8 GHz)
Single Core:
1247
Multi Core: 9364

Cinebench R23 Scores (Ryzen 7 7700X, Stock)
Single Core:
1960 (57.17% improvement)
Multi Core: 19099 (103.96% improvement)
 
I havent used Armory Crate for awhile now, I dont have fond memories of it. I am runnig it now and it is showing all of my cores and threads being active. Geez what a pig. But idle temps are still low so I dont think its actually working the CPU too hard? The colors it puts out are so much nicer than OpenRGB lol..

If I bought a 5v RGB cable would I be able to connect my GPU to a mobo header and control its lights like that? It really messes with my OCD that I cannot get the timing just right when I try manually :D
 
A word of warning for those setting per core Curve Optimizer values: if you're testing with CoreCycler, always check CoreCycler logs in addition to Event Viewer logs. I've had Core 6 crash (PC reboot) during testing, but Event Viewer had Core 0 logged as the culprit. Took me a few days of experimenting before I figured out that I need to adjust Core 6 CO value.
 
Anyone know where to check RAM version?

QVL list says CMT32GX5M2X6400C38 ver 5.43.13 is supported at 6400 with XMP. Started off with BSODs and SOC voltage 1.3v in the red :D

EDIT: doesn't matter, no way this could work at 6400, at least not with my mobo. Not sure what ASrock was smoking to put this totally Intel kit with only one XMP profile on the QVL :laugh: At least 6200 cl36 works at 1.35v with better aida64 numbers than the kingston 5600 cl40 @ 6000 cl30 that wouldn't do 6200 for some reason. Will think about which one to keep, not sure if blindly trusting QVL and buying an outright intel kit is grounds for RMA. Maybe if the Corsair is cooler as it's around 1/2" taller might be better :rolleyes:
 
Last edited:
Why is ryzen 5800x3d considered to be so good? How do we know its not just a handful of selected games that see big gains in FPS?

I just watched this video here... I don't know... I got a ryzen 5600 at a 4.7 all core for $129 - and it matches and beats the 5800x3d in several games. At first I was skeptical this video was accurate, but then it showed games like Watch Dogs Legion and the 5800x3d smothering the non-x3d chips, which is accurate... but still that's only only one of a handful of games in reviews I have read... not sure if 5800x3d is worth $215 more... (I only looked at the 1440p benchmarks)

because dozens of websites reviewed it and they all came up with amazing results, other than the few who deliberately crippled it with 2x8GB ram setups

Theres no way that one reviewer got 'the truth' over everyone else


Got it: He's doing bullshit to get attention, the post itself has DOZENS of affiliate links - it's clickbait bullshit
1674021768978.png
 
I was under the impression that games really don't use more than 6 cores anyway, so that's not a big deal really. Hardware Unboxed seems to say that, benches seem to prove it as well, since a 7900x matches a 7600x in basically every game

Minimum framerates are typically much better on CPUs with more cores even when averages aren't, it reduces stutter as well.
 
IMG20230118032021.jpg


From a 1920x to a 3900x for the bench. Handles everything perfectly so far, even some really, really messed up ram that some dingus wrote over the XMP profiles. (not shown)

Titan Xp and GTX690 for BOINC at the moment. It's cold.
 
I havent used Armory Crate for awhile now, I dont have fond memories of it. I am runnig it now and it is showing all of my cores and threads being active. Geez what a pig. But idle temps are still low so I dont think its actually working the CPU too hard? The colors it puts out are so much nicer than OpenRGB lol..

If I bought a 5v RGB cable would I be able to connect my GPU to a mobo header and control its lights like that? It really messes with my OCD that I cannot get the timing just right when I try manually :D

My problem with Armory Crate is not so much it being there but it is borderline impossible to remove cleanly. Their uninstaller tool still left shit behind iirc.

It's one of the few things that can control G.skill's stupid RGB, but it's inconsistent and I'm pretty sure the unholy matrimony of G.skill RGB and Armory Crate nuked more than one Windows install last year.

I've been slowly trying to rid my entire system of RGB. Trident Z I put up with because it's B-die but you can bet there will be no lights on my next DDR5 kit
 
barely anything can touch it there where it's a CPU bound scenario
I saw significant gains at 4k
Remember that a big part of it's advantage is that it doesnt care about low latency ram any more, so running 32GB+ of slower ram becomes a non issue


Helped a person remotely build an AM5 system over facebook the last few days, was "fun" at times
1674118659931.png


Key thing came up here here: over tightening the AM5 cooler resulted in slow boots, DRAM light, VGA light etc with an after market AM4/AM5 universal cooler
Any time you hear of people with AM5 issues, remind them to loosen the damn cooler! universal ones can go tighter on AM5 than they need to, and cause some weird issues
 
Remember that a big part of it's advantage is that it doesnt care about low latency ram any more, so running 32GB+ of slower ram becomes a non issue
I‘m running 32GB of 3600 CL16 so can’t really do much better tho it doesn’t bench well anymore of course. I’m at 3440x1440 now I can’t say I’ve seen any gains but then I just upgraded from a 6700XT to a 6950XT. My Time Spy score is where I’m losing with a low CPU score
 
Remember that a big part of it's advantage is that it doesnt care about low latency ram any more, so running 32GB+ of slower ram becomes a non issue
I had a friend of mine say that a lot of the DDR5 memory has high latency and he was going to stay away from it until it matures. Are you saying that Ryzen doesn't care about this? If so, how? Why?

I got this kit of RAM for free when I bought my Ryzen 7700X. Normally costs about $150. The timings are 36-36-36-96.
 
I had a friend of mine say that a lot of the DDR5 memory has high latency and he was going to stay away from it until it matures. Are you saying that Ryzen doesn't care about this? If so, how? Why?

It's not that it doesn't care, it's that the Vcache size greatly diminishes Ryzen's usual reliance on Fabric speed. Works the same way as 16MB Zen 3 (Cezanne) vs 32MB Zen 3 (Vermeer), the APUs should be running highest possible Fabric (2000-2400) to try and mitigate the loss of L3. There is still some gains from Fabric speed and timings for X3D but very minor.

The Intel IMCs at least and good DDR5 (Hynix A and M) have clearly proven that there is no "high latency" problem with DDR5. AMD's DDR5 UMC is just plain weak in 1:1 mode, which is why positively no one is picking AM5 for higher DDR5 OC. But if the trend continues, that should at least not be a problem for X3D gaming perf, thanks to Vcache.

5800X3D suffered a +3ns or so latency penalty compared to the regular CPUs. Wouldn't be surprised if Raphael-X is the same.
 
I saw significant gains at 4k
Remember that a big part of it's advantage is that it doesnt care about low latency ram any more, so running 32GB+ of slower ram becomes a non issue


Helped a person remotely build an AM5 system over facebook the last few days, was "fun" at times
View attachment 279860

Key thing came up here here: over tightening the AM5 cooler resulted in slow boots, DRAM light, VGA light etc with an after market AM4/AM5 universal cooler
Any time you hear of people with AM5 issues, remind them to loosen the damn cooler! universal ones can go tighter on AM5 than they need to, and cause some weird issues
Interesting. I wonder if am4 suffers that or if noctua coolers can??
 
I mean this is kinda apples and oranges but with this same RAM on my 5600X OC to 3800 CL18 no other timings adjusted it would get sub 60ns just barely tho so lets say it got 60ns
This is now at total stock speeds obviously. I really wish I could OC the IF just a bit on this but then @Mussels pretty much said it's not really so important
mem3D.jpg
 
I mean this is kinda apples and oranges but with this same RAM on my 5600X OC to 3800 CL18 no other timings adjusted it would get sub 60ns just barely tho so lets say it got 60ns
This is now at total stock speeds obviously. I really wish I could OC the IF just a bit on this but then @Mussels pretty much said it's not really so important
Here's my result with 3600 b-die with tight timings, 5600x vs 5800x3d
AIDA64 run 3.png
AIDA64 run 3.png
 
Back
Top