5800x3d vs 12900k tuned gaming benchmarks

Finners · 31 May 2022 at 16:49

Are you not leaving performance on the table in these CPU bottleneck scenarios only have 2 single rank dimms on the Ryzen?

Is the 12900k losing much having the e-cores turned off?

Edit*
4:02 in this video

Ryzen 5000 DDR4 Memory Performance, XMP vs Manual Timings, Single vs Dual Rank

Join us on Patreon: https://www.patreon.com/hardwareunboxedJoin us on Floatplane: https://www.floatplane.com/channel/HardwareUnboxedAMD Ryzen 5 5600X: https:...

youtu.be

CAT-THE-FIFTH · 31 May 2022 at 16:55

@Robert896r1 Will be interested to see if you can get the top score in the Fallout 4 thread with your Ryzen 7 5800X3D. Your friend already had the best score in the thread with their Core i9 12900K.

humbug · 31 May 2022 at 16:57

Chapchap27 said:
I got my 5800x3D on Friday and have been able to play BF2042 now finally after 6 months- before with my 5800x the fps went from 80-90 to 120 but now it’s a solid 120.

I’m also looking forward to when the free fly event in Star Citizen is over to get a better understanding the improvements there too. I tried Star Citizen out over the weekend but it was too frustrating with all the crashes etc but then they got over 110k new account sign ups over the past week and a bit so the servers took a hammering and performance generally degraded.

Overall it’s a great CPU and glad I could find a decent UK reseller to get one.

This guide really helped with temps too.

How-to-undervolt-AMD-RYZEN-5800X3D-Guide-with-PBO2-Tuner/README.md at main · PrimeO7/How-to-undervolt-AMD-RYZEN-5800X3D-Guide-with-PBO2-Tuner

Get the Most out of your 5800X3D using PBO Curve Optimizer! - PrimeO7/How-to-undervolt-AMD-RYZEN-5800X3D-Guide-with-PBO2-Tuner

github.com

Star Citizen is in a critical phase of development.

3.17.X (Current live build) is the first patch with most of the Gen12 / Vulkan API in, its also the first patch with the Shard Manager up and running, this is critical for Server Meshing, its also the first patch with partial Persistent Entity Streaming, again this is needed for Server Meshing.

So there is a lot of very critical back end tech being stress tested in this build and its playing havoc on stability, Star Citizen is at an impasse, its on the brink of expanding into something closer to a true MMO and with multiple Star Systems, Full Persistent Entity Streaming testing starts in a few weeks and in Q3 this year full Server Meshing testing, there after we get the first of many more systems to come, Pyro, Nyx ecte....

https://www.youtube.com/watch?v=kZWT4aNY6lU&t=2160s

Robert896r1 · 31 May 2022 at 17:01

@CAT-THE-FIFTH I can't. I already ran it. S1 I top out around 114fps and S2 around 133ish. I didn't post it since it's close enough to JZ to not matter much

CAT-THE-FIFTH · 31 May 2022 at 17:06

Robert896r1 said:
@CAT-THE-FIFTH I can't. I already ran it. S1 I top out around 114fps and S2 around 133ish. I didn't post it since it's close enough to JZ to not matter much

But it's still another data point - so might as well put it in there! But it does seem that for the Corvega test,which is mostly a graphics rendering test(loads a single core very highly), AMD can almost match a very tweaked Intel CPU. However,in the second test which is a combination of drawcalls and NPC AI,Intel seems to be having an edge. Interesting!

humbug · 31 May 2022 at 17:14

CAT-THE-FIFTH said:
But it's still another data point - so might as well put it in there! But it does seem that for the Corvega test,which is mostly a graphics rendering test(loads a single core very highly), AMD can almost match a very tweaked Intel CPU. However,in the second test which is a combination of drawcalls and NPC AI,Intel seems to be having an edge. Interesting!

Bearing in mind that the 5800X3D is also clocked near a Ghz lower than the 12900K, the 3D cache can't always make up for such a large clock deficit.

The fact that its able to match or beat the 12900K more often than lose is not bad considering the 12900K is clocked a Ghz higher and is running DDR5 at 2X the speed of DDR4.

Robert896r1 · 31 May 2022 at 17:14

Finners said:
Are you not leaving performance on the table in these CPU bottleneck scenarios only have 2 single rank dimms on the Ryzen?

Is the 12900k losing much having the e-cores turned off?

Edit*
4:02 in this video

Ryzen 5000 DDR4 Memory Performance, XMP vs Manual Timings, Single vs Dual Rank

Join us on Patreon: https://www.patreon.com/hardwareunboxedJoin us on Floatplane: https://www.floatplane.com/channel/HardwareUnboxedAMD Ryzen 5 5600X: https:...

youtu.be

ecores aren't harmful for the most part but can be left off for primarily gaming setups. One of our guys did extensive testing here on the various core configurations for ADL: https://kingfaris.co.uk/blog/12900k-core-configs/summary We also felt 8 vs 8 was more direct as I mentioned in the OP.

DR Bdie would help to a degree but this was a what I'd call budget build to have some fun on Zen. The current kit I'm using is from my 9900k setup so no additional build cost. A new DR Bdie kit is about 190 so not really up for spending that much. https://www.overclockers.co.uk/team...00c14-3200mhz-dual-channel-kit-my-003-8p.html

If you got one for me to test with, happy to!

Chapchap27 · 31 May 2022 at 17:15

Robert896r1 said:
Yes. Get the PB02 Tuner, go -30 (that's the max possible on x3d) on all cores and stability test from there.

I’ve got mine -20 all cores except the top two with -15. Have done zero testing other then BF2042, Star Citizen,Hell Let lose- which are three games which would show stability problems I am sure.

My question is what would I benefit from going all -30 over what I have set ? Half a degree less heat or an extra fps - I say that as Ive found the last couple of % performance is got the hardest.

CAT-THE-FIFTH · 31 May 2022 at 17:16

humbug said:
Bearing in mind that the 5800X3D is also clocked near a Ghz lower than the 12900K, the 3D cache can't always make up for such a large clock deficit.

The fact that its able to match or beat the 12900K more often than lose is not bad considering the 12900K is clocked a Ghz higher and is running DDR5 at 2X the speed of DDR4.

Well it does very well at stock clockspeeds at lower power consumption so if I had the money,I would certain consider one for my PC which is SFF!

humbug · 31 May 2022 at 17:19

CAT-THE-FIFTH said:
Well it does very well at stock clockspeeds at lower power consumption so if I had the money,I would certain consider one for my PC which is SFF!

For gaming i don't see the point in spending 2X the money for something that's no better, i mean you really are just buying all the most expensive stuff and then pushing it all to its limits to keep up with it, that seems Bat-Poo crazy to me

humbug · 31 May 2022 at 17:23

Its like the 5KW water chiller thing, you know what i'm talking about?

Finners · 31 May 2022 at 17:27

Robert896r1 said:
ecores aren't harmful for the most part but can be left off for primarily gaming setup mainly. One of our guys did extensive testing here on the various core configurations for ADL: https://kingfaris.co.uk/blog/12900k-core-configs/summary We also felt 8 vs 8 was more direct as I mentioned in the OP.

DR Bdie would help to a degree but this was a what I'd call budget build to have some fun on Zen. The current kit I'm using is from my 9900k setup so no additional build cost. A new DR Bdie kit is about 190 so not really up for spending that much. https://www.overclockers.co.uk/team...00c14-3200mhz-dual-channel-kit-my-003-8p.html

If you got one for me to test with, happy to!

another 2x8 to go with yours would probably cost you ~£60 second hand and going by benchmarks seems like it would be a decent uplift in these situations.

KompuKare · 31 May 2022 at 17:37

CAT-THE-FIFTH said:
But it does seem that for the Corvega test,which is mostly a graphics rendering test(loads a single core very highly), AMD can almost match a very tweaked Intel CPU. However,in the second test which is a combination of drawcalls and NPC AI,Intel seems to be having an edge. Interesting!

With ProcessExplorer I think I did notice two main threads, so context switching can't really be the explaination. However, it is possible with two intensive threads 96MB cache is no longer enough. Any rumours about how much cache Zen4 3D will bring to the table?

Considering the cost, the fact that it works reasonble well without a lof memory tuning and consumes a lot less power I too would take 5800X3D over a 12900K. But it really is no surprise that there are workloads where one does better than the others. In true hardware and gamer hyperbole, any of those small % wins get hyped up as Thrashing, Stomping, and similar!

CAT-THE-FIFTH · 31 May 2022 at 17:41

humbug said:
For gaming i don't see the point in spending 2X the money for something that's no better, i mean you really are just buying all the most expensive stuff and then pushing it all to its limits to keep up with it, that seems Bat-Poo crazy to me

In terms of cooling,etc would be easier with a Ryzen 7 5800X3D at stock clockspeeds. But its a bit too pricey for me,like a Core i9 12900K or Core i7 12700K so I think my Ryzen 7 5700X will suffice for now!

KompuKare said:
With ProcessExplorer I think I did notice two main threads, so context switching can't really be the explaination. However, it is possible with two intensive threads 96MB cache is no longer enough. Any rumours about how much cache Zen4 3D will bring to the table?

Considering the cost, the fact that it works reasonble well without a lof memory tuning and consumes a lot less power I too would take 5800X3D over a 12900K. But it really is no surprise that there are workloads where one does better than the others. In true hardware and gamer hyperbole, any of those small % wins get hyped up as Thrashing, Stomping, and similar!

Its two threads for sure. Not heard about how much cache Zen4 3D will have ATM. However,what I would like to see is how a Core i5 12400F looks like in the game.

humbug · 31 May 2022 at 17:46

KompuKare said:
With ProcessExplorer I think I did notice two main threads, so context switching can't really be the explaination. However, it is possible with two intensive threads 96MB cache is no longer enough. Any rumours about how much cache Zen4 3D will bring to the table?

Considering the cost, the fact that it works reasonble well without a lof memory tuning and consumes a lot less power I too would take 5800X3D over a 12900K. But it really is no surprise that there are workloads where one does better than the others. In true hardware and gamer hyperbole, any of those small % wins get hyped up as Thrashing, Stomping, and similar!

Its also why exceptionalism becomes normalised in the minds of those doing it, like Intel hacking a 5,000 watt water chiller in to the loop of an 18 core CPU to stop the 600 watts + its sucking down causing it to melt a hole straight through where the socket is just to say "See.... we still win"

Ok well i don't need to do any of that #### to get that level of performance. I just plug in and play.

"But i win"

Troezar · 31 May 2022 at 17:53

humbug said:
Its also why exceptionalism becomes normalised in the minds of those doing it, like Intel hacking a 5,000 watt water chiller in to the loop of an 18 core CPU to stop the 600 watts + its sucking down causing it to melt a hole straight through where the socket is just to say "See.... we still win"

Ok well i don't need to do any of that #### to get that level of performance. I just plug in and play.

"But i win"

The great thing is if you can see past all that stuff there are some great options. My criteria are stable, cool(er), and as fast as matters. Whilst great for benchers 10% extra speed, extra power and heat and possibly some tweaking is a no-no for me. The best out-of-the box price/performance will do me, I may push price if it's easy to obtain, hence drop in a 5800X3D to replace my 3700X. 10 minute swap and a big performance boost

humbug · 31 May 2022 at 17:59

Troezar said:
The great thing is if you can see past all that stuff there are some great options. My criteria are stable, cool(er), and as fast as matters. Whilst great for benchers 10% extra speed, extra power and heat and possibly some tweaking is a no-no for me. The best out-of-the box price/performance will do me, I may push price if it's easy to obtain, hence drop in a 5800X3D to replace my 3700X. 10 minute swap and a big performance boost

Yup.

The reason simracers like it is because the performance is excellent with a 4 year old cheap motherboard and some inexpensive RAM, sure the 12900K will match it highly tuned with £300 RAM but...... really? Does that make these simracers wrong?

Robert896r1 · 31 May 2022 at 18:49

Gears of War 5 results added to OP. That wraps up all the games we'll test.

I'm sure we'll do something for Zen4/RPL as well.

Troezar · 31 May 2022 at 19:17

humbug said:
Yup.

The reason simracers like it is because the performance is excellent with a 4 year old cheap motherboard and some inexpensive RAM, sure the 12900K will match it highly tuned with £300 RAM but...... really? Does that make these simracers wrong?

I better not point out my 32GB of 8 Pack B-die C14 RAM then :cry:

Looks like massive overkill with the 5800X3D

LtMatt · 31 May 2022 at 19:47

Hmmm, I have Gears 5 (but never played it before). Will download and give it a go.