AMD Ryzen 7 5800X3D gets benchmarked ahead of embargo lift

Do not rush with a single result you can not draw a conclusion,as AMD has said the difference in SOTR will be around 10%, the system that ran with INTEL had memories at 4.8GHz so there is room for greater performance on the other hand of course and AMD's system ran memory at 3.2 at stock settings so in this system there will be an increase with higher frequencies 3733-3800 if the memory controller works as in 5800x,it also does not matter what graphics card they used,at 720p all you need is to have the details game at ultra as well as the shadows so you can get the graphics card out of the equation,we also have to wait for AGESA 1.2.7 which will also bring better performance and the new driver for the AM4 chipsets,Gigabyte has already released a new beta driver for its motherboards there are many things to be done up to the reviews.

 

Show me
Aida 64 memory benchmark maybe?


I'm using some cheap green Dell 4800 mhz that I overclocked a bit, so nothing special.

Result with "stock" cpu:

 

So apparently that V-cache is much more worth than stupid eight E-cores on top. Who would have thought.
Really weird how that mediocre Alder Lake got hyped just because Intel didn't screw up entirely...

 

Asus sucks especially the apex

 

Yep, the worst performing motherboard in history

 

"Asucks"

 

Hehe, mines plain, MSigh? My previous system was Gigashyte

 

take it from someone who built a 3960x 5950x and 12900k in the last 3 years
fps especially premade ingame have become completely useless, they say 150fps whatever but it doesn't look nor feel like it on AMD and it's not
if you look at the 1% you'll see intel is noticeably higher and if you look at the frametimes you'll know why AMDs even 5000 have less, lots of small spikes when Intel is more like a straight line

I'll believe the 5800x is better if I see the tomb raider benchmark running with zero hiccups nor stuttering (something I see all the time on big channels ltt,jayztwocents,level1techs they think it's normal but it's not) I had those on 5950x they are completely gone on 12900k (same with borderlands 3 benchmark if it freezes in some parts..it also doesn't on intel)
what to watch for ? when the camera pans left to right on the ruins and then goes to the market..if you have any stutter or freeze it comes from the plateform
edit : on one of the leaked pic you can see a very nice straight horizontal frametimes line...interesting, very...I still have an AM4 motherboard and it's way better than garbage z690

there was a big jump in smoothness going from 3000 to 5000 I really hope for 7000 they will finally do it because I hate Intel way to manage lanes and bandwith so much

 

Sorry, but that's just a bunch of nonsense. BL3 not freezing on Intel, lol. The game thread stalls for hundreds of milliseconds when some enemies drop epic loot, no matter the CPU...

 

I did a little 720P low detail with motion blur set to off was the only change and I avg 283 fps with my setup in Tomb Raider.

Asus z690 Extreme MB
32 gb Gskill DDR 5 6000 c36 @ c32 timings
Zotac 3090 ti Amp Holo
Intel 12900k 5.1 ghz all core

 

Using same settings , as tested . 185 vs 231 .

 

i wsill wait for ryzen 6600x later this year be more worth it i think

 

I just ran that bench thanks to them having a demo version of the game and at "low" preset with bloom off I get 187 and it only pegs my 3090 at 57% so Shadow of the Tomb Raider is quite CPU limited it appears.

 

They aren't using the ingame benchmark. Looks like they did a coustom run, just to make it harder to compare they're result....

We have 370fps+ with 12900k and 7200mhz c30 tweaked ddr5 in this benchmark.

This is why it's pretty epic fail to use 4800mhz c40 ddr5 in this benchmark, to compare with 5800x3d.....

 

Why is it a failure? Not everyone can buy expensive DDR5 and push it to 7.2.

 

Why are you showing 7000 CL32 in Aida and 7200 CL 30 in Tomb Raider?
Do you have individual overclock profiles depending on what benchmark or game is used?

I am at 112 FPS with my old 1700 with 3200mem, but even with that old CPU the game is only using 60-70W power compared to 120-130W at full load, so the benchmark run is not using enough CPU cores, even at that stupidly low resolution.
That is probably why they chose to do a custom run, just like hardware unboxed also does.

 

You are correct intel is smoother with less stuttering but in saying that intel has been riding the ringbus since the i7 920 so it's got all the kinks out, I know the 12900k isn't technically a ring bus CPU but it sort of uses a lot of design from it. AMD has only just started with Zen architect and going from 3000 to 5000 was a pretty big change from going for 4X4 each CCD to 8X8 CCDS so yea it will be interesting for the future.

 

Pretty good results from a last gen amd CPU going up against a next gen intel cpu. Makes me wonder what AMD's next gen CPU will do if it's that much faster then a 5800x3d at the same price point which also uses ddr5.

 

very nice,but im afraid this technology isn't going to make it to 200eur cpus,and this is where the best perf/dollar gaming cpus are.
I'm not sure it'll make it into 250-300eur segment even.

it is,however,the best way to use resources to make a modern 8 core gaming cpu faster.
better than putting extra 8 e-cores or 4 normal cores on an 8-core cpu.

I think once people have a chance to choose an 8 core with extra cache over a 12 core, they'll go for 3d cache and put this silly idea of futureproofing an 8/16 with 4 extra cores to bed.

 

Future proofing is a bit silly in general if you have something decent now . It does make sense in case you are buying now a whole system and you are in dilemma 250 euros for the 6 core or 310 for the 8 core ... In such case it does make sense to go for the 8core for "future proofing ". Otherwise you are spending today to get performance you might need tomorrow and as a result you are getting less performance for more money!