***The Official 5900X \5950X owners thread***

[von Humboldt]

Lads, which website/YT channel would be most reliable for gaming benchmarks? I want to compare the 5900X with the 5800X3D at 3440P UW.

 

[kully1972]

Lads, which website/YT channel would be most reliable for gaming benchmarks? I want to compare the 5900X with the 5800X3D at 3440P UW.

Try gamers nexus or jay2cents or overclock3d for starters

 

[Brutos]

Even then, I have my doubts. Zen 4 processors may have a slight performance edge over Zen 3, but they aren't as power efficient eg. a 7950X can hit 300W ( vs half that for a 5950X ). Same situation for the latest Nvidia GPUs - increased performance but at the cost of a big increase in power consumption. In the current climate, where the price of energy is going up, who wants to afford the running costs of this new hardware ?

You are correct actually, the power draw of these new cpus and gpu too high and not worth the extra cost at all.

 

[kully1972]

Just installed my 5950x.
I think it hates my Corsair dominator ram and refuses to boot into windows with 3200 mhz xmp docp profile,
I’ve tried inputting values manually as well with no luck
Stress testing only seems to like 2133 mhz.
I have the latest bios installed on ASUS impact viii Itx mobo.
Any ideas?

 

[uscool]

Just installed my 5950x.
I think it hates my Corsair dominator ram and refuses to boot into windows with 3200 mhz xmp docp profile,
I’ve tried inputting values manually as well with no luck
Stress testing only seems to like 2133 mhz.
Any ideas?

Updated to the latest bios ?

 

[InfiniteNoise]

Just installed my 5950x.
I think it hates my Corsair dominator ram and refuses to boot into windows with 3200 mhz xmp docp profile,
I’ve tried inputting values manually as well with no luck
Stress testing only seems to like 2133 mhz.
I have the latest bios installed on ASUS impact viii Itx mobo.
Any ideas?

I would say reset the BIOS, but you probably already did that when the motherboard prompted you to reset settings after you installed your 5950X, but try that again just in case, but this time leave everything else on default settings and use XMP/DOCP and see how it goes.

Also try upping the voltage a bit for the RAM, I had a similar issue with my previous motherboard, it was unstable at 3200MHz XMP until I set it at 1.4v, you could try going up in 0.01v increments (starting at 1.36v), but don't go above 1.5v, but that's really only for extreme overclocking so you won't need to get to that point.

 

[kully1972]

I would say reset the BIOS, but you probably already did that when the motherboard prompted you to reset settings after you installed your 5950X, but try that again just in case, but this time leave everything else on default settings and use XMP/DOCP and see how it goes.

Also try upping the voltage a bit for the RAM, I had a similar issue with my previous motherboard, it was unstable at 3200MHz XMP until I set it at 1.4v, you could try going up in 0.01v increments (starting at 1.36v), but don't go above 1.5v, but that's really only for extreme overclocking so you won't need to get to that point.

Thanks for the advice infinitenoise but upping voltage did not help
I heard some 5000 series CPU’s have poor memory controllers
I’m tempted to buy new micron g skill memory as the Corsair is Samsung b die I think

 

[uscool]

Poor memory controller ?, CPU officially supports effective memory speeds up to 3200 MHz

 

[InfiniteNoise]

Thanks for the advice infinitenoise but upping voltage did not help
I heard some 5000 series CPU’s have poor memory controllers
I’m tempted to buy new micron g skill memory as the Corsair is Samsung b die I think

If the new RAM doesn't work, it's possible that 5950X is the issue, though CPU being the issue is very rare, I don't think I've heard anything about 5000 series having poor memory controller, I know 1000 series had poor memory controller, especially if you go above 3200 MHz.

Looking at your sig, I'm assuming your 3900X worked fine with the RAM at 3200 MHz?

I'm running 3600 MHz on the RAM with my 5900X without any issues.

 

[kully1972]

Yes 3900x worked fine with same ram.
I’ll give the g skill a try and if that fails it’s not the end of the world.
The increase in fps is ten percent at best and I’m happy to take the hit.

 

[Ross Thomson]

Just became a member of this club. Got a 5900x for £360. Its replacing a 1700 in an MSI x470 gaming M7 AC board.

 

[InfiniteNoise]

Yeah, I grabbed a 5900X for 371 a month ago just before the price went back up, and I see it's gone back down again, 5800X3D is also at the same price too, I probably would have went for that if I'd just waited.

I'm impressed at how much cooler 5900X runs compared to 5800X, even with a worse cooler, I went from Scythe Fuma 2 (80c max with 5800X, with some CO adjusting) to Noctua U12S Redux (I haven't seen it go above 74c with 5900X, at stock settings).

 

[Ross Thomson]

Yeah, I grabbed a 5900X for 371 a month ago just before the price went back up, and I see it's gone back down again, 5800X3D is also at the same price too, I probably would have went for that if I'd just waited.

I'm impressed at how much cooler 5900X runs compared to 5800X, even with a worse cooler, I went from Scythe Fuma 2 (80c max with 5800X, with some CO adjusting) to Noctua U12S Redux (I haven't seen it go above 74c with 5900X, at stock settings).

I want the 12 cores more than the extra gaming fps. I'm on a custom loop

 

[Minstadave]

Yeah, I grabbed a 5900X for 371 a month ago just before the price went back up, and I see it's gone back down again, 5800X3D is also at the same price too, I probably would have went for that if I'd just waited.

I'm impressed at how much cooler 5900X runs compared to 5800X, even with a worse cooler, I went from Scythe Fuma 2 (80c max with 5800X, with some CO adjusting) to Noctua U12S Redux (I haven't seen it go above 74c with 5900X, at stock settings).

Same TDP but the heat is spread across two chiplets instead of one, makes a big difference to temps.

 

[Trick]

I've done about 35 hours of Curve Optimiser testing on my 5900X now, which CTR reports as a Silver sample. this is what I've finished with and is 12+ hours of CoreCycler/Prime95 stable:

CCD0 CCD1
(0) -6 (6) -30
(1) -8 (7) -30
(2) -4 (8) -30
(3) -22 (9) -30
(4) -23 (10) -28
(5) -18 (11) -30

PPT = 185W
TDC = 125A
EDC = 140A

Boost is set to +150Mhz, there doesn’t appear to be any clock stretching when observing clock and effective clock in hwinfo64.

At all core load I'm getting 4575Mhz on each core. At single core load i can occasionally touch 5100Mhz and routinely get 4950Mhz on cores 1 and 2. All core temps dont get above 71 degrees using an Arctic Freezer II 240mm AIO.

Since AGESA 1.2.0.5 the boost/undervolt/power-handling/overclocking/call-it-what-you-will is nerfed compared to what benchmark scores people were getting when Ryzen 5000 series was first released, and I encountered the EDC 'bug' that was reported where setting EDC >140A resulted in boost voltage being artificially throttled. With my settings PPT sits at around 98%, TDC at 98% and EDC 100%. I havent much tried backing off PPT and TDC yet but a few tests seemed to indicate a noticable performance drop.

One thing I haven’t been able to work out though is why power-reporting (HWInfo64 PRD) is 112% at full load when stock. This and the AGESA power-bug might account for some of the poor stock performance, but I see no way in the bios to dial-in compensation for this for my MSI X570 Tomahawk without knowing the boards rated current delivery. As it stands the system might think it’s drawing 12% more power than what it actually is hence throttling boost a bit. Iirc GN did a video on this a few years back.

Anyway, my Cinebench scores and boost seem ok for both single and multi-core given what amd have done to agesa >1.2.0.5 (I’m on 1207), so I think I'm done with tuning for a while. With PBO turned off or set to auto my cpu performance is pretty poor compared to where i think it should be, so i guess the effort was worth it.

Cinebench R20 m = 8745 s = 630
Cinebench R23 m = 22254 s = 1612
CPU-Z v17.x m = 9999 s = 670

 

[Ross Thomson]

Got my 5900x a couple of days ago and paired it with a msi B550 unify and CL16 4000 ram.

PPT 210 (195)
TDC 140 (130)
EDC 190

-30 on all cores

Hits 4650-4674 on all cores during a R23 run. Temp 75C (out the box was 63C). Single core R23 is 4900.

Score 23801 is my best.

Ram isn't so good. Can't hit 3800 at all and only above 3600 with GDM on.

 

[Finners]

I've done about 35 hours of Curve Optimiser testing on my 5900X now

Nice detailed post, I'm just starting to play with CO now but trying to get the best efficiency I can vs outright performance. What RAM do you have.

I'm jealous of people who seemingly can just run -30 on everything or my threshold for stable is higher. My numbers per core are slightly worse than yours and I'm not even trying to increase the core speeds so my max is 4950.

 

[Trick]

@Finners yeah I’d be surprised if the all core -30 setups are fully stable. I can set mine at fully -30 and it’ll not crash in windows/games and a couple of minutes on corecycler, but if it’s tested properly it soon gives up.

Regarding ram: nothing special, just TeamGroup 3200 CL16 running at 3600 CL14

 

[MrFrazzles]

I'm only getting 9300 on Cinebench R20 Stock with my 5950x Why is that?

I get around 24,000 on R23, which is actually about normal for stock.

 

[gbgt2]

I'm only getting 9300 on Cinebench R20 Stock with my 5950x Why is that?

I get around 24,000 on R23, which is actually about normal for stock.

RAM speed? Have you definitely applied DOCP? You should be scoring a bit over 10K I think