Moving from 3800x to 5900x any comments?

[narsbars]

I am on a 570 platform with a 3070 vid card, I never OC. I grabbed a 5900X to replace my 3800X, also replaced my Wraith Prism with an Assassin 120. My 4K monitor is limited to 60 FPS, but under testing I have seen it drop to low 40s and not always maintain a steady 60 FPS. Will I see any difference in 4K performance, 1080, or 1440? Have I bought any time prior to NEEDING to upgrade?
Basically hoping I didn't waste money just because the upgrade bug bit my butt.

 

[Zepher]

Do you have the 5900X installed? If you do, just run games in those resolutions you asked about and see for yourself.

 

[Mloot]

OP, if you are upgrading solely for gaming and still have a chance to return your 5900x, you may want to get the 5800x3D instead. Hardware Unboxed did a nice video pitting it against the Ryzen 3600 and Ryzen 5600x, both with a 6600xt and a 6950xt at a couple of different resolutions. You may be able to assume that the 5600 will have somewhat similar performance to the 5900x (within reason) and the 3600 with your 3800x. In some games, the 5800x3D saw some nice gains in average FPS and again in 1% lows.

Here is the link to that video: 5800x3D benchmarks

However, if you are doing productivity along with gaming, then the 5900x is likely the better choice over the 3D part.

 

[pitingres]

Try dropping graphics detail levels to low, and run the game(s) that have fps drops. If the drops go away, you're probably GPU limited, not CPU limited.

A 5900X should give you some improvement over a 3800X. I do agree though that if you can switch to a 5800X3D you'll see more improvement for gaming, in most games.

 

[jlbenedict]

Basically hoping I didn't waste money just because the upgrade bug bit my butt.

That upgrade bug gets more ass than... umm... well.. it never runs out of teeth



Is it safe to say that at 1440p and higher though... maybe the upgrade should have been something more than the 3070? I would think you would get more benefits there, at 4k

But, as typical...there are those specific game titles that scale with bumps in CPU, more so, than just bumps in GPU alone..

 

[narsbars]

Sounds like I fed the upgrade bug for very little gain. Well at least I can move the old 3800X to offset the cost. FYI, this was a used CPU so I saved substantially from new, counting on the resale of the 3800x to offset some of the cost.

 

[GotNoRice]

I am on a 570 platform with a 3070 vid card, I never OC. I grabbed a 5900X to replace my 3800X, also replaced my Wraith Prism with an Assassin 120. My 4K monitor is limited to 60 FPS, but under testing I have seen it drop to low 40s and not always maintain a steady 60 FPS. Will I see any difference in 4K performance, 1080, or 1440? Have I bought any time prior to NEEDING to upgrade?
Basically hoping I didn't waste money just because the upgrade bug bit my butt.

The bottom line here is that you need to monitor your GPU usage while gaming/testing. This is infinitely more informative than simply looking at your FPS and making assumptions. Any time your GPU usage is at 100%, you are GPU limited, and at that point you could double your CPU power and still not see an increase. That's not the CPU's fault.

But the 5900X is indeed a good step-up from the 3800x. I upgraded from a 3900X to a 5900X at one point, and in situations where I was not GPU limited, the performance increase was noticeable. If you are GPU limited now, at least you know what you need to upgrade next.

 

[pitingres]

Also, if you're GPU limited in a game, you have options: change resolution or change graphics quality settings. If you are CPU limited, it's usually just the way the game engine is written and there's rarely much you can do about it, other than throw hardware at it.

 

[chino069069]

The bottom line here is that you need to monitor your GPU usage while gaming/testing. This is infinitely more informative than simply looking at your FPS and making assumptions. Any time your GPU usage is at 100%, you are GPU limited, and at that point you could double your CPU power and still not see an increase. That's not the CPU's fault.

But the 5900X is indeed a good step-up from the 3800x. I upgraded from a 3900X to a 5900X at one point, and in situations where I was not GPU limited, the performance increase was noticeable. If you are GPU limited now, at least you know what you need to upgrade next.

i'm about to make a jump from a 1600 and a rx 580, assume it would be cost effective to go straight to a 5 series or make a pit stop in the 3 series.. ?

 

[riev90]

i'm about to make a jump from a 1600 and a rx 580, assume it would be cost effective to go straight to a 5 series or make a pit stop in the 3 series.. ?

jump directly to 5000 series.
the vanilla 5600 (non x) or 5700x are a lot cheaper now and will have substantial performance bump from 1600x.

If your usage is solely for gaming, then ryzen 5600 + gpu of rx 6700xt / 6800 / gtx 3070 / 3080 will do nicely.

 

[funkydmunky]

i'm about to make a jump from a 1600 and a rx 580, assume it would be cost effective to go straight to a 5 series or make a pit stop in the 3 series.. ?

Depends on budget and what you want out of it as far as longevity. I think going to a 5600 and whatever GPU fits your price would be a very economical and large performance boost.

 

[RecentlyAdded]

I've been running the 5900X for 2 years and not going back to under 12 cores!
I would say learn how to do the precision boost for each individual core. The Ryzen master tool will figure out the optimum underclock for each core(0 to -30) and I set my in bios to what Ryzen master suggests and give it +3 cushion.

 

[GotNoRice]

i'm about to make a jump from a 1600 and a rx 580, assume it would be cost effective to go straight to a 5 series or make a pit stop in the 3 series.. ?

You should see if your board is compatible with the 5800X3D. It's the fastest AM4 gaming CPU and can actually go toe-to-toe or even beat next-gen chips in many games. I'm not sure why you would bother with the 3 series (Zen 2) unless that is all your motherboard will support or you get a really really good deal.

 

[funkydmunky]

I've been running the 5900X for 2 years and not going back to under 12 cores!
I would say learn how to do the precision boost for each individual core. The Ryzen master tool will figure out the optimum underclock for each core(0 to -30) and I set my in bios to what Ryzen master suggests and give it +3 cushion.

Again a non descript work use. If gaming then BS. We have enough benchies to nullifuy your 12 core or bust hypothosis. Detailed bemchmarks welcomed of course.

 

[kirbyrj]

Anything other than strictly gaming should see a noticeable bump in performance.

Gaming at 4k/60Hz with a 3070 is probably going to be a wash. Might see more of a difference with faster refresh monitor and a higher end video card.

 

[narsbars]

Anything other than strictly gaming should see a noticeable bump in performance.

Gaming at 4k/60Hz with a 3070 is probably going to be a wash. Might see more of a difference with faster refresh monitor and a higher end video card.

Now that I have had the 5900X in for long enough to test it I find that I get far fewer dips in fps. My monitor is limited to 60fps but was dropping to very low 30s in various games but now it maintains in the mid 40s at the lowest point. This is not a scientific measurement but I usually wait to upgrade anything until I can "feel" the difference. Went from a 1080 to a 3070, no OC and could feel it even if not overwhelmed. I got a good buy here on the H and feel like I made the right choice. Now waiting until next year for a 3090TI hoping the prices will drop. Thanks to everyone for the input.

 

[GotNoRice]

Now that I have had the 5900X in for long enough to test it I find that I get far fewer dips in fps. My monitor is limited to 60fps but was dropping to very low 30s in various games but now it maintains in the mid 40s at the lowest point. This is not a scientific measurement but I usually wait to upgrade anything until I can "feel" the difference.

Unfortunately the 5800X3D trounces all other AM4 CPUs (including the 5900X and 5950X) when it comes to feeling "smooth" aka reduction of micro-stutter. Being able to visualize it like this is very informative IMO, because this is almost never represented or shown in typical reviews. All of those tall orange lines = Microstutter. Note that in this comparison, the 5900X even has a slight RAM speed advantage.

It's also important to think about how RAM dependent regular Zen 3 chips are. CPUs like the 5900X show big gains from upgrading your RAM to at least 3600Mhz. But the 5800X3D, with it's huge cache acting as a buffer between the CPU and RAM, isn't really held back much even if you still have slow DDR4 from many years ago (which is likely the case if you are upgrading from an older-generation Ryzen CPU).

We have enough benchies to nullifuy your 12 core or bust hypothosis. Detailed bemchmarks welcomed of course.

Pretty much... 8 Cores vs 12 Cores is almost irrelevant even for productivity unless you are doing very very specific CPU intensive workloads that can actually load down more than 8 cores. It wasn't that long ago that 4 cores was still considered high-end; 8 cores is still a lot. In fact, even the new Intel top-dog 13900k only has 8 performance cores. The other 16 efficiency cores are irrelevant in any gaming scenario (many people simply disable the E cores). In reality, most people (even users who do use their systems for more than gaming) aren't doing stuff that would load down more than 8 cores, or if they are, it's very occasionally. Going from a 5900X to a 5800X3D for example might make it a little slower once or twice per day when you do something like compress a ZIP file, but in games the 5800X3D dominates. Does it really matter if a ZIP file takes 30 seconds to compress instead of 20 seconds? Maybe if you are doing it hundreds of times per day, otherwise...

The performance difference for gaming is kind of insane actually.

And even Intel is forced to admit how well the 5800X3D does against it's top-end chip. This is a slide from their 13-series launch presentation, where they even went so far as to represent the data from the 5800X3D in a rather odd way in order to make the beatdown seem less severe. They also include numbers from the 5950X, which should be pretty close to if not identical to a 5900X in terms of gaming performance.

That isn't to say that the 5900X is bad, not by any means, but the 5800X3D is not only faster than the 5900X in gaming, it's in an entirely different league. Even with a 5900X maxed out via PBO, etc, it's not going to make up the difference, not even close.

 

[narsbars]

Unfortunately the 5800X3D trounces all other AM4 CPUs (including the 5900X and 5950X) when it comes to feeling "smooth" aka reduction of micro-stutter. Being able to visualize it like this is very informative IMO, because this is almost never represented or shown in typical reviews. All of those tall orange lines = Microstutter. Note that in this comparison, the 5900X even has a slight RAM speed advantage.

View attachment 528698

It's also important to think about how RAM dependent regular Zen 3 chips are. CPUs like the 5900X show big gains from upgrading your RAM to at least 3600Mhz. But the 5800X3D, with it's huge cache acting as a buffer between the CPU and RAM, isn't really held back much even if you still have slow DDR4 from many years ago (which is likely the case if you are upgrading from an older-generation Ryzen CPU).



Pretty much... 8 Cores vs 12 Cores is almost irrelevant even for productivity unless you are doing very very specific CPU intensive workloads that can actually load down more than 8 cores. It wasn't that long ago that 4 cores was still considered high-end; 8 cores is still a lot. In fact, even the new Intel top-dog 13900k only has 8 performance cores. The other 16 efficiency cores are irrelevant in any gaming scenario (many people simply disable the E cores). In reality, most people (even users who do use their systems for more than gaming) aren't doing stuff that would load down more than 8 cores, or if they are, it's very occasionally. Going from a 5900X to a 5800X3D for example might make it a little slower once or twice per day when you do something like compress a ZIP file, but in games the 5800X3D dominates. Does it really matter if a ZIP file takes 30 seconds to compress instead of 20 seconds? Maybe if you are doing it hundreds of times per day, otherwise...

View attachment 528700

The performance difference for gaming is kind of insane actually.

And even Intel is forced to admit how well the 5800X3D does against it's top-end chip. This is a slide from their 13-series launch presentation, where they even went so far as to represent the data from the 5800X3D in a rather odd way in order to make the beatdown seem less severe. They also include numbers from the 5950X, which should be pretty close to if not identical to a 5900X in terms of gaming performance.

View attachment 528701

That isn't to say that the 5900X is bad, not by any means, but the 5800X3D is not only faster than the 5900X in gaming, it's in an entirely different league. Even with a 5900X maxed out via PBO, etc, it's not going to make up the difference, not even close.

Thanks for the great and informative answer. Luckily for me they didn't compare the low price I paid for the 5900x. I ended up saving $150.00 over the lowest price 5800X3D I could find on the used market. I am definitely saving your answer as reference material. Thanks again.