Will the 5900X or 5800X be your next gaming CPU?

Will the 5900X or 5800X be your next gaming CPU?

  • 5900X

    Votes: 73 55.3%
  • 5800X

    Votes: 25 18.9%
  • Neither/I'm waiting for next gen

    Votes: 34 25.8%

  • Total voters
    132
Going for less than 8 cores/16 threads is stupid, if you plan on keeping the computer for more than 2 years.

According to steam stats 8 cores are reaching 50% share in about 2 years. Consoles are using 8 cores/16 threads starting this year.
Yes, that's what I was going to say... PS5/xbox series x are already 8 core. I'm not planning on upgrading kids PCs for the next 4-5yrs, thus I'm leaning toward 5900x (if the price difference is indeed $100USD) or 5800x.
 
After watching this, I'd better be blown away by something at 1440p or I'll be "cheaping out" on a 3600 instead.

 
Going for less than 8 cores/16 threads is stupid, if you plan on keeping the computer for more than 2 years.

According to steam stats 8 cores are reaching 50% share in about 2 years. Consoles are using 8 cores/16 threads starting this year.

that's a meaningless metric in the grand scheme of things.. the xbox 1 and ps4 both had 8 core processors in them and the vast majority of games never used more than 4. on top of that 1 of the 8 cores on the xbox 1 was purely dedicated to OS and apps. so based on your theory all games released in the last 7 years should be scaling to 7 cores/threads minimum because consoles dictate how game engines are developed right? yet that hasn't been the case in the last 19 years since the original xbox released.

anyways we're getting off topic and this has been discussed to death but the thing to learn here is don't let consoles dictate what you choose to buy, buy based on your needs/wants.
 
that's a meaningless metric in the grand scheme of things.. the xbox 1 and ps4 both had 8 core processors in them and the vast majority of games never used more than 4. on top of that 1 of the 8 cores on the xbox 1 was purely dedicated to OS and apps. so based on your theory all games released in the last 7 years should be scaling to 7 cores/threads minimum because consoles dictate how game engines are developed right? yet that hasn't been the case in the last 19 years since the original xbox released.

anyways we're getting off topic and this has been discussed to death but the thing to learn here is don't let consoles dictate what you choose to buy, buy based on your needs/wants.
You said it yourself. PS4 is 8 threads, and thanks to game engines shifting towards 8 threads, 4 core 4 thread cpu's are completely outdated by now.

Just about the same thing will happen now, if you settle for 12 threads instead of 16 threads.
 
Going for less than 8 cores/16 threads is stupid, if you plan on keeping the computer for more than 2 years.

According to steam stats 8 cores are reaching 50% share in about 2 years. Consoles are using 8 cores/16 threads starting this year.

Let's be real here, until you see actual game benchmarks showing that 6 cores aren't enough then you don't actually have any constraints. People talked for years about how the I5 series with 4 cores and no hyper threading was a bottle neck but when benchmarked you could clearly see that 4 cores were enough and hyper threading actually reduced game performance in half the games out there. Buying a 6 core cpu now and saving $150 USD is an easy choice, it's going to be years before you need 8 cores and if you game at 1440p or above then it's going to be an even longer wait. The 5600x is a banger of a value, it even includes a cooler in the box.

Edit: For pure gaming I know I would much rather have a 5600x than a 3700x. The argument that games are becoming more multi-threaded is only sort of true, look at Flight Sim 2020, that game only really uses a single CPU core which is why Intel is kicking AMDs ass in that game (until the 5xxx launch).
 
Last edited:
More interested in Zen 3 Threadripper options whenever they come out. Hopefully no issue with using them on the TRX40 plate form.
 
I wonder if they'll wedge a 10 core 20 thread with 70MB cache between the R7 and R9. They didn't leave a lot of room in numbering, maybe it would be a 5850? I'd be happy with the 8/16 but I wanted that extra cache that the R9 has 😋
 
Last edited:
I wonder if they'll wedge a 10 core 20 thread with 70MB cache between the R57and R9. I'd be happy with the 8/16 but I wanted that extra cache that the R9 has 😋

that's what i was hoping the 5800x was going to be with the 5700x being the 8 core but looks like that's not going to happen.. :(
 
that's what i was hoping the 5800x was going to be with the 5700x being the 8 core but looks like that's not going to happen.. :(
The 5800 is odd from a naming standpoint. I can't imagine what a 5700 would be in terms of specs. So I'm guessing there won't be one this time around. I think the 5800 could easily have been named 5700, and then a 10 core be named 5800
 
The 5800 is odd from a naming standpoint. I can't imagine what a 5700 would be in terms of specs. So I'm guessing there won't be one this time around. I think the 5800 could easily have been named 5700, and then a 10 core be named 5800
probably 8c/16t with 5600x clocks OR they just decide to kill the sku for this series because of the RX5700 confusion.
 
Having a two CCD (two chiplet) 8 core I can see, 4 core each enabled for the failed full CCDs, 5700x?. Having a 10 core, 5 core each CCD would be an interesting skew. The 5800x has one CCD, does not have to use infinity fabric between the cores, maybe the best gaming chip, if PBO is effective and can bump up single core clock speed would be very nice. A full 200mhz bump => 5ghz speed. 3000 series never seemed to bump up past an increase of 100mhz.
 
Waiting for benchmarks before buying. 5600x or 5700x. Using a 2600 at the moment and it doesn’t brake a sweat. Next generation graphics card is what I am more interested in. Zen 4 and ddr 5 comes about I will make the jump as a early adopter I have no problem with this. I will chunk money
 
I already have a X570 motherboard all i need to do is swap a 3900x for a 5900x. Easy upgrade.
 
I deciding to spare myself the launch headaches after going through nvidias deal the other month. 3900x it is.
 
I just built my 3900x rig earlier this year. I tend to run my computers for a couple years before rebuilding. However, if real-world performance is a decent leap between the 3900x and the 5900x then I may upgrade the CPU next year. I probably wait until a year after the 5900x releases.
 
I may just buy a RX 6000 card because I have a B550m already and 32Gb of 3600Mhz memory .. I sold both Ryzen 5 3600 that I had and kept the 3700x as I always factored in the cooler which made the chip $280 MSRP release price day one and appears to be one of AMD's fastest chips in 65watt form it still can lead the pack as a flag ship that all chips are judge by in power usage .. I have thought of pairing it with some 4400Mhz dimms that the Mortar supports .

 
Between the two, I'd go 5900X, but I think I'll probably sit out this generation. The improvements are really nice, I just don't expect I'd notice any difference at 4K where the CPU is largely irrelevant.

I'm leaning toward sticking with my 3900X, it's certainly not holding me back in my non-gaming workloads (Photoshop/Lightroom, Fusion 360, Visual Studio).

Whenever I do upgrade, I'll replace the 3600 in my FreeNAS box with the 3900X 😎
 
Stuck between the 5800X and the 5900X myself. The 5900x seems like it should be a "decent" (couple of %) better gaming processor as it has a higher clock speed AND only has 6 cores per CCX, which would mean higher core/cache ratio. One would imagine both of those increase performance.

I have a X570 Tomahawk, which I believe has pretty decent VRM's so I should be ok to throw it in here.
 
Stuck between the 5800X and the 5900X myself. The 5900x seems like it should be a "decent" (couple of %) better gaming processor as it has a higher clock speed AND only has 6 cores per CCX, which would mean higher core/cache ratio. One would imagine both of those increase performance.

I have a X570 Tomahawk, which I believe has pretty decent VRM's so I should be ok to throw it in here.

Nah, the 5800x will have more efficient core/cache with all 8 cores in the same CCX. The difference are the boost clocks with the 5900x having higher boost clocks. But that all doesn't mean the 5800x is limited, just turn up the wick on PBO.
 
Im waiting a bit to see the other offerings from AMD. There is no way that there is no 5600/5700x, and most likely they will be as good as the 5800x in gaming!
 
Im waiting a bit to see the other offerings from AMD. There is no way that there is no 5600/5700x, and most likely they will be as good as the 5800x in gaming!

It's been said before that the other chips will come later.
 
Yes, everyone please wait to get a CPU so I can get one on launch day. Last thing I need other than a GPU upgrade to finish this build.
 
If we are talking high end AMD gaming CPUs for next gen. Why the heck is 5950x on there? If the new boost is any indication the highest binned chip will likely get most all core boost that games engage. I will probably be going all out this time and getting the 5950x and since zen 4 is new motherboard I will likely wait for zen 5 after this upgrade. Should control my upgrade itch for 3 years minimum lol. Plus will be a drop in upgrade to my zen 2. only time I hold off yearly CPU upgrade is if I have to change board and memory. Unless zen 4 is 50+% increase I should be fine for 3 years until zen 5 ish.
 
Nah, the 5800x will have more efficient core/cache with all 8 cores in the same CCX. The difference are the boost clocks with the 5900x having higher boost clocks. But that all doesn't mean the 5800x is limited, just turn up the wick on PBO.

But the 5900X has twice the L3 cache. 32MB for each 6-core CCX instead of 32MB total for all 8 cores on the 5800X. What I don't understand is how the 5900X will handle core prioritization. Let's say you have a program that uses 2 cores. Would there be more benefit from using 2 cores on the same CCX (less infinity fabric latency but having to share a single 32MB block of L3 cache) or would there be more benefit to using one core on each CCX (more infinity fabric latency but each core would potentially have access to it's own 32MB block of L3 cache)?
 
But the 5900X has twice the L3 cache. 32MB for each 6-core CCX instead of 32MB total for all 8 cores on the 5800X. What I don't understand is how the 5900X will handle core prioritization. Let's say you have a program that uses 2 cores. Would there be more benefit from using 2 cores on the same CCX (less infinity fabric latency but having to share a single 32MB block of L3 cache) or would there be more benefit to using one core on each CCX (more infinity fabric latency but each core would potentially have access to it's own 32MB block of L3 cache)?
I seriously doubt you'll see much difference between any of the AMD Zen 3 chips in games. I suspect the benchmarks will have the 5600x right at the top of charts above ALL 3xxx chips and just below the higher end 5800x and 5900x on launch day. If all you do is game, then saving money on a 5600x and getting a 3080 makes the most sense.
 
I seriously doubt you'll see much difference between any of the AMD Zen 3 chips in games. I suspect the benchmarks will have the 5600x right at the top of charts above ALL 3xxx chips and just below the higher end 5800x and 5900x on launch day. If all you do is game, then saving money on a 5600x and getting a 3080 makes the most sense.
I agree 100%. However since this is it for AM4 line and I do not upgrade every 2yrs, I'm leaning toward buying the best I can afford which at the moment is 5900x (assuming MSRP price will hold up).
 
But the 5900X has twice the L3 cache. 32MB for each 6-core CCX instead of 32MB total for all 8 cores on the 5800X. What I don't understand is how the 5900X will handle core prioritization. Let's say you have a program that uses 2 cores. Would there be more benefit from using 2 cores on the same CCX (less infinity fabric latency but having to share a single 32MB block of L3 cache) or would there be more benefit to using one core on each CCX (more infinity fabric latency but each core would potentially have access to it's own 32MB block of L3 cache)?

You've not been paying attention. Look at the recent synthetics bench leak. It's ideal to have all data inside a single CCX. Anytime you have to move across CCX dies you take a hit. Granted that hit is a lot less now than previously but still is a loss. You seem to be fixated on the caches too?

Anyways if ya look at the synthetics in the other IPC thread you'll see the 5800x has the greatest gain over its predecessor.
 
Last edited:
I agree 100%. However since this is it for AM4 line and I do not upgrade every 2yrs, I'm leaning toward buying the best I can afford which at the moment is 5900x (assuming MSRP price will hold up).
Looking at your current PC I would 100% get the 5600x and save the cash for a 3080. That 1070 is going to hold you back hard, I doubt you'll even see a frame rate increase over the 5ghz 8700k you're running now.

Better yet, just keep using your excellent 8700k and buy a 3080 as soon as you can find one. This is speaking purely from a gaming perspective.
 
Looking at your current PC I would 100% get the 5600x and save the cash for a 3080. That 1070 is going to hold you back hard, I doubt you'll even see a frame rate increase over the 5ghz 8700k you're running now.

Better yet, just keep using your excellent 8700k and buy a 3080 as soon as you can find one. This is speaking purely from a gaming perspective.
that's my rig, I'm building 2 pcs for my kids. (my old 3570k and 6500 based systems). Once they will be able to buy 3080s; I'll be getting back my 1070 to drive (downgraded) 1080p@60 display :) I don't game that much anymore...
 
You've not been paying attention.

I'm waiting for 3rd party benchmarks (not random leaks) just like everyone else. My question was not about 5800X vs 5900X, but how the 5900X will prioritize it's cores. If a program uses 6 cores or less, a 5900X could potentially run them all on a single CCX also - if that's how it makes sense to prioritize it's core usage.

It's ideal to have all data inside a single CCX.

"All data" meaning what exactly? Even if their is a benefit to having your game only use cores from one CCX and sharing the same cache, it probably won't help if your OS and everything else in your system is also on that same CCX and also using up some of that same cache. With a 5900X, you could potentially have your game running with up to 6 cores on a single CCX, with it's own cache, separate from other things like your OS which could all use the other CCX. That's why I asked about core prioritization, because on one hand the 5900X could act very much like a 5800X up to 6 cores (if it tries to fill one CCX first) or it could act totally different (if it tries to spread the load evenly between each CCX).

You seem to be fixated on the caches too?

Double the cache is hardly a tiny difference. I don't think that asking a question about it means that I am "fixated" rolleyes.gif
 
I'm waiting for 3rd party benchmarks (not random leaks) just like everyone else. My question was not about 5800X vs 5900X, but how the 5900X will prioritize it's cores. If a program uses 6 cores or less, a 5900X could potentially run them all on a single CCX also - if that's how it makes sense to prioritize it's core usage.



"All data" meaning what exactly? Even if their is a benefit to having your game only use cores from one CCX and sharing the same cache, it probably won't help if your OS and everything else in your system is also on that same CCX and also using up some of that same cache. With a 5900X, you could potentially have your game running with up to 6 cores on a single CCX, with it's own cache, separate from other things like your OS which could all use the other CCX. That's why I asked about core prioritization, because on one hand the 5900X could act very much like a 5800X up to 6 cores (if it tries to fill one CCX first) or it could act totally different (if it tries to spread the load evenly between each CCX).



Double the cache is hardly a tiny difference. I don't think that asking a question about it means that I am "fixated" View attachment 290579

AMD's algorithms will keep all the data within the CCX if possible to reduce hits incurred if/when it has to move across to the other CCX. Thus with a CCX that is only 6 cores per, it will have greater misses thus incurring a hit versus an 8 core CCX. You are misunderstanding how that cache works and you are making the assumption that 6 cores per 32mb caches is somehow faster or such. Bottom line the smaller cores per ccx will incur more penalties faster. The 5900x is actually the worse cpu in the line up because it suffers from only 6 cores per ccx. The 5950x is 8 cores per ccx so it's hit is less but obviously nothing you can do about it, though its offset by the highest bins.
 
AMD's algorithms will keep all the data within the CCX if possible to reduce hits incurred if/when it has to move across to the other CCX. Thus with a CCX that is only 6 cores per, it will have greater misses thus incurring a hit versus an 8 core CCX. You are misunderstanding how that cache works and you are making the assumption that 6 cores per 32mb caches is somehow faster or such. Bottom line the smaller cores per ccx will incur more penalties faster. The 5900x is actually the worse cpu in the line up because it suffers from only 6 cores per ccx. The 5950x is 8 cores per ccx so it's hit is less but obviously nothing you can do about it, though its offset by the highest bins.
It would be so much easier if we had independent reviews out a day or two before launching CPU lineup. Damn you if you not order it asap as you may risk to have it out of stock and damn you if you order the "wrong" one... LOL However making educated decision after reading some reviews makes more sense. For me it is between 5800x and 5900x. I'm willing to pay extra to have piece of mind of having at least 8 core CPU, which might be stupid, unnecessary and a total waste of money.
 
It would be so much easier if we had independent reviews out a day or two before launching CPU lineup. Damn you if you not order it asap as you may risk to have it out of stock and damn you if you order the "wrong" one... LOL However making educated decision after reading some reviews makes more sense. For me it is between 5800x and 5900x. I'm willing to pay extra to have piece of mind of having at least 8 core CPU, which might be stupid, unnecessary and a total waste of money.

I think for 98% of users the 5800x or the eventual 5700x is the sweet spot. Just raise the PPT and booyah.
 
but for only $100USD (if true) you can score an extra 4 cores!!! LOL

But it comes at a cost that's why the gap isn't large. Ya need to realize that the two most optimized cpus will be the 5800x and the 5950x as they booth max out the the cores per ccx ratio.
 
AMD's algorithms will keep all the data within the CCX if possible to reduce hits incurred if/when it has to move across to the other CCX. Thus with a CCX that is only 6 cores per, it will have greater misses thus incurring a hit versus an 8 core CCX. You are misunderstanding how that cache works and you are making the assumption that 6 cores per 32mb caches is somehow faster or such. Bottom line the smaller cores per ccx will incur more penalties faster. The 5900x is actually the worse cpu in the line up because it suffers from only 6 cores per ccx. The 5950x is 8 cores per ccx so it's hit is less but obviously nothing you can do about it, though its offset by the highest bins.

I'm just wondering why that logic doesn't seem to apply to current high-end Zen2 chips?

For example, when we compare the 3800X vs the 3900X:
https://cpu.userbenchmark.com/Compare/AMD-Ryzen-9-3900X-vs-AMD-Ryzen-7-3800X/4044vs4047

Single-core and Dual-core speeds are the same, and presumably both the 3800X and 3900X would do this using only one CCX.
Quad-core speed is also the same, despite the fact that the 3900X only has 3 cores per CCX, so by definition the 3900X would be using at least 2 different CCX at this point. The 3800X could run it on a single CCX, yet that doesn't do anything to help it.
Octa-core speed is faster on the 3900X by a significant amount, even though the 3900X would have to use at least 3 different CCX at this point. The 3800X would only be using 2 CCX because that's all it has, but again, that doesn't seem to help it.
 
I'm just wondering why that logic doesn't seem to apply to current high-end Zen2 chips?

For example, when we compare the 3800X vs the 3900X:
https://cpu.userbenchmark.com/Compare/AMD-Ryzen-9-3900X-vs-AMD-Ryzen-7-3800X/4044vs4047

Single-core and Dual-core speeds are the same, and presumably both the 3800X and 3900X would do this using only one CCX.
Quad-core speed is also the same, despite the fact that the 3900X only has 3 cores per CCX, so by definition the 3900X would be using at least 2 different CCX at this point. The 3800X could run it on a single CCX, yet that doesn't do anything to help it.
Octa-core speed is faster on the 3900X by a significant amount, even though the 3900X would have to use at least 3 different CCX at this point. The 3800X would only be using 2 CCX because that's all it has, but again, that doesn't seem to help it.

I'm not following what you are asking? On Zen 2 the dual CCD chips have a big advantage over the single CCD chips. I dunno where you're going but if you really want a 5900x, go for it shrugs. The 5900x won't have the ideal cores to ccx arrangement by the mere fact of it's core count.
 
Back
Top