AMD Ryzen 5000 Builders Thread

[DasFox]

Amazon has the 5600X at $199, I just bought one, for the specs, this is a pretty decent deal.

 

[Dave3000]

Between a Ryzen 5950X and a Ryzen 5800X3D, which CPU will be better in the long run for gaming, at least for the next 3 years. I'm right now using an 5800X with an Asus B550-F (Wi-Fi) motherboard and I'm trying to decide whether to upgrade to a 5950X or a 5800X3D. I'm mostly gaming on this PC but I also do office work (not that I need better productivity performance). I do know that if I upgrade to a 5950X, everything that is CPU intensive will be faster but if I upgrade to a 5800X3D, only some CPU intensive games will be much faster based on the reviews I read, and I will take a little performance hit in raw CPU performance and productivy.

 

[scannall]

Between a Ryzen 5950X and a Ryzen 5800X3D, which CPU will be better in the long run for gaming, at least for the next 3 years. I'm right now using an 5800X with an Asus B550-F (Wi-Fi) motherboard and I'm trying to decide whether to upgrade to a 5950X or a 5800X3D. I'm mostly gaming on this PC but I also do office work (not that I need better productivity performance). I do know that if I upgrade to a 5950X, everything that is CPU intensive will be faster but if I upgrade to a 5800X3D, only some CPU intensive games will be much faster based on the reviews I read, and I will take a little performance hit in raw CPU performance and productivy.

What you have is just fine. Are you sacrificing children for a 4090ti? Yeah, it will still be fine. 500 FPS mean nothing in day to day use.

 

[DasFox]

Are you sacrificing children for a 4090ti?

LOL, that’s a good one! 👍

Anytime you buy something, ‘Best’ really just depends on your ‘Needs’.

I mean, do we need a 4090ti to play WOW? LOL of course not, so buy whatever suits your purpose, and if you don’t know what that is, all you do is research specs on gamimg.

At the moment, all I mainly play is Apex on a 1060 6gb, don’t need a 3080 for Apex...

To many people buy to much overkill and just waste money.

And last, the hardest part to figure out is ‘Future Proof’, making it last, but when you game a lot, the trends in gaming don’t always change extremely radical from one year to the next, and neither does hardware, usually in small increments and then performance gain sometimes only slight boosts too at times... USB2, USB3, PCI-E3, 4, 5, etc...

P.S. My gaming box I just built is a Rog Strix B550-F Gaming, Ryzen 5600x, 2x8gb ram, corsair rm750x, and looking to get a EVGA GeForce RTX 3070 XC3 BLACK GAMING, 08G-P5-3751-KL. For 2K Gaming this should still be pretty sweet for sometime, or next year, slap in a 4070, or 4080. I also bought a MSI Optix G273QPF, it’s an ISP panel Gsync, 165hz, still sitting in the box.

 

[StinkyPinky]

I have a 3080 12GB card and usually spend my days playing 10 year old games like Kings Bounty and Civ 5 and 6.

I think I just got old and can't be bothered learning new games anymore.

One of these days I'll play something newer

 

[igor_kavinski]

Between a Ryzen 5950X and a Ryzen 5800X3D, which CPU will be better in the long run for gaming, at least for the next 3 years.

DDR5 is the future due to massive bandwidth increase with higher RAM speeds. Future games/game engines will likely be developed to take advantage of this bandwidth. The V-cache on 5800X3D provides 2+ TB/s bandwidth. I think for gaming at least, the 5800X3D will age more gracefully.

 

[Makaveli]

I have a 3080 12GB card and usually spend my days playing 10 year old games like Kings Bounty and Civ 5 and 6.

I think I just got old and can't be bothered learning new games anymore.

One of these days I'll play something newer

Never too old to learn something new friend!

 

[eek2121]

Never too old to learn something new friend!

Hah, wait until he discovers something like Factorio. That is the game that dominates my time. I have a 3090 and Factorio isn't even 3D...

(I use the 3090 for other, more productive stuff, but still)

 

[DasFox]

I have a 3080 12GB card and usually spend my days playing 10 year old games like Kings Bounty and Civ 5 and 6.

I think I just got old and can't be bothered learning new games anymore.

One of these days I'll play something newer

Well you better starting player Newer Gen Games, or you wasted money on that 3080....

Like I always say; ‘The best hardware, only depends on your needs, nothing else’

 

[gdansk]

Well you better starting player Newer Gen Games, or you wasted money on that 3080....

Like I always say; ‘The best hardware, only depends on your needs, nothing else’

I can't speak for him but I'm in a similar situation. I like having a good graphics card in case I try new games. But I usually end up going back to the classics I know I'll enjoy.

Plus some emulators really work much better with overpowered hardware.

 

[DasFox]

Ahhh yes, now using Emulators, that’s a different story, but true, higher end to a point, does allow better gaming emu performance.

 

[Hans Gruber]

After further testing of my 5600 with the B2 stepping. I have concluded that there is no benefit over the B0 stepping. The 5600x still probably is on better silicon. I say this because they say the 5600x can do 4.8ghz on 1.35v and my newer process B2 5600 needs 1.375v. It runs fine @ 1.35v but during heavy BF5 gaming the computer restarts. At 1.375v the 5600 is completely and totally stable. I have never even got a blue screen with it.

As for where my 3600 ended up. Everything is recycled parts. My original B350 has the latest AGESA bios. It runs my Hynix CRJ 16GB kit 16-20-20 @ 3800mhz with 1.37v My ancient GTX 970 has brand new fans. My Cooler Master 240 Master Liquid that I picked up for $40 in 2018 is still runs like new. I can put any Zen 3 CPU in this machine and it has good VRM's.

Anyway, look at the stats for a B350 board with a Ryzen 3600 and running ram @ 3800mhz. I could only get the 98th percentile (no precision boost on B350)
MSI B350 GAMING PRO CARBON (MS-7B00) Performance Results - UserBenchmark

 

[Det0x]

Little comparison between 5800x3d @ 4550/4450mhz with ~4066MT/s memory and 5950x @ ~5200/5050mhz with ~4000MT/s memory

Spoiler: 5800x3d

5800x3d
SuperPi - 1M with BenchMate - 7sec 686ms - Ryzen 7 5800X3D @ 4567.2MHz
SuperPi - 32M with BenchMate - 6min 4sec 610ms - Ryzen 7 5800X3D @ 4567.2MHz

PYPrime - 2b with BenchMate - 8sec 397ms

y-cruncher - Pi-25m - 0sec 501ms - Ryzen 7 5800X3D @ 4461.1MHz
y-cruncher - Pi-1b - 32sec 75ms - Ryzen 7 5800X3D @ 4461.1MHz
y-cruncher - Pi-2.5b - 1min 32sec 459ms - Ryzen 7 5800X3D @ 4450MHz

wPrime - 32m with BenchMate - 2sec 340ms - Ryzen 7 5800X3D @ 4461.4MHz
wPrime - 1024m with BenchMate - 1min 4sec 98ms - Ryzen 7 5800X3D @ 4466.8MHz

7-Zip - 105612 MIPS - Ryzen 7 5800X3D @ 4467MHz

PiFast with BenchMate - 17sec 400ms - Ryzen 7 5800X3D @ 4567MHz

Spoiler: 5950x

5950x
SuperPi - 1M with BenchMate - 6sec 696ms - Ryzen 9 5950X @ 5213.3MHz
SuperPi - 32M with BenchMate - 5min 32sec 276ms - Ryzen 9 5950X @ 5213.4MHz

PYPrime - 2b with BenchMate - 9sec 779ms

y-cruncher - Pi-1b - 20sec 433ms - Ryzen 9 5950X @ 4775.2MHz
y-cruncher - Pi-2.5b - 58sec 362ms - Ryzen 9 5950X @ 4750.2MHz

wPrime - 32m with BenchMate - 1sec 486ms - Ryzen 9 5950X @ 5059.6MHz
wPrime - 1024m with BenchMate - 29sec 504ms - Ryzen 9 5950X @ 5034.5MHz

7-Zip - 205233 MIPS - Ryzen 9 5950X @ 4950.1MHz

PiFast with BenchMate - 15sec 170ms - Ryzen 9 5950X @ 5234.9MHz

No way around it, in these memory benchmarks cpu clockspeed is still king in everything except in PYprime
(dual CCD with twice the cores and much higher write bandwidth also helps )

Did also some Geekbenches but results seem to be all the place, not really sure why
Geekbench3 - Single Core - 7390 points, Multi Core - 86068 points (ST and especially MT seems to be too low)


Geekbench4 - Single Core - 8104 points, Multi Core - 62264 points (Normal ST, MT score way to low)


Geekbench5 - Single Core - 1859 points, Multi Core - 17831 points (good ST, MT score way to low)

Think i have to do some more runs to figure out what going on here.. Could be down to whea throttling above 1900:3800 in the MT tests.

 

[Dave3000]

Does the 64MB L3 cache on the 5950X get fully taken advantage of in gaming (games that use 8 cores or less) or the 32MB L3 cache on the 2nd CCD will be ignored until a game is using more than 8 cores since each CCD has 8 cores on the 5950X, therefore, the 2nd 32MB L3 cache of the 5950X will only be used in a game if the game is using more than 8 cores? I see that there is not much of a difference in gaming in the reviews between a 5800X and 5950X despite twice the L3 cache on the 5950X and it seems that it's only the little higher boost clocks that are making that little bit of difference in gaming. However, the 96MB L3 cache the 5800X3D is making a huge difference in some games as shown in the reviews but that's understandable since that 96MB L3 cache is for the whole CCD of the 5800X3D.

 

[moinmoin]

Does the 64MB L3 cache on the 5950X get fully taken advantage of in gaming (games that use 8 cores or less) or the 32MB L3 cache on the 2nd CCD will be ignored until a game is using more than 8 cores since each CCD has 8 cores on the 5950X, therefore, the 2nd 32MB L3 cache of the 5950X will only be used in a game if the game is using more than 8 cores?

Cores can only access the L3$ of the same CCX, which for Zen 3 / 5000 series equals CCD. Also the use of cores depends on the scheduler and doesn't work like linearly filling one CCX first and only using another CCX thereafter, Windows scheduler is actually randomizing it to spread heat between different cores. That's why not only does the second CCD in 5900X and 5950X most often not help with games but occasionally even have adverse effects due to threads communicating and migrating back and forth between different CCDs.

 

[Det0x]

Little comparison between 5800x3d @ 4550/4450mhz with ~4066MT/s memory and 5950x @ ~5200/5050mhz with ~4000MT/s memory

Spoiler: 5800x3d

5800x3d
SuperPi - 1M with BenchMate - 7sec 686ms - Ryzen 7 5800X3D @ 4567.2MHz
SuperPi - 32M with BenchMate - 6min 4sec 610ms - Ryzen 7 5800X3D @ 4567.2MHz

PYPrime - 2b with BenchMate - 8sec 397ms

y-cruncher - Pi-25m - 0sec 501ms - Ryzen 7 5800X3D @ 4461.1MHz
y-cruncher - Pi-1b - 32sec 75ms - Ryzen 7 5800X3D @ 4461.1MHz
y-cruncher - Pi-2.5b - 1min 32sec 459ms - Ryzen 7 5800X3D @ 4450MHz

wPrime - 32m with BenchMate - 2sec 340ms - Ryzen 7 5800X3D @ 4461.4MHz
wPrime - 1024m with BenchMate - 1min 4sec 98ms - Ryzen 7 5800X3D @ 4466.8MHz

7-Zip - 105612 MIPS - Ryzen 7 5800X3D @ 4467MHz

PiFast with BenchMate - 17sec 400ms - Ryzen 7 5800X3D @ 4567MHz

Spoiler: 5950x

5950x
SuperPi - 1M with BenchMate - 6sec 696ms - Ryzen 9 5950X @ 5213.3MHz
SuperPi - 32M with BenchMate - 5min 32sec 276ms - Ryzen 9 5950X @ 5213.4MHz

PYPrime - 2b with BenchMate - 9sec 779ms

y-cruncher - Pi-1b - 20sec 433ms - Ryzen 9 5950X @ 4775.2MHz
y-cruncher - Pi-2.5b - 58sec 362ms - Ryzen 9 5950X @ 4750.2MHz

wPrime - 32m with BenchMate - 1sec 486ms - Ryzen 9 5950X @ 5059.6MHz
wPrime - 1024m with BenchMate - 29sec 504ms - Ryzen 9 5950X @ 5034.5MHz

7-Zip - 205233 MIPS - Ryzen 9 5950X @ 4950.1MHz

PiFast with BenchMate - 15sec 170ms - Ryzen 9 5950X @ 5234.9MHz

No way around it, in these memory benchmarks cpu clockspeed is still king in everything except in PYprime
(dual CCD with twice the cores and much higher write bandwidth also helps )

Did also some Geekbenches but results seem to be all the place, not really sure why
Geekbench3 - Single Core - 7390 points, Multi Core - 86068 points (ST and especially MT seems to be too low)


Geekbench4 - Single Core - 8104 points, Multi Core - 62264 points (Normal ST, MT score way to low)


Geekbench5 - Single Core - 1859 points, Multi Core - 17831 points (good ST, MT score way to low)

Think i have to do some more runs to figure out what going on here.. Could be down to whea throttling above 1900:3800 in the MT tests.

As expected, were running too high infinity fabric clock, resulting in whea-errors and throttling in the MT workloads..
When running my normal everyday 1900:3800 memoryspeed the scores are back to normal

Geekbench 3: ST = 7651points and MT = 95277points


Geekbench 4: ST = 8062points and MT = 75117points


Geekbench 5: ST = 1859points and MT = 20477points

 

[Asterox]

Solid CPU upgrade, all that on five years old B350 motherboard.