4 Years of Ryzen 5, CPU & GPU Scaling Benchmark

NightAntilli

Posts: 657   +821
Interesting test and results, but I can't help but see it as incomplete. Firstly... Let's start with this;

"If you're wondering why we didn’t use more Radeon GPUs like the RX 6900 XT and 6800 in place of the RTX 3090 and RTX 3070, there are two main reasons: most of you are interested in buying a GeForce 30 series GPU and most of you have, at least according to a recent poll we ran."

Can't say I agree with this. If you want to cover as large a range of configurations as possible, you'd have to use one high end and one mid range card from each vendor, I.e. RTX 3090 + RTX 3070 and 6900 XT + 6800 or 6800XT + 5700XT. I would understand wanting to keep the 5700XT in there considering how popular that card is, and wanting a lower tier card in the list. Even using the 6800 instead of the 5600XT would have been a better choice imo. The choice to use 5700XT + 5600XT is a very odd one, considering those two cards are not that far from each other in terms of performance, and they are the same die, and the 5600XT isn't that popular.

Secondly... That it "shouldn't matter as they represent a similar performance tier" is also not true, because GPU drivers influence CPU usage quite a lot. So even if you have two high end cards in the same performance tier, the fact that they are from different vendors (which approach things very differently mind you) means that they should be tested individually. Just like you can't use an Intel CPU to represent AMD CPU performance or vice versa, you can't do it for graphics cards either.

Lastly... I am perfectly aware that this is about CPU scaling, and not comparing GPU drivers performance on the CPU. However, we currently have no scaling of the most popular last gen nVidia card, and no scaling for the latest high end AMD cards. That's two groups of people that have a lack of information regarding their Zen CPU performance.
Yes, I am perfectly aware that this takes a lot of work. But ideally, it would have been something like RTX 3090 + RTX 3070 + RTX 2060(S) and 6900XT + 6800 + 5700XT. This would cover the high end cards of each vendor, the more mid range offering from each vendor, and the most popular mid range card of the previous generation. It would have been a more complete picture.

 

ScottSoapbox

Posts: 88   +105
Interesting test and results, but I can't help but see it as incomplete. Firstly... Let's start with this;

"If you're wondering why we didn’t use more Radeon GPUs like the RX 6900 XT and 6800 in place of the RTX 3090 and RTX 3070, there are two main reasons: most of you are interested in buying a GeForce 30 series GPU and most of you have, at least according to a recent poll we ran."

Can't say I agree with this. If you want to cover as large a range of configurations as possible, you'd have to use one high end and one mid range card from each vendor, I.e. RTX 3090 + RTX 3070 and 6900 XT + 6800 or 6800XT + 5700XT. I would understand wanting to keep the 5700XT in there considering how popular that card is, and wanting a lower tier card in the list. Even using the 6800 instead of the 5600XT would have been a better choice imo. The choice to use 5700XT + 5600XT is a very odd one, considering those two cards are not that far from each other in terms of performance, and they are the same die, and the 5600XT isn't that popular.

Secondly... That it "shouldn't matter as they represent a similar performance tier" is also not true, because GPU drivers influence CPU usage quite a lot. So even if you have two high end cards in the same performance tier, the fact that they are from different vendors (which approach things very differently mind you) means that they should be tested individually. Just like you can't use an Intel CPU to represent AMD CPU performance or vice versa, you can't do it for graphics cards either.

Lastly... I am perfectly aware that this is about CPU scaling, and not comparing GPU drivers performance on the CPU. However, we currently have no scaling of the most popular last gen nVidia card, and no scaling for the latest high end AMD cards. That's two groups of people that have a lack of information regarding their Zen CPU performance.
Yes, I am perfectly aware that this takes a lot of work. But ideally, it would have been something like RTX 3090 + RTX 3070 + RTX 2060(S) and 6900XT + 6800 + 5700XT. This would cover the high end cards of each vendor, the more mid range offering from each vendor, and the most popular mid range card of the previous generation. It would have been a more complete picture.

I look forward to your review of all of those combinations.

Drop a link to it here when you're done.
 
"Stepping up to the RTX 3070, we see the 5600 XT outperform the 3600X by an 8% margin when comparing 1% low data"

5600XT? Think you meant to the 5600X? It's the products names getting too similar!
 

Nobina

Posts: 3,133   +3,009
Interesting test and results, but I can't help but see it as incomplete. Firstly... Let's start with this;

"If you're wondering why we didn’t use more Radeon GPUs like the RX 6900 XT and 6800 in place of the RTX 3090 and RTX 3070, there are two main reasons: most of you are interested in buying a GeForce 30 series GPU and most of you have, at least according to a recent poll we ran."

Can't say I agree with this. If you want to cover as large a range of configurations as possible, you'd have to use one high end and one mid range card from each vendor, I.e. RTX 3090 + RTX 3070 and 6900 XT + 6800 or 6800XT + 5700XT. I would understand wanting to keep the 5700XT in there considering how popular that card is, and wanting a lower tier card in the list. Even using the 6800 instead of the 5600XT would have been a better choice imo. The choice to use 5700XT + 5600XT is a very odd one, considering those two cards are not that far from each other in terms of performance, and they are the same die, and the 5600XT isn't that popular.

Secondly... That it "shouldn't matter as they represent a similar performance tier" is also not true, because GPU drivers influence CPU usage quite a lot. So even if you have two high end cards in the same performance tier, the fact that they are from different vendors (which approach things very differently mind you) means that they should be tested individually. Just like you can't use an Intel CPU to represent AMD CPU performance or vice versa, you can't do it for graphics cards either.

Lastly... I am perfectly aware that this is about CPU scaling, and not comparing GPU drivers performance on the CPU. However, we currently have no scaling of the most popular last gen nVidia card, and no scaling for the latest high end AMD cards. That's two groups of people that have a lack of information regarding their Zen CPU performance.
Yes, I am perfectly aware that this takes a lot of work. But ideally, it would have been something like RTX 3090 + RTX 3070 + RTX 2060(S) and 6900XT + 6800 + 5700XT. This would cover the high end cards of each vendor, the more mid range offering from each vendor, and the most popular mid range card of the previous generation. It would have been a more complete picture.
The amount of time you spent writing that comment you could've made your own forum and benchmarked every GPU out there.
 

NightAntilli

Posts: 657   +821
I look forward to your review of all of those combinations.

Drop a link to it here when you're done.
The amount of time you spent writing that comment you could've made your own forum and benchmarked every GPU out there.
Next time that you get any sort of faulty product, remember to not complain nor look for warranty, but instead to tell yourself to make a better one instead.
 

dirtyferret

Posts: 684   +862
"We should also quickly discuss core counts, since that came up a lot in our previous CPU/GPU scaling feature which focused on Zen 3 parts. We want to reiterate what we said in our Ryzen 5 5600X review, and that is that gamers shouldn’t focus on CPU core count and the only real consideration should be CPU performance and how much CPU performance you actually need.

For gaming all you need from the Zen 3 range is the 6-core/12-thread model right now, but that’s not to generalize by saying any 6-core/12-thread processor is all you need for gaming because as we’ve seen here, 6-core/12-thread performance can mean just about anything with the 5600X up to 100% faster than the 1600X in the most extreme scenarios, so just focus on CPU performance and get over this core count obsession for gaming."


Damn it Steven, stop posting such factual nonsense. Children need to believe in Santa Clause, your drunk friend at the bar who still lives with his parents needs to believe he can score with a super model, and fanboys need to believe more cores equal future proofing!
 

Tom Yum

Posts: 100   +231
"We should also quickly discuss core counts, since that came up a lot in our previous CPU/GPU scaling feature which focused on Zen 3 parts. We want to reiterate what we said in our Ryzen 5 5600X review, and that is that gamers shouldn’t focus on CPU core count and the only real consideration should be CPU performance and how much CPU performance you actually need.

For gaming all you need from the Zen 3 range is the 6-core/12-thread model right now, but that’s not to generalize by saying any 6-core/12-thread processor is all you need for gaming because as we’ve seen here, 6-core/12-thread performance can mean just about anything with the 5600X up to 100% faster than the 1600X in the most extreme scenarios, so just focus on CPU performance and get over this core count obsession for gaming."


Damn it Steven, stop posting such factual nonsense. Children need to believe in Santa Clause, your drunk friend at the bar who still lives with his parents needs to believe he can score with a super model, and fanboys need to believe more cores equal future proofing!

Steve's point is that what matters is the level of performance provided in total, not how many cores provided. That doesn't mean that more cores doesn't future proof performance, just that GPU's right now don't require more than a 5600X to remain GPU limited. Future GPU's may become limited by a 5600X, just like the 7600k can't push the latest GPU's to their max performance.

Having said that, if you upgrade your CPU alongside your GPU, you'll likely never experience CPU limited framerates if you stick to that mid-tier CPU (Ryzen 5/1x700 level), so buying a 5950X/10900K for gaming alone could be silly unless you plan not to update it for the next 6-8 years (but plan to still update your GPU during that time). Mid-tier CPU combined with the fastest GPU you can afford remains the best combo for gaming-centric performance.
 

dirtyferret

Posts: 684   +862
Steve's point is that what matters is the level of performance provided in total, not how many cores provided. That doesn't mean that more cores doesn't future proof performance, just that GPU's right now don't require more than a 5600X to remain GPU limited. Future GPU's may become limited by a 5600X, just like the 7600k can't push the latest GPU's to their max performance.
So what you are saying is a higher performance CPU would perform better than a lower performance CPU but that is different than Steve explaining a higher performance CPU performs better than a lower performance CPU. Perhaps if you spent more time learning what the term "duplicate" means you would not have to learn how to stop repeating what other people say.
 
Last edited:

hahahanoobs

Posts: 3,426   +1,570
Interesting test and results, but I can't help but see it as incomplete. Firstly... Let's start with this;

"If you're wondering why we didn’t use more Radeon GPUs like the RX 6900 XT and 6800 in place of the RTX 3090 and RTX 3070, there are two main reasons: most of you are interested in buying a GeForce 30 series GPU and most of you have, at least according to a recent poll we ran."

Can't say I agree with this. If you want to cover as large a range of configurations as possible, you'd have to use one high end and one mid range card from each vendor, I.e. RTX 3090 + RTX 3070 and 6900 XT + 6800 or 6800XT + 5700XT. I would understand wanting to keep the 5700XT in there considering how popular that card is, and wanting a lower tier card in the list. Even using the 6800 instead of the 5600XT would have been a better choice imo. The choice to use 5700XT + 5600XT is a very odd one, considering those two cards are not that far from each other in terms of performance, and they are the same die, and the 5600XT isn't that popular.

Secondly... That it "shouldn't matter as they represent a similar performance tier" is also not true, because GPU drivers influence CPU usage quite a lot. So even if you have two high end cards in the same performance tier, the fact that they are from different vendors (which approach things very differently mind you) means that they should be tested individually. Just like you can't use an Intel CPU to represent AMD CPU performance or vice versa, you can't do it for graphics cards either.

Lastly... I am perfectly aware that this is about CPU scaling, and not comparing GPU drivers performance on the CPU. However, we currently have no scaling of the most popular last gen nVidia card, and no scaling for the latest high end AMD cards. That's two groups of people that have a lack of information regarding their Zen CPU performance.
Yes, I am perfectly aware that this takes a lot of work. But ideally, it would have been something like RTX 3090 + RTX 3070 + RTX 2060(S) and 6900XT + 6800 + 5700XT. This would cover the high end cards of each vendor, the more mid range offering from each vendor, and the most popular mid range card of the previous generation. It would have been a more complete picture.
With all due respect to Techspot and the replies to this mans' comment, he was not disrespectful and is not speaking out of turn. What good is a comment section if we don't ever question or give opinions to things because someone might feel offended or unnapreciated? Last I checked the comments were for comments and feedback and not designed to be used by fans. Steve is a celebrity in the tech space, but he's also one of us.

I believe Steve when he stated his reasons for the cards he used for these GPU scaling tests, but as a reader without the resources Techspot has, I would also like to see the configurations dude suggested in a future article just so I can see it for myself. Obviously not every configuration can be tested, but we can further limit how much perf we have to calculate ourselves for untested parts and configurations.

This does not take anything away from the work Steve did and will continue to do. This is all I will say.
 
Last edited:

Tom Yum

Posts: 100   +231
So what you are saying is a higher performance CPU would perform better than a lower performance CPU but that is different than Steve explaining a higher performance CPU performs better than a lower performance CPU. Perhaps if you spent more time learning what the term "duplicate" means you would not have to learn how to stop repeating what other people say.
I'm largely repeating Steve's point because from your comment you seemed to take it to mean more cores doesn't provide future proofing, which isn't what Steve is saying.
 

dirtyferret

Posts: 684   +862
I'm largely repeating Steve's point because from your comment you seemed to take it to mean more cores doesn't provide future proofing, which isn't what Steve is saying.
here is literally what Steve said

" We want to reiterate what we said in our Ryzen 5 5600X review, and that is that gamers shouldn’t focus on CPU core count and the only real consideration should be CPU performance and how much CPU performance you actually need.... so just focus on CPU performance and get over this core count obsession for gaming. "

So it seems you repeatedly don't understand what either Steve or I said
 

Tom Yum

Posts: 100   +231
here is literally what Steve said

" We want to reiterate what we said in our Ryzen 5 5600X review, and that is that gamers shouldn’t focus on CPU core count and the only real consideration should be CPU performance and how much CPU performance you actually need.... so just focus on CPU performance and get over this core count obsession for gaming. "

So it seems you repeatedly don't understand what either Steve or I said
More cores provides more performance, more IPC provides more performance, performance = frames, we are in agreement. Cores can provide future proofing through higher performance, but for most people the GPU is the limit, hence don't go crazy with CPU power if your main focus is gaming. U
I completely agree with Steve, I don't agree with your statement that more cores don't provide a level of future proofing, 7600K owners can attest to that.
 
Good job, thank you, Steve.
Having all three Rs 1600, 2600 and 3600, with the best of GPUs being RX 5700 in desktops used at home, I was considering upgrades, obviously.
At those crazy GPUs' prices these days, gonna keep waiting for 2022 to bring back some "normality".
Great support from your review, just in time, thx again.