AMD Ryzen 7 2700X 3.7 GHz

[Vayra86]

Can I ask a dumb question here? What clock speed was the 8700K running at when many of these tests were being conducted?

Considering all-core load, 4.3 Ghz

 

[Charcharo]

Can I ask a dumb question here? What clock speed was the 8700K running at when many of these tests were being conducted?

4.3 Ghz boost for all cores. In some games, likely 4.7 Ghz boosted single core.

 

[cucker tarlson]

As to your comment on the 4c/4t, what CPU was it exactly?

As for PurePC, even though I can't read Polish for shit, their choices in benchmarks to use is ridiculously good, arguably the best on the net right now. Their CPU reviews tell me more than any others I can find.

I was running a 4790K and experiencing a CPU bottleneck in some games. They came up with an updated review for 5775c, this time they did it on more modern games on most importantly gtx 1080 (the previous ones were on gtx 980, which was the fastest on the market at the time broadwell-c made it, which may explain a large portion of why it recieved a lukewarm welcome). I was somewhat sceptical,but I got a good deal on that CPU, I installed it and I was like "hoooly crap" when I tested it. It lined up almost perfectly with what I saw in the review. Even made a little thread here on TPU,although in the thread I was rather feeling the limits of this cpu than making a "in most cases" scenario.
https://www.techpowerup.com/forums/...c-edrams-impact-on-gaming-performance.236514/
Their cpu testing methodology is top notch.

 

[Charcharo]

I was running a 4790K and experiencing a CPU bottleneck in some games. They came up with an updated review for 5775c, this time they did it on more modern games on most importantly gtx 1080 (the previous ones were on gtx 980, which was the fastest on the market at the time broadwell-c made it, which may explain a large portion of why it recieved a lukewarm welcome). I was somewhat sceptical,but I got a good deal on that CPU, I installed it and I was like "hoooly crap" when I tested it. It lined up almost perfectly with what I saw in the review. Even made a little thread here on TPU.
https://www.techpowerup.com/forums/...c-edrams-impact-on-gaming-performance.236514/
Their cpu testing methodology is top notch.

Well, seems like something I need to check on a game by game basis then.
Of course I can always go full PCMR and mod my areas to show what CPU benchmarking really means for all the filthy casuals.

 

[Vayra86]

Well, seems like something I need to check on a game by game basis then.
Of course I can always go full PCMR and mod my areas to show what CPU benchmarking really means for all the filthy casuals.

Look at Total War: Warhammer for a good example of CPU destruction. Its a modern day Crysis in that sense - and it even uses threads quite well...

I'm actually curious what it was running in FC5. Dunia engine is heavily single threaded.

FC5 will load multiple cores and will probably not keep the CPU at 4.7, but rather 4.3

I cba to remove my OC to check it for ya... Board makers apply multi core enhancements from the factory for good reason, because CFL will lose the 4.7 Ghz in most gaming scenarios.

 

[cucker tarlson]

4.3 Ghz boost for all cores. In some games, likely 4.7 Ghz boosted single core.

I'm actually curious what it was running in FC5. Dunia engine is heavily single threaded.

Look at Total War: Warhammer for a good example of CPU destruction. Its a modern day Crysis in that sense - and it even uses threads quite well...

I found that not all games that thread well actually give ryzen all that big a boost. I guess more threads is not always equivalent of faster IPC, even on games that put the CPU load evenly across the cores.
I think you should prioritize IPC and ram speed still, even in 2018, and then find a cpu that offers that with enough cores/threads that won't choke the performance. Actually video tests are good for that, you can see the CPU load across the cores change. If it stays under 75% it's absolutely fine. +80% is higher than my liking. At +90% you may experience fps dips and stutter, how bad it's gonna be depends on the game.

 

[Vayra86]

I'm actually curious what it was running in FC5. Dunia engine is heavily single threaded.


I found that not all games that thread well actually give ryzen all that big a boost. I guess more threads is not always equivalent of faster IPC, even on games that put the CPU load evenly across the cores.

Correct, my experience as well. The constant swapping of threads across cores doesn't help with monitoring either. But there is for sure a tremendous amount of overhead as core counts go up and the usage you see on HT/SMT threads is almost exclusively overhead. Even in DX12. Another thing that happens with virtually every game is that FPS is always limited by the biggest thread; so essentially no matter how well it uses threads, there is still a single threaded limitation.

 

[Charcharo]

Look at Total War: Warhammer for a good example of CPU destruction. Its a modern day Crysis in that sense - and it even uses threads quite well...



FC5 will load multiple cores and will probably not keep the CPU at 4.7, but rather 4.3

I cba to remove my OC to check it for ya... Board makers apply multi core enhancements from the factory for good reason, because CFL will lose the 4.7 Ghz in most gaming scenarios.

I do not have it. From the very CPU heavy and multi-threaded games I have Metro, Witcher 3, BF1, Wolfenstein 2, DOOM 2016, Crysis 3.

I told you my Metro bench. For Wolfenstein I use either the Panzerhund sequence in New Orleance or one of the late game arenas and wait a minute for most of the AI to spawn before going all in on them. That is what I used to test my CPUs, and STALKER Call of Chernobyl. Which uses only 2 threads and is CPU heavy to a wacky degree. Is superior to Far Cry 5 as a game though, by a gigantic degree.

 

[cucker tarlson]

. Another thing that happens with virtually every game is that FPS is always limited by the biggest thread; so essentially no matter how well it uses threads, there is essentially still a single threaded limitation.

Good observation, I was thinking the same thing but wasn't sure.

 

[Vayra86]

I do not have it. From the very CPU heavy and multi-threaded games I have Metro, Witcher 3, BF1, Wolfenstein 2, DOOM 2016, Crysis 3.

I told you my Metro bench. For Wolfenstein I use either the Panzerhund sequence in New Orleance or one of the late game arenas and wait a minute for most of the AI to spawn before going all in on them. That is what I used to test my CPUs, and STALKER Call of Chernobyl. Which uses only 2 threads and is CPU heavy to a wacky degree. Is superior to Far Cry 5 as a game though, by a gigantic degree.

You just play the wrong type of games to see it

What you need is an isometric ARPG, an RTS, stuff like that. Starcraft 2 - it hinges almost entirely on single thread performance and was the bane of the FX processors back in the day. Games like Torchlight, Grim Dawn, and almost every single indie game out there all rely heavily on single thread performance. These are games that cannot push lots of work towards the GPU because the entire battlefield needs to be responsive at all times, there is no option to 'stream' game content based on the player's position in the world, you can't really use LoD and dynamic ways to spawn assets, etc..

There are also games that eat every thread you throw at it and also want the highest clock, and that is the MMO. Guild Wars 2, WoW are good examples (though WoW is so well optimized you won't notice in a lot of cases and it just runs into network limitations before it hits a CPU wall - I had 50-60 FPS in player hubs even on my 3570k which is a tremendous feat of coding).

 

[GoldenX]

You know, we didn't have such discusions with FX.
Next thing is Threadripper, right?

 

[Charcharo]

You just play the wrong type of games to see it

What you need is an isometric ARPG, an RTS, stuff like that. Starcraft 2 - it hinges almost entirely on single thread performance and was the bane of the FX processors back in the day. Games like Torchlight, Grim Dawn, and almost every single indie game out there all rely heavily on single thread performance. These are games that cannot push lots of work towards the GPU because the entire battlefield needs to be responsive at all times, there is no option to 'stream' game content based on the player's position in the world, you can't really use LoD and dynamic ways to spawn assets, etc..

There are also games that eat every thread you throw at it and also want the highest clock, and that is the MMO. Guild Wars 2, WoW are good examples (though WoW is so well optimized you won't notice in a lot of cases and it just runs into network limitations before it hits a CPU wall - I had 50-60 FPS in player hubs even on my 3570k which is a tremendous feat of coding).

Oh... trust me, when STALKER's modified A-Life for 1000 NPCs is turned on (simulates all of them real time) and the bullets with my addons with fixed ballistics start flying, the CPU screams. Since grenade shrapnel is also a physical objective... you get the idea. Still one of the most WTF tier CPU benches. I play Starcraft 2 and ... even it can not compete. But good choice still.

IMHO every game should be as multi-threaded as humanly possible since otherwise it is leaving performance on the table. Lets be real, both Intel and AMD are climbing a mountain now with IPC and clock speeds. It will still get higher, but not much higher less we switch to a new material for our CPUs. So until/when that happens, game developers have no real choice but to git gut.

 

[Vayra86]

You know, we didn't have such discusions with FX.
Next thing is Threadripper, right?

You must have missed them, then

If anything this shows how incredible the jump was from FX > Ryzen

 

[GoldenX]

Man, the Phenom II has aged.

 

[trparky]

Starcraft 2 - it hinges almost entirely on single thread performance and was the bane of the FX processors back in the day.

That's one of the games that really has me questioning what processor to buy. I currently have a 3570K@4.4GHz and I keep coming back to the question of whether Ryzen is good enough to be able to run that game decently when compared to my current processor that struggles late into the game.

 

[Charcharo]

Man, the Phenom II has aged.

We all age. The abyss awaits. But muh ALife and ballistics are eternal.

That's one of the games that really has me questioning what processor to buy. I currently have a 3570K@4.4GHz and I keep coming back to the question of whether Ryzen is good enough to be able to run that game decently when compared to my current processor that struggles late into the game.

Depends on whether you have an AMD or Nvidia GPU with Starcraft 2.

 

[Vayra86]

That's one of the games that really has me questioning what processor to buy. I currently have a 3570K@4.4GHz and I keep coming back to the question of whether Ryzen is good enough to be able to run that game decently when compared to my current processor that struggles late into the game.

Decide for yourself - but Ryzen is in a very good place all around, its not remotely comparable to FX. Note that this is without an OC - so this is a pretty good scenario to gauge relative Ryzen / Intel worst case gaming performance.
https://www.techporn.ph/amd-ryzen-7-1800x-am4-cpu-review/

 

[trparky]

Well alrighty then... since I lock my FPS to 60Hz (because my monitor is 60Hz) to reduce screen tearing it looks to me like it doesn't even matter which CPU I get.

Edit
Yeah I know... I play games like a pleb here at 1080p@60Hz. With GPU prices being in the stratosphere and with good 1440p monitors still being quite expensive when compared to 1080p monitors that can be had for under $100 USD I just can't justify the added cost of going 1440p.

Edit #2
I can't be the only gamer who thinks this way.

 

[Charcharo]

Pretty much.

Honestly memes aside, I am pretty happy with how Zen has turned out. It has some funky weird faults in some titles, and does not play well with Nvidia at times (or maybe it is the other way round? ) but it is fast in everything. Not always the fastest, but generally pretty good engineering. Good at everything, real good at some things, not quite the best, small CPU cores, upgrade path, efficient.

I do hope for some real upgrades with Zen 2 and Ryzen 3000, but CPUs are exciting again. Things are actually interesting and people can pick any CPU (bar the pentium and arguably lowest end R3 or i3) and expect great performance in 99% of titles, and passable in the remaining 1% at the least. Is this not what we all want in the end ?

 

[trparky]

I do hope for some real upgrades with Zen 2 and Ryzen 3000

I have a feeling that Zen 2 is going to be where Ryzen will take off like a rocket. Not only is Zen 2 (Ryzen 3000 Series) going to be based upon GloFlo 7nm process node but AMD has also partnered up with IBM and they have some seriously smart people at IBM. If there's anybody in the business that will know how to take Zen to the next level, it will be the people at IBM.

What's really funny about Intel is what I was reading over at AnandTech's forums regarding Intel's struggles to get to 10nm. It seems that Intel is completely drunk on their own Kool-Aid and refuses to reach out to outside experts despite the fact that the likes of TSMC and Samsung have both successfully made the leap to 10nm (and beyond) while Intel is still struggling at 14nm.

David Schor: Intel 10nm in big problems

This is why I have a very good feeling that AMD is going to thoroughly kick Intel's ass with Zen 2.

 

[Charcharo]

I have a feeling that Zen 2 is going to be where Ryzen will take off like a rocket. Not only is Zen 2 (Ryzen 3000 Series) going to be based upon GloFlo 7nm process node but AMD has also partnered up with IBM and they have some seriously smart people at IBM. If there's anybody in the business that will know how to take Zen to the next level, it will be the people at IBM.

What's really funny about Intel is what I was reading over at AnandTech's forums regarding Intel's struggles to get to 10nm. It seems that Intel is completely drunk on their own Kool-Aid and refuses to reach out to outside experts despite the fact that the likes of TSMC and Samsung have both successfully made the leap to 10nm (and beyond) while Intel is still struggling at 14nm.

David Schor: Intel 10nm in big problems

This is why I have a very good feeling that AMD is going to thoroughly kick Intel's ass with Zen 2.

To be fair, the nm does not refer to the node's size nowadays. It is just a marketing term. Zen's 14nm is actually more like 20nm with Finfets. Intel's 14nm is more like 16nm. TSMC's 16nm is like 20nm with Finfets...

 

[kanecvr]

Than 7600K ? What ?
Here Ryzen 2700X is 1350, 8700K is 1400, and 8700K is just better overall.

8700k = 1633 lei // 2700x = 1364 lei. Z370 board (obligatory for overclocking) ~ 550 lei while a decent x450 board is about 480 lei (not to mention one could go for a B450 witch is cheaper and still allows for overclocking, or the older X370/B350 chipsets). All in all, the price difference is about the value of a reasonably fast 8GB kit of dual channel DDR4. That coupled with it's workstation level performance makes it a best buy for me, since I do both games and work on the same computer.

I don't understand why regular consumers would buy the intel chip. As an enthusiast, it makes sense for me to go for the 8700k as most samples will reach 5GHz, and at that speed the performance gap in games wides, while the x2700's lead in productivity apps shrinks - and here comes another BUT - the 8700k gets STUPID HOT even at 4.7GHz... at 5GHz you need to de-lid it, replace the TIM with liquid metal, spend 100-200$ on a decent water cooling solution, and use a mother of a PSU and an expensive motherboard... so... not really my can of worms...

All and all, I'll stick with my 3930k for a while longer. At least until DDR4 prices settle.

 

[trparky]

I don't understand why regular consumers would buy the intel chip. As an enthusiast, it makes sense for me to go for the 8700k as most samples will reach 5GHz, and at that speed the performance gap in games wides, while the x2700's lead in productivity apps shrinks - and here comes another BUT - the 8700k gets STUPID HOT even at 4.7GHz... at 5GHz you need to de-lid it, replace the TIM with liquid metal, spend 100-200$ on a decent water cooling solution, and use a mother of a PSU and an expensive motherboard... so... not really my can of worms...

There's no denying that Intel overclocks like nobody's business but at what cost? This is the part that I've never understood about some people in the Intel camp. At some point the overclocking is going to have diminishing returns.

 

[Melvis]

You want 1080p or higher ? You can't handle the truth.

Well clearly I do as I live in 2018 where 1080p is the most used and most common res for gaming, this isnt 2003 anymore....

My 2 cents, based on my thousands of hours spent gaming on 3570K + 980Ti and 4790K + 1080. Even at 1440p you'll find plenty of places across various games that are CPU heavy. Plenty. It is impossible for the reviewer to recreate those, taking into account they bench 10+ games, they'd spend hundreds of hours just looking for good testing places. I know one guy from a Polish website who is responsible for gpu and cpu tests exclusively, and he always goes through the routine of finding them - look how different his results are from W1zzard's or the majority of other reviews for that matter.

https://www.purepc.pl/procesory/test_procesorow_amd_ryzen_7_2700x_vs_intel_core_i7_8700k?page=0,34
https://www.purepc.pl/procesory/test_procesorow_amd_ryzen_7_2700x_vs_intel_core_i7_8700k?page=0,35
https://www.purepc.pl/procesory/test_procesorow_amd_ryzen_7_2700x_vs_intel_core_i7_8700k?page=0,36
https://www.purepc.pl/procesory/test_procesorow_amd_ryzen_7_2700x_vs_intel_core_i7_8700k?page=0,37
https://www.purepc.pl/procesory/test_procesorow_amd_ryzen_7_2700x_vs_intel_core_i7_8700k?page=0,38
https://www.purepc.pl/procesory/test_procesorow_amd_ryzen_7_2700x_vs_intel_core_i7_8700k?page=0,39
https://www.purepc.pl/procesory/test_procesorow_amd_ryzen_7_2700x_vs_intel_core_i7_8700k?page=0,40
https://www.purepc.pl/procesory/test_procesorow_amd_ryzen_7_2700x_vs_intel_core_i7_8700k?page=0,41
https://www.purepc.pl/procesory/test_procesorow_amd_ryzen_7_2700x_vs_intel_core_i7_8700k?page=0,42
https://www.purepc.pl/procesory/test_procesorow_amd_ryzen_7_2700x_vs_intel_core_i7_8700k?page=0,43
https://www.purepc.pl/procesory/test_procesorow_amd_ryzen_7_2700x_vs_intel_core_i7_8700k?page=0,44

Forget about 3-10% percent that most sites usually show, this tells a completely different story. Now you may rightfully wonder how those results can be reproduced without all this time spent pciking the right locations. 720p testing. It won't show you what fps you'll be running at, let's say 1440p, when you find a cpu heavy scene, but they will show how CPUs would cope in such scenario.

So far you have showed me that you actually agree with my statement? with all the links that show exactly what I am talking about, that yes even at 1080 the res most used shows that there are actually differences in performance because of the stress on the CPU, its good to see that you got what I meant and agree

Going all caps doesn't make gaming benchmarks about the game itself in a CPU review, and it still doesn't change a thing about the relative performance gaps you see there. Its fine if you don't (want to) get it, so skip the 720p page and move right on to your preferred resolution to see what's the CPU for your use case. No one is stopping you...

Seems like I have to to get the point across for those who clearly dont get it and still live back in 2003. You just said it yourself, its a CPU review....but we are going to show you gaming benchmarks on a res from 2003? cant you see how irrelevant that is? it still is gaming benchmarks........at a res no one uses! with said hardware, so please stop spreading fud and get back to the yr 2018 for god sake.

Again im still waiting for that person that games at 720p with a i7/Ryzen 7 and a GTX 1080 Ti.

 

[cucker tarlson]

Well clearly I do as I live in 2018 where 1080p is the most used and most common res for gaming, this isnt 2003 anymore....



So far you have showed me that you actually agree with my statement? with all the links that show exactly what I am talking about, that yes even at 1080 the res most used shows that there are actually differences in performance because of the stress on the CPU, its good to see that you got what I meant and agree




Seems like I have to to get the point across for those who clearly dont get it and still live back in 2003. You just said it yourself, its a CPU review....but we are going to show you gaming benchmarks on a res from 2003? cant you see how irrelevant that is? it still is gaming benchmarks........at a res no one uses! with said hardware, so please stop spreading fud and get back to the yr 2018 for god sake.

Again im still waiting for that person that games at 720p with a i7/Ryzen 7 and a GTX 1080 Ti.

I don't think 720p represents what you will see at 1080p, except in CPU limited areas of games, and that is the whole point. It shows which CPU is best so I don't understand people whining about it, it helps you get the most bang for your buck. You know you don't have to pay AMD for their product if you don't like it. I mean "you know" more figuratively, not specifically about you, cause you seem to totally discredit that 720p results with an argument that you repeat like a broken record. We know no one plays at 720p anymore. We figured it out. But no one looks at 720p tests cause they're running a 720 rig, and that is what you don't get. We look at it and see CPU limited scenario.

BTW how stupid are AMD for not letting me use an aftermarket cooler on my CPU and voiding my warranty ? I mean I know they can't tell but what they're claming is beyond retarded.

https://support.amd.com/en-us/search/faq/147