AMD Ryzen 7 2700X Rears Its Head On Geekbench

[TheinsanegamerN]

So, IPC is barely higher than of the seven years old Core i5 2500? I still don't understand why people are so excited about Ryzen. AMD kinda ended Intel's monopoly but not so much.

Wake me up when Zen can achive at least 5500 score in single threaded mode in GeekBench 4, so that I'd be finally tempted to upgrade my aging PC.

Because AMD went from "laughably behind" to "quite close" to intel's performance, and a healthy AMD is good for competition.

Look at the pathetic IPC increase from sandy bridge to skylake. Barely any performance increase 90% of the time. Without AMD, intel slumbers and does jack.

Alternatively, even if ryzen was only about on par with sandy (it isnt, its on par with haswell) there has been so little activity by intel that this new generation could surge ahead of intel given another year or two. That is exciting. Oh, and double the cores for a lower price helps. And now I can build a PC without giving intel my money.

 

[RejZoR]

Zen2 wont be out until 2019+... Zen+ is coming out soon (weeks).

"Ryzen 2" is the product of the idiotic media (no amd slide/anything said ryzen 2)

I meant Zen+.

 

[Vego]

does it outperforms 7820X?

 

[Fx]

I'm just waiting for the reviews and money shots... Geekbench and such is probably no more helpful than XTU for benching... Not something proven I'd say...

Agreed. I couldn't care less about any artificial bench for anything. I never use them. Give me some numbers of performance for software used in production or gaming (something useful) for real perspective.

 

[Slizzo]

You all seem to be forgetting he 8700k has better min frame rates at every resolution across the board than ryzen by about 5-10 fps depending on the game, and thats stock clocks... sorry but min fps is king, I like how you all only focus on max fps though Kappa

Actually, 99th percentile frames are what's important. If the framerate isn't consistent, then you have stuttering and that by far is the most impactful thing to gaming experience.

 

[ratirt]

No, you can't... that is what I am trying to say... it DOES NOT SCALE. Also, your GPU should be running 100% in most titles regardless. Its only the jenky 720p low, 1080ti at 1080p playing CS:GO where it doesn't get a full load.

What is the point of knowing CPU A and CPU B difference is 20% at 720p with low settings, but a couple % at 1080p with 'normal' settings?? What use is that 720p dataset when its settings are fabricated to show an exaggerated difference?

Anyway, this isn't about 2700X... so I digress. You are welcome to reach out through PM though.

If we follow your premise if you need to keep the GPU at 100% to sustain the max FPS at a given resolution. That's fair. But on the other hand if we say that the Ryzen 1080p res is not that great cause the FPS drops down maybe it's because the utilization of the CPU? lower FPS but the CPU is not being utilized correctly. It still has a lot of headroom.
CPU 50% GPU 88%. Does this mean the Ryzen's slow or the game is not using the resources? Similar situation happens in the 720p. The GPU so as the CPU are not being utilized fully. If we conclude that the GPU must stay at a 100% for the best FPS the CPU must use as many resources to feed the GPU which isn't happening in some games.

I don't think that the 720p res is a good indictor. Especially if we see games that gain no FPS difference or marginal difference in FPS moving from 720p to 1080p despite of the CPU brand

 

[Gerolf]

CPU 50% GPU 88%. Does this mean the Ryzen's slow or the game is not using the resources?

If you check task manager during such situation (CPU or GPU are not fully loaded) most likely the reason is bottleneck in singlecore performance. 99% of games can't load all cores equally, so 1 core will be loaded at it's max and other cores will be loaded much less. That's why all games benefit from singlecore performance but only some games from additional threads/cores, and only few for more than 6 cores.

 

[ratirt]

If you check task manager during such situation (CPU or GPU are not fully loaded) most likely the reason is bottleneck in singlecore performance. 99% of games can't load all cores equally, so 1 core will be loaded at it's max and other cores will be loaded much less. That's why all games benefit from singlecore performance but only some games from additional threads/cores, and only few for more than 6 cores.

I know what it means but thanks for sharing.
You call it single core performance issue and I call it games lack the ability to distribute task with given resources. I look at this issue from a different angle but you can blame whatever or whoever you want.

 

[Gerolf]

You call it single core performance issue and I call it games lack the ability to distribute task with given resources.

Well, I will prefer CPU that can run all games, no matter how badly they are optimized. CPUs with more than 6 cores are installed in about 4% of computers, so from game publisher's point of view it's really not worse the effort to optimize for large number of threads.

Edit:

Typical game have large parts of code that can't be calculated simultaneously, for example because further calculations depends on the result of previous.
Amdahl's law explains that

 

[Vayra86]

I know what it means but thanks for sharing.
You call it single core performance issue and I call it games lack the ability to distribute task with given resources. I look at this issue from a different angle but you can blame whatever or whoever you want.

That statement is as old as the first dual core and even today, we still see that almost every game can not be fully threaded at will, they still all have a main thread that is the 'weakest link' in terms of performance and this is where RAM bandwidth/latency, clocks and IPC shine. This is also why Ryzen is almost consistently behind Intel with a very small set of exceptions, notably all of them games with non-standard threading. Manycore CPUs do offer a performance boost because the other threads no longer saturate the core that also runs the main thread. And SMT/HT becomes more of a benefit these days in gaming because of, sadly, the use and coding of DRM, but also due to the use of (web) overlays in games and other plugins that do not directly use game code.

As for the 720p testing discussion... I'm not getting into that one again, it means immovable objects meeting unstoppable forces @EarthDog but I will say that I still firmly believe it to be an informative and useful test to gauge relative CPU performance.

Regardless, the idea of games ever becoming completely multithreaded is probably never going to become a reality because gaming is 'sequential' of nature. If this, then that, and 'this' is entirely up to player input thus hard to 'predict', and predicting all possible outcomes is often much more costly than just 'doing it'.

 

[trparky]

Not counting the CPU cooler, motherboard or RAM. But yeah I just picked a Ryzen 1700 up brand new for $150 off eBay during the 20% off sale. At the end of the day you paid more than double for 5% more performance, grats. I just love people who tout their "on up ism" attitues all the while support a shitty company. I guess the company reflects the buyer.

Actually it's more like about $150 more for an Intel-based system versus an AMD-based system in my neck of the woods.

Taking into account the same case, power supply, RAM, video card, CPU cooler, and SSD the AMD system with a R5 1600 is priced at $1145.92 before tax and an equivalent Intel-based system with an 8700K is priced at $1295.92 before tax. The difference is just $150. So the question is, when I take this into account what kind of system do I want to buy? Intel or AMD?

 

[ratirt]

That statement is as old as the first dual core and even today, we still see that almost every game can not be fully threaded at will, they still all have a main thread that is the 'weakest link' in terms of performance and this is where RAM bandwidth/latency, clocks and IPC shine. This is also why Ryzen is almost consistently behind Intel with a very small set of exceptions, notably all of them games with non-standard threading. Manycore CPUs do offer a performance boost because the other threads no longer saturate the core that also runs the main thread. And SMT/HT becomes more of a benefit these days in gaming because of, sadly, the use and coding of DRM, but also due to the use of (web) overlays in games and other plugins that do not directly use game code.

As for the 720p testing discussion... I'm not getting into that one again, it means immovable objects meeting unstoppable forces @EarthDog but I will say that I still firmly believe it to be an informative and useful test to gauge relative CPU performance.

Regardless, the idea of games ever becoming completely multithreaded is probably never going to become a reality because gaming is 'sequential' of nature. If this, then that, and 'this' is entirely up to player input thus hard to 'predict', and predicting all possible outcomes is often much more costly than just 'doing it'.

I don't think the Ryzen IPC is worst than Intel's. The latency between CCX's is what brings Ryzens behind Intel's CPUs. The differences between intel an ryzen cpus in this matter are pretty big. Ryzen's IPC is very good i think.

 

[trparky]

The Infinity Fabric is what kills performance, if they can get the latency down when going across it it would help in boosting performance.

Come the end of April I'll have the money saved up to build the system that I want to build. What processor that system will have is still very much up in the air. Do I go with the tried and true Intel solution (8700K) or go with AMD? That I have no idea. I'm looking forward to the Ryzen 2000-Series (not Zen 2) benchmarks, maybe that will make up my mind.

 

[ratirt]

The Infinity Fabric is what kills performance, if they can get the latency down when going across it it would help in boosting performance.

Come the end of April I'll have the money saved up to build the system that I want to build. What processor that system will have is still very much up in the air. Do I go with the tried and true Intel solution (8700K) or go with AMD? That I have no idea. I'm looking forward to the Ryzen 2000-Series (not Zen 2) benchmarks, maybe that will make up my mind.

Got similar idea here. Although i want Ryzen if i'm willing to change anything if not i'll just stay with my 3770k for some more time. Still waiting for benchmarks and then we will see.

 

[evernessince]

Actually it's more like about $150 more for an Intel-based system versus an AMD-based system in my neck of the woods.

Taking into account the same case, power supply, RAM, video card, CPU cooler, and SSD the AMD system with a R5 1600 is priced at $1145.92 before tax and an equivalent Intel-based system with an 8700K is priced at $1295.92 before tax. The difference is just $150. So the question is, when I take this into account what kind of system do I want to buy? Intel or AMD?

It's really up to what you need. If you need high FPS gaming, Intel will be better. Otherwise I don't see a scenario where you would pick the Intel system if the AMD system is that much cheaper.

 

[HTC]

2 other GB stock results for 2700X: one with RAM @ 2133 and the other with RAM @ 3200:

https://browser.geekbench.com/v4/cpu/compare/7494872?baseline=7504447

 

[phill]

Shame it wasn't the same setups used bar the ram speed but still, nice increase for Geekbench.. I hope it's across the board that the performance goes!! I'm looking forward to the reviews of the CPU very much...

 

[HTC]

More performance news (from videocardz):







NOTICE the board used: could it have a negative impact on the performance?

 

[EarthDog]

A motherboard typically has VERY little impact on performance (stock).

Wondering why those benchmarks are on paper in a book?

 

[Joss]

Wondering why those benchmarks are on paper in a book?

It's the current paper edition of Canard PC Hardware.

 

[ikeke]

2 other GB stock results for 2700X: one with RAM @ 2133 and the other with RAM @ 3200:

https://browser.geekbench.com/v4/cpu/compare/7494872?baseline=7504447

https://browser.geekbench.com/v4/cpu/compare/7494872?baseline=7643957

Compared to my personal R7 1700@stock with 2933cl14 memory.
https://valid.x86.fr/8f6fru