Discussion Is Intel actually better than AMD for gaming?

[BrandonFitzpatricc]

I’ve seen benchmarks of Intel’s 9th gen CPU’s vs AMD’s Ryzen 3000 CPU’s, and Intel’s CPU’s usually get around 10 more FPS, but I’m not sure if AMD’s CPU’s are fully optimized for these benchmarks. They normally perform better with faster RAM (DDR4-3600 or above is usually best), and precision boost overdrive generally provides better results than manual overclocking. Intel doesn’t need these optimizations as far as I know, you mostly just need to overclock it. I’m technically making an assumption by saying that the Ryzen 3000 CPU’s aren’t fully optimized during CPU benchmarks, but I feel like the gap between the 3700x and the 9700k (for example) aren’t as large as benchmarks show, even if Intel still has a slight upper hand

 

[tennis2]

Ryzen 3000 is pretty much matching Intel IPC now, but Intel still has the frequency advantage.

 

[KhouhenStercie.Bacus]

Both of the CPU's are good and evenly match don't compare the CPU's only on one thing, both of the CPU has it's own pros and cons.

Intel are leaning towards Frequency they are starting to move on multiple cores. after Ryzen shows prowess on Multiple cores.

Now there are a lot of things you have to consider, since you mentioned gaming Intel has the upper hand since the Games are more reliant on good Frequency boost most of the games are not yet utilizing the multiple cores. but since Ryzen's release 1st gen are not yet well optimized since it was their first release since intel was doing good at the time before however 2nd to 3rd gen they are getting good updates and well optimized on some games as well getting good performance so you have to take time for the product to be introduced and well optimized by the developers.

now if you compare Intel and AMD for me since I'm a streamer do games and video editing multi-tasking will take be better on Ryzen CPU's since it has more cores and threads to process multi-tasking loads as you can see on the reviewers

however if you game only without these and frequently edit videos what so ever Intel will do its job delivering you good performance on Games. not that I hate INTEL or AMD because my laptop is INTEL and Desktop is a Ryzen 7 1700

Good thing with the Intel is that they are more efficient like Nvidia delivering less power with good performance but the problem is the Price that's the other thing

AMD or Radeon they are good they are more on Raw Power with power consumption which is high bills to pay now good thing is that the AMD is shifting to efficiency with good performance and a lot of people getting AMD because of their price to performance which is really good. than intel's offering.

The reason why the Intel can price that much its because shortage of the CPU's and manufacturing process and there are no competition from AMD before that's why the Intel keeps refreshing the CPU until AMD was revived through Ryzen and these are business of course why spend more money where you can get good performance with better deal.

 

[Kastytis]

I feel like the gap between the 3700x and the 9700k (for example) aren’t as large as benchmarks show, even if Intel still has a slight upper hand

All the benchmarks are done with a 2080 ti and 9700k does come on top in most games but the moment you start putting lower end GPUs the difference will get smaller and smaller and by 2060/2070 (I'm guessing here) they will perform the same (in benchmarks). The thing is that benchmarks usually don't show how smooth the performance is. While the 9700k will have better FPS on average since it is not hyper-threaded it might experience stuttering in some games while the 3700x might not if it's extra threads are being utilized

 

[sizzling]

To exploit the extra performance a 9700k/9900k can provide you have to use a 2080Ti and run 1440p or lower. For most setups you would never tell the difference.

 

[TerryLaze]

While the 9700k will have better FPS on average since it is not hyper-threaded

Which shows that benchmarks are not optimized for intel as well and not only for ryzen.

it might experience stuttering in some games while the 3700x might not if it's extra threads are being utilized

This might as well be because the ryzen will clock up to 4.2 and stay there for the whole duration of the benchmark to keep up while the intel one might swing from 3.6 to 5Ghz depending on the scene.

To exploit the extra performance a 9700k/9900k can provide you have to use a 2080Ti and run 1440p or lower. For most setups you would never tell the difference.

Yes are you going to spend ~ $500 on a CPU for current GPUs only?A normal person might hope to keep a $500 CPU for 5-6 years.

 

[geofelt]

Yes and no.

Yes,
Intel is better for gaming. but not by enough to make a difference you will notice.
Perhaps more so on cpu centric games like sims, mmo and strategy games.

No.
Really, any intel 9th gen or ryzen 3000 processor over $200 or so will be a decent performer.
How good will be determined more by your graphics card than the processor.

A balanced gamer will budget about 2x the cost of the processor for the graphics card.

 

[digitalgriffin]

I’ve seen benchmarks of Intel’s 9th gen CPU’s vs AMD’s Ryzen 3000 CPU’s, and Intel’s CPU’s usually get around 10 more FPS, but I’m not sure if AMD’s CPU’s are fully optimized for these benchmarks. They normally perform better with faster RAM (DDR4-3600 or above is usually best), and precision boost overdrive generally provides better results than manual overclocking. Intel doesn’t need these optimizations as far as I know, you mostly just need to overclock it. I’m technically making an assumption by saying that the Ryzen 3000 CPU’s aren’t fully optimized during CPU benchmarks, but I feel like the gap between the 3700x and the 9700k (for example) aren’t as large as benchmarks show, even if Intel still has a slight upper hand

That all depends on what your budget it. If you compare a $330 Ryzen 3700X to a baseline i7-8700 ($320 ) The AMD would walk all over it in every category.

If you compared a Ryzen 3800X ($399) to an Intel 9700K ($364 + $60/cooler) It would be a more close match if you ran it at STOCK. If you overclock the Intel, the intel would win. But then you have consider you would need a more robust power supply and better cooling as the intel runs hotter. (It's listed as a 95W part, but those numbers are shown to be wildly off when you push it.)

So in the end, if you overclock them, then yes Intel is the fastest. But you pay for that privledge of hitting 5GHz+ all core. And when you use processors like these, you are USUALLY aiming for 4K gaming. At these resolutions, the GPU is a more important consideration as games become GPU limited (NVIDIA 2080/2080ti) A Fast CPU will help, but not as much as a fast GPU.

Now if you game at 1080, then the CPU has to keep up with setting up frames for the GPU. In this case you become more CPU bound. But at these resolutions, unless you want stupid fast frame rates, it really doesn't matter.

It's all about balance.

 

[drea.drechsler]

Which shows that benchmarks are not optimized for intel as well and not only for ryzen.
..

To be clear... I think gaming 'benchmarks' aren't really 'benchmarks' that can be optimized. They are actual gaming performance using actual scenes, characters and action scripted from within the game. So if the 'benchmark' is optimized for one CPU or GPU then the entire game is.

That's what makes GTA V an Intel "benchmark" favorite as it heavily favors Intel. (But also considered by some as irrelevant today, more an example of sloppy, poorly optimized game coding. LOL.)

Why that comes about is pretty interesting. In some recent Intel charts that map out Intel's competitive advantages (in a Ryzen 3000 world, with comparable IPC, more cores at cheaper cost) they refer to their Software group as one of their 'six pillars'. That suggests (strongly) to me they intend to keep on helping developers with highly optimized code and cheap (free?) compilers for game development. To a small developer, getting cheap help like that must be a boon and getting locked into one installed base of gaming systems isn't as bad as it may seem if it's the largest.

 

[NightHawkRMX]

In gaming, when gpu bound, ryzen is on par with intel. Your average gamer will have a gpu bound system and the performance will identical with intel or amd, but you would save money with amd. Also amd with its core advantage would have a more stable framerate.

If you have a lot of money to spend, are using a 2080ti, and play at 1080p 240hz, intel will indeed be noticably better performing. Unfortunately, nobody actually plays like this.

 

[digitalgriffin]

To be clear... I think gaming 'benchmarks' aren't really 'benchmarks' that can be optimized. They are actual gaming performance using actual scenes, characters and action scripted from within the game. So if the 'benchmark' is optimized for one CPU or GPU then the entire game is.

That's what makes GTA V an Intel "benchmark" favorite as it heavily favors Intel. (But also considered by some as irrelevant today, more an example of sloppy, poorly optimized game coding. LOL.)

Why that comes about is pretty interesting. In some recent Intel charts that map out Intel's competitive advantages (in a Ryzen 3000 world, with comparable IPC, more cores at cheaper cost) they refer to their Software group as one of their 'six pillars'. That suggests (strongly) to me they intend to keep on helping developers with highly optimized code and cheap (free?) compilers for game development. To a small developer, getting cheap help like that must be a boon and getting locked into one installed base of gaming systems isn't as bad as it may seem if it's the largest.

Demonstrably false. NVIDIA was caught years ago hand optimizing a specific benchmark where they knew ahead of time what calls were to be made when because it was "canned" As a result the results were artificially high and no way reflected the actual gameplay.

 

[drea.drechsler]

Demonstrably false. NVIDIA was caught years ago hand optimizing a specific benchmark where they knew ahead of time what calls were to be made when because it was "canned" As a result the results were artificially high and no way reflected the actual gameplay.

GTA V isn't just the benchmark, it's the entire game. Not the same thing.

And demonstrably false...how? GTA V clearly performs better on Intel WAY out of proportion to any other game. Whether or not it was INTENTIONALLY optimized for Intel could be disputed, but whether or not it performs BETTER is not up for debate.

That there's little incentive for the developers to 'fix' it is also rather obvious: it's an old, niche game and there's precious little money in it. But that being the case, why keep trotting it out as a somehow relevant and modern gaming 'benchmark'? And I'm also talking 'optimizing'. I believe what Nvidia did went WAY beyond simply optimizing and more like Intel has got BAPCO doing with SysMark.

 

[digitalgriffin]

GTA V isn't just the benchmark, it's the entire game. Not the same thing.

And demonstrably false...how? GTA V clearly performs better on Intel WAY out of proportion to any other game. Whether or not it was INTENTIONALLY optimized for Intel could be disputed, but whether or not it performs BETTER is not up for debate.

That there's little incentive for the developers to 'fix' it is also rather obvious: it's an old, niche game and there's precious little money in it. But that being the case, why keep trotting it out as a somehow relevant and modern gaming 'benchmark'? And I'm also talking 'optimizing'. I believe what Nvidia did went WAY beyond simply optimizing and more like Intel has got BAPCO doing with SysMark.

GTA V benchmark is a canned benchmark. What I mean by canned is that it's repeatable and known ahead of time. That means specific sections of that canned script can be analyzed and hand optimized by video card companies. This optimization may not apply to other parts of the game.

For example: I may write a 1024 case switch statement. This would take considerable time for the CPU too determine which branch of code to execute as it test each condition. Not only this but jumps are no longer in local buffers which causes a load penalty on the jump. But if I hand optimize the switch statement at time X:Y:Z:, because I know condition K will always be true, then I can skip a lot of code evaluation. They also employ other dirty tricks like lowering quality of the render.

In my own engineering code at work, I will quite literally execute millions of lines of code over and over again as the solution settles to a steady state. If I knew ahead of time the end design target it will be about 2000 of hard coded settings with no evaluations.

Just google: video cards caught cheating on benchmarks. https://www.bing.com/search?q=video cards caught cheating on benchmarks&qs=n&form=QBRE&sp=-1&pq=video cards caught cheating on benchmarks&sc=0-41&sk=&cvid=009114F9C3474679A392A92DF447F8D7

Good review sites will create their own scripted benchmarks to avoid this kind of cheating.

Now whole game optimizations is accomplished by pre-loading shader programs into video vard memory, or textures into memory. These are hand optimizations that apply to the entire game and done by the video card company driver team usually. That's where you see a lot of these performance improvements come from. Compiling shaders takes time. Transferring textures from main memory takes time. So doing these actions in advance benefits the game as a whole. But this is NOT the same as hand optimization for benchmarks.

 

[drea.drechsler]

GTA V benchmark is a canned benchmark. ...

I don't know what you're getting on about exactly... are you saying the benchmark script in GTAV doesn't represent actual game play? And so, if that's the case, was it "hand-coded" to favor Intel processors and that's why it works so well for them?

I'm not trying to allege anything similar...just that it obviously DOES favor Intel processors, however it came about. Sloppy coding, hand optimization, use of Intel compilers with switches enabled, whether unknowingly or knowingly, that favor Intel processors or jiggering with a benchmark script. However it was done it obviously favors Intel.

But if the script WAS jiggered to favor intel, that changes things dramatically. These 'independent' review sites e.g., GamersNexus, should be made aware of it so they can go riteously ballistic as it effectively makes them Intel shills if they continue to use it.

 

[NightHawkRMX]

Its just that benchmark is predictable and things csn be optimized for it.

 

[drea.drechsler]

Its just that benchmark is predictable and things csn be optimized for it.

Then I'm going to say they're no longer a benchmark...but merely a demo. And probably a bad one since the game may not play nearly as smoothly as the demo shows it.

Reputable review sites should be able to identify and filter out those demos that have been 'jiggered' to play smoother than game play will and never prop them up as a benchmark for reviewing hardware.

EDIT: Not that many people have 9900K's or 3900X's and 2080TI's that allow you to just crank it all to 'ultra' at 4K and go for it. So my understanding of the reason those 'benchmark' scripts are included with games are to allow you to adjust quality settings to assure you have smooth gameplay. Now it seems to me it pretty much makes it pointless (unless you like watching smooth demos) if you get smooth running benchmarks but gameplay turns out to be jerky.