3700X vs 9900K, that is the question...

[cucker tarlson]

It's helpful to know, sure...but doesn't Witcher 3 have an integrated benchmark? I don't run it...

I also thought he has it listed somewhere... just not in the review? I agree a section in each GPU review listing the testing settings would be helpful.

It also, to me, mind numblingly stupid to benchmark any game at 720p in the first place for the reason of the results you just posted... because it doesn't scale the same and exaggerates a difference not found at higher resolutions.

using regular resolution like 1080p and finding a cpu heavy place is ideal.
witcher 3 doesn't have a benchmark,but big cities (novigrad and beuclair) are extremely heavy on the cpu.tested myself

What I found about 5775c EDRAM's impact on gaming performance.

I found the option in bios to OC my EDRAM as well as disable it. I chose WatchDogs2 as my benchmark because of how CPU dependent that games is. Resolution is 2560x1440 but graphics settings are low to rule out GPU bottleneck. The results are staggering. EDRAM @2GHz - 118 FPS EDRAM disabled -...

www.techpowerup.com

digitalfoundry's recent vid,novigrad abolutely destroys cpus

 

[mbeeston]

So I'm sitting here thinking about building a new system, I already have an nVidia GTX 1060 so that saves me some cash in building the system.

I have two systems with parts chosen...

Config #1: Intel 9900K, Gigabyte Z390 Aorus Ultra, and a Corsair Hydro H115i Platinum
Config #2: AMD 3700X and a Gigabyte X570 Gaming X (this config would use the AMD Prism cooler)

Both systems would have 16 GBs of RAM and a Samsung 970 EVO 500 GB SSD.

Of course, if I go by price, I could save myself almost $400 by going with the AMD config but as with everything in life, it's really not that easy.

Here's some background... I play Diablo 3 and Starcraft 2 casually, I game casually and I really don't care about extremely high frames per second. I may get back into World of Warcraft at some point but that's a maybe. Hell, I still have two 1080p-class monitors locked at 60Hz. The most I would probably get is a 75Hz Freesync monitor that's again a 1080p-class monitor. Again I don't care about extremely high refresh rates for I'm not into the whole competitive gaming scene. I also like to spin up Hyper-V virtual machines for testing stuff in... just because. Right now I have a Hyper-V VM that I test Windows 10 Fast Builds in just to see what Microsoft is up to.

Yes, I have posted in other threads about how Intel is best for gaming IF you are into highly competitive gaming and you absolutely must have the highest FPS. But again, I'm not the kind of person.

So with all of that said, the question of course is... Do I need an Intel build for my needs or should I save some cash (nearly $400) and go with the AMD build?

i'd get the 3700x and put the extra money towards a new gpu, with a 1060 you'll get the same performance out of both anyway.

 

[EarthDog]

using regular resolution like 1080p and finding a cpu heavy place is ideal.

Is it? See last sentence (I edited).

Anyway, a bit OT, this silly thread in the first place... I digress.

 

[cucker tarlson]

Is it? See last sentence (I edited).

Anyway, a bit OT, this silly thread in the first place... I digress.

it's not about ppl moaning.
cpu test in gaming should reflect what you will experience at normal resolution but in places the cpu is the limiting factor,that's just common sense.ppl can say whatever they want frankly,opinions can't change the fact.

OT,but still impressively civil for an intel-amd thread.

 

[EarthDog]

but in places the cpu is the limiting factor,

We'll have to agree to disagree... because cherry picking a specific area like that is but a snapshot of the whole game. A worst case scenario. People bitch if it is that, and people bitch if it is another way. Reviewers cannot please everyone.

 

[cucker tarlson]

We'll have to agree to disagree... because cherry picking a specific area like that is but a snapshot of the whole game. A worst case scenario. People bitch if it is that, and people bitch if it is another way. Reviewers cannot please everyone.

well they can bitch if they want to,but this is what cpu testing is.what your cpu will perform like when it's the limiting factor.worst scenario-yes,but how else are you going to learn how game behaves in cpu heavy locations.cpu reviews are for ppl who can interpret them too.

 

[Vayra86]

Alright, maybe you guys are right. Keep my 8700K and just do a cheap upgrade with a Ryzen 5 2600 along with a cheap B450 board.

I’m looking at the Gigabyte B450M DS3H. I need compatible memory, preferably DDR4-3000 CAS16, and I’ll do a cheap upgrade for him.

That was my thought on page 1. Sensible choice, well played. Keep at it

well they can bitch if they want to,but this is what cpu testing is.what your cpu will perform like when it's the limiting factor.worst else scenario-yes,but how else are you going to learn how game behaves in cpu heavy locations.cpu reviews are for ppl who can interpret them too.

That again huh. @EarthDog and I don't agree on that either even though my practical experience when the CPU gets pushed in games, proves him wrong every time. Its hard to convince him, trust me

But yeah, we can at least enjoy W1zzards' 720p benches

 

[cucker tarlson]

That was my thought on page 1. Sensible choice, well played. Keep at it



That again huh. @EarthDog and I don't agree on that either even though my practical experience when the CPU gets pushed in games, proves him wrong every time. Its hard to convince him, trust me

But yeah, we can at least enjoy W1zzards' 720p benches

I kinda feel like W1zzard could do so much more by so much less.He's testing 10 games at 3 resolutions.Look at what gamersnexus or purepc do,they test 5-7 games at one resolution only,but do a lot more research as far as cpu limited areas.They come up with results that better reflect reality and they do it much quicker.

 

[Vayra86]

using regular resolution like 1080p and finding a cpu heavy place is ideal.
witcher 3 doesn't have a benchmark,but big cities (novigrad and beuclair) are extremely heavy on the cpu.tested myself

What I found about 5775c EDRAM's impact on gaming performance.

I found the option in bios to OC my EDRAM as well as disable it. I chose WatchDogs2 as my benchmark because of how CPU dependent that games is. Resolution is 2560x1440 but graphics settings are low to rule out GPU bottleneck. The results are staggering. EDRAM @2GHz - 118 FPS EDRAM disabled -...

www.techpowerup.com

digitalfoundry's recent vid,novigrad abolutely destroys cpus

This video man... I watched it. That is 90% total BS though. He is actually advocating to play on different resolutions to gain consistency?! What is he smoking. The overall conclusion is no CPU guarantees flawless gaming and shit code will be shit code regardless. This really goes nowhere. The cherry picked frame drops are also remarkably pro-Intel across the whole thing, if you ask me. That Crysis one... whut? All CPUs drop there and its common when the viewport shifts to a large open area. Not just in Crysis... And it doesn't even hamper gameplay, its a checkpoint load/save point as well (which induces that stutter alongside the other load!)...

Goes to show you need to actually play games to know what examples to pick, instead of stare at charts

 

[EarthDog]

That again huh. @EarthDog and I don't agree on that either even though my practical experience when the CPU gets pushed in games, proves him wrong every time. Its hard to convince him, trust me

It doesn't show ANYTHING except at that abhorrently low resolution. That is the ONLY thing it is good for. The proportions DO NOT SCALE and the proportions are different as it goes up. So those who play at 1080p+, that type of exaggerated testing is largely not relevant as the result simply does not apply anywhere close to the same manner.

What can you take away from a result that says... zOMG holy shyte this CPU holds back FPS by 20% WTFBBQ???? When at the resolution you game at it, the difference is nill? If people used 720p results as The Gospel, we'd be in a whole world of hurt.

You're practical experience, respectfully, is nothing compared to actual empirical testing which you can easily reference and see my point. But hey, its hard to convince some people of the facts, trust me. I can be convinced, it just takes facts and empirical testing to do so, not unsupported butt dyno experience.

EDIT... weird... people are disappearing off my ignore list. I don't think I've seen a vayra86 post (or was it Vaya dumas?) in MONTHS! But I surely see why!!! LOL!

 

[cucker tarlson]

This video man... I watched it. That is 90% total BS though. He is actually advocating to play on different resolutions to gain consistency?! What is he smoking. The overall conclusion is no CPU guarantees flawless gaming and shit code will be shit code regardless.

no,he's saying that 4K gives better consistency in frametimes and he's right.Don't know why you misinterpreted it.Look at the frametime graph when it's running 1080p - even 9700k can stutter here and there.4K - no stutter at all.

the problem EarthDawg has is thinking that testing e.g. witcher 3 in novigrad specifically is cherry picking while it's absolutely not.

It doesn't show ANYTHING except at that abhorrently low resolution. That is the ONLY thing it is good for. The proportions DO NOT SCALE and the proportions are different. So those who play at 1080p+, that type of exaggerated testing is largely not relevant.

What can you take away from a result that says... zOMG holy shyte this CPU holds back FPS by 20% WTFBBQ???? When at the resolution you game at it, the difference is nill?

You're practical experience, respectfully, is nothing compared to actual empirical testing which you can easily reference and see my point. But hey, its hard to convince some people of the facts, trust me. I can be convinced, it just takes facts and empirical testing to do so, not your butt dyno experience.

yeah man but that's what I said earlier,cpu testing in games actually requires the user to know their stuff or else they come up with conclusions like the "omg omg" sentence you wrote.

 

[Vayra86]

no,he's saying that 4K gives better consistency in frametimes and he's right.Don't know why you misinterpreted it.Look at the frametime graph when it's running 1080p - even 9700k can stutter here and there.4K - no stutter at all.

the problem EarthDawg has is thinking that testing e.g. witcher 3 in novigrad specifically is cherry picking while it's absolutely not.

Right but how does that affect either the CPU choice, or the monitor you are stuck with anyway? Sure it gives better consistency, but what value does that statement have? Its been obvious since the dawn of gaming. And meanwhile, we constantly need to hear the 9700K leads a bit. Every time.

 

[cucker tarlson]

Right but how does that affect either the CPU choice, or the monitor you are stuck with anyway? Sure it gives better consistency, but what value does that statement have? Its been obvious since the dawn of gaming. And meanwhile, we constantly need to hear the 9700K leads a bit. Every time.

he's just speaking from the perspective of reviewing it at different resolutions.
again,for the one who watches it to interpret.

 

[EarthDog]

yeah man but that's what I said earlier,cpu testing in games actually requires the user to know their stuff or else they come up with conclusions like the "omg omg" sentence you wrote.

ANd if you don't test that way....... the unadorned don't have to walk away misinformed.

 

[Vayra86]

It doesn't show ANYTHING except at that abhorrently low resolution. That is the ONLY thing it is good for. The proportions DO NOT SCALE and the proportions are different as it goes up. So those who play at 1080p+, that type of exaggerated testing is largely not relevant as the result simply does not apply anywhere close to the same manner.

What can you take away from a result that says... zOMG holy shyte this CPU holds back FPS by 20% WTFBBQ???? When at the resolution you game at it, the difference is nill? If people used 720p results as The Gospel, we'd be in a whole world of hurt.

You're practical experience, respectfully, is nothing compared to actual empirical testing which you can easily reference and see my point. But hey, its hard to convince some people of the facts, trust me. I can be convinced, it just takes facts and empirical testing to do so, not unsupported butt dyno experience.

EDIT... weird... people are disappearing off my ignore list. I don't think I've seen a vayra86 post (or was it Vaya dumas?) in MONTHS! But I surely see why!!! LOL!

It does not scale and yet, the performance gaps return in the worst case scenarios. In other words, it can be used as a stand-in for a worst case in-game scenario wrt CPU load. You're at liberty to believe or disbelieve that, I really don't care. Go go ignore button quick!

Anyway, topic hijacked... Sorry bout that. Dropping it.

 

[cucker tarlson]

ANd if you don't test that way....... the unadorned don't have to walk away misinformed.

I never said it's not needed,I said it's not representative of real world cpu bottleneck.

EDIT... weird... people are disappearing off my ignore list. I don't think I've seen a vayra86 post (or was it Vaya dumas?) in MONTHS! But I surely see why!!! LOL!

 

[dirtyferret]

I kinda feel like W1zzard could do so much more by so much less.He's testing 10 games at 3 resolutions.Look at what gamersnexus or purepc do,they test 5-7 games at one resolution only,but do a lot more research as far as cpu limited areas.They come up with results that better reflect reality and they do it much quicker.

GN tests at 1080p and 1440p, one less resolution then TPU.

 

[cucker tarlson]

GN tests at 1080p and 1440p, one less resolution then TPU.

ah yes,they do.still,less work and more informative end result.

 

[dirtyferret]

I'm going to put my two cents in and try to avoid the flying bullets.

I like that TPU tests at 720p, it gives me a basic understanding of what CPU headroom may be in certain games.

I like that GN tests at .1 even though it's hardly the end all be all of micro stuttering

I like that pcgamer shows a composite of 97th percentile in their gaming suite

I like that anadtech shows 95th percentile

I like that techspot/hardware unboxed does 36 game reviews

I understand fan boys will cherry pick these results to justify why "their" CPU is the best of the best regardless how it may fly in the big picture of things.

I think it would be rather boring if every site ran the exact same tests with the exact same hardware and games. After all, who here plays the exact same games on the exact same hardware? Now if you excuse me, I going to hide in my bunker.

 

[Papahyooie]

You guys really don't understand benchmarking do you lol? Funnily enough, it literally says why TPU does 720p tests IN the review, so you're just too lazy to find out.

They test games at 720p resolution because that effectively removes the GPU from the equation. A high end GPU will barely be working at 720p, so it allows the differences in the CPUs to come out. At 4k, the CPU doesn't matter as much because the GPU is loaded to the point that it's not waiting on the CPU. At 720p, the GPU is almost always waiting on the CPU, so you get a true representation of how a CPU performs. It also gives a good indicator of how the CPU will perform with a large range of GPUs, because that's an indicator of how many frames you will get MAXIMUM with a particular CPU, regardless of what GPU you use. If a CPU cannot hit 144 fps in a particular game at 720p, it will never EVER hit 144 fps in that game at realistic resolutions, even if you upgrade to a faster GPU. As such, it is a true representation of the CPU's performance in that game, uncolored by GPU choice.

But you all will probably still argue with me lol...

 

[trparky]

I've changed my plans for my upgrade. The plan has been completely scrapped, the change from an 8700K to a 9900K (like many of you in this thread have said) would be so negligible an upgrade it wouldn't be even noticeable.

I am however planning on upgrading my father's system soon. I plan on getting a cheap AMD B450 motherboard, a Ryzen 2600 (last generation), and some RAM for less than $250. I already have a case, SATA SSD (which is more than enough for him), power supply, and a video card we can reuse. So that leaves only the processor, motherboard, and memory to replace which should make for a quick and cheap upgrade that will last him for years to come.

Hopefully in a year or two, when Ryzen gets even better, I'll do the same. Keep my SSD and power supply to do a cheap upgrade instead of having to do a full system rebuild.

 

[cucker tarlson]

You guys really don't understand benchmarking do you lol? Funnily enough, it literally says why TPU does it IN the review, so you're just too lazy to find out.

They test games at 720p resolution because that effectively removes the GPU from the equation. A high end GPU will barely be working at 720p, so it allows the differences in the CPUs to come out. At 4k, the CPU doesn't matter as much because the GPU is loaded to the point that it's not waiting on the CPU. At 720p, the GPU is almost always waiting on the CPU, so you get a true representation of how a CPU performs. It also gives a good indicator of how the CPU will perform with a large range of GPUs, because that's an indicator of how many frames you will get MAXIMUM with a particular CPU, regardless of what GPU you use. If a CPU cannot hit 144 fps in a particular game at 720p, it will never EVER hit 144 fps in that game at realistic resolutions, even if you upgrade to a faster GPU. As such, it is a true representation of the CPU's performance in that game, uncolored by GPU choice.

But you all will probably still argue with me lol...

and yet you see completely different results at tpu testing 720p and other sites testing 1080p but in cpu limited locations,the latter actually being worse for the cpu despite 1080p/Ultra settings.
choosing a testing location is the alpha and omega of cpu testing in games.It's all you need to do.

 

[Papahyooie]

and yet you see completely different results at tpu testing 720p and other sites testing 1080p but in cpu limited locations.
choosing a testing location is the alpha and omega of cpu testing in games.It's all you need to do.

You're suggesting TPU doesn't test their games in specific locations, consistent over their tests?
You're suggesting that the random fluctuations and circumstances that a PC happens to be in at any given moment, including changes in game state, memory loaded, human error by looking at a slightly different spot, etc are more reliable than blanket limiting the GPU out of the equation by lowering the resolution?

Both nonsense.

 

[cucker tarlson]

You're suggesting TPU doesn't test their games in specific locations, consistent over their tests?
You're suggesting that the random fluctuations and circumstances that a PC happens to be in at any given moment, including changes in game state, memory loaded, human error by looking at a slightly different spot, etc are more reliable than blanket limiting the GPU out of the equation by lowering the resolution?

Both nonsense.

1. no,but they don't mention locations and the results actually look like they're not.
2.you're blowing the in-game testing margin of error out of propotion.way,way,waay out of proportion.in games that include a quicksave option and for a reviewer that knows how to do it it's not that difficult to do a hundred identical runs.

look at ac odyssey results.
if they tested in-game in Phokis those results would be 50% lower,1080p,1440p,doesn't matter.My 5775c gets so hammered in this location it can drop below 50,and it's a place you go to very,very often.

matter of fact,wait a second......

 

[Papahyooie]

1. no,but they don't mention locations and the results actually look like they're not.
2.you're blowing the in-game testing margin of error out of propotion.way,way,waay out of proportion.in games that include a quicksave option and for a reviewer that knows how to do it it's not that difficult to do a hundred identical runs.

look at ac odyssey results.
if they tested in-game in Phokis those results would be 50% lower,1080p,1440p,doesn't matter.My 5775c gets so hammered in this location it can drop below 50,and it's a place you go to very,very often.

What in the world does that have to do with anything? It doesn't matter if your 5775c gets hammered at a specific location. If the runs are exactly the same, then any CPU will have to do the exact same calculations that your 5775c did, and therefore would be "hammered" the exact same amount. So in Phokis, the results would be 50% lower... so what?? They'd be 50% lower across ALL CPUs, so the ratios would be maintained. The delta between the processors listed would be identical. Benchmarks are never about what the raw fps number is. It's about COMPARISON of the processors tested. You're completely misunderstanding the point of benchmarks here. It's not to give you a reliable indication of a specific performance target. That's not how that works, because your setup will ALWAYS be slightly different, and you'll always have slightly different numbers. The fact that the framerates would be lower in a specific location across the board is entirely irrelevant. The point of doing 720p tests is to provide an accurate comparison between the chips, while taking the GPU effectively out of the comparison.