[ 3 / biz / cgl / ck / diy / fa / g / ic / jp / lit / sci / tg / vr / vt ] [ index / top / reports / report a bug ] [ 4plebs / archived.moe / rbt ]

Due to resource constraints, /g/ and /tg/ will no longer be archived or available. Other archivers continue to archive these boards.Become a Patron!

/g/ - Technology

View post   

[ Toggle deleted replies ]
File: 111 KB, 1240x695, 1491938810343.jpg [View same] [iqdb] [saucenao] [google] [report]
59871625 No.59871625 [Reply] [Original] [archived.moe] [rbt]

With Intel Processors(TM) you can expect a beautiful and stutter filled experience the developers envisioned for you when playing.

Buy Intel now, buying AMD is supporting communism, and you know that communism is the reason China and Russia have the most malicious hackers.

>> No.59871650

This sounds like Bate madlibs.

>> No.59871701

>not 720p
baka OP, everyone knows that you need to test games in 720p to eliminate GPU bottlenecks.


>> No.59871720

No, we should test at 160×120.. It's only fair

>> No.59871795

This bait is bad and you should feel bad.

>> No.59871809

Who needs resolutions anyway? It should be run headless and see how much mails per second it can parse before choking.

>> No.59871851

720p? what do you need 480p for?

>> No.59871930

The only fair CPU test is seeing how a system runs full of malware, since that's the most common Windows user workload.

>> No.59871986
File: 148 KB, 1019x1024, 1492016432266m.jpg [View same] [iqdb] [saucenao] [google] [report]

That is good

>> No.59872022

Ice cold.

>> No.59872057

>6 FPS minimum

b-but muh GHz

>> No.59872058

Oy vey, its annudah shoah

>> No.59874065


>> No.59874086 [DELETED] 

fake news

>> No.59874103
File: 234 KB, 882x758, 1491330680978.png [View same] [iqdb] [saucenao] [google] [report]


>> No.59874120 [DELETED] 

Jews are the master race and super intelligent. Id rather buy my shit from a Jew

>> No.59874125


>> No.59874139

m-muh cockspeed...

>> No.59874163 [DELETED] 
File: 532 KB, 745x782, 1487547970515.png [View same] [iqdb] [saucenao] [google] [report]


>> No.59874329
File: 51 KB, 1280x720, ARIGHT.jpg [View same] [iqdb] [saucenao] [google] [report]


>> No.59874354

What is the reason for the better min frames exactly? Has anyone discussed or even theorized what part of the Ryzen architecture is responsible for this?

>> No.59874383


All memes aside this is true. When you're doing a CPU benchmark you want to eliminate all other variables. It makes zero sense to benchmark with a GPU bottleneck

>> No.59874407

The more threads genius. This isn't 2012 anymore. AAA Games routinely take use of 4 threads and up.

>> No.59874416
File: 343 KB, 1024x609, intel.jpg [View same] [iqdb] [saucenao] [google] [report]

>no 7700K @ 5.5GHz cooled by horse semen
>not testing at 640x480

Delete this clickbait now. I can't believe AMDPoorfags think this is a fair test, making Poozen look better. Mom WILL be hearing about this!

>> No.59874417

game mode fixes this

trust me goy

>> No.59874420

having 2 more cores and 4 to 8 threads helps

>> No.59874432

Than its a good thing it's a CPU bottleneck, obvious by how the i5 is choking on its tounge.

>> No.59874456
File: 104 KB, 582x1048, forhonor4CPU.png [View same] [iqdb] [saucenao] [google] [report]

>quad core CPUs

>> No.59874469
File: 45 KB, 500x353, image.jpg [View same] [iqdb] [saucenao] [google] [report]

>it actually makes Ryzen better
I'm waiting for the Firelake owners to off themselves when this happens

>> No.59874487

Did he eat it?

>> No.59874532

Intel still faster :)))))

>> No.59874540
File: 495 KB, 1070x601, 1489122339460.png [View same] [iqdb] [saucenao] [google] [report]


>> No.59874589

Is that all it is though? Even in games that aren't heavily multithreaded, Ryzen seems to be better at the min frame rates. It can't be only the extra cores.

>> No.59874601

>m-muh butter smooth 20 fps

Intel still has higher averages, minimum means nothing

>> No.59874643

Pro tip:
Games can only make so many frames per second.
There's calculations that go across many clocks, and calculations that are time limited.
Unless you go back to the days of dos where games literary ran faster with clock speeds/IPC.

>> No.59874711

>Is that all it is though?
pretty much, it just offloads OS/background noise to other threads
not to mention a lot of games take advantage of 6 cores already

6900K/5960x/6850K for example are much better CPUs for gaming than 7700K even though they get lower averages.
it costs $1 000/500 which makes it stupid purchase just for games
you can get 1600x for 250 though

>> No.59874718

The more a given workload is spread between multiple threads the less of an impact any interruption of any given thread will have on the overall performance.

An interruption could either be by OS scheduling with different programs, or even within the game itself due to the multithreading design (ie where one thread is 'interrupted' because it has to wait for another thread to finish).

In simple terms:

Say you have a single-threaded game and the OS wants to interrupt it. For the time that the game is interrupted, performance drops from 100%. The result is potentially a very big drop in FPS while the game is interrupted - say from 100FPS down to 0FPS. In reality you're unlikely to perceive a drop all the way to 0FPS because the interruption is normally very short - likely only a few milliseconds - but technically for the period of time the program is interrupted it is no longer running at all, so I will use this as an example anyway.

If the game uses 2 threads, and the OS interrupts one of them, performance would drop 50%. The result is a significant drop in FPS, but also drastically less than it would be on the single-thread game - say 100FPS down to 50FPS.

If the game uses 4 threads, and the OS interrupts one of them, performance would drop 25%. Here we have doubled the number of threads, and again halved the performance loss. Notice however that in reference to the maximum speed however, the change is only 25%. Here we might be going from 100FPS down to 75FPS.

Double this again to 8 threads and you get to 12.5% for a performance drop, or 100FPS down to 87.5FPS.


Obviously this isn't a perfect example of how parallel processing works, so the real-world % change will be much more case-by-case, and most multi-threaded programs are not perfectly balanced, but this is a simple version of why people make such a big deal out of having more threads for multitasking and why more threads will improve minimum framerates.

>> No.59874762

I can see game mode helping Intel's lower core count and crappier SMT since it stops background tasks from fucking around while you game.

>> No.59874783

funny enough, by early accounts, it helps more cored platforms more

>> No.59874793
File: 245 KB, 800x612, 1489160516428.png [View same] [iqdb] [saucenao] [google] [report]


>> No.59874816

your day of reckoning has arrived, corelets

>> No.59874829
File: 224 KB, 799x540, 102cee2d7faf13b5744942deda834ad9.png [View same] [iqdb] [saucenao] [google] [report]

Isn't that ironic, dontcha think?

>> No.59874839
File: 604 KB, 800x523, good goy.png [View same] [iqdb] [saucenao] [google] [report]

>10 FPS higher average at 100+ FPS
>turns into a stuttery mess

Intel wins this round, goys.

>> No.59874847

source that isn't pulled straight out of your ass?

>> No.59874849
File: 307 KB, 915x678, corelets.png [View same] [iqdb] [saucenao] [google] [report]


>> No.59874854

>moar coars not a meme anymore

AMD is avenged.

>> No.59874892

>that huge difference in results
>gpu bottleneck

That looks pretty damn characteristic of cpu bottlenecking, to me

>> No.59874907
File: 393 KB, 801x720, [HorribleSubs] Gabriel DropOut - 10 [720p].mkv_snapshot_11.58_[2017.03.13_11.24.19].png [View same] [iqdb] [saucenao] [google] [report]

>Even the FX series has comparable minimum framerates

>> No.59874914


>> No.59874924


>turns into a stuttery mess

Yeah if you're watching your chink shit hentai on one monitor, erping with your steam bf on the other and gayming on your third monitor, then yes, it will stutter. But if you close other apps while gaymen it will outperform ryzen

>> No.59874952

I hate how they're pitting the quad cores against the Ryzen in all of these tests.
Quad core is growing obsolete at an increasing rate and 6 cores is going to be the new sweet spot.
I want to see how these Ryzens hold up against overclocked 6800k/5820k.
So far I've only seen stock results of them.

>> No.59874990

Why do I have to close other "apps" (lel) to run a game well? I thought we moved past 2004 already?

>> No.59874991

This is an explanation for why minimum framerates improve with more threads.

On the other side, pic related is Amdahl's law, which shows the theoretical improvements in maximum framerates with more threads.

Notice in this case how software that runs 50% in parrallel has very limited improvements even going from 1 to 2 threads, and has already achieved around 85% of maximum at 4 threads. This is why you see so many benchmarks following a similar curve as comparable CPU architectures go up in number of threads.

Ideally balanced software will show meaningful improvements all the way up to 128 threads (although notice that the relative jump from 64 threads to 128 threads would only be around 12%).

This is also why MOAR CORES is a relatively limited solution - many types of software may not be able to reach such ideal levels of parrallel processing within the foreseeable future, and you'll notice that even a 95% parallel program has already reached 50% of it's maximum potential at 16 threads. This means that with all else equal, even in the best-case scenario a 16-thread processor that run a game at 100FPS would likely not see any real merit in jumping up to a 1024 or 2048-thread processor to run a game at 200FPS.

>> No.59874997

6800k = $400
5820K = ~$380 if you can find one
1600 = $219
1500x =$190

6800k=1600x anyway though

>> No.59875000

Because of these fucking stutters I will never consider an Intel CPU for gayming ever again.

>> No.59875005
File: 34 KB, 648x486, Amdahl.png [View same] [iqdb] [saucenao] [google] [report]

Whoops, pic included.

>> No.59875034
File: 162 KB, 633x900, 1489103005792.png [View same] [iqdb] [saucenao] [google] [report]


>> No.59875045


There is a test from pooccftech, but im not sure how reliable they are. They did show the 6800k at 4.1ghz outperforming the 1600x at 4ghz by a significant margin in gaymen, and a small but noticeable lead in productivity workloads. Especially the last part makes me think that AMD lied about the IPC, it should be closer to Haswell IPC than Broadwell, which is really disappointing imo.

That said, price performance wise the 1600x is a no brainer. If only the gaymen performance was closer then there would be no reason to get a 6800k.

Here is the link for it in case anyone wants to check it out https://youtu.be/doSoWiMrnhU

>> No.59875072

What if calculation a requires calculation b and c to be done?

>> No.59875108


Intel has been the only option for like 7 years now, did you just suddenly notice this when Ryzen was released?

Come on dude, this is shilling if i have ever seen it. You dont need to have 500 applications open in the background. Close your apps and lock your framerate to a number you're happy with, and boom, no stutters anymore

>> No.59875112
File: 305 KB, 2560x1440, 1481537763816.jpg [View same] [iqdb] [saucenao] [google] [report]

>They did show the 6800k at 4.1ghz outperforming the 1600x at 4ghz by a significant margin in gaymen


>> No.59875119
File: 73 KB, 850x469, Screenshot from 13-04-2017 20:09:18.png [View same] [iqdb] [saucenao] [google] [report]

IPC isn't static, it depends on the application, aka what kind of shitty code it's running.

>> No.59875132
File: 1.20 MB, 2484x3052, Ryzen bench.png [View same] [iqdb] [saucenao] [google] [report]


>6800k=1600x anyway though
Not so sure about that.
A stock 5820k at a puny 3.6Ghz actually beats the 1800x 4.1Ghz in some situations and a good 5820k can OC all the way to 4.8Ghz, where as every single one of them should hit at least 4.5Ghz
I haven't really seen any benches of these CPUs overclocked to those levels and going against Ryzen.

That's the first bench I see with any overclock on these 6 core Intels.

>> No.59875144

t. shlomo goldenstein the devourer of foreskins

>> No.59875162

do you know what stutters are?
it's when frame rendering goes beyond 30ms
no fps lock will help you with that

>> No.59875171
File: 10 KB, 400x210, 1492022790561.jpg [View same] [iqdb] [saucenao] [google] [report]

(sub)people creating such shit bait threads should be permabanned by our beloved moderation :^)

>> No.59875200



They ran multiple tests, in most of them the 6800k had a significant lead


I know, but we see a consistent trend accross all the tests. The only explanation for it is lower IPC. Still, Haswell IPC isnt terrible, but its disappointing that they lied about it. This also further widens the gap between Ryzen and Kaby Lake in single core, which again is disappointing.

Ryzen still is great for the price, but it cant match Intel per clock and core in pure performance

>> No.59875220

There's plenty of single threaded benchmarks out there, go look at them.

>> No.59875224
File: 315 KB, 2560x1440, 1465976042912.jpg [View same] [iqdb] [saucenao] [google] [report]


>> No.59875240


It will help. If you lock your FPS at say 120, the CPU wont be performing at 100% usage anymore as it wont be trying to push to 140+ frames anymore. Stuttering is caused by the CPU running at full load and a background process being activated, which forces the CPU to allocate resources to that task, dropping frames in the process. If you lock your frames you wont be running at full load anymore and those background processes wont interfere with your gaymen anymore.

>> No.59875241
File: 148 KB, 1259x875, 1484377586832.jpg [View same] [iqdb] [saucenao] [google] [report]

also it's on 2600Mhz SR RAM
I'd imagine it would run even better on 3200 or dual rank RAM.

>> No.59875293
File: 70 KB, 1398x438, ARMA.png [View same] [iqdb] [saucenao] [google] [report]


To be fair, Intel benefits from faster RAM in many of the same games and programs as Ryzen does.

>> No.59875302
File: 25 KB, 1350x244, Screenshot from 13-04-2017 20:21:10.png [View same] [iqdb] [saucenao] [google] [report]


>> No.59875314


Oh lol, a single threaded game.
Is Fallout 4 your next choice?

>> No.59875337

I wonder how much they are paying AMD for the privilege. More than Nvidia i'd bet. Will know q4 or q1 next year.

>> No.59875358

sure, but degree is much smaller
ryzen gets benefit across the board

>> No.59875369
File: 25 KB, 1299x818, HandBrake.png [View same] [iqdb] [saucenao] [google] [report]


No, Handbrake is.
Look I don't give a flying fuck about the war between Jewtel and AMD, but to say that faster RAM offers no benefit or only benefits one side, is fucking retarded.
It benefits all systems across the spectrum.

>> No.59875389

Then they all have to be on one thread

>> No.59875390

>non game that fetches from memory and does massive amounts of compression all the time depends on a faster memory

Are you just retarded or pretending?

>> No.59875411
File: 393 KB, 462x497, [HorribleSubs] Kobayashi-san Chi no Maid Dragon - 10 [720p].mkv_snapshot_03.26_[2017.03.15_14.21.42].png [View same] [iqdb] [saucenao] [google] [report]

Are you saying that's what the guy doing the benchmark was running in the background while he took those tests?

>> No.59875420

That's part of the point.

Software can't become infinitely parallel because certain elements will always depend on others, hence the relevance of Amdahl's law. However, we're probably also still a long ways away from reaching the limits on parallel software design itself. One of the big reasons is that parallel software is simply more difficult to design and develop.

What the chart tells us is that even if we get software that is able to run extremely well in parallel, it wouldn't be too difficult for us to reach the practical limits of performance by increasing available threads. If we want to continue to improve performance, we will have to look at something other than MOAR CORES.

What this information and Ryzen's release should tell people is that there is LIKELY not much more to be gained in gaming performance by going above 8-12 threads due to the linear/blocking nature of games themselves. The biggest improvement more threads adds after that is in reducing OS interrupts and thereby increasing minimum frames and multitasking ability. On the other hand, a lot of server, enterprise and productivity software can be designed to run in parallel much more efficiently - hence why the R7 benchmarks for those applications came out so much stronger.

Other improvements can still be had with CPUs other than the core/thread configuration itself of course, such as pipeline design, branch predictions (parallelizing predictions), cache design and size, etc, but the practical reality is that if 16/32 consumer chips become available, that's pretty much endgame for that territory.

Basically, my next CPU upgrade is going to be a 6-core/12-thread with the best singlethread performance I can get in that configuration.

>> No.59875425


There's scaling in FPS for both Intel and AMD when faster RAM is used.

>> No.59875428

wow what the hell intel. ryzen might be "behind" with its 104 - 121 fps, but fuck does it destroy intel in min fps and that min fps matters a hell a lot more. who cares if you get 145!!!! if the overall experience is garbage. i'd rather get 121 than 145 that behaves as if it had parkinsons.

>> No.59875435

New substrate when?

(Spoilers: never)

>> No.59875441
File: 313 KB, 2560x1440, 1478750031622.jpg [View same] [iqdb] [saucenao] [google] [report]

for irony sake, results are identical in wildlands

IPC is fine, check this

>> No.59875465

Yes, sadly. They have already been talking about it for what, at LEAST 10 years now?

>> No.59875480
File: 227 KB, 871x475, Screenshot from 13-04-2017 20:36:08.png [View same] [iqdb] [saucenao] [google] [report]

In most games it just doesn't even matter.
Having a dual rank memory kit brings higher performance than a jump from 2600 > 3000

>> No.59875496

6+ core Intel is not even worth considering at all, unless you're scrooge mcfuckoff and have enough money to sail a yacht down main street.

>> No.59875505
File: 10 KB, 450x346, 1462289264344.png [View same] [iqdb] [saucenao] [google] [report]

>branch predictions (parallelizing predictions)
this reminds me

I'd like to see how ryzen performs after 2 hours of playing a game
for science reasons msotly

>> No.59875521

It's too bad reviewers / benchmarking never really gets into the interesting parts about CPU architecture.

>> No.59875534

>Bullshit: The Post

>> No.59875577

if you do that with $120 cheaper 1600
it won't matter either by your logic since it can get 120 fps anyway

i7 is literary worthless then and i5 goes down to single digits which amkes it annoying as hell
good to know

>> No.59875586
File: 480 KB, 493x342, aSTJbOerwCKqc.gif [View same] [iqdb] [saucenao] [google] [report]


It literally is true though

>> No.59875635


120 was just an example. As long as you lock it about 10fps under its max full load capabilities you'll have enough resources left for background processes to not interfere with gaymen. This is for i7's btw, there is no reason to get an i5 now unless you do literally nothing but gaymen while not even running discord or something comparable in the background, which is basically using your pc as a gaymen console.

>> No.59875768

Congrats, after a decade of insisting on MOAR COARS despite it being pointless there are finally a couple of games that actually benefit from it.

>> No.59875785

more like a few dozens

>> No.59875811

>Quad corlet mad his quads are a dying race.

>> No.59875824


>> No.59875839


Just wait for coffeelake 6 core, intel is gonna win after all

>> No.59875873
File: 73 KB, 600x865, 1488294400896.png [View same] [iqdb] [saucenao] [google] [report]


not exactly true

also whole point of this thread is to say that i5 is worthless, i7 is barely worth it for the money and you should get 1600/1600x for new build

>> No.59875896

No reason to be mad since my i5 has worked great and was the superior choice for years on end. Though I think I'll still wait for 1 or 2 more CPU series from Intel and AMD before making any upgrade, having to buy not only a new cheap but also new mobo+DDR4 RAM for a minor increment in performance doesn't seem worth it.

>> No.59875898

unless they would be able to get it to 4.5Ghz under 100w TDP on 14nm - no

6800k is 140w at 3.7 already

>> No.59875997

>No graphs
>No links
>No evidence
Enjoy your shoah, shekelchaser :^)

>> No.59876018
File: 9 KB, 250x250, 325.jpg [View same] [iqdb] [saucenao] [google] [report]


>> No.59876040

>No graphs
>No charts
>No links
>No evidence at all
Kill yourself, Intel shill.

>> No.59876054


>muh watts

Who cares nigger. If it can hit 4.5 stable at coffeelake IPC (it most likely will) it will demolish Ryzen

>> No.59876084

As has been pointed out, the workload (especially in games) is not static. More cores doesn't just allow you to do the same workload faster, it also allows you to do a larger workload. For example even if you can't squeeze any more physics performance out of more cores, you can still build larger games with more AI or more polygons instead.

>> No.59876114
File: 28 KB, 650x337, 1492070406450.png [View same] [iqdb] [saucenao] [google] [report]


>> No.59876115

GTA V loves threads. i5 just can't do it unless you lower draw distance and AI density. Then it works OK.

>> No.59876158
File: 61 KB, 600x713, Carltoncock.jpg [View same] [iqdb] [saucenao] [google] [report]

>just wait

>> No.59876190

It's a fucking CPU.
You guys need to get over yourselves.
-t.1600 user

>> No.59876219

If you have sixteen threads, as on a Ryzen processor, there's a massively reduced chance that the operating system or background programs (or even subthreads within the game, like threads to clean up data structures or autosave) will interrupt system critical things like the render thread.

Additionally if (and this is a big if) the game is spreading it's load between as many threads as it can get it's hands on (as is possible in tasks like rendering w/ Vulkan, or certain parts of physics simulation), the interruption of any one thread holds up a smaller proportion of the overall workload.

Also if the game is spreading the work as described above the impact of any particularly tough scene is likely to be lessened by the greater horsepower at work. For example if the game engine is spreading the physics collision detections across all available threads and a whole pile of small objects get stuck in a dent in the terrain, the continual collision checks will cause a drop in frame rate, lessened by having more cores to work on the problem.

>> No.59876230

Get out, we want to measure our epeens in peace.

>> No.59876275

>not having a room full of 32 core/64 thread workstations dedicated to running Waifu2x

>> No.59876311

This. It's the main reason Microsoft is coming up with gaming mode for windows. So the os doesn't decide to do some background stuff and steal the CPU for a bit. As for your Norton antivirus deciding to do a database. Well...

>> No.59876465

Is the guy who was talking about "static FPS" in here? Didn't get to finish calling him a retard.


>> No.59876488
File: 105 KB, 622x626, 1393883809444.png [View same] [iqdb] [saucenao] [google] [report]

Your favorite youtuber just made a video about this.

>> No.59876535

Ryzen is so fucking based. I never anticipated this level of assmad from one CPU release.

>> No.59876592
File: 136 KB, 915x487, 1491692531227.png [View same] [iqdb] [saucenao] [google] [report]


Intel is still better

Buy intel

>> No.59876662
File: 265 KB, 1064x948, 14919189563783.png [View same] [iqdb] [saucenao] [google] [report]


>> No.59876686

So intel objectively stutters more. Good to know.

>> No.59876699

Intel superior single core performance > ahmemed moar cores

>> No.59876701
File: 254 KB, 990x652, 14919902833690.png [View same] [iqdb] [saucenao] [google] [report]

Intel can't even hanlde 4k

>> No.59876722

>some frames went up to 180ms on the Intel
>while AMD's worst frames didn't go over 22ms


>> No.59876748



>b-b-b-but the human eye can't see past 3.5 GB

>> No.59876749

>that everything after 06:25


>> No.59876761

B-b-but It doesn't matter if my game freezes every 15 seconds, muh 10FPS

>> No.59876797
File: 315 KB, 664x602, it can't be helped - Haruhi.png [View same] [iqdb] [saucenao] [google] [report]

I pity people like myself who have quad cores.
I'll be grabbing a 1700 this year before quads yet really fucked over in 2018.

>> No.59876854
File: 77 KB, 637x685, Most Curious - Eru.jpg [View same] [iqdb] [saucenao] [google] [report]

Someone have graphs of CPU utilization on the quad cores? I bet they're being used to the limit, having no headroom left.

>> No.59876855

Now this is a good video dealing with a very specific issue.

>> No.59876879

Frametimes are completely irrelevant. I heard about it from some dude on /g/ who said all that matters is "static FPS."

>> No.59876924


>> No.59876952

I heard that "perceivable FPS" is all that matters though

>> No.59876954

I like this.

>> No.59876957
File: 121 KB, 704x512, subject ffffffffffffffff.jpg [View same] [iqdb] [saucenao] [google] [report]

>Coffee Delayke
>le epib new socket, time for a new mobo goy
PLEASE stop getting cucked. It's embarrassing to watch.

>> No.59876976

Enjoy buying another new motherboard.
Also, delid it.

>> No.59877133

Save yourself a hundred and get the 1600/1600x instead

>> No.59877172

I'm not sure that tech works in games. I don't think they're predictable enough for it to work. You'll always be doing different shit CPU wont be able to really do more then what's currently asked for.

>> No.59877210

I've thought about it since I might go with Zen+ on 2019, but I'd rather have more cores for building these two years. I'm also betting on rpcs3 getting good at the end of this year and it seems to like having more cores.

>> No.59877456


>he is too poor to spend 150 dollars on a new mobo

AMDpoorfags btfo

>> No.59877495

Nice try, Applefaggot.

>> No.59877712



Stop right there

>> No.59877943

that's why you use quad sli Titan XP's

>> No.59878089

Can you even SLI more than 2 cards this gen? I thought you couldn't.

>> No.59878164
File: 1.63 MB, 994x1012, 1473658457281.png [View same] [iqdb] [saucenao] [google] [report]

I run gtav at max 1080p with a 4770k never stutters not even once or has fps drops while Firefox is open with music mixes/ constant alt- tabbing

you guys are retarded arguing over CPUs when people will buy what they want

>> No.59878488

Sure thing corelet.

>> No.59878612

are you going to ever play in 720p? No? then it doesn't matter, those are real life results there, and there is no way that a gpu could bottleneck a processor only on the min FPS

>> No.59878641

>tfw not even king of corelets because of shitty amd apu with 4 "modules" which perform like 3 real cores

>> No.59878783

That's some impressive delusion there, anon.

>> No.59878790



>> No.59878909

>for the exclusive price of 1999 USD*

>*only applicable if shekel account contains 100 shekel account monies

>> No.59878949

not seeing axis on the thumbnails makes this bait/10.
when will they ever learn.

>> No.59879068
File: 273 KB, 1032x774, 1490522506611.jpg [View same] [iqdb] [saucenao] [google] [report]

Enjoy burning your house down

>> No.59879126

t. ITT Tech graduate Linux user

>> No.59879150

this thread makes no sense

>> No.59879170

Your existence makes no sense. Off yourself.

>> No.59879177

Can't read?

It shows, quite convincingly, that (((Intel))) makes shitty product/performance ratios

>> No.59879410

>have i5 3470
>Can't run 4k 60 fps without video card hardware acceleration. It stutters and shit itself
>Get Ryzen 1600
>Can run multiple 4k 60 fps videos with plenty of room left
Feels good.

>> No.59879669



>> No.59879700

>Have a mid range of the mid range processor from years ago
>It performs shitty on things that weren't relevant in its time
>Buy new processor that is slightly higher than mid range
>It performs well on things that are relevant in its time

Really stimulates my senses. 5 gum.

>> No.59879732

This. Well let the Ryzentards enjoy their 3 weeks of relevance I guess.

>> No.59879792

Yeah just buy the i5 6400 good goy!
Best value in a 4 core chip!
10% more game performance at gayms when locked down to only that? oy vey!
Workstation intention? Fuck you!
CPU cooler?
Haha get the fuck out of here you greedy nazi :^)

>> No.59879954

>are you going to ever play in 720p?
>Not emulating 240p games in 720p for a perfect 3x integer scale

>> No.59879994

>CPU cooler?
Do people really look forward to using stock coolers on their system? are people that poor that they can't afford the basic cm 212?

I've always thought of stock coolers as the industry min, and it seems like we're so far from that today even with cheap coolers.

>> No.59880070

if it's pre agesa update this is worthless. the microcode update decreased memory latency by 20%

>> No.59880285

The fact that Ryzen's CPU coolers are actually decent is a bonus.
Not an expectation but another benefit to a company's product in terms of competition with the consumer (especially those that aren't that into overclocking).

>> No.59880599

>4.5 stable at coffeelake IPC
Won't happen, bro.

>> No.59880608

>coffeelake 6 core
Overpriced and underperforming, guaranteed.

>> No.59880684

>Well let the Ryzentards enjoy their 3 weeks of relevance I guess.
What's Intel released in 3 weeks. i7-7700K part two housefire boogaloo?

>> No.59880858

Not even once

>> No.59880918

This. Coffee lake is 14nm AGAIN the fourth one now. The kaby lake fiasco shows that the core architecture is played out at least on 14nm and they cannot sqeeze anymore blood from that stone. Thats why they just binned higher and pushed stock clocks a few hundred mhz higher over skylake. All they can do on cannon lake is binn even higher get maybe one or two hundred mhz more tops and probably slash prices to compensate now that they have to compete with ryzen

Ryzen second gen will be even more catastrophic because they already logged most of the areas for improvement on Zen and they are easy as fuck. Smaller process (maybe 7nm) tweak the infinity fabric and make sure the new AM4 boards support higher DDR4 memory at least 4000mhz and ideally they would match the 4500mhz thats on the top end for the next series of intel boards.

first gen Ryzen with 3600mhz DDR4 at 4ghz is running hot on the heels on 5ghz intel chips with 3200-3600mhz so second gen zen with more IPC and higher clicks even just in the 4.3-4.4 range with 4000mhz DDR4? thats going to decimate them.

>> No.59880969

That feeling when you purchase an new i7-6950X for $800.

>> No.59880995

>promise we're not cherrypicking
Then why do I see reviews from benchmarks that when negative are called "day 0 benchmarks of course the drivers suck" but when they're not cpu limited games and the stats are the same for intel and AMD the memers jump all over it like it's a peer reviewed scientific paper?

All these threads are the same joke. How are you kids entertained by this unless you're legitimately fucking autistic?

>> No.59881023

>underestimating hyperscaling

>> No.59881372

Not just cores, guys. There's been speculation that AMD's SMT works better than Hyper Threading, which means that in the event of cache misses, the cores are still kept fed with work. More significantly, Ryzen has two large L3 caches (8MB each), whereas a 7700k, for example, has just one 8MB cache. When the L3 cache is accessed on Ryzen from within its own CCX, it's significantly faster with lower latency than Intel's L3 implementation. Granted, things are complicated when the CCX wants to hit the other cache, but still. AMD's L2 cache is also faster with lower latency. So not only does Ryzen suffer less from cache misses due to superior SMT, it's also less likely to have cache misses and when it does, the caches will reorganize data faster, in most cases.

The Infinity Fabric (IF) kind of screws things up a bit, but it seems to me that Ryzen is AMD's beta test for the technology. The Zen cores are okay, but the real winner here is IF. I'm not sure how Nvidia or Intel can compete with an AMD MCM APU that uses IF, since it seems pretty clear to me that if it's fast enough to connect two quadcores on the cache level, it's fast enough to connect a GPU to CPU. I'm very interested to see what AMD plans to do with the technology. So far it seems to be in Ryzen, Project Scorpio, and the PS4 Pro, but I wouldn't mind seeing a Zen CCX beside Polaris 10 for ~$250. I think that's completely reasonable and with good Polaris 10 chips drawing around 90W and an 8c Ryzen part's TDP at 65W, we're definitely looking at sub 140W TDP for a chip like that. Very interesting. It's probably why AMD spent money developing such nice coolers.

>> No.59881591
File: 35 KB, 580x386, 1489426555471.jpg [View same] [iqdb] [saucenao] [google] [report]

>Current year
>recording Min framerates
This is why 0.1% and 1% lows are a Superior testing method

Minimum framerates are statistical outliers.

>> No.59881674

both are important. If you have constant microstuttering, it's annoying. If your game instead hangs for a split second every few minutes of gameplay, it's also annoying.

What's happening right now is that because AMD has finally released a competitive product, shortcomings in Intel and Nvidia's products are being highlighted. There are people who consider themselves PC enthusiasts that have been building PCs in what is essentially an Nvidia/Intel marketplace for their entire adult lives (talking 18-24 year olds here), so I'm sure there's going to be just tons of shitposting on /g/ over it

>> No.59881833

>tfw king of corelets arrived today
>tfw stoked as fuck because it's a 500% increase in performance from my laptop I've used for 6 years

It's all relative brah

>> No.59882156
File: 9 KB, 248x203, Nope.jpg [View same] [iqdb] [saucenao] [google] [report]

It really isn't, businesses don't care about it, consumers with the money don't care about it, video editors don't care, etc. It's just a small minority of "stick it to the jew" or kids who see "more cores=better". Unless you want to stare at your cinebench scores all day then that's fine.
I'm out

>> No.59882187

t. corelet Intelcuck

>> No.59882190
File: 22 KB, 175x164, 1445291359018.jpg [View same] [iqdb] [saucenao] [google] [report]

>AMD goes for MOAR COARS in FX
>AMD fans admit it was a mistake knowing AMD won't make the same mistake again
>Makes the same mistake again
>Fuck it we'll just call everyone corelets cause this will never change

>> No.59882197

>more cores
>still get's shit on by a 7700k in rendering videos
gotta be real kick in the nuts

>> No.59882206

I missed you, Salazar.

>> No.59882222

>No graphs
>No charts
>No links
>No evidence of any kind
Shekelchasers BTFO

>> No.59882238

>Reduced to shitposting without any evidence at all because you can't rely on R7 day 0 benchmarks anymore
Intelcucks on literal suicide watch.

>> No.59882250

Go to your nearest movie studio. or really any major business dealing in CPU shit and guess how many Ryzen PCs you will find

>> No.59882270

>taking this long to reply

Are you stuttering?

>> No.59882273
File: 277 KB, 853x1280, 4500 CL16.jpg [View same] [iqdb] [saucenao] [google] [report]

Imagine Ryzen running pic related.
What's the highest RAM speed it has been tested on, 3200? 3600?
Can't wait to see what happens when people slap +4000Mhz monster RAM to it.

>> No.59882279

Prove it, Shiester McShekelstein.

>> No.59882312

Geee its almost as if ryzen isnt the actual enterprise chip. Actually it is, AMD are the real corelets who cant pass 8 cores and will flounder because anyothing over 8 is impossible for themt o fab.

>> No.59882339

>Intelshills reduced to embarrassing themselves
What a great day to be alive.

>> No.59882362

Fkn kek

>> No.59882712

>Can't wait to see what happens when people slap +4000Mhz monster RAM to it.
Past 3,200mhz, ryzen sees little gains.
Gamers Nexus's 3,600mhz tests showed little or no gains from 3200mhz

>> No.59882851

this has to be some of the shittiest hardtubing work i've ever seen, even early 2000's hardocp forum shit had better bends than this good lord

>> No.59882891
File: 10 KB, 236x238, me.jpg [View same] [iqdb] [saucenao] [google] [report]


>tfw still running a 8-9 year old Core Duo E8500

>> No.59883216

knowing how they test most undemanding parts of games I wouldn't consider those numbers as real

>> No.59883254

>cpu bottlenecking gpu
>intel is good for gayming guise!

>> No.59883464


Why have I never seen gameplay this stuttery until Ryzen came out?

Seems real convenient.

>> No.59883470


>> No.59883474

Is this stutter limited to Kaby Lake?

Because I am watching these GTAV videos and they are stuttering like mad and nothing like this happens on my older 4790k. Hell I'd fucking uninstall the game if it was playing like these videos do.

>> No.59883521

I am of russian and i7 7700 (locked) is my favorite CPU on the Internets.

>> No.59883531
File: 14 KB, 659x382, rip.png [View same] [iqdb] [saucenao] [google] [report]

>> No.59883612

It has to be
I have a 6700k and I have never ever experienced these issues ever.

>> No.59883664

my theory is that kaby micro throttles due to how much intel pushed skylake arch
and since it's at 100% all the time it's that noticeable

>> No.59883703


here is huge BUT to this theory though

>> No.59883814

It's just strange to me how all of a sudden after the Ryzen release all of the video came out the woodwords talking about how the 7700k has microstutter issues

Did windows release an update or something with the new bios for z270

>> No.59883843

kek, how the tables have turned. Intelfags were extremely smug with Sandy Vagina and for good reason, but they can't seem to take the heat from Ryzen or lack thereof.

>> No.59883880

it's not strange, nobody had or considered 6900K
it has exactly same benefits but for $1 000

and for years people thought that those stutters are software related, apparently you can bruteforce it on the cheap now
stutter is not a new thing, people just accept it as part of PC
FX series btw are absolutely terrible in that regard

>> No.59883990


>average thread load is 25.438%
>GPU load during most intense sequences is 88%

>average thread load is 55.625%
>GPU load during most intense sequences is 97%

THIS IS WHY the minimums are so much higher on Zen compared to Kabylake, and why there's GPU stalls which leads to those 100-300ns latency spikes and noticeable hitches in the game.

My understanding (correct me if I'm wrong), is that the primary thread in any CPU in Windows is CPU1. On Zen, even under the most intense of situations, the thread load does not exceed 36%, whereas on the 7700K the primary thread (CPU1) hovers around 50% and maxes out at 79%. While there's more breathing room present on both primary threads, due to the greatly distributed load on Zen, any interrupts don't cause you to perceive the GPU stalls visually.

Yeah, in the video framerate is tanking for BOTH CPUs under graphically intense situations, but it appears that the framerate is more evenly dropping on Zen--thus the transition from a high fps state to a low fps state isn't as jarring, as it is on Kabylake, where the drops are catastrophic, before it recovers.

>> No.59884000

Except the stutter issue literally just popped up for the 7700k
before nobody complained about anything but now all of a sudden the 7700k has a stuttering problem. And I mentioned in another thread that me and my friends who also have 6700ks cannot replicate this stuttering. We've tried making our overclocks unstable, underclocking, fucking with the memory, switching GPUs, nothing. It's just weird how all of a sudden these problems are coming up, it has to be some stupid botnet update intel put out or something for the 7700k

>> No.59884072

gta 5 behaves strangely on my 7700k, sometimes its fine, other sessions i notice very slight "ticks"

i honestly think its more to do with the game engine, rather than the cpu. Capping the framerate seems to remedy it.

>> No.59884095

It's just weird is all, did you have a 6700k before? Cause like I mentioned before I genuinely cannot for the life of me re-create this stuttering. I've even tried having a bunch of processes open in the background.

>> No.59884127

pretty much


after six minute or so he talks exactly about that

>> No.59884138

I went from Phenom 940 to g3258 last year and experienced stuttering and uneven frame delays despite much higher single thread performance, and equal or higher multithread performance. Office/shitposting tasks felt more responsive on g3258 but in games stuttering and jerkiness was extremely annoying. Felt like bad framerate interpolation.
Upgraded to 4670 and it got better but nowhere near as smooth as my old Phenom 940 (in old games that were not cpu-bound below 60fps there)
Maybe most people just never not experienced that shit to really notice?

Gotta get Ryzen as soon as decent motherboards get in stock locally.

>> No.59884175

you are fucking stupid

>> No.59884183

itt: AMD lies and FUD

>> No.59884201

This shows that no matter what, you are better with more cores.

>> No.59884223

>graphically intense situations
That one hickup shown on video happens when he is jumping over a wall with only two big flat surfaces and the character model on the screen. Thatäs not graphics related, more likely some shitty periodic garbage collection algorithm that Ryzens happen to handle better than the corelets

>> No.59884231

If a drop in quality settings, be it resolution, post processing, or the selection of which ingame assets to load and display changes the way the game uses the CPU, then that's a variable you want to keep at a "fair" level for the CPUs' intended market. It makes zero sense to benchmark the CPU in a setup where it's not doing all the same things it would be doing in practice.

>> No.59884252

Having more headroom with the number of threads available is certainly the most likely suspect (I remember this argument was one of the best reasons for getting a Q6600 back in the day), but I wonder how much of it can be attributed to the meme neural network branch predictor

>> No.59884399


What makes you think that if a core stalls, another core can take it over? This entire concept of only losing 12.5% performance is inaccurate I would say.

Spreading runtime over more cores could hide stutter in the instances though, as the render time is limited by the longest thread. So that any stutters in the fast treads might not been noticed. This however, would never hide the ~200ms interruptions seen on the intel systems.

There must be some unique issue with the intel systems as far as I am concerned. Maybe interuptions by the os could cause this (again, note that this would not result in an effective 12.5% performance penalty for ryzen, as some threads must be pretty much idling).

>> No.59884424

It's got more to do with issues that were always there, but people generally don't pay it much attention if it's not being highlighted. Take frame time consistency as an example. A couple years back it was all the rage with FCAT and other tools being used to analyze them when AMD's (at the time) poor CrossFire implementation caused micro stutters, but after a while people stopped caring and reviewers stopped analyzing frametimes when it faded and became a "non-issue".

The issue this time is very minor by comparison, but of the same nature. People will forget about it and reviewers won't think to look for it after a while, even though the issue might still be there for years or have the potential to crop up again.

>> No.59884550

I fucking love this video. Why won't more """""tech experts"""" benchmark frametimes? It is not that difficult to do.

>> No.59884563


People have been talking about stutters on pc's for ages. Typically in console vs pc discussions. Its just after the release of ryzen that we know that it is related to intel systems.

>> No.59884596
File: 973 KB, 801x1500, 1490399409123.png [View same] [iqdb] [saucenao] [google] [report]

It is bentium 4 all over again

>> No.59884627

You still can return it senpai

>> No.59884648

>comparing a CPU which was out barely a month ago to a huge installation base of (((intel)))
Normies will realize that intelis shit after a period of time. Specially that intel CPU's are housefires compared to Ryzen AMD has a very good chance to make deals with laptop manufacturers for better performance and longer battery life.

>> No.59884680

>It's just weird is all
competition didn't exist before ryzen, that's all

>> No.59884684
File: 36 KB, 413x395, 1310483412100.jpg [View same] [iqdb] [saucenao] [google] [report]

>7700k with 1080 is faster than a 1080x with a 1080ti in GTA V


>> No.59884689
File: 992 KB, 801x1500, 1492122290568.png [View same] [iqdb] [saucenao] [google] [report]


>> No.59884702

saved, thank you based anon

>> No.59884732


>> No.59884898
File: 93 KB, 216x241, 50462012.png [View same] [iqdb] [saucenao] [google] [report]


>> No.59884950

You should also edit the "bentium 4 HT inside" from last image.

>> No.59885069

Because they don't actually play the games they benchmark, they just look at the numbers.

>> No.59885406


>> No.59885429

Why doesn't someone bench systems with tons of malware installed? I think it is an untapped market.

>> No.59885637
File: 57 KB, 1000x1000, 2109cae3ddd1d6a4ffc20de85ab1b607.jpg [View same] [iqdb] [saucenao] [google] [report]


Name (leave empty)
Comment (leave empty)
Password [?]Password used for file deletion.