why is amd so popular now when intel cpus still have better single

10% higher single core performance with a GPU bottleneck is irrelevant.
10% higher single core performance with a CPU bottleneck is nearly close to irrelevant.


tl;dr Intel's single core advantage doesn't do much outside of a few games

>>61201155
FUCKING BLUE BOARD YOU STUPID ANIME VIRGIN FUCK

>>61201155
>>>/v/

Also quit with your anime bullshit

Only the lastest and greatest intel CPUs actually have better single core performance. If you have an intel CPU that's less than one socket (IE year) old, it's worse than Ryzen in every single way.

>>61201155
Because gamers are retarded and follow whatever the latest trend is, in this case Ryzen

>>61201264
i have a 6700k, bought it in april and it crushes any ryzen cpu in performance.

>>61201345
I think you're kinda confused here, Intel is the one depending on the retarded to stay relevant.

>>61201206
No nipples were shown, so it's safe.

>>61201155
>i thought gaymers only cared about single core performance?
replace gamers with outdated retards

>>61201353
No it doesn't. It roughly matches it, in single core, and loses big time in multicore.

>>61201366
IT IS NOT.

>>61201155
Cuz poor fags, enterprises still buy Xeons.
Science foundations with mega clusters still buy Xeons.

>>61201399
No nips, no genitals, no cum, it's '4chan worksafe'.

Because most modern AAA titles are multi-threaded because it's not 2006. Single core performance only matters at sub-1080p resolutions and on old as fuck dx9 games. AMD shits on Intel in value and performance, that's why they've taken so much market share.

>>61201393

Quit shilling dumbfuck

Ryzen is literally nothing but die shrunk Bulldozer.

And no, a die shrink does not increase per clock performance you illiterate monkey.

File: 2sperf.png (65KB, 620x326px) Image search: [Google]

65KB, 620x326px

>>61201402
No they don't. They're buying Epyc.

File: 1444173404336.png (429KB, 619x706px) Image search: [Google]

429KB, 619x706px

>>61201446
O-rly, name a single server cluster running Epyc.

>>61201393
it does lad.

15% - 25% is not matching. i don't need the multicore speed that ryzen offers because i'm not compiling or encoding. most games don't even utilize the extra cores.

look up every gaming benchmark, intel wins almost every single game in terms of fps. ryzen 5 is good for budget gaymers and ryzen 7 is good for workstations that occasionally game.

intel is still the recommended cpu for vidya.

>>61201480
Sure:
https://www.amd.com/en-us/press-releases/Pages/amd-epyc-datacenter-2017jun20.aspx

Okay this isn't exact datacentres, but it's a list of the companies that have bought into Epyc.

>Dell, HPE, Lenovo, Mellanox, Samsung Electronics, Supermicro, VMware, Xilinx, and many others
>Cloud datacenter customers Microsoft Azure and Baidu announce deployments

People aren't buying Xeons. They're buying Epyc. Now, they still HAVE xeons from before when Epyc was available. But the next set of server purchases are going to be remarkably short of Xeons.

Because places like this still buy 16 million Xeon based Supercomputers.

Quad Socket Xeons with terabytes of ram.

They still got whole empty floors waiting to be filled with more racks, but will they fit them with Epyc based ones? It's doubtful.

https://www.youtube.com/watch?v=3RqF8m65r8g

>>61201513
>15% - 25% is not matching.
It's more like 5-10%, although it does matter heavily what kinds of games you play.
If you play the games that are intel favoured then go for it, intel is the choice.

>>61201527
Well it doesn't matter what we think, we're in speculation land. What matters is at the end of 2017 there will be a flood of AMD IS FINISHED and INTEL IS FINISHED threads based on some news article showing that Xeon purchases are higher/lower than Epyc. I'll eat my words if in the future Epyc hasn't sold more than Xeons, but unless you have data we're just pumping hot cum into our own mouths. Pointless, and you regret it afterwards.

File: 1489214401416.png (1MB, 2484x3052px) Image search: [Google]

1MB, 2484x3052px

>>61201155
The ryzen processors are more than good enough for games and single threaded stuff.

Pic related is old, do someone knows if someone re-did benchamrks after all the updates the whole ryzen platform has gotten?

>>61201534
i have no dog in this fight.

i buy whatever the better value at the time is, intel to me right now is the better value. amd cpus in the past have been notorious to have short lives. this skylake cpu should stay relevant for 5-10 years

File: bf1.png (102KB, 1604x1080px) Image search: [Google]

102KB, 1604x1080px

>>61201513
>muh average fps

Enjoy your stuttering, corelet.

Different platforms for different needs.
Sure, we have Epyc topping out at dual socket and 4TB total capacity ram.

On the other hand there are companies and foundations buying Quad socket and Octa socket Xeons-EX running 6 channel memory and 6TB of ram per socket

These are totally different markets.

Hey you're that other guy who likes to post toobie. I'm not sure I'm okay sharing my waifu with you but I don't have this image yet so I'll let it pass this time.
In regards to your OP, I'm not sure. It seems that Zen actually made the intel consumer products obsolete with their core count because right before release the i5 was fine.
My friend bought a r5 system and I have a highly overclocked i5 box while I get much better fps in games he almost doubles me in cinebench

>>61201415
>4chan worksafe
I think I get it. Thanks

>>61201598
4S+ is under 10% of the market.

File: 5014e83299ac3c8b8ff177fea54fff31--girl-guns-madoka-magica[1].jpg (60KB, 500x708px) Image search: [Google]

60KB, 500x708px

>>61201604
My Xeon 1231-V3 only gets 571CB which is less than a i7-3720QM @ 590CB according to Cinebench R15

I fell for the "buy a Xeon instead of an i7 meme" back then.

>>61201438
>>61201513

Not him but ryzen is definitely not a die shrunk bulldozer.
Comparing the the 1800x and the 6700k, we're looking at a +-5% single core performance and in multicore it doesn't even come close.
Even if you get the cheapest ryzen r5 it may be 15%-20% slower in single core benchmarks at most.

Now your purchase of a 6700k was by no means a bad one, but you need to get off your high horse and accept that ryzen isn't as bad as it seems.

>>61201595
only thing that stutters are the cute anime girls i watch lad.

i also got this 6700k and a z270m board for $370 dollardoos. i compared that to amd and a 1700x was not only more expensive but a 1600 and 1600x was roughly only $10-$15 cheaper.

not really worth it when you have to overpay for motherboards to support ram that won't oc to 3200.

>>61201571
give me a 1080 ti and i'll do all the 1800x benches :^)

>>61201648
They are also the market that pays millions of dollars per purchase. That's 1000 $1000 CPUs in a million.

Oh, just like how the worlds wealth is up there in the single digit percentage of people who give no fucks about the masses?

>>61201693
i'm not saying ryzen is bad, i was looking hard at ryzen before i made my decision.

i was able to get my 6700k for $270. i'm sure had i not been able to get it for so cheap i would've bought a 1600 or 1600x.

>>61201199
If you just take stock for stock, ryzen is anywhere from 7% to 15% worse than intel's 7700k

Once you overclock, amd is about 5-7% worse then stock 7700k

and once you overclock both amd is worse by around 17-22% single core, assuming you go to 4.8ghz which seems to be the norm outside of deciding, if you get golden intel silicon its possible to push the difference closer to 30%

however, once you push 4k you are gpu bottlenecked, and on games that amd can't push passed 144hz near lock at 1080p, intel isn't fairing much better

Then you have to make a large caveat to if a cpu is pushed to 100%, your experience playing games or even useing it becomes shit as any little thing that happens in the background, on my 1700 at 3.0 that variance is between 1-3%, on intel that would be 2-6% disproportionately affects how you use the computer. you would much rather have overkill, which for now is a 6 core or an 8 core, as 4 cores are fully utilized more often than not.

>>61201595
Depends on the game. In some games like ROTR, Ryzen frametime spikes are much higher than intel

>>61201580
Saying this unironically should be an offense worthy to be jailed.

>>61201726
you may end up regretting buying that come zen2, if I was in your position I would have gotten a 1600 hold over cpu just so I didn't have to buy a full new system for an upgrade.

zen may be within 7% clock for clock of a skylake cpu some loads it surpasses it by quite a lot but some its lower, 7% seems to be the average (granted a bios update game zen an across the board 3% higher ipc, so that gap is narrower now then at launch) but zen 2 seems to either be an unlocking of how the infinity fabric works, or tweaks that add up to low hanging fruit, intel's low hanging generation got 30-50% faster, amds early estimates were 15% but I don't know if that took into account the possibility of 7nm node and a node that can clock higher or if it was just purely the mechanical side.

still not a bad price on the i7 by any means, that would have had me questioning should I wait or not before zen came out.

>>61201682
It's fine we're entering the golden age of CPUs. You could jump on a new HEDT platform next year and be set for a while. I'm legitimately an AMD fan so if you do anything aside from gaming I'm going to guess Threadripper would serve you best however I strictly only play games and do other nondemanding tasks and at high refresh so intels' current shit serves me better.
Personally excited for the new GloFlo 7nm process, they're claiming 5ghz operation,

>>61201733
Source for those numbers because they are certainly not the ones I've seen.

>>61201231
ANIME WEBSITE

>>61201940
>>61201231
>>61201206
It's vidya shit, don't give /a/ a bad name.

>>61201534
its 15-25% when you overclock,

>>61201513
not all games utilize 2+ cores, but for ones that do, the extra cores offer load balancing, in most benchmarks you are in a sterile environment, with no background fucking anything, it's skews the results from what you would see real world.

>>61201871
funny enough, there is justification for that statement. sandy bridge came out 6 years ago and if you aren't demanding 144 fps from games, it's still very useable today, and in all likelihood will be useable when it hits 10 years old. around 10 ish years ago we hit a parody point, where cpus were good enough for normal tasks and anything that use to be really demanding like video is offloaded to a gpu rather then the cpu, for sandybridge to not make it to 10 years would either require a new video format not doable with the gpu but compresses so good places are willing to toss low end users under the bus for the lower bandwidth, or all programs are going to need to come with crypto miners. for games, if its unplayable on sandybridge then in all likelihood it is unplayable on skylake as skylake isn't a 100%+ improvement over sandy bridge.

>>61201751
the difference between devs who are good at engine work and ones that just make the engine work.

2b more like 2 cute tits

>>61201911
I only deal with base clocks, not really taking boost into the equation, so a ryzen base is 3ghz and a 7700k is 4.2
ryzen is between 4-7% slower, i'm not sure where more recent bioses put ryzen along with how good memory is handled now but affect how ryzen compares overall.
Then you have the magic of cross ccx to deal with and this is where you get large discrepancies from, as cross ccx is a straight 20% downtick in performance for amd at its worst, I can no longer find the video where the person forced cross ccx comparisons in a tech demo but it was a demo that demanded to talk between cpus, and when they were on the same ccx he got 20fps and when they were on different ones he got 16fps

Ideal to ideal amd may be trailing by 5% or less at this point, but real world they are trailing 15-22% without golden silicon.

>>61201550
epyc will cut into intels data center market share significantly, but won't likely outsell it till next year as the people in charge of buying shit have had plans for 1 year + with resources allocated and are not in a position to change their setup.

>>61202308
Post your AMD stock

File: my sides.png (173KB, 640x400px) Image search: [Google]

173KB, 640x400px

>>61201438
>Ryzen is literally nothing but die shrunk Bulldozer.
Holy shit the /v/ spillover is real.

>>61202451
He's not wrong

>>61202486
Um sweety he's extremely wrong. How would a die shrink of bulldozer have hyperthreading...>>61202486

>>61201155
Amd lower clocked and out preforming higher clocked CPUs.

They are not that different except for AVX-512 and it's an irrelevant extension to gaymen.

>>61201155
Gaymers are a small market, it doesn't matter if gaymers keep using Intel, it AMD dominates the server and mobile markets, Intel is pretty much fucked.

>>61203503
>he cares about a company's profits

It's true, /g/ really is sponsored by AMD.

>>61201155
Big AAA games become more and more complex, and with incoming VR piling all the work on one core will not work anymore.

>>61201155
The difference is not noticeable if you are just playing mainstream games (not using emulators and shit like that) and it's a lot cheaper.

>>61203527
we dont want them to fail

>>61203565
You clearly don't know games work. There is no need to use more than 2 or 3 cores, devs hardly use more than this because it's too hard and expensive to break the engine in pieces (audio, physics, post rendering...) and run those pieces on many cores, the end result is not that great and would be even worst if we could compare the same thing running on 2 very strong imaginary cores.

>>61203745
Games are becoming exponentially more demanding and so the need for optimisation rises. VR is quite demanding already, and it's become worse when AAA devs will want to make a real VR game. You are gonna hit the limit of pulling all shit on top of 1-2 cores sooner or later. Vulcan already gives us a taste of what is next, future would be behind developing an efficient game engine that can split up work among various cores, especially if we will see a resurrection of mmos that will need to be efficiently split across various cores to look nice and run smooth. Just because everyone dumped shit one 1-2 cores doesn't mean that this trend will continue. Also you don't need to explicitly break game engine completely into many pieces, just redirecting tasks will be enough of an improvement.

>>61201155
>why is amd so popular now when intel cpus still have better single core performance?
Because they are cheap.

>>61201155
>i thought gaymers only cared about single core performance?- 61 posts and 6 image replies shown.
And still do.

because it's not the fucking '90s and scp is fucking irrelevant and ryzen fucking destroys anything by intel in multicore performance

> But muh competitive market

Most people in 2k17 still don't realise, that intel and amd probably run a cartel and play their game on and on to cheat us out of their money.
I bet quantum cpus will be a thing in personal computing and be just as expensive as our regular old shit now is, as soon as intel and amd will get whooped by the government

>>61205306
>qunatum cpus
>muh cartel

How to spot a fucking retard who eats too much nigger e-celeb shit from youtube

>>61201402
>>61201480

I'm looking forward getting some epyc servers in my 48U server rack and/or threadrippers in our workstations. As I generally plan our purchases, albeit smallish in a datacenter scale, I dont think I'm the only one looking into replacing dual xeon e5's with epycs.

I sold my intel stocks few days before zen dropped. Its looking like it has been the right choice.

>>61205370
You seem pretty sure that what this guy is saying is wrong. Care to explain?

>>61203745
https://www.youtube.com/watch?v=v2Q_zHG3vqg

>>61202451
You should go back to /v/
Ryzen is a respin of Bulldozer on a die shrink and a newer process. They took each fake core on Bulldozer and beefed them up to replace the CMT they had before, then stole Intel's hyperthreading technology for more threads.

>>61206417
>Intel's hyperthreading technology
Top lel.

File: 134495534714.gif (412KB, 320x240px) Image search: [Google]

412KB, 320x240px

>>61206417
>stole Intel's hyperthreading technology
Is this level of stupidity even legal?

>>61206442
>>61206467
How else did AMD suddenly come up with their own version of hyperthreading in such a short time?

>>61201206
Calm down, kiddie. Your mommy is not here to get mad at you.

>>61206486
https://en.wikipedia.org/wiki/Simultaneous_multithreading#Historical_implementations

>>61206417
>then stole Intel's hyperthreading technology
DEC Alpha 21464 had it long before them.

>>61206486
Because Jim Keller designed it in the first fucking place. He worked on the DEC Alpha 21464, and he did work on Zen as well.

>>61206486

Go ask Intel why their version of SMT is so shit - both AMD's implementation and IBM's (hmm, delicious 8-way) crush Intel's efforts on every single front. Hell Hyperthreading can stil leasily give negative scaling, something that is (in practical terms) never going to happen on ryzen and POWER.

tl;dr Intel has spent a decade "perfecting" a technology to make it worse than what others are doing.

File: into the trash.gif (3MB, 400x311px) Image search: [Google]

3MB, 400x311px

>>61206690
No proof to back your claim?
then guess where your post belongs

File: Piledriver-Die.jpg (511KB, 800x982px) Image search: [Google]

511KB, 800x982px

>>61201438
>>61202486
>>61206417
I'm going to show you two pictures now.

This is Piledriver, the die used for the FX-8350 (Bulldozer 2)

>>61207754
and this is Zen

Do you see ANYTHING. ANYTHING AT ALL that is reused or similar.

File: 950px-amd_zen_octa-core_die_shot.png (1MB, 950x417px) Image search: [Google]

1MB, 950x417px

>>61207784

>>61207784
No, I literally don't see ANYTHING.

>>61207798
They changed the colors
Whoop-de-fucking doo

File: 1492443338178.jpg (97KB, 720x1280px) Image search: [Google]

97KB, 720x1280px

>>61201438
Hahahha
WHAT
Hands off that fucking crack.

>>61207784
>dumb fuck doesn't even know how to post properly

Yeah, no. Educate yourself.

>>61201733
>>61201911
http://cpu.userbenchmark.com/Compare/Intel-Core-i7-7700K-vs-AMD-Ryzen-5-1600X/3647vs3920

7700K vs 1600X = ~ 10% effective speed difference

The main issue with AMD isn't IPC, its the low clock. Intel can OC to ~5+ Ghz. Meanwhile AMD's stuck at around 4.1Ghz at best. So there's a 20% clock difference from OC on both side.

On stock 1600X is 3.2Ghz and 3.7 turbo.
On stock 7700K is 4.2Ghz and 4.5 turbo

20% clock difference stays even on stock.

With IPC being effectively similar/same, the real difference is the clock. 20% difference.

The guy is saying if AMD OC and the Intel 7700K stays same, (4.1 vs 4.5) then the difference is minimal. And that's true. Its reduced down to <9%. And these are pure low thread application. For any multithread, the AMD will be competitive at stock and with AMD OC, it will overtake stock 7700K.

>>61208011
>cinebench
Thought it was about gaming tho.

>>61208067
>gaymen
This is /g/, it's irrelevant.
Zen2 will clock extremely high anyway, fmin-fmax curve for 7nm LP will be nuts because it's a node developed by IBM.

>>61201155
This was true years ago, but we've had the same old quad cores with barely any improvements for the past 6-7 years since sandybridge. Recent games can max out i5s now and I don't see quad core i7s lasting as long as they once did, especially when we get more powerful gpus with volta/navi

Now you can get good ipc 6-8 cores there's literally no reason to buy quad cores that are the same price. Not to mention even Intel is now ditching quad cores on mainstream i5/i7 with coffeelake in September.

>>61208089
>Zen2 will clock extremely high anyway
Not guaranteed. Its a hope and projection at best. But there's no guarantee.

Hopefully AMD can improve their clock limitation and improve upon it. In that case, it would totally ruin Intel's lineup. However AMD needs some special sauce, more RD for something special other than simply performance/price. They need innovation.

>>61208112
Well it depends, they might not need higher clock speed if they can improve the ipc further. They're already almost on par with Intel, and Intel had years in the lead

>>61208089
>AMD is bad at it so it's irrelevant

Once again, /g/ is sponsored by AMD

File: current zen node vs zen2 node.png (129KB, 1562x332px) Image search: [Google]

129KB, 1562x332px

>>61208112
>not guaranteed
The fuck are you smoking? 7nm LP whitepaper lists 5ghz operation AND lower drive voltage.

>>61208089
no guarantee it's good or not other than ibm made it and some 'its suppose to be good's' ill hold off on sucking the process off till we get some initial silicon shown publicly.

>>61208011
You also have cross ccx latency to contend with which will fuck most programs that aren't effectively using more than 4 cores. Its an adea that the userbenchmark doesn't take into account. Its why some people when reviewing the cpu were wondering what the fuck everyone else was doing wrong when they published numbers that were significantly better than the rest.

>>61208164
>cross-CCX latency
HELLO KYLE!
HOW ABOUT YOU JUST FUCKING KILL YOURSELF.
LIKE DO IT.

>>61201580
Skylake cpus won't last that long.

Sandybridge lasted long because of the stagnation in the cpu market, you bought a quad core that's only 25% faster than sandybridge with the same core count when the cpu market is finally changing.

>>61208161
Not guarantee. AMD could bankrupt, they could delay, etc. in all that time.

Until it's announced by AMD, its, once again, a hopeful projection.

>>61208205
Its gonna be better than current ryzen which is already really good

File: Untitled.png (67KB, 1217x231px) Image search: [Google]

67KB, 1217x231px

>>61201155
AMD isn't popular in that market at all. In fact AMD's marketshare among the gaming market has decreased every month since Ryzen's release. Don't confuse /g/ with the general population.

Holy shit I love this thread
>MUH SINGLE CORE PERFORMANCE
>ITS JUST A SHRUNKEN BULLDOZER
>AMD stole MUH HYPERTHREADING REEEEEEEEE

Fucking intel shills are literally tech illiterate autists on damage control. Ryzen is good. It'll only get better. The better AMD does,THE CHEAPER YOUR PRECIOUS JEWTEL PROCESSORS WILL BE GET THAT THROUGH YOUR FUCKING HEAD.

>>61208228
This will be obvious come Intel's earning report, but steam is opt in, and most of the hardware there is OEM and laptop junk.

And most importantly, I doubt ryzen owners would even take the survey.

>>61208228
>steam laptop survey
(You) tried.
>>61208205
>l-lies! These Nazis from IBM can't totally can't make a better node than Chosen People of Intel
That's you. And no, companies don't lie in whitepapers unles you're nvidia.

>>61208225
That statement is a matter of fact. However there is nothing that backs up your statement.

AMD could release a rebrand for all it care. Please think about how you use your language.

>>61208254
AMD hasn't announced it will use GloFo's 7nm manufacturing for Zen2. This is your own projection. What has happened is GloFo has released their white paper on 7nm. AMD has released Zen.

Once again, don't confuse your own wishful thinking and your own projection with reality. Its possible AMD will choose GloFo 7n for Zen2. Possible is not what is real or what has happened. At this very point in time, AMD hasn't released any statement regarding them using GloFo 7nm for Zen2.

>>61208289
>AMD hasn't announced they are going to use GloFo's 7nm
Oh shieeet please stop.
I am dying of laughter.

>>61208254
It's steams overall marketshare survey for windows. What do you think happened, people waited till Ryzen came out to buy fancy new Intel laptops? Get real. Be in denial all you want my friend, doesn't change reality.

>>61208089
But the argument was gayming. I know that's stupid, but you moved the goalpost then.

>>61208329
Anon please stop using steam like something that's in any way reflect real market share in CPU market. Don't be a /v/irgin.

>>61208353
The OP specifically was talking about gaming. Try to follow the quotes.

>>61208182
sorry, he has autism and can't handle the fact amd is not better than intel in every way shape and form

>>61208353
>Thread is about gaming performance
>DEH LARGEST GAMING COMMUNITIES HARDWARE SURVEY IS NOT AN ACCURATE SOURCE

literal retardation.

>>61208551
Steam hardware survey is opt-in and in no way is indicative of actual CPU marketshare.

>>61208551
Easily 60-70% of steam users are playing on craptops.

File: IntelShillHavingBuyersRemorse.png (162KB, 633x900px) Image search: [Google]

162KB, 633x900px

>>61201438
>>61202451

R-ryzen is shit! I-it's just bulldozer again!
You m-must believe in me! My 6700k was a g-great buy!

>>61208562
Okay so what is indicative of actual CPU marketshare of the gaming market? Your ass?

>>61208562
>A survey of millions of people is not statistically important
Right on bro, all us AMDheads are smart enough to opt out of the survey, we wouldn't want devs to know that people have more than 4 cores so they might actually make games that utilize them!

File: future.png (727KB, 1764x782px) Image search: [Google]

727KB, 1764x782px

>>61208289
Shut up, please.

>>61208593
They don't have a source, they're just shitposting because /g/ is AMD sponsored

>>61208562
>survey is opt-in so it doesn't count

Do you actually have an actual argument here with evidence to back it up - Something like Intel or AMD users are more/less likely to opt-in or are you just moving the goalpost because AMD has poor market share?

>>61208638
>blindly believing marketing slides
And you're the dumbass who fell for the Bulldozer and Fiji hype, aren't you

File: 1495812218308.png (944KB, 1228x1502px) Image search: [Google]

944KB, 1228x1502px

>>61208963

File: cpuz_x64_2017-07-04_00-56-07.png (29KB, 403x402px) Image search: [Google]

29KB, 403x402px

>>61208963
Unfortunately, no.

People want competition so Intel stops being so damn greedy.

>>61208244
>intel
>cutting prices
lol