>bad multicore benchmarks >"our engineers will optimize

File: 1471139996247.jpg (80KB, 1280x720px) Image search: [Google]

80KB, 1280x720px

>>58009042
literally incomprehensible

>>58009081
I'm talking about AMD's stream where they showed Ryzen. It had all those elements described in the OP. The woman in the picture is the CEO of AMD.

what did you expect? they had to cherrypick scenarios that hid the fact that zen only has clock for clock perf equivalent to sandy bridge while making it appealing to gaymers.

>>58009081
If you are this fat you should consider suicide.

>>58009042
so? what's your point? back to >>>/pol/ if you want 100 guaranteed replies under one hour.

File: 1481628334629.png (7KB, 300x500px) Image search: [Google]

7KB, 300x500px

>>58009042

what's with the lesbian style? She would be qt witl normal long hair

>>58009109
>>58009042
Equal amount of cores between both CPU's, they're raping Intel IPC wise, and probably clock wise too
With so few replies you won't get paid

>>58010673
why? at that size you only have a few years to live anyway

>>58011056
>madkek

>>58009042
>That's what you get with a female CEO.
still better than what was happening with previous limpdick CEO

>>58009042
>bad multicore benchmarks
uh, I'm pro Intel but AMD had the best multicore til 2013

>>58009042
>dota streamer who can't into hardware encoding

Hardware encoding is shit for streaming.

You need at least over 15Mb to achieve decent quality compare to CPU x264 with 3Mb encoding.

File: 1480370064647.jpg (62KB, 498x460px) Image search: [Google]

62KB, 498x460px

>>58009042
I really see all that equipment that can do H265 HW encoding...

>>58012638
How well do newer 2-pass HW solutions fare in comparison to x264?

>>58010957

DELET THIS NUMALE!

>>58009042
>building a fucking computer in VR
wat

>>58009042
>stock value quadrupled in a year
>huge ass SoC contracts
>company making actual money for the first time in forever
>GPU market share up 10%
>caught up on massive technical debt

by all metrics she is doing a terrific job

File: Capture.png (1MB, 1920x989px) Image search: [Google]

1MB, 1920x989px

Zen isn't even faster than a stock 2600k in AMD's own cherrypicked benchmark.

Let's hope AMD at least sells off RTG to someone competent before they go bankrupt.

>>58009042
Who was that obese negress doing the battlefront demo? She couldn't even work out how to start the game. Reminded me of that ugly fat one from Lady Ghostbusters who had her nudes leaked

File: simians_tracy2.jpg (38KB, 355x367px) Image search: [Google]

38KB, 355x367px

>>58015612
>that ugly fat one from Lady Ghostbusters who had her nudes leaked
the one on the right here?

>>58009042
that's not a female.

>>58015554
>implying /v/, I mean /g/, knows or cares about anything aside from muh gaymes

>>58015554
Their market strategy is great. Basically market their cpu's on the basis that the majority doesn't know anything about the technology, and are only looking at "buzz" word metrics and comparing them to other products "buzz" word metric and the cost.

"ooooo, dis one has 4.5 gb speed and is $200 cheaper than that one with 3.5 gb. It must be really good! Look, it even has more cores!"

File: RUEBXQn.gif (3MB, 219x396px) Image search: [Google]

3MB, 219x396px

>>58009042
>>bad multicore benchmarks
>beating a $1.2k Intel CPU
>bad

>>"our engineers will optimize it"
Happens all the time with any semiconductor company, so yes. they will optimize the final silicon.

>>no single-core benchmarks
This is worrying indeed.
Still, assuming the chosen multicore benchmark workload is somewhat representative of Zen's general performance and their SMT implementations are about the same, you can assume that Zen's single core performance is in the ballpark of Intel's.

>>building a fucking computer in VR because it's The Future™
The demo chosen was retarded, but VR/AR definitely is The Future™

>>playing battlefield as a CPU benchmark
It's something many potential buyers will do.
The event was a "Fan Event" after all.

>>dota streamer who can't into hardware encoding
It's a CPU benchmark.

>>a fucking obese female nigger "gamer" playing battlefront
pic related

>>58016000

>This is worrying indeed.

Pick your poison since Intel reccomend blender as a benchmarking tool.

1) Lower IPC but a superior SMT implementation
2) Worse SMT but better single core

You can't have both worse SMT and worse ST while beating a 6900k in a piece of software AMD's older architectures get absolutely slaughtered in - even when they were new.

File: sandy Dec 2016.png (186KB, 993x718px) Image search: [Google]

186KB, 993x718px

>>58009109

>>58010673
thanks for the advice
i will consider it

>>58016376

http://www.ebay.com/sch/i.html?_from=R40&_sacat=0&_nkw=2600k&LH_Complete=1&LH_Sold=1&rt=nc&_trksid=p2045573.m1684

you can get a 2600k for $100-140 these days, which is pretty good considering the terrible stagnation in newer uarchs put out by intel and the fact that zen is already disappointing

>>58015556
>Zen isn't even faster than a stock 2600k
Source?

>>58009042
malti core is a meme. They should figure out a way to reduce the latency from shoving your lambda to the threads and getting the result down to at most 1 microsecond. This latency is 20-50 microseconds at the moment which is good enough for the things that need to be parallelised but slow enough that you can't just parallelise everything by default and expect it to be faster than the single core version.

What most developers want is to say: I want to run these four lambdas on different cores RIGHT FUCKING NOW, not after the OS decides to shedule the damn threads.

>>58016454

His rectum.

The intel defence force has been working overtime these last few days and now the hot new thing to claim intel's superiroity with is power usage delta.

/g/ is the same as OCN, anandtech and the like (but better than Hardocp) in that its currently trying to disprove what AMD has shown of zen with memes. Hardocp is actually convinced zen is slower than sandy vagina.

>>58016489
>anyone who doesn't blindly believe AMD's marketing BS is shilling for intel

kek, imagine the AMD fanboy butthurt if it was NVIDIA or Intel putting out these hilarious demos that don't represent real world perf.

>>58016485
every CPU with any sorts of caches is optimized for the case of hitting peak performance after "warming up" the critical event processing loop.

you're never gonna see the day where threads can be spawned, scheduled, and execute meaningfully in a few thousand cycles, since even normal single threaded code isn't expected to warm up in a core that fast.

File: Capture.png (26KB, 808x263px) Image search: [Google]

26KB, 808x263px

even AMD shills are starting to admit that Zen is DOA.

soon they'll switch to their shilling tactic of 'it was never meant to beat Intel in the first place, AMD wanted to make an inferior product'.

>>58016526

>vlc

>>58016526
Huh, it's almost like EVERY company biases benchmarks in favor of their own products...

Maybe instead of sitting around arguing over made-up numbers, we could wait until some third-party reviewers have got hold of the thing and given us less-biased comparisons?

>>58016617
Just up the TDP to 140W like the i7 6900K. AMD has XFR which automatically adjusts voltages for temperature.

Remember how AMD cherrypicked Blender benchmarks with Bulldozer too?

>>58016896
This, it's like everyone has the memory of a bird lol

>>58011308
That's what you think. My grandfather has been this big for my whole life and he's still alive at 74.

>>58015716
it makes sense that they would try to get whatever cash they could out of faildozer while they scrambled to make zen for half a decade. it's still pretty staggering the amount of excavator chips they sold, considering how awful it was. I guess we should thank consolefags for ryzen.

but I don't think the "market strategy" you're talking about was particularly effective. they lost nearly all their market share in the PC market. that isn't great by any stretch of the imagination.

>>58016068
my money's on better SMT

either that, or their branch prediction is fucking amazing and zen "ramps up" ridiculously well

>>58016942
>>58016896
Things that never happened: The Posts

The Bulldozer arch did terribly in Blender. It never favored their arch at all.

>>58016896
>Intel recommends Blender to benchmark their high end CPUs
>B-B-B-B_B_B_B-B-UUUUUUUUUUT AMD CHERRYPICKED

Also, Bulldozer was never competing against a CPU with the same core/thread amount.
It was always 8c Bulldozer vs 4c/4t or 4c/8t Intel, where BD will obviously have the advantage.

File: blender.png (7KB, 504x474px) Image search: [Google]

7KB, 504x474px

>>58017049
Proof

>>58017178
>less is better
are you trying to be retarded?

>>58017031

Well at the very least ramping up can't mean adjusting clocks as AMD demonstrated zen locked to 3.4ghz.

File: BROADWELL-E-41.jpg (149KB, 614x523px) Image search: [Google]

149KB, 614x523px

>>58017178
Piledriver doesn't even compete with Sandy Bridge 4modules vs 4 cores without HT.
Blender isn't a cherrypicked AMD favoring benchmark. It isn't a synthetic, its a real world workload, and it heavily leverages FPU ops that the entire BD family of architecture did terribly in.

Intel shills are totally BTFO over Ryzen performing so well in Blender.

>>58017189
You're illiterate.

>>58017217

Then I guess handbrake must be AMD cherrypicking because integer workloads are a strong point for bull- i'm sorry, I can't keep a straight face while typing that.

>Intel shills are totally BTFO over Ryzen performing so well in Blender.

Pretty much.

any price leaks?

>>58009109
This

>>58009042
>Woman as CEO
No wounder their drivers are terrible.

>>58017592
Nothing substantiated, but its been expected for a long time that the top end SKU would sell for around $500.
Some Chink "leak" showed as much, though it was nothing but a collection of rumors and already known info, so its entirely baseless.

Given how they positioned it for streaming I don't think they're going to go near the upper i7E prices. It'll probably be between $350 and $500.

File: 1453195944945.jpg (43KB, 797x799px) Image search: [Google]

43KB, 797x799px

>>58016997
I think he meant from Lisa Su onwards.

There is not much AMD could have done on the CPU segment without Zen, as Bulldozer and derivates were a disaster, and they didn't quite have the money to force OEMs to use it, like Intel did with Pentium 4.

They have recovered a bit on the GPU segment though, with a different approach to it.

>>58017178
You are fucking retarded

>>58018113
butthurt intel shill

>>58016813
>Just up the TDP to 140W like the i7 6900K.

The Broadwell-E chips only have higher TDPs from their massive (but rarely active) AVX units, and integer workloads won't even come close to them.

Zen is decent, but it's not like magically 33% more energy efficient than Broadwell.

>>58012610
This. My 955BE from 8 years ago still eats the i3-6100 for breakfast.

Battlefield 1 is very CPU intensive.

By the way, some guy that was in the presentation and got to play BF1 on both Intel and Zen systems commented that the Zen system had FPS drops below the 60s high 50s unlike Intel which drops to the 60s and staying in the 70s mostly.
Both systems were running Pascal Titan X

It looks like it will have lower single core performance.

>>58019160
There is more to the TDP disparity than just AVX. The Broadwell-E line typically pull 100-110w under a nominal heavy load. Big AVX will stress them to 120-130 and it will drop clock rates.
The logic is simply dense, and takes more heat sink to cool effectively, so TDP rating is higher. Lighter workloads they can pull 90w~

Zen based parts do have lower TDP, but Zen cores are also smaller than intel's core i arch. Intel has native 256bit datapaths. Zen is only 128bit. The FPU isn't nearly as big.
Its just a smaller core. Though its still the biggest thing AMD has ever made.

File: Road-to-Zen_Architecture-Brief-6-1440x811.png (225KB, 1440x811px) Image search: [Google]

225KB, 1440x811px

>>58020471
>Intel has native 256bit datapaths. Zen is only 128bit.

which datapaths are you even talking about?
Everything I saw for Zen showed 32B/clock transfers except L1D loads.

File: 7-1080.1262858105.png (101KB, 1920x1080px) Image search: [Google]

101KB, 1920x1080px

>>58020850
>cache
LOL I see you're confused.
I'm referring the the FPU itself. Zen has two 128bit FMACs.
Intel's latest core i arch can natively handle 256bit ops. Zen cannot do the same, it has to take a big performance hit by spreading them across the two 128bit FMACs.

>>58020850
>>58020909
See this is why I preferred the real world software demo. I have no fucking clue what any of this means, and I couldn't give two shits either.

Does it run the programs I intend to use better than an i7 for the same price?

Yes - I buy it. No - I buy Intel.

>>58010673
Being that fat is suicide

>>58020909
people don't usually refer to ALUs as datapaths, put whatever floats your boat, champ.

>>58009042
Software should always be designed to utilise multiple cores, ever wonder why servers and supercomputers mostly use Linux? Because Linux offers better performance in multiple core systems, for commercial systems evolution has completely disregarded single core perf., Intel and brainwashed gamers are obsessed with a false economy of improving single core performance encouraged by lazy game design,

multiple cores has greater potential than single core optimisation, only lazy programming holds the current dynamic in place, it will be broken soon since both major consoles are following multiple cores now, themselves able to detect the bullshit of single core

>>58021100
My grandfather was that fat until he was 87 years old and smoking killed him, a habit he had since he was 16. I was obese all my life, and my arteries were tested totally clean a few months ago, and my blood tested like that of a sportsman were it not for the high uric acid content.

Don't believe everything jew doctors tell you.

>>58021114
Its not just ALUs, champ. Its everything in the pipeline that feeds instructions to executing logic from decode to retire.
The entire physical path that said ops take. The datapath.

Google the term and learn something.

>>58021175
But the doctors told you there was nothing wrong with you?

>>58021213
Tests told me there was nothing wrong with me. Values on a paper do not lie. Doctors who tell you being fat is a health risk just by looking at you do.

WHY MUST EVERYTHING BE SO STUPID NOWADAYS

I REALLY NEED TO PUNCH SOMETHING SOON

>>58010957
awww a new pepe how cuuuuuuute

>>58016489
Tbh Hardocp was on point with the 480 RX prediction.

>>58012638
Not everyone has a dedicated encoding/capture rig. You'd be mental to encode in software on the same system that you're using to play games. It's just asking for frame drops.

>>58021304
>4 terabytes of captures in OBS the last 4 years
>0 frames dropped
At least you tried...

>>58021325
Yea okay, maybe if you play some games which don't use the CPU at all. Even I get 15% CPU usage with some really simple settings.

File: 15_kek.jpg (38KB, 313x218px) Image search: [Google]

38KB, 313x218px

>>58021526
Replace your shit CPU then. Chrome with shockwave eats more CPU than OBS.

>>58021565
Okay, I'll get right on buying a 1800 dollar CPU.

>>58021590
Don't you see the arithmetic here? If Chrome eats way more CPU with a shitty flash plugin than OBS does, then you are doing something very wrong in OBS. This is a core i5 on base clock for god's sake.

>>58021643
Then why don't you reveal your magical settings that not only use low CPU but also produce good results. An i5 at base clocks will drops frames like crazy on SW encoder if you run GTA5 BTW. The game alone will tap out 100% on all 4 cores.

>>58021565
>>58021643
Retard

>>58021725
No it doesn't, it runs with 30-60fps on an i3 dualcore with highest texture settings, rest medium and if you're too jewish to buy a broadcasting software with GPU encoding support then it's just your own fault.

I tolerate a lot, but talking shit about GTA V? Not on our watch, kiddo.

>inb4 hurr durr what are multipliers

>>58021725
Maybe you should stop trying to deliver the same thing you see on your monitor to the stream? I mean give me a break, it's a fucking stream. You don't need to show people what 60FPS with 1440p feels like in a fucking window that is not even fullHD in size. Do you know how many people watch streams in full screen?

Zero point fucking zero something percent.

>>58009042
>4k gayming meme

I want it to stop.

>>58021818
We are talking about CPU encoding here scumbag. I can encode on GPU just fine if I want to.

And yea it runs 30-60 FPS.. if you turn the CPU intensive settings down.

>>58021827
I've never watched a stream not full screen.

>>58021845
>>58021845
>We are talking about CPU encoding here
How fucking stupid are you?

>he has no quicksync

lmaoing@your life

>>58021862
The fuck is your problem, retard? Why do you butt into conversations you have nothing to do with and which don't interest you? You fucking autistic or something?

you all are retards. i want amd to do extremely well with zen, just so intel has to do actual something rather than just the bare minimum.

for too long intel has done just enough, so they produce shit slightly better each time. if amd delivers, i want intel to go full fucking force.

and not even considering how prices would drop. everything is looking bright.

>>58021238
>Tests told me there was nothing wrong with me.
Tests that doctors did for you, moron.

Also, it doesn't matter if you think you're healthy. You're still a fat sack of shit.

>>58021925
Because I have no respect for you, shithead.

>>58021070
>I have no fucking clue what any of this means, and I couldn't give two shits either.

They're arguing about terminology for the floating-point SIMD for the processors.
It barely matters since more and more things are moving to GPU offload (for fp32) at least.

>>58021955
More like you're trying to save face because you have no idea what was being talked about until it was pointed out to you.

>>58021939
If you can't see the difference between drawing conclusions from a personalized test result, and someone just generally assuming something and forcibly trying to apply it to you without ever having looked at your test results, then I am sorry for you.

You deserve your Dr. Shekelberg.

>>58017178
>>58017217
Less is better. Those both show the AMD CPU's doing the worst of all the tested ones for those graphs.

Retard.

>>58022003
How are you this fucking illiterate?
How is everyone in this thread so illiterate?

>>58020266
>Battlefield 1 is very CPU intensive.
yeah nah its not my i5 can handle it just fine its more gpu intensive

>>58021827
>Do you know how many people watch streams in full screen?

That's how I watch them.

>>58009042
Intel shills in full damage control mode.

>>58009042
You are just a sexist asshole

Hope Zen/Ryzen will be competitive in performance and in price. Best of luck, Lisa!

>>58022021
You sat there saying Bulldozer benchmarks were cherrypicked, and your only proof are benchmark charts showing it doing worse than everything else.

So you are the mongoloid here apparently.

>>58022021
Welcome to the interwebz, where people with a tenuous grasp of english from all over the world will try to argue with you, whether you like it or not.

>>58022055
As if a set of people of a specific community where the multi-screen meme spreads like wildfire; where people have two dozen screens because fuck window switching would be the 99,999%. Think again, you're that tiny smidge that has more money than sense, because having a dedicated screen for your Spotify playlist and another one for your streams is something you absolutely can't live without...

>>58022087
Good job proving how retarded you are.
I explicitly stated that AMD didn't cherrypick benches, because Blender has never favored their arch. The whole Bulldozer family of arch performs poorly in Blender.
I then supplied to images to back up my claim. Again in my third post explicitly stating my position, which couldn't possibly be any more clear and direct.

Your IQ is too laughably low to follow a few simple posts. You're an honest to god special ed retard.

>>58022113
>you're that tiny smidge that has more money than sense

I'm the guy that didn't throw out his old monitor when he upgraded to a new one.

One of my 3 monitors is my old 5:4 1280x1024 monitor I used through the 2000's

>>58022135
>Remember how AMD cherrypicked Blender benchmarks with Bulldozer too?

>>58022145
That wasn't my post, special ed.
I responded to that post saying that it never happened.

Again you're just proving you're a legitimate, in the flesh, drooling retard.

look how fucking mad anon is

top! kek!

>>58022156
That's your fault then. You posted that proof to your own post as if it was the other guy trying to call you out.

>>58022214
Its not my fault that you're too much of a malformed subhuman to coherently follow a few simple posts. The benches I posted directly reflected the statements I made. Anyone with a nominal IQ over 80 could see that.

I bet you're a fucking Brazilian. Only a third world shitter could be this dumb and still manage to be on the internet.

Could be far worse, at least they're not infested with social justice jews.

>>58022241
Not my problem you don't know how to link posts in an anonymous image board.

>bad multicore benchmarks
stopped reading there

>>58009042
She looks like a man with lipstick??

>>58022404
Welcome to 99% of all Asian people.
Unless they get tons of cosmetic surgery they're usually rough looking bitches.

>>58009042
Why did you use a picture of Iwata?

>>58009042
>hardware encoding
>wanting an objectively inferior quality image

>>58012638

Its still shit in quality wise. you need high amount of bitrate to achieve SW quality with HW encoding.

>>58021304

Its all depend on the game you play.
If the game you stream doesn't hog all your CPU core then it won't be a problem.

Most of streamer used i7 with low bitrate SW encoding.
The quality still looks far more better than average 6Mb bitrate with HW encoding and that bitrate is what youtube recommended for 1080p60fps stream.
At 6Mb with HW you got artifact in any fast pace games like first person shooter.

I was streaming an esport tournament for a local event and with one PC solution and HW AMF encoding, I need at least 10Mb to make it looks decent.

>>58022664
>Its still shit in quality wise. you need high amount of bitrate to achieve SW quality with HW encoding.

was meant for >>58015398

>>58016526
>real world = muh gaymes
>>>/v/

>>58009042

>this is a female in Asia
No wonder you all have to fap to anime schoolgrills

>>58023153
I'm kind of glad I'm a dude

I can be ugly and nobody gives a flying fuck

>>58016617
>reddit
Oh, so that's where all these retards and shills are coming from.

>>58023138
Oh wow, you figured out who the vast majority of the market for these products is.

/clap

File: 1481852983460.gif (2MB, 700x394px) Image search: [Google]

2MB, 700x394px

>>58023153

>>58015556
Sorry to shatter your dreams but your 2600k isnt a 6900k

>>58009042

Your alt-right white nationalist is showing.

>>58024254

no doubt, all that means is AMD found a way to misrepresent performance of both chips because the zen chip couldn't outperform any of intel's offerings on a level playing field.

>>58024357
When using the same settings that AMD used in the demo its performing right on par with Haswell-E and Broadwell-E systems from users on Anandtech.

https://forums.anandtech.com/threads/post-your-ryzen-blender-demo-scores-amd-clarifies-blender-benchmark-confusion-run-150-samples.2494600/

Though I'm sure a buttmad little shill like you will still find a way to contort the truth to fit your childish fanboy narrative.

>>58024421

except AMD never showed the settings used for the render on the stream, all we know is that a random AMD employee said 150 samples. very likely that they used different settings or outright faked the result given that we already know zen is sub-sandybridge IPC from the AotS benchmark leak.

>>58024457
WRONG. They showed the settings on screen before the test was run. You could clearly see them.
There you go spreading childish bullshit lies, exactly as expected.

>>58022113
Monitors are cheap as fuck dude. 1080p for less than 100.

>>58024467
>WRONG. They showed the settings on screen before the test was run. You could clearly see them.

they only showed settings for the power consumption demo, which was 100 samples.

>There you go spreading childish bullshit lies, exactly as expected.

being skeptical about the validity of a benchmark done by AMD (who should be making their product look unrealistically good just like Intel and NVIDIA would) is 'childish' and 'lies'? are you sure you're sane?

>>58009042
Who is that semen demon

>>58024507
Pretending like a questionable 2.8ghz early engineering sample is indicative of final silicon performance is something only a little shitposting child shill would do.
Pretending like a shit tier game bench is a tell all for performance, but multiple Blender runs are faked just screams /v/irgin shitposter.

Lie and contort the truth, faggot shill.

>>58024563
>Pretending like a questionable 2.8ghz early engineering sample is indicative of final silicon performance is something only a little shitposting child shill would do.

never claimed it was indicative of final performance

>Pretending like a shit tier game bench is a tell all for performance, but multiple Blender runs are faked just screams /v/irgin shitposter.

not taking AMD's word at face value isn't shitposting, it's common sense. waiting for independent reviews and benchmarks is what a logical personal would do.

>Lie and contort the truth, faggot shill.

didn't know i was a shill for not unconditionally supporting your favorite company.

>>58024507
For fuck's sake.
The benchmark isn't fake, it's cherrypicked.
Learn the difference.

Why 6mbit 1080p video isn't enough for shitty streaming platform like Twitch?

>>58024648
>The benchmark isn't fake, it's cherrypicked.

semantics. the real world performance will not be the same as the demoed performance, which is all that matters.

>>58024629
>that we already know zen is sub-sandybridge IPC from the AotS benchmark leak.
Those are your words. Speaking as if we factually know final performance from an ES chip in a game. Trying to pass off a lie as an accepted fact. Good job contradicting yourself, child faggot shill.

>>58024648
Blender has never favored AMD's arch. Saying its cherrypicked is just as fallacious and childish.

>>58024673
>Trying to pass off a lie as an accepted fact.

so now you're claiming the ES leaks never happened? fact is that several independent benchmarks showed zen as between westmere and sandy bridge IPC, whereas AMD's idealistic demo showed them near haslel IPC. most likely the final result will be somewhere in the middle, which is abysmal.

>>58024673
>Saying its cherrypicked is just as fallacious and childish.
It's cherrypicked until proven otherwise.
How do you prove it otherwise? By having an entire benchmark suite, many programs, many test situations.

It's a single benchmark tool, ergo it is cherrypicked.
Who knows, they may have fucked up and accidentally released a representative benchmark.
But unless you think AMD is incompetent they didn't.
Or if you think AMD would purposefully underrepresent their own fucking product on their own release announcement.

It's bewildering why you trust any company's own release, it's like you were born yesterday.

>>58021175
Good goy, buy our Mcdonalds and Coke! You are healthy like virile sportsman!

>>58021159
yes and no. some things are easy to make parallel, some things are not.

lazy programmers, not so much, incompetent ones that get driven out when software works followed by maintenance programmers more likely.

>>58021839
I want it to go full swing ahead, will be nice to stagnate graphics again while im on 1080 144 and have no need to upgrade period for a number of years.

>>58022467
Anyone know why hardware encoding shits out worse image quality?

>>58022003
>>58017189
didn't he just post proof that bulldozer did shit in blender?

>>58017592
>>58017722
current leaks point 350 and 500 for 2 skus at 8 core, but honestly I don't believe them as people started thinking there would be 2 8 core skus and magically we get several unsourceable part lists that list benchmarks and prices? some of them pointing it to being equal to an intel 6 core shortly after the rumor mill put 8 core zen to 6 core intel on the supposed price.

the best we have is they were showing us an 8 core cpu at a fan event.

they showed us games, and they played this up to gamers, so there isnt a fucking chance it will be the 1000$ range no matter if it crushed a 6900 or not.

the streaming table was the most interesting for price though. they pair a 6900 a zen and a 6700 up against each other... why? why pair up a 6700? why not an i5 or an i3? What I believe it was there to do was say, you are getting this (6900) performance for (6700) price range.

They also made a BIG point to say that 3.4 was the minimum you can expect, so a higher up sku is likely.

>>58025025
Pepsi CEO doesn't drink pepsi, Mcdonalds CEO doesn't eat at mcdonalds, and KRAFT CEO doesn't eat processed cheese, and Tobacco company CEOs don't smoke.
Least off camera, anyway.
They run regularly, too.

You know what the biggest health problem affecting ceos and bigwigs of these fast food companies is?
Knee cartilage damage from running too much.

>>58020266
unless you have the exact same scene benching, however, just found this.
http://www.gamersnexus.net/game-bench/2673-battlefield-1-cpu-benchmark-dx11-vs-dx12-i5-i7-fx/page-2

you should be looking at a 6900 to push 107 fps at minimum, hell, amds 8370 pushes 66 fps as a minimum, there is no reason the game went sub 60 in the demo unless the titan was shitting itself, which at 4k i would expect.

Honestly seems like someone is lying, or they stupidly used a dx12 with not only nvidia parts, but using dx12 inb bf when its basicly broken (dont know if they ever fixed it)

>>58021936
I want amd to do well just so I can see if there IS a performance wall we are hitting.

>>58025228
Looked more like the guy before him trying to refute him with a benchmark that showed the opposite of what he was intending.

>>58025362
yea, I can see that, but I also see the same person finding some benches and posting them as proof they didn't cherry pick.

>>58024652
Personally would go 720p and 60, but thats my internet being shit and 2.5 would be the best i could pull out. less shit pixels the better.

>>58025276
>They run regularly, too.
>You know what the biggest health problem affecting ceos and bigwigs of these fast food companies is?
>Knee cartilage damage from running too much.

Good. When the proletariat rise up against the bourgeoisie they are going to need that skill. Nice to see them planning ahead.

>>58025351
>there is no reason the game went sub 60 in the demo

It's BF1, that games broken performance wise still and CPU limited.

>>58022072
oh yeah. SEXISM is the issue here, right? fucking moron

>>58015398
Pure shit. There's not a single hardware decoder that comes close even to single-pass x264

They're in a different order of magnitude altogether.

>>58021304
Normal people have multi-core systems, and most games still run on 1-2 cores at most. You can very easily dedicated a core on a quad core system to x264. But even that might be overkill, since x264 with a fast preset is very light on the resources and still better than any HW decoder ever produced.

>>58009042
no multicore benchmarks...
handbrake literally is the definition of multicore benchmarks..
no single core benchmarks..
blender is literally one of the best single core per ipc benchmarks
playing battlefied to showcase vega who knew amd wanted to shocase their new cpu with their new gpu SHAME ON THEM

because we all know that a 6700 can handle a 1080p game PLUS encoding PLUS streaming..
shame to every serious streamer that has a dedicated encoding and streaming pc...

typical /g/ user

>>58025850
They showcased Vega on Battlefront actually. Not even with a particularly graphically demanding scene either.

>>58009042
Shalom, friend. Is this the meeting place?

>>58020909
>128bit FMAC
are you literally stupid? amd literally told that the blender test was on avx 256 built...
also i dont really think you understand how modular the desing amd created is..

this is literally the only good think bulldozer had they could fuse pause or switch the fpus midcycle on the fly hence the good results on the heavy multithread scenarios..

AMD is utter shit

>>58025955
Eh when you can get 6900k level performance for nearly the price of a 6700k, maybe not.

Lolmd doesn't even beat an i3 in single core performance!

Who cares? Do you use single core software like Russian games?

>>58025392
>Good. When I get my fatmobile they are going to need that skill. Nice to see them planning ahead.
>>58021987
>I am healthy at every size, one of my head-mates is actually a doctor, xir can confirm this. You are all idiots, and probably racists/homophobes as well, for believing the lies of the fat-shaming healthcare industry.
kek at this fatty

>>58015716
You're still under the assumption that their major customers are buying one CPU at a time for a simple system.

You're not clever in the slightest. Do you seriously think some enterprise customer of theirs wouldn't know what they're buying and what their systems need?

Do you honestly think Google signed a contract with them because 'OMG AMD's CPUS GO AboVE 4 JIGAHERDS!!!11".

Buzzwords in technology only work on consumer retards.

>>58026192
well they named their HT treadripper...
i suggested their boost feature to be named "obliterationizer"

>>58025926
>I'm a shit eating moron and I don't know what I'm saying
>I'm also entirely incapable of reading

The FPU in the Zen core is two 128bit FMACs. They do not have the throughput potential of a native 256bit datapath.

>>58026265
i think you need to go back and read how the fpus on zen works...

>>58015716
kek implying that ONLY AMD uses that market strategy

go away

>>58026282
Stop pretending like you know what you're talking about. Or you can just go right back to plebbit with your tech illiterate ellipses shitposting.

Spliting larger FPU ops over two FMACs is not a new concept, and it is not one for performance. It is saving die area and power consumption while taking a substantial performance hit. It takes more cycles to process the same op. This will never change no matter the implementation. The Zen core's FPU cannot compete with the larger FPU in the core i arch.

>>58026310
some times i think im talking to apes around here
go and read what AMD did with it before you actually think that "its the same as before"

>>58010957
>short hair is lesbian
go back to texas, cletus

>>58026353
You can't bluff your way through a conversation with someone in EE, kid.
Stop pretending like you know what you're talking about.

>>58026375
yeap you got me finally i dont know how dynamic the new system is despite the fact that amd already told us so
but you know your truth is always the right truth

File: 11-1080.1838658295.png (81KB, 1916x1080px) Image search: [Google]

81KB, 1916x1080px

>>58026422
Stop talking out of your ass, and actually educate yourself.
https://www.youtube.com/watch?v=sT1fEohOOQ0

Using vague words and trying to make yourself sound informed is the most pathetically transparent bullshit ever. AMD moved on from the FlexFPU because there was no longer a need, but that doesn't change how segmented FPU resources operated when joined to handle a single large op. This is a principle more certain than gravity. You cannot pull blood from a stone. Separated data paths take more cycles to process a single op than a data path of native width all else being equal.

You're too dumb to fool anyone, even on /g/ you stick out like a sore thumb.

File: 12-1080.2968623445.png (72KB, 1920x1079px) Image search: [Google]

72KB, 1920x1079px

>>58026476
wrong slide

>>58026364
Short hair is not exactly indicative of healthy female sexuality, Jason.

>>58009109
At least they are going to force Intel to innovate (which they won't be able to because they are lazy fucks too reliant on the tick tock marketing). Just matching Sandy Bridge but at a fraction of the cost would undercut Intel to the point of them forfeiting the CPU race but would harm long term development as AMD becomes what Intel is now. (unless Intel has fleeced us all and kept a secret all this time)

HOW ABOUT YOU ACTUALLY FUCKING SHOW ME THE FPS AND SCORES INSTEAD OF SHOWING ME TWO FUCKING SCREENS WITH NO ACTUAL CONTEXT.

What I can leanr from this thread is that blender and hanbrake don't count and in no way show what a cpu is capable of (let us not forget Intel uses blander to show off their cpus).

Also: styreaming is fine o na 6700k as its more than you will ever need despite it being well known a lot of top streamers have a 2nd machine entirely dedictaed just to streaming alone making ryzen and the 6900k good choices for the job.

>Cocktease Zen since their last huge failiure for almost 3+ years now
>Finally finish it
>O-oh we have to delay it to put 'finishing touches'
>Finally Demo it
>Dont provide actual benchmarks or meaningful tests
>Just BF1 a GPU heavy game thats heavily multi threaded
>No Single Thread/IPC tests
Its going to be junk isnt it?

I cant wait for the first price drop.

Where the fuck is Vega?

>>58025399
not at 4k, that is a gpu issue, not a cpu.

>>58020909
Why don't the MUL and ADD on the FPU have their own schedulers like the integer ALU and AGUs do?

>>58026037
single core literally just needs to be good enough, and guess what, the worst case scenario for zen is sandybridge, and guess what processor had more than good enough single core.

>>58009042
> VR
> playing battlefield
> dota streamer
> female nigger "gamer" playing battlefront
/v/ pls.

>>58026689
>meant to be a multi core based CPU
/v/irgins complaining....
again.

I have a freesync monitor, and I am not buying a Nvidia card, neither will I buy the fucking useless RX480.


>no Vega consumer chips until March/April next year

Fuck off

I don't even know what gaymes you are playing. What kind of game still use just one core?