INTEL BTFO

INTEL BTFO
MEMELAKE IS FINISHED

Could you be more of an inflammatory faggot?

Intel will rerelease Memalake with multithreading enabled i5 @4.5Ghz
that's their only way to compete

Lets wait til we actually see how it performs in actual games eh

With most games, definitely not as good as synthetic seeing as core performance isn't very good.

>3.3, 3.2 and 4.0 GHz have the same per-core performance


|
|>
|
|3
|

I'm hoping that's driver and chipset maturity, unless someone has a better idea.

How about its performance with running 3D programs, game engines, mining cryptos and other stuff aside from video meme games?

Nobody cares. There are better cards for 3D and mining is dead.

>ryzen 3.2ghz BTFO 6900k 3.2Ghz
>6800k 3.4Ghz BTFO Ryzen 4.0 Ghz
???????

that chart is just the overall physics score divided by the number of cores. There is no single core benchmark in firestrike.
the suspiciously perfect scaling might be an inidcator that the Zen scores are faked to begin with.

That's called an i7, dipshit

Not for long, fuccboi

You missed the news of Intel readying a Kabylake-X i5 SKU with SMT enabled and housefire TDP.

Don't forget the days of engineering sample excuses are long gone
If it's a shit clock it's a shit clock, this is what AMD is putting on the market

KabylakeX is gonna be on x99 right?
1151 can't actually go past 4 cores and 2011-3 CPUs already have housefire tdps for their housecooler heatsinks

...

im pretty sure 1151 can go past 4 cores its just that intel are jews

Kaby X will be LGA 2066

A socket change a year keeps the goyim in fear

Apparently KabylakeX tops out at 4 cores as well for whatever reason.

And for whatever reason I have no idea why anyone would buy it compared to the 1151 socket, quad channel memory? Would that even affect anything with only 4 cores?

15% improvement soon lads

Is pretty fucking accurate, AMD working their asses off and still making crap, and intel making all the money

15% uplift *in Sysmark*
Kaby Lake also showed a 15% uplift over Skylake in Sysmark, so that looks incredibly disastrously bad for intel. Its basically the same as saying
>Look, we have absolutely fucking nothing, all over again!

...

Thank you based Jim

Mfw Zen is released and all real world tests with compression, encoding, decoding, gaming and physics in general will be better than Intel.

Even if 1800X turns out to be 80% of the performance of 6950X for 50% of the price, I'll buy one.

Just purchased a 6600k rip

I can't say that I know the limits of 1151 pins on a socket but I know Intel said their 1151 socket only supports quad cores
I bet there isn't a single 5+ core CPU on 1150, 1155, or 1151, my point is I doubt Intel would go back on their word when all they have to do is release a new socket next year

If the 1700/1700x/1800x are all 8/16 core threads is there any reason why anybody wouldnt just get the 319$ cpu over the 499 one and just overclock the hell out of it since all Zen cpus are unlocked from the start?

>15%
>nothing
Congrats, enjoy never feeling like your CPU is too slow for 8 more years
Unless you need multicore but that was a decision you knew long ago, irrelevant of brand loyalty

That's nice and all but what are the temperatures?

15% in one synthetic benchmark that is compiled to favor your arch is nothing. Skylake and Kaby Lake are less than 1% apart in IPC, and just a few days ago during an intel keynote an engineer stated that they are internally the same arch. The Kaby Lake core *is* Skylake only with some hardware bugs fixed.
Intel stating that their new arch will be another 15% uplift in the same synthetic bench is saying that they have nothing. 15% uplift in Sysmark is actually fucking nothing.

>few months ago by AMD
LOOK OUR IPC IS SIMILAR TO BROADWELL LOOK EVERYONE LOOK
>a bit latter by frogs
LOOK EVERYONE WE TESTED WHAT YOU ALL WANT ZEN'S IPC IS SIMILAR TO BROADWELL LOOK EVERYONE LOOK
>now
LOOK EVERYONE WE TESTED FOR WHAT EVERYONE'S WAITING FOR ZEN'S IPC IS SIMILAR TO BROADWELL LOOK EVERYONE LOOK

I FUCKING GET IT ZEN'S IPC IS SIMILAR TO BROADWELL JESUS FUCK

I'm not favoring anything, I'm saying that a 15% increase (and this is CPU, synthetic or not, it's relevant) isn't nothing, especially when the official statement from years before Kaby lake existed is that Kaby lake is basically skylake, but apparently 15% better according to you
You can't say a CPU gen is 15% better and then ridicule it

Uhh, he's saying the intel slide of 15% improvement is bullshit because it is bullshit, Kaby and Skylake are literally the same shit and Intel wants to sell you that one is faster than the other.

>synthetic CPU benchmarks
You mean non-gaming, boring old-man multicore performance?

lol

It's the same guy.

it's a prank

DELETE THIS!

>tfw upgrading my 2500k to Zen next month
>tfw upgrading my 780Ti to VEGA in a few months
The future is Red my friends.

You sound like the kinda guy who would actually believe the geekbench scores and declare ARM the best architecture on the planet, beating everything else ever made. Protip, not all benches are equal.

geekbench scores are valid for the same architecture, not for arm vs x86 you autist...
you sound like hte kind of guy who thinks he knows everything when in reality you're just another dumbass looking at the same charts we all are

You mean the same ISA, not architecture.

Geekebench isn't valid because it tests a bunch of SIMD some arches have and some don't, pulling them in an aggregate score which literally means nothing after that.

But geekbench is a subjective benchmark, comparing systems to systems in arguably CPU-dependent benchmarks.
You're saying that two CPUs with the exact same clock and core and architecture can perform differently in a benchmark and that's not relevant?

>but apparently 15% better according to you
15% better in precisely one benchmark
0-1% better in literally everything else.

Yes he means ISA
So it is valid to use these benchmarks to compare any modern Intel or AMD CPU against each other?

But if they're just generic x86 workhorses doing generic x86 tasks, then what makes one benchmark so special where the raw maxed multicore CPU power is so much better on a CPU that is the same?

It will be literally retard to keep being NVIDIA or Intel after 2017

Wow, you can't even read.
Sysmark is compiled specifically to favor intel's arch, they even listed it in a footnote on their marketing slide. A 15% uplift in Sysmark is nothing. Literally nothing.
Skylake to Kaby Lake was nothing.
The generation of arch coming after Kaby Lake is promising the same nothing performance uplift.

Hes genuinely too stupid to understand.

Kaby Lake wasn't an architecture change though

I'll stick to my E5-2620v4 Sure it's only 3GHz but quad channel is important to my usecase.

Geekbench? Yes, but only if both support the same SIMD geekbench tests.

Even with that Geekbench is flaky at best, a lot of microbenches that in no way simulate real life performance.
You should test it with a test suite of at least another few benchmarks, its score alone isn't worth much, a CPU can't be tested with a few scripted tests.

>You mean the same ISA, not architecture
>Geekebench isn't valid because it tests a bunch of SIMD some arches have and some don't
>You mean the same ISA, not architecture

>Intel still on top
The only thing that matters.

>10 core, 140W TDP, $1700

I love context.
Wait til the 32 core Naples get tested against the 24 core Broadwell, that should be hilarious.

Mainly because it's a 10 core 20 thread part vs the AMD one below it which is a 8 core 16 thread part.

But what about the 72 core Xeon?

>Zen+
Man, 2026 will be a great year for AMD

AMD is hotter. Waste more energy. Fuck AMD.

That's an accelerator, running fucking Silvermont cores.

SILVERMONT CORES, jesus.

>5% more performance for only 1200$ more! such a good deal goyim get it now!

Intel has 32 core Skylake Xeons, not sure about their availability, but they're a clear response to Zen.
Though they're absurdly priced Xeon came it at a 2.1ghz, and word is that Naples clocks a lot higher at 180w.

That is a Phi.
Its small Atom cores made to do nothing but process AVX.

>availability
Considering Intel just released their NEWEST AND GREATEST 24 core Broadwell I don't think we'll see that 32 core Skylake until late Q3 or Q4

>clearly a response to zen
>selling xeons and focusing on multicore performance for xeons since 2007
Shit, Intel has had multicore on lock for a while now, a few years ago their multicore xeons started getting close to their core-series IPC.
You know, years before zen was announced

Try agian, little shill.
Intel didn't make the jump from a 24c/48t Broadwell to 32c/64t Skylake by coincidence.
They did it because AMD's Naples is 32c/64t.

Skylake Xeons are a direct response to Zen.

Zen+ won't exist for many many years though.
You can't "respond" to nothing, that's what people on Sup Forums do

Zen is a direct response to Coffee Lake
AMD must be scared shitless to be releasing something like that

Naples is a direct response to optical non-silicon processing

>95W AMD vs. 140W Intel
>AMD wastes energy

umm... sure

Intel has a bigger high-end turbo and performance range
I mean sure if it topped out at the performance level of zen of course it would be consuming much less power, but it's better, goes farther

Geekbench 4 is relevent as fuck these days. It tests a lot of both synthetic and real world CPU-based tasks. ARM and x86 both run the exact same tests too.
You have to look past the initial score and you can see the figures of each individual sub-test.

There hasn't been a better CPU benchmark.
Now queue that blog post by Linus ragging on the older GB2 and GB3, they were rightfully shit though, but GB4 fixed all of the issues.

>I mean sure if it topped out at the performance level of zen of course it would be consuming much less power, but it's better, goes farther
p r o v e i t
r
o
v
e
i
t

What do I need to process to get my CPU running at a "synthetic benchmark level"?

Read the AMDtech graph of them crying, cucked in the corner

>Zen+ won't exist for many many years though.
AMD is releasing new arch every year. So your statement is flat out false.

>You can't "respond" to nothing, that's what people on Sup Forums do
Are you brain damaged, or just clueless.
Naples is Zen, not Zen+. Naples Opterons are releasing shortly this year.

Coffee Lake is actually a mainstream response to the lower binned Ryzen chips. Intel can't address the multithreaded performance of a 6c/12t Ryzen SKU so they had to tack two more cores on to Kaby Lake.

Intel has no new core arch coming out until 2021 at the earliest.

Passmark, kek.

A bag of dicks, because synthetic benchmarks are full of fucking shit.

No single application runs the entire frontend of a modern core at 100%.
Literally
Not
A
Single
One

There literally does not exist a single useful application that runs the FPU, ALUs, fetchers, prefetchers, predictor, caches and decoders at full tilt at the same time, that's fucking nonsensical and if such a program exists it won't be anything but 20 inline loops running infinity.

In conclusion, synthetics are shit

Then why the fuck does a lot of users here on Sup Forums keep quoting Passmark as the best CPU test when the clearly superior Geekbench 4 exists?

Forkbomb!

But programs do run each one right? These decoders and shit do have functions right?
So why is a cumulative test wrong?
If it's "stressful" on the CPU isn't that a good indication of its minimum single-application performance?

Autism: the post

Because people are fucking idiots, how do you not know this when you've been on this board longer than a week.

Personally I think a laptop CPU boosting to 4GHz for a single-core application is a bigger lie than a CPU score being gimped by running at 100%

>not just sticking with 4790k and DDR3 until all of this stabilizes a bit

Also HDMI 2.1 (integrated variable refresh rate) going to make gsync/freesync obsolete but both the GPU/Monitor need HDMI 2.1... so I hope you lads arent wasting money "upgrading".

Want a car analogy? synthetics are like redlining your car over 10km in first gear.
Normal stuff like gaming is regular traffic with dropping from fifth/fourth gear to third and overtaking someone from time to time.

But retards here will call AMD shit even if the one processor that beats it is something they'd never actually buy.

>want a car analogy?
No
Nobody does
Not a single person on Sup Forums thinks your car analogies are anything other than flat out retarded.
Like 100% complete fucking retard shitposting
Continue on with your argument, omitting the fucking "im a giant dipshit everybody look at me!" Analogies

Too bad, I like my car analogies.

Synthetics usually test a single feature isolated, or an abstracted performance metric even harder to draw parallel to. It would be like measuring how fast the wheels themselves could rotate before becoming unstable.

>These wheels can roll an astounding 300mph!
>The car can accelerate to 300mph?
>No, but the wheels could theoretically roll that fast!

No I don't
Car analogies for tech are fucking stupid I don't see why people think anybody wants them.
What programs are "high gear optimized"? What the fuck does that even mean?

You're reading too much into it, if you drove a car it should be easy to understand that the first scenario is wasteful and completely unrealistic.

Actually cpus are more like self driving electric cars because they don't have gears, have nobody to pass, can't be controlled or directed for more power, and don't use the "power" in ratio to output "speed"
Come to think of computers are nothing like cars, weird concept

No actually I didn't read into it at all
I saw the first sentence then completely disregarded everything you said as ignorant shitposting, like I do with every technology=car analogy

But what program is the first scenario? And why?
Does a benchmark make my CPU slow? Does using the CPU wear it out?
Can't you shift gears in a video game and suddenly use half as much "power" and keep going the same "speed"? Video games don't have to use any CPU power from what you're saying

And that's why you're a fucking retard. Case closed.

>trunk closed

He was saying that synthetics use too much of your CPU, not that they use too little of your CPU
I'm trying to figure out why he thinks a holistic test is worse than benchmarking each little individual feature

The first scenario is all cylinders firing like lasers at insane RPM, something like wPrime or IBT
^
That's not close to real workload.
A benchmark(synthetic) makes your CPU run at its theoretical limit.

Your computer is slow because you don't got one of these bad boys