FUCK AMD INTEL RULEZ

I can't wait for Skylake v4 in Q2 2018!

Other urls found in this thread:

danluu.com/cpu-bugs/
servethehome.com/amd-ryzen-7-1700-linux-benchmarks-zen-buy/
cpu.userbenchmark.com/Compare/Intel-Core-i7-6950X-vs-AMD-Ryzen-7-1800X/3604vs3916
solid-run.com/marvell-armada-family/armada-8040-community-board/
twitter.com/NSFWRedditVideo

>thinking Intel will touch its arch or even memory hierarchy when they can barely get 10nm working on 100mm2 dies.

bump

Jews gonna jew, also

>squares with hexcodes
Fix your shit.

>wait for some shit
>as soon as it comes out, start waiting for the next one
How pointless

...

a 6 core version of the 7700k would be nice

but...

I just don't know how Intel manages to revolutionise computing each year with such enormous IPC uplifts!

- Tom's

This thing is getting more rebrands than a 8800 GT

2015 - Skylake
2016 - Skylake refresh
2017 - Skylake refresh refresh +2 cores
2018 - Skylake dieshrink
2019 - Icelake?

>Icelake?

Also known as Skylake dieshrink refresh

>L2$

Level 2 cash???

What the fuck. Normies in tech...

you mean
>2015 - Haswell Refresh Dieshrink
>2016 - Haswell Refresh Dieshrink refresh
>2017 - Haswell Refresh Dieshrink Refresh refresh +2cores
>2018 - Haswell Refresh Dieshrink Refresh Refresh + 2 cores refresh
>???

>2016
>hope that zen is uncompetitive
>2019
>be in denial that zen is uncompetitive
>2021
>50% market share

even 25% would be nice enough to make AMD profitable again

Sup Forums would flip its shit - the shitposting would be immense because the Sup Forums spillover (which makes up roughly 80% of Sup Forums's content) would not be able to comprehend such a reality.

>a 6 core version of the 7700k would be a literal housefire

Fixed that for you.

Maybe we'll finally get those legendary 15% performance gains.

>2013 - Haswell (Haswell)
>2014 - Haswell Refresh (Devils Canyon)
>2015 - Haswell Refresh Dieshrink (Broadwell)
>2016 - Haswell Refresh Dieshrink Refresh (Skylake)
>2017 - Haswell Refresh Dieshrink Refresh Refresh (Kaby Lake)
>2018 - Haswell Refresh Dieshrink Refresh Refresh +2 cores (Coffee Lake)
>2018 - Haswell Refresh Dieshrink Refresh Refresh +2 cores Dieshrink (Cannonlake (Laptop only :^) ))
>2019 - Haswell Refresh Dieshrink Refresh Refresh +2 cores Dieshrink Refresh (Ice Lake)
>2020 - Haswell Refresh Dieshrink Refresh Refresh +2 cores Dieshrink Refresh Refresh (Tiger Lake)

Ftfy.

kaby lake-x is a really incredible level of kikery for intel

>2021 - NEW ARCHITECTURE! Maybe?

They have no other option other than doing nothing which would be suicide. They HAVE to try and stop AMD gaining a foothold.

Intel uses desperation launch! It's not very effective ..

...

>those little boxes instead arrows/like/dislike/favorite
kek linux users

I used to sell weed to this dude's brother

Intel's found a safe spot with Skylake because it's very likely they can't squeeze out any more IPC, now they'll probably just shrink, and sell cheaper until their new arch in 2021 comes.

The ultimate "Just wait", now brought to you by Intel.

Cannonlake will be 15% faster than Kabylake!


[spoiler]
Mobile only in Sysmark [/spoiler]

That's why we need Ryzen to BTFO Intel, but how do we make Intel babbys understand this?

Intel's management is killing the company.


danluu.com/cpu-bugs/


Holy shit

It's more like the current Intel-released technology was so far ahead of AMD that they didn't have any competition.

Intel just needs to have the fastest chips on the market as long as Windows is locked to x86-64 and has a very large user-base.

Keep in mind that Intel has had 24c/48t chpis since 2011 (at insane prices). They can release consumer 8c/16t chips and price them as their current 4c/8t chips. They can release chips with a higher IPC too. Doing so does not make financial sense, though, since people are willing to pay what they are currently charging for current chips.

If AMD starts eating away at Intel's market share then they will simply take something they already have ready off the shelf and release it. I don't see that as very likely, though. AMD would need to do more than give comparable bang for the buck, they need to offer significantly more bang for the buck. Say R7 1800X cost $200 not $500, for example. That would force Intel's hand. But that's not happening.

The cheap shit Intel sells has higher TDP and lower performance than ARM, and the good shit is too fucking expensive for most people.

You are smoking some premium shit if you believe any of that.

Intel is getting skullfucked by ARM, ARM chips ship out 10 times higher number than x86 CPUs per quarter.
It's complete chaos in the datacenter group since a month ago when Microsoft of all people gave Intel the middle finger and embraced ARM

Intel doesn't know what to do, and considering their size their response is going to be slow

>Intel
>higher IPC
Oh hell my sides that's some quality shitposting.

Why is everyone bullying poor Intel?

>poor
>Intel
For years these kikes hoarded money. Not their retribution comes.

>Otellin leaves
>Intel collapses

Lmao

Isnt Skylake a new Architecture though?

Because it's been roughly 8~9 years of AMD bullying, paybacks a bitch.

It's as 'new' as skullfucker P6 core can get.

It is, but not all that a big change.
The server Skylake is new new, consumer Skylake is a smaller arch change, unlike Haswell that was a pretty decent change.

Their data center cannon lake chips couldn't be die shrunk and we're canceled.

Meanwhile Naples is not only more power efficient, it's faster to boot.

Intel is fucked until they figure out their node shrink.

>muh 60% margins forever and ever and ever in troubles

Thank fucking God.
There's probably not a more jewish company out there.

But it is. They literally own the market.

It's not about beliefs. Go take a look at Intel's website and check out what kind of CPUs they are offering right now if you're willing to pay as much as you would for a luxury car for a CPU.

I really hope Microsoft does release some Windows server version that runs fine on ARM. I use GNU/Linux myself so for me ARM would be just fine. What ARM actually lacks for my personal use is a good motherboard with connectivity. There's no affordable ARM system with 8 SATA connectors, for example. There's also no ARM motherboard with 3 DP or 3 HDMI outputs and no way of adding a GPU with PCI-e.

If it was possible to buy a motherboard with these things and an ARM or RISC V CPU then I'd do it in a heartbeat.

>>muh 60% margins forever and ever and ever in troubles

Oh they actually are in trouble, but it's not that apparent. The computer market is saturated. Everyone who wants a computer has one. And they are "good enough" for most people. If you have a 3 year old laptop and all you do is browse Facebook then you have little reason to buy a new one.

Even if you have a 80% market share you're in trouble if that market shrinks by 50%.

>muh 60% margins forever and ever and ever in troubles

To be fair Sup Forums is stupid enough to believe this sort of thing. The best one in recent times is that AMD's stock will never ever climb significantly despite the writing being in the wall since mid to late last year. A few smart anons on Sup Forums said if you can into stocks, always bet on duke and go all-in with AMD. Most of Sup Forums simply laughed and declared AMD to be bankrupt by 2017 (y'know, the 5th straight year they have been declared bankrupt) and look what happened - their stocks exploded and very little of it is attributed to zen.

If anyone actually followed AMD they'd know how aggressive AMD has been in securing large long term contracts (especially in china) along with a lot of diversification of their technology IP's (though that is just an extension of their HSA initiative started 2014 or something). AMD has spent a lot of time and effort building up a framework to get the dream of HSA a reality. We are not there yet but the likeso f zen's infinity frabric, the high bandwidth cache going into vega and those radeon pro boards that bolt-on SSD all fall under the same umbrella. Who knows when we will see a complete product that can tie these all together without enormous latency but the groundwork is there and its going to be relevant for some of the biggest growing markets - like machine learning.

Its also why Nvidia has started throwing their weight around as they are trying to keep Intel out as well as prevent AMD gaining a foothold.

tl;dr Sup Forums doesn't know shit and is constantly blindsided by reality.

How do we save Intel Sup Forums?
We can't let those dirty commies win.

By murdering Krzanich and appointing someone actually competent, also with engineering degree.

You want Dr. Lisa Su?

That's AMD's CEO. Intel needs their own.

It's almost like I was making a point or something.

>What the fuck. Normies in tech...

$ has been the denomination for cache memory since at least the 80s.

>tfw Nehalem
When will I need to change ?

Zen2.

Intel does rule tho

Name one achivement of amd, intel has this

AMD makes a smaller core that's within 5% IPC of Intels best big core while using way less power and Intels response is Skylake until 2019?

What is this company smoking?

Literally years ago.

>What is this company smoking?

Diversity quotas.

>best big core
The 1700X and 1800X don't even come close to touching the 6950X in terms of performance.

Great reading comprehension.

Sometimes they are competing against dual xeons.

servethehome.com/amd-ryzen-7-1700-linux-benchmarks-zen-buy/

>b-but muh gaymes, muh single core

It's 20-30% faster and costs 3 1/2 times as much. Sounds like a steal. Also, it has two more cores, it fucking better be faster, dumbass.

cpu.userbenchmark.com/Compare/Intel-Core-i7-6950X-vs-AMD-Ryzen-7-1800X/3604vs3916

>that's within 5% IPC of Intels best big core
You're legit retarded. Kill yourself.

yes they do, they come really close in a lot of things. which is fucking sad.

This, workloads that can leverage AMD's 4 complex decoders and where SMT can fully stretch its legs it can get very close to the 6950X even with two less cores.

ah ryzen looks pretty good but unfortunately my job is sitting in a room and repeatedly RARing and unRARing homosexual pornography so it looks like i'll have to spend $1500 on the 6950x.

I know that pain.

>$ for cache
>xtor for transistor
legit triggered

jokes aside, i wonder what the cause is there for such a discrepancy. is it purely the extra cores? or do instruction sets come into play heavily with winrar

>still buying Intel after they switched to not solding the heat-spreader in order to save chump change per overpriced CPU
fuck you. you're the reason why we can't have nice things.

>muh teeny tiny dies

It's a launch day benchmark, any large discrepancy and scores that make no sense will be the fault of poor CCX utilization.

Here's one from 3 days ago.
Also look at the graph scales and actual numbers.

To be fair single core performance is still the grand daddy of cpu design but (and this is the big one) without a truly commanding lead in it for the majority of workloads (remember zen is only going for 80% TAM) it is those secondary features that become very important - some of which AMD is leagues ahead of Intel's current designs. Zen is nothing short of wicked being a SoC which opens all kinds of doors for semi-custom designs based on the chip since so much I/O is handled by the chip itself. You could do some crazy things in the console space (like prepherial support) via usb for example as you wouldn't need chipset to handle a million devices.

Watch this space - zen is going to ensure a lot of future contract wins in the semi-custom market precisely because AMD has had to design a chip that scales from tiny TDP envelopes (sub 10w) all the way up to massive server farms where its all about I/O and power draw. What I am curious to know is exactly how close microsoft (and probably sony too) got to zen as it neared the final silicon - there is no way either of those companies would let zen slip them by for future console deployments.

Muh games though, that is just the response of the uneducated, unwashed masses.

Possibly the absolutely enormous cache per core the 6950x has.

Winrar is 50% slower on Ryzen in SINGLE THREAD on that chart, against a Broadwell-E
Even if that BDW-E had 90MB of L3 and 5.0GHz clocks that isn't something that's supposed to happen.

Solder is good but not the best.

Broadwell isn't known to be very efficient at those clocks.

The huge slab of L4 is a major contributor to broadwell-e's clock and power draw limits.

>L4

Only the desktop 4 core broadwell's have eDRAM

solid-run.com/marvell-armada-family/armada-8040-community-board/

soon

can't wait for ARM DOMINATION

>using rar instead of 7z
when will you homos learn!

>not using targz

>not using peazip

>unencrypted and not gpg

pls kid
7z just werks for encryption across os's and has better compression

This is why you need competition.

Regardless I'm quite happy improvements are slow now. I'm sick of what the software industry has become. I know people are gonna be angry with me for thinking the old days were better but the restrictions in the 90's made software engineering less about herding developers like they were sheep towards not extremely slow solutions and more about working with actual coworkers with minds of their own.

I can't be the only one who feels this way.

$ for cache is industry standard though. I'd say xtor is less so. But it's perfectly readable and is just a sign of how problematic Twitter is for communication, not his fault.
I can't fathom why people use Twitter with the limit they have.

I will eat my own socks if Project Scorpio isn't rocking a Zen APU.

>I can't fathom why people use Twitter with the limit they have.
This, whenever I try to talk some tech on twatter I get too verbose I'd need to split it into 5 tweets, because there's nothing worse than some badly explained hardware technology.

>1990+27
>using (((social media)))
shiggy