>people waiting for Skylake-X
>it was a turd
>people that waited got Ryzen
Meanwhile what is happening to Intel is gonna happen to AMD's GPU division as well, thanks to Vega, but who cares, Nvidia's GPUs aren't TOO overpriced like Intel's top end gear is.
Skylake based Xeons' where it's at
There are actually a few that are good with CPUs, like Montero..
And EPYC(which aren't counted in the OP's chart obviously) set enormous lead world record thanks to how well the coins fit in the cache.
>I’m going to send you an e-mail. I’d love help bringing this to our cluster. We only have around 3000 dual socket Xeons now but we expect to grow.
>only 3000 dual socket Xeons
>only 3000 dual socket Xeons
>only 3000 dual socket Xeons
>for mining coins
>O N L Y
THE WORLD IS FUCKING INSANE
>>people that waited got Ryzen
As if buying a $1000/600 CPU, with $300 motherboards, needing 4 channel memory and a fucking liquid cooler would ever be a choice(unless you have more money than sense and/or crazy) against a $300 CPU on a $100 motherboard that's only 20-30% slower
>SFF desktop sitting on top of my stereo
>rendering tower behind desk
>Toughbook sitting on shelf connected to monitor
>everything connected to NAS
>HTPC in the living room
I know this feel.
>This graph counts the baselines submitted to us during these time period and therefore is representative of CPUs in use rather than CPUs purchased.
the new itx boarda are really tempting me though thing is all my computers are atx and id need a desk just for all of them and no where for the monitors unless i use a 2nd desk i wish manufacturers would modernize old platforms like socket 478, 775, am3 onto itx boards but itd be a niche market for people like me
>OUR POO PROCESSOR IS B-BETTER THAN INTEL R-REALLY
This is the best damage control you can do? Sad!
Please provide citation that the ringbus is actually a problem.
>You can't post steam hardware survey of over 50 million people to show AMD cards aren't selling, it's not proof of market share, you're wrong!!!!!!!!!!!
>Oh here's this graph from a benchmark that has not even a single million users, this is proof that Intel is dead
Which one is it, shills?
>faggots with 5 accounts with only CS:GO
>loads of accounts that not even have a profile
>the number of accounts somehow translates the ammount of CPUs being used for gaming
Eat shit and die of syphilis
anytime amd gets close to overtaking intels marketshare intel introduces something that absolutely rekts amd. intel has had over a decade just shitting out camrys of the cpu world shit like skylake x is a band aid
Scales right up to the badnwidth of the actual memory controllers.
>The graph showing AMD and Intel market share is made of up thousands of PerformanceTest benchmark results and is updated daily
>made of up thousands of PerformanceTest benchmark results
people who believe this shit are unbelievable.
Intel's new arch is Linus Tech Tips tier. They're taking x86 and chopping parts of it off to make it fit into a smaller space. It's going to be garbage with backwards compatibility issues out the ass.
The difference in perf of AMD and Intel during late 2018/2019 will be far larger than it is now, unless Ice Lake is another miracle, which I doubt since their new arch is in 2020 and IL is another skylake.
All your damage control won't help here
Intel is going to be finished & bankrupt in 2018
Mark my words and screencap this
>intel introduces something that absolutely rekts amd
like OEM bribes and fake benchmarks?
The X86 deal dials back when IBM asked Intel for a secound suplier for X86 chips, so Intel made it so that only AMD can make it, if they somehow file bankrupcy even if another company buys AMD they will not be able to produce X86 CPUs. Newly acquired Advanced Micro Devices would be reduced to ARM CPUs and GPU manufacturing wich are not their best at the moment.
Just this and still the biggest market makers (ryzen 3/APUand mobile APUs) are not even out yet.
This is gonna get hilarious over the next year.
If anybody was going to buy AMD, it would have happened a couple of years ago when they were at their lowest ebb and barely worth the change down the back of Brian Krzanich's couch.
Personally I think stock boost clocks will be slightly over 4.4GHz for the 10 core models, but the IPC to be over 15%, 10% seems kinda low knowing how many things the Zen1 didn't include.
>but just the fact that the architecture was only as good as the """" enterprise class """" programmers that wrote software for it
Fixed that for you. Next time, Intel won't make the mistake of expecting Java programlets to have half a clue of what the fuck they're doing.
If the reason people are still buying Intel is higher clockspeed gaming performance, their 6 core isn't even going to sell as well as their 4 cores.
Even amongst competent programmers, however, unless you're literally only dealing with extremely regular code with extremely regular memory access patterns, hardware ability to deal with variable instruction latency is a good thing.
>10% seems kinda low knowing how many things the Zen1 didn't include
Zen didn't include those things because they weren't worth the power budget. They still won't be worth the power budget with Zen 2, because anything that gives less benefit for the same power as adding another core isn't worth bothering with.
Intel doesn't sell products based on overclocking performance, that's why I think the 6 core coffeelake will be around the 1600X's clocks and around $350, the 6 core Skylake-X is naturally clocked a bit higher to make a perf difference.
Why would they cannibalize their own products? Intel never does this.
Why would they even want the market to think their mainstream products are FASTER than HEDT products?
Also Intel's hard market segmentation. Itanium was just too expensive for any enthusiast community to form.
Yeah, let's blame programmers for a shitty CPU design.
Then enjoy thermo conductive cum in your CPU.
Larrabee was just an in-order pipeline, though. If you're going to say that it was designed to require special compiler support, then that's true of literally any architecture ever.
>Itanium was just too expensive for any enthusiast community to form.
True, but the fact that it didn't perform very well on common workstation software probably didn't help either.
it's only "best seller" bc AYMMMD is practically giving it away for free you shmuck
Hello netburst my old friend
I've come to stoke your flames again
Because market share is slowly leaking
AMD released Ryzen while I was sleeping
And the bipelines that were burned in my brain
Within the flames, of netburst
Imagine coming up with damage control this weak.
>Tfw got R5 1600X for $250
God fucking damn it AMD, why the fuck couldn't you just wait a month before slashing prices? I finally got burned by buying shit on release, never again.
I have 150 games. I'm online only a few nights a week, or sometimes even go weeks without logging in as I happen to have a job which requires me to travel. Sometimes I play in offline mode, as well.
I dont understand your 12m claim. I'd guess the number as a whole is less than 100m but way higher than 50m.
h-hh-heh, Just Wait™ for next arch AMDrone, then you'll be s-sorry!
In the land of the free and home of the brave shit is cheaper, I just didn't do the conversion in my head. I assumed it would be closer to this, what was the launch price in pounds?
>Some theories are they're going to remove the unlocked multiplier from their desktop CPUs entirely
They'd be really stupid not to do that, they've been needing to simplify their product line-up for a long time anyway and getting rid of the "K" series would be amazing.
Former intel engineer here. Not gonna lie it's hilarious watching them crash and burn right now, but seriously, we can't let them reactive the pipelines again.
Why do you insist that MMX or x87 or BCD instructions must be with us until the end of time? We should be happy if these finally go away. Especially the BCD instructions that take up one-byte decoding space.
Are you trying to say that Intel's and AMD's interconnects are the same? I myself haven't even seen any technical description of the IF whatsoever, so please enlighten me if that is the case.
They are not.
Intel's mesh is an advancement of the ringbus design which allows it to scale more, because it's not a ring but a mesh (duh), i.e. every core is connected to every core.
It scales pretty well but increases the initial latencies from the ringbus even between adjacent cores.
Zen has CCXes with 4 cores (as of now, different CCXes might follow) which are connected through the inifinity fabric which is coupled with the memory bus in some shape (the admittedly only questionable part about the otherwise pretty good architecture). Internally the CCXes are connected via a ringbus(?).
This means internal CCX latencies are lower than overall mesh latencies but once you hit the IF they skyrocket.
Regular low-core ringbus latencies are still the best, but as has been pointed out multiple times, they scale like shit.
AMD's IF also has the advantage that it allows them to scale the binning by disabling faulty cores inside the CCX (i.e. sell 8C, 6C and 4C SKUs fromm the same silicon). The same was definitely not possible with Intel's Ringbus (since one faulty core obviously trashes the whole chip). I'm not sure how the Mesh architecture scales in terms of binning though, I guess that remains to be seen.
>Zen has CCXes with 4 cores which are connected through the inifinity fabric
>Internally the CCXes are connected via a ringbus(?).
You have no idea what you're talking about. IF is what is connected the CCXs and the dies in MCM configurations. The interconnect between the cores inside a CCX hasn't been disclosed, but it's probably some kind of crossbar.
>IF is what is connected the CCXs and the dies in MCM configurations
>Zen has CCXes with 4 cores which [the CCXes] are connected through the inifinity fabric
that's what I said
>The interconnect between the cores inside a CCX hasn't been disclosed, but it's probably some kind of crossbar.
which is why I put a question mark behind my statement
Posted on the wrong thread sorry 4chinners.
This can't be correct.
You're telling me we've gone back to Q2 2007 levels of Marketshare overnight?
If you look at pic related, AMD is doing BETTER than pre-Sandy last time it had slightly superior products with Athlon II and Phenom IIs.
This cannot be right.
I'm building one anyway
I have an i7-7700K gaming PC, a 6700K back-up gaming PC/shitposting station, a 5775C HTPC for the living room, a 4790K for my bedroom TV, and a 3770K acting as a file server for the 5775C and 4790K.
I have a 2700K on its Z68 motherboard hanging on my wall if that counts.
It appears to be too good to be true only because AMD did disastrously since then (Sandy).
This is actually not that great a healthy marketshare in a 2 horse race desu.
AMD has to rise back up to 40ish % or can't compete. Vicious circle.
>Aren't prices in America before taxes?
Yes. Amerifats cheat when bragging.
And btw you illiterate morons, nominal exchange rate conversion is meaningless.
If your currency de/appreciates overnight, you're not paying more or less just because the price in USD or whatever other currency are btter or worse NOMINALLY.
Jesus, it's like no one in here has graduated HS or something.
I'm not replacing anything, I'm going to start making Youtube videos as a hobby and I need a video capture/re-encoding PC.
I'm saving up for an i9-7820K and a X299 once the BIOS and hardware for those become stable enough.
>b-b-b-but muh delid
So what? I lap and delid all of my past i7s (with the exception of the 2700K and 5775C which are soldered).
>Meanwhile what is happening to Intel is gonna happen to AMD's GPU division as well, thanks to Vega, but who cares, Nvidia's GPUs aren't TOO overpriced like Intel's top end gear is
yeah less than a month to see what the consumer vega is all about
>Tfw when you're a part of the market share increase
Sorry to say it to Intel, who've I've been pretty loyal to for almost 10 years now, but the way you've handled your business decision this past year and the upcoming quarters is deplorable. I'm glad I helped give AMD its current boost. Maybe you'll rethink your strategy and actually make products with us in mind rather than trying to save pennies a processor.
I feel like Ive seen this graph somewhere before...
>Intel is so far ahead in market share, they're not even thinking about AMD!
To be fair, yelds of fully functional 8 cores die are 80% and they can successfully salvage the remaining 20% into gimped parts, hence the previously mentioned 99% die usage.
It's a great idea, almost no waste. IF is a great idea if only for this. Next earning call will be dope for AMD, investors like this kinda shit.
the year of the AMD desktop has finally arrived
fucking retards, any new consumer cpu would be disproportionately popular in a benchmark application, because you test something when it's new and then you forget about it unless you're some overclocking autist, and the sample size is tiny "thousands"
>ucking retards, any new consumer cpu would be disproportionately popular in a benchmark application, because you test something when it's new
Right, the point is to compare it with NEW intel sales idiot.
it's not very relevant, anyone who would buy an intel cpu already has for example 4790k/6700k/7700k, ryzen fits a different use case, mainly prosumers who do video rendering and such, and a minority of gamer kids
We all know how this ends
memers have been waiting for ryzen for at least 2-3 years, while during this time intel has had a healthy business, so now that ryzen recently came out the popularity in new purchases is temporarily higher than it would have been in an apples to apples comparison
I'm going with this so far: https://fr.pcpartpicker.com/list/rc8csJ
I was going to buy a RX 570 or 580 to go with the super cheap freesync monitor, but due to miners I guess I'll do without
Yeah, I might end up buying a 1050ti or a secondhand card from my local craigslist...
Thankfully I'm looking more for a CPU heavy build than pure gaming, so I'll just take what I can get
I think I'll buy the GPU at the last moment, in case I get lucky and Ethereum crashes in the meantime.
I'm moving to a new place sometime this month, so I'll buy the heavy stuff after moving.
Otherwise yeah I guess I'll get something good secondhand.
it's not even sales stats, it's benchmark stats, if you buy for example a i7-7700k chances are you won't bother to test it on this literally who benchmark, because the result is already known
there is more variation with ryzen depending on your ram and ram clock speeds so you would have a greater need to test your cpu, whereas with intel if your cpu clock speed is good you know your cpu is good
and this is a literally who pajeet benchmark with a sample size of "thousands"
why am i even wasting my time on this fuck off
>muh secret value
If Intel had any real answer for Zen, we wouldn't have this god awful X299 platform.
No, Intel doesn't have a bunch of 10 niggahurtz 50 watt god chips just sitting in a warehouse waiting to end AMD
Do you really think their duplicate detection is so shit it can't tell if you've changed your RAM or not and just assumes it's a completely new build?
Someone has been watching sales on some german aggregate site, mindfactory IIRC and Intel/AMD sales have literally been 50/50, which means if that's true AMD would catch up to Intel in no time, reminder that AMD before Zen sold 1 CPU for every 8 Intel sold
GPU buyers are much more prone to fail for funky nvidia marketing.
Like, can someone explain me that graph?
>amd driver bug
>nvidia driver bug
but that's not the point because you would mention both nvidia and amd ("GLSL shader not working on AMD/ATI, but works on NVIDIA "), look at the actual issues, amd drivers are shit in performance too, that's why amd shills say they age like wine, because amd drivers are poorly optimized and they take years to improve the drivers
EVERY GODDAMN TIME
FUCKING IGNORANT 56% """"WHITE"""" RETARDS
the graph clearly states that after a certain point you get diminishing returns with increased power