>16 cores
>hyperthreading
>ddr4 support
>arm64
>32 threads
>FinFET 14nm
It's over, intel is finished
[citation needed]
>15 unused cores
>still shit performance per clock cycle compared to intel
Yeah, no. I'll never get another AMD for a gaming build ever again and professionals already know better than to buy AMD.
>>54114446
>jim keller architecture
>low serial performance
Ok.
>>54114446
>shit performance per clock
It's one of their main focuses, AMD said. They'll focus on making it faster per core. The main reason why it has so many cores and threads is because this is initially planned to be a server grade CPU, but AMD will also market it for the high end desktop market. They also have been designing the CPU architecture from scratch for about 5 years now, with some of the biggest names in the CPU industry like Jim Keller. So you know it's gonna be good.
>>54114541
>They'll focus on making it faster per core
And if anything they've done in the past 6 years is any indication they'll fail miserably.
Well shit, I need to get a job before these land.
>>54114557
>took an underperforming, power hungry chip
>turned it into a small, vastly more efficient chip with still lower than nominal performance
Yeah, I'd believe they can do it. Bulldozer was a shitshow of an architecture but Kaveri and Carrizo chips are actually pretty efficient in wattage draw and often come in under their rated TDP under full load.
>>54114446
>professionals already know better than to buy AMD
Professional gamers?
Professional shills?
Linus tech tips?
Literally who?
>>54114617
>professionals
>Linus tech tips?
Now you're just trolling m8
>>54114557
I'm not telling you to buy it the moment it lands or anything. We'll have to check benchmarks, and AMD is kinda notorious for generating a ton of hype and making a lot of blank marketing statements in the past. However, if the specs of this processor alone are any indication then this is going to be a beast of a processor. Plus the fact that they finally added ddr4 support after so long is nice.
>>54114446
>the professionals are laughing at your post right now
>>54114541
>So you know it's gonna be good
Brand loyalty fags, pls
>>54114495
>Jim Keller
He ran off even before ZEN was halfway done.
How many cores is it going to have? 8 per die isn't going to be enough unless it's cheap and sells for 300 per cpu.
>>54115032
It's AMD, probably 64 cores or something. And then the welfare queens will start complaining why their favourite gaymen studio doens't optimize games for 64 cores.
>>54115015
He didnt run off, he did his job and went elsewhere, as he always does
>>54115045
He didn't stick around for the first prototype to surface, like he did with Apple for example.
>>54114350
RIP INTEL
>>54114350
If ZEN fails to deliver performance close to haswell then AMD is lost,forever.
>>54115089
They are aiming to bring Sandy Bridge (januari 2011) performance.
>>54115053
because that's how he rolls now.
he comes in, blows some minds, makes a glorious new architecture, drops the mic and leaves.
all while wearing a life jacket to not drown in pussy.
>>54115053
Didn't Apple have him onboard for a few generations of A* CPU?
>>54115007
>brand loyalty
I've never built an AMD system. Not once. This isn't about shilling, this is just me getting hyped over what could potentially be an amazing high-end processor. I've been going intel for a long time, but if zen lives up to expectations (which AMD claims it does, as they tested it and found zero bottlenecks, of course according to them), then I'm going AMD this year.
>>54115093
Seriously? Why would you want to aim for 5 year old architecture and call it brand new?
>>54115147
They're not, that guy was just trolling.
>>54115147
Because Intel has kind of a leap when it comes to single core performance and it's nearly impossible for AMD to just beat Intel at their own game.
Besides Sandy Bridge single core performance should be sufficient.
>>54115107
Don't know, don't care. My only point was that Keller just ran off after delivering a design whereas with Apple he stayed untill the release of the products.
>>54115157
I was not. The single core performance of AMD Zen is supposed to be on par with Sandy Bridge. Which is 6 years old by now.
>>54115172
Sufficient doesn't mean it's good. They need to step their game up
>>54115183
No, hence the moar corez game AMD plays.
Obviously 4 cores Sandy Bridge wouldn't be good. But 8 cores Sandy Bridge is great. Don't forget that most of the edgelords on /g/ still use the i5-2500K which is kind of timeless.
>>54115172
>supposed to be on par with Sandy Bridge
According to who? Your expert opinion?
>>54115197
According to thousands of threads about ZEN.
What do you expect AMD to do then? Beat Skylake at single core performance?
I think it's safe to say nobody actually expects zen to beat any Intel products. It'll probably cost slightly less while being 10~20% slower. A nice alternative for those on a budget and that's about it.
A performance war is out of the question.
AMD can't win a price war with Intel either.
What other realistic options are left?
>>54115205
Well yes, since it runs on the same node process, has more cores, has a better multithreading technique than hyperthreading. I think it's safe to assume it will at least beat skylake. Dunno about haswell e or broadwell e, but I'm 99% sure it'll beat skylake.
>>54115206
Focusing on desktop CPUs was a stupid move anyway. The mobile market is a billion times bigger than the desktop market by now.
Meanwhile nobody in their right minds will ever buy a laptop with the AMD logo on it. They should've used that IPC increase to make better mobile processors.
>>54115218
>I think it's safe to assume it will at least beat skylake. Dunno about haswell e or broadwell e, but I'm 99% sure it'll beat skylake.
How much does AMD pay you? I'm talking SINGLE CORE performance here, right?
Obviously they can beat Skylake if they just add 20 more corez and find some program that manages to utilize all of them. That was never the problem anyway, what AMD is trying to do is increase single core performance.
And no, I do not believe they have a shot at beating Skylake at single core performance. Nobody in their right minds believe that.
>>54115234
>doing all this shit and getting mad for free
man you suck dick
>>54115245
I'm not mad, I just clearly say AMD has no chance at beating Skylake in single core performance, and you go on talking about more cores, multithreading, etc. Totally missing the point.
>>54115234
>How much does AMD pay you
>Says the guy who outright believes a product will be shit before it's even released
lol ok. I've been going intel for years as I've posted here>>54115123 , in case you missed that. I'm just hyped because all the facts point to this being good. They have fucking Jim Keller on it, the same guy who made intel have a run for their money with the athlon 64, and a lot of other former intel engineers. They also said that single, per core performance is a huge focus of zen, and they designed a whole new architecture from scratch so it's not gonna be bulldozer, or steamroller all over again, this is something new.
>>54114350
ADD MOAR COARS XD
>>54115015
>He ran off even before ZEN was halfway done.
He was there for 5 years.
That's plenty of time to build a new arch from the ground up.
>He didn't stick around for the first prototype to surface
So there being leaks or rumours about something is what helps something cross the bridge between reality and fantasy?
AMD are getting VERY good at keeping their lips tight. Look at Fury, all we knew about it before it's release was that it would use HBM. We didn't even know it's name.
I don't know where you got such fed such blatant misinformation. But that's just what it is, misinformation.
Hell. Keller's reputation alone should be enough for you to see the insanity in claiming he left before a project was completed.
>>54115107
A6 to A8 where Keller's work.
Apple are probably pulling an AMD here and just rehashing the same arch for a few years after Keller leaves in an effort to get some degree to performance gain from it.
>>54114617
>Literally who?
Literally anyone with half a brain who's not a broke ass negro living on welfare. AMD is shit in the CPU space and they know it. Microsoft is poised to buy them out they suck so bad at it. The ONLY thing that keeps them alive is their decent GPUs
>>54115356
>AMD is shit in the CPU space and they know it
Somebody obviously wasn't here around 200-2006, back when AMD dominated both the CPU and the GPU market from nvidia and intel.
Yeah, AMD has its ups and downs just like every other company, but it's worth giving zen at least a chance before jumping to conclusions. Plus, zen apus are confirmed t have hbm powered gcn video cards, which is leaps and bounds better than intel's integrated gpu. plus it can actually run in ccrossfire with other amd cards you buy.
>>54115421
>Somebody obviously wasn't here around 200-2006
>200-2006
fucking mortals REEEEEEEE
>>54115205
>According to thousands of threads about ZEN.
Yes, because /g/ is clearly the be-all-end-all of tech journalism and analysis.
Zen is expected to hit a 40% IPC uplift over Excavator. AMD's latest iteration of the Piledriver core arch. The only Excavator products available right now, are APU's. Under the code-name Carrizo. So let's compare that to a Sandy Bridge i5.
>AMD A10 7870k
>Single Thread Rating: 1514
>Intel Core i5-2500k
>Single Thread Rating: 1898
Both from PassMark ofcourse.
So if we were to speculate of Zen's performance, we could simply just do this;
>1514 x 1.4 = 2119
A Haswell i5 (4670k) pulls in a Single Threaded Rating of 2200 according to PassMark and their measurements.
While I realise this is the roughest way possible of estimating the performance of Zen. It's better than just saying "There's been thousands of threads about it so it must be true!" going by that logic, that means 9/11 was an inside job, and the moon is made of Gorgonzola.
>>54115421
>AMD dominated the GPU market from 2000-2006
AMD didn't have GPUs till late 2006
>>54115356
>Microsoft is poised to buy them out
Wow.
I remember back in 98 when Microsoft were going to buy AMD.
And in 99.
And 2000.
And 2001.
And 2002.
And 2003.
Then in 2004 Apple were going to buy them.
Then in 2005 it was IBM.
Then in 2006 nVidia were going to buy AMD.
Companies would be fools to buy AMD, not because of their underperforming products, but because the CPU marketspace is a hell hole. Only a fool would want to get into it.
>>54115439
>>54115439
Anon, 9/11 is an inside job. Take it like this. America blew up a few of out own ships to start conflicts to seem justified in public eyes
>>54115442
ATI did though.
>>54115519
And ATI was a separate company from AMD entirely, with no ties in anyway.
AMD bought ATI in 2006.
>>54115532
>AMD bought ATI in 2006
Yeah exactly my point.
Even though they bought ATI in 2006, for most of the year and every prior year ATI was an independent company
>>54115543
Perhaps you should revise this post in that case (>>54115421) specifically, this part.
>Somebody obviously wasn't here around 200-2006, back when AMD dominated both the CPU and the GPU
AMD didn't have GPUs until '06. So to say that dominated the GPU market UNTIL '06 is wrong by virtue of the fact that they did not have the means to until '06.
>>54114446
>professionals
>gaming
Wow. Just. Wow.
>arm64
>>54114617
Linus shill tips
I'm on a 3570k at 4.3Ghz. If they somehow can top that with more cores, then I'm jumping ship immediately.
>tfw make -j32
>tfw ffmpeg
>tfw more VMs
Hurry the fuck up, AMD. I want to build a server as well.
>>54115616
Meant amd64. I'm on a spartphone with an awful touchscreen keyboard.
What is the point of buying an amd? Isnt it just for more chores nowadays anyways?
>>54115695
kill yourself
>>54115714
But I'm already dead inside anon.
https://www.youtube.com/watch?v=-0BRWkjY4kg
>>54115438
lol
Even Jim Keller thinks Zen is great. https://www.youtube.com/watch?v=SOTFE7sJY-Q
I hope those 25k€ in AMD stocks will finally get me a Lamborghini
>>54116011
top lel
>>54116011
This fucking amdman.
>tfw didn't buy at 1.6
>>54116011
Its like 2.8 a share now, if he bought at 1.8$ then he would have around 13k now
>>54115697
Dunno about where you live but the only cheaper intel CPU at the time (~1.5yrs ago) I built my pc was the g3258. This was compared to an fx6300, so of course I'd buy the fx. Still is pretty much the same, tho I haven't been looking at prices every day like I did right before buying.
Not that I'd defend my 6300 to death, but it really does its job well, and is priced /competitively/, it isn't just cheaper than alternatives, its both considerably better and cheaper. Also oc potential is fun as fuck.
>>54116059
>amdman
>madman
damn
Bought and i5-6600k and OCed it to 4.6ghz, fuck the waiting game and everyone who thinks that Zen will change the game, I'll never need to upgrade.
>>54116011
I have about 1k€ on the side, is it worth investing in AMD? Serious replies only.
>>54116651
Doing it right now is stupid. You should have bought when it was at 1.6-1.8.
Zen is the thing we've been waiting for for years.
This is going to be just like every previous AMD chip for the last 10 years.
It will be trash when it comes to most applications and will only do well when it comes to programs specifically programmed to make use of extra cores.
As for MS buying them out, why would they ?
There is so little synergy to be gotten here, spend a few billion to make your chips for your xbox just a little cheaper..why ?
By the time Zen comes out , there will be broadwell-E, new skylake and kabylake stuff too
AMD's time window to release this stuff is getting smaller and smaller, by the time it releases it will feel like its 2 generations behind intel.
>>54114350
if i recall correctly they hired the guy who designed the phenom 2 series to make zen.
im actually optimistic this time around.
phenom was a great series and dominated intel on some grounds.
i still have my old 1090T running my freenas box which runs 2 linux vms and plex with transcoding.
its still pretty capable.
>>54116619
This, people are delusional
>>54117117
>On 18 September 2015, Keller departed from AMD to pursue other opportunities, ending his three-year employment at AMD
Zen will surely take off !
>>54115727
Be more dead on the outside then.
>>54117170
m-maybe that was enough to work his magic?
I'm just bracing myself for disappointment.
>>54116651
It's an OK buy, but not a great one right now (you can probably get better returns elsewhere if you look hard enough).
If it drops to 2.50 or so, I'd buy. There is a lot of upside long-term.
Like Intel is gonna improve anything.
For gaming at least
>>54117243
One of the most taxing things I do on the computer is emulation. Upgrading to Skylake from Sandy solved most if not all problems.
>>54114350
Intel already has CPUs with 18 cores, and they're making some with 22.
How is that special?
>>54114350
Sounds pretty good to me. Can't wait to render videos on this. It's probably going to be pretty cheap.
>>54116619
Uh, you have to consider there are people who have nothing to upgrade from in the first place and are looking to build a new rig. Personally for me, zen is fantastic news. I mainly only play emulator games and run virtual machines of different old operating systems to play old games. Those 16 cores are gonna come in handy when I'm trying to run my virtual machines. Fuck paying $1000 for anything more than quad core.
Will probaly be as fast as a 4690k. Not an intel fanboy, but amd needs to step up their game. Amd shitdozer needs a real good upgrade. Currently rocking a 4790k at 4.8 ghz
>>54117342
Yeah man, let's pay thousands and thousands of burger shekels for that.
>>54117342
Thats a low end CPU anon.
They ain't even shitting out the higher end CPU's and their making more cores and less power consuption cheaper then the next.
They literally inovated intel out of the playingfield and they are having the second coming of the wonderfull world of server architectures.
>>54117385
Bulldozer is the reason AMD is at its current state. Their GPUs are great, but holy shit is Dozer bad. A Zen matching Haswell with more cores would be fantastic, but that's definitely not happening.
>>54117389
>>54117397
You realize if this is even remotely competitive with Intel's offerings, Intel would just drop their prices, right?
>>54117411
I agree with you at the gpus. The 7970 is a great gpu, using that one atm
slembul
>>54117420
You do realize that would only work for the retail market right?
If the Zen cores deliver even half of the promissed processing speeds every data center from here to timbaktu will be buying them like there is no tomorow.
And honestly there is that anty trust law that makes them not being able to dip as low as AMD due to market competition laws.
>>54117411
The fact that it's on a 14nm FinFET node to begin with means that it's likely going to be better than haswell e.
>>54114350
>tfw just bought a skylake i5
Did I mess up /g/??
>>54115438
>>54117420
The funny thing is, this is exactly how AMD is trying to stay competitive.
Just go take a look at AMD's financials, AMD hasn't made any profit in the last few years, not zero profit, literally a loss
if Intel were to drop prices in a significant way, it would definitely crush AMD at this point.
>maybe even trigger some anti-trust/anti-competition shit too
>>54117498
Depends on what you use processors for. Gaming? Then nah, a skylake from this generation can you last you pretty for much for the next 2 decades before you should ever feel the need to upgrade.
JIM KELLER STRONG
>>54117513
The CPU market is literally the shittiest market in technology right now to want to compete and work in.
>>54115439
>A10-7870k
>Excavator
That's still a Godavari
You should be looking at the X4 845 for Excavator CPUs
>Athlon X4 845
>Single Thread Rating: 1794
That's not too bad considering that the X4 845 is a 65W TDP CPU and the 2500k is a 95W TDP CPU.
>>54117874
Thanks for the correction I guess.
My mind slipped when I was trying to think of Excavator based APUs.
>>54117874
And a slightly conservative 30% IPC increase for single-threaded performance means that it will best a stock Skylake i5-6600k by about 9.6%
BUT
That IPC increase doesn't mean it will apply to single-threaded performance. That could mostly apply to multi-threaded performance, which means that Zen might match the 6600k in 4-threaded performance only.
>>54117874
the 95W TDP is iGPU right?
shouldn't it be compared with an APU for a more fair assessment?
I want to upgrade AMD. I don't want to wait so much. At least make Polaris and Vega a killer to quench my thirst.
>>54117986
There is no Excavator APU out yet, plus the A10-7860k is a 65W TDP APU that stays within 93% of the 95W A10-7850k, which is a 95W TDP APU
You could try comparing the X4 845 with a Xeon E3-1230 V5, which is the quad-core i7-6600 without the iGPU. That ranks in at 1956 for Single-Threaded performance.
>>54115439
No excavator chip has an L3 cache fyi.
>>54117996
Same here. Really, the only thing holding me back from going AMD was the fact that they still use ddr3 in 2016, and their processors still don't run on FinFET. Now that that's finally going to be over with zen, amd can be an actual option beside intel.
>>54118234
I don't know about you people, but I will only switch to an AM4 platform if it supports M.2 and USB3.1 natively WITH ITS OWN DEDICATED LANES. I do not want to sacrifice SATA ports, PCIe slots, or rear I/O ports because I end up needing to use all of them. The reason why I'm skipping Skylake and waiting for a X99 replacement is because they either take away PCIe3.0 lanes from the slots to provide those functions, or they don't support them natively and use PCIe lane multipliers that run at a hefty price premium.
>>54117513
Its over, they are finished. They should make graphics cards.
>>54118599
>They should make graphics cards
They tried twice. The second attempt became the Xeon Phi
>>54118386
>I don't know about you people, but I will only switch to an AM4 platform if it supports M.2 and USB3.1 natively WITH ITS OWN DEDICATED LANES
It probably will, because they have no reason not to. It's fucking 2016 already, intel has been using ddr4, pcie, usb 3.1 for years now while amd is still stuck in the past. If they don't add native support for the latest technologies they'd be deliberately gimping themselves. Nobody, and I literally do mean nobody would buy amd if they don't support the latest technology standards, except for people who want to build a pc below $999 I guess.
Really, intel's main issue is the fact that they can't stop being jews for a moment and release their products at a competitive price to combat AMD. If only intel stopped being literal jews for a moment and put out their processors at a good price, they would've blown AMD out of the water ages ago.
>>54114350
how long until this comes out?
>>54118804
USB3.1 and M.2 ports use the PCIe lanes from the chipset rather than having their own dedicated lanes. So for most Z170 motherboards, if you use an M.2 port, it takes away x4 PCIe3.0 lanes from your PCIe slots, which more often than not are the only PCIe3.0 slot remaining because those lanes are shared with the gigabit Ethernet NIC, USB3.0/2.0, and USB 3.1
Kaby Lake and the Z270 chipset will put USB3.1 on their own dedicated lane from the chipset, but probably not for the M.2 port.
AMD can win me over by offering dedicated PCIe3.0 lanes for the M.2 and USB3.1 so that I can still use all the PCIe slots on the motherboard without having to worry about lane budgeting.
>>54118804
>they would've blown AMD out of the water ages ago.
Intel doesn't want to do that, the need AMD to stay afloat
>>54114350
Please be true.
>>54118954
I hope so too bro.
>>54118932
Actually, not really. If AMD dies (which it probably never will, people kept saying it will many times, but it never did), then intel takes over the entire market and basically monopolises the CPU market. Consoles will have to resort to intel for CPUs. So will data centres. So will pretty much everybody, since intel is the only choice.
>>54119033
>intel takes over the entire market and basically monopolises the CPU market
Exactly what they want to avoid.
>>54119081
Why?
>>54117243
DELETE THIS
>>54119110
cuz they can pick whatever prices they want and u have no option but to buy it famalam
>>54119169
Why the fuck would intel want to avoid that? Isn't that good for them?
>>54119110
https://en.wikipedia.org/wiki/Breakup_of_the_Bell_System
>>54119204
sry lmao thought why a customer wouldnt want that
afaik for intel if amd goes belly up they'd have to buy a billion patents or some shit, quite a lot of the cpu tech is patented by amd
I just bought a used E5 2648 v3 Xeon for 200 burger coins.
Fuck me if I ever have to upgrade, ever.
Are they still planning to make that ambidextrous cpu?
I really want to see a amd64-arm64 hybrid.
>>54119233
>quite a lot of the cpu tech is patented by amd
AMD sits on a true mountain of patents relating to cpus and gpus which makes the scenario of them dying a clusterfuck as suddenly a shitload of much, much bigger companies no longer hold the rights to make their shit.
>>54114350
I'm not a gaymer so I don't care.
>>54119262
What benefit would that be? I can't think of a single reason why a desktop PC would use x86 over ARM apart from running some obscure software that only runs on ARM.
>>54114617
>>54114821
Where did the "professionals use AMD" meme come from? In my office not a single one of the computation guys use AMD, only one of the rendering guys use AMD (He is going to be fired soon too as it takes longer for him to render shit than others), and none of the system admins use AMD. This stays consistent with every office I go to, where did this meme come from?
>inb4 guy getting fired
Guy refuses to get a better processor, we have no other option. Dudes a fucking idiot.
>>54119081
>monopolises
How stupid are you people to think intel is a monopoly, you are literally just butthurt that they charge you more for the best. That's how business works, if you dominate the CPU market you can charge more for better.
>>54119361
I'd imagine I could assign certain processes to run on the ARM cores exclusively.
For example a torrent client, that way I can keep my torrents running and when I'm away the x86 cores can shut down.
>>54119363
It might have something to do with AMD being used in a special effects company back in the day
I still have no idea and I've been here for years
>>54115096
He just works more or less freelance now for the highest bidder and if Zen is any good he can charge a good chunk more for future projects. What a glorious cunt.
Is there actually any competition besides Intel/AMD? Some sort of hipster-cpu brand?
>>54119392
There are three companies producing x86 CPUs and VIA's virtually nonexistent, if AMD goes down Intel's in trouble
>>54119480
VIA?
Do they still exist?
>>54119480
VIA
>>54119480
MCST from Russia
>>54119410
But does Zen have dedicated ARM cores rather than emulated or virtual ARM core support? That seems like a waste of die-space for a function not many people would be willing or able to use since ARM CPUs are starting to proliferate the server and workstation space at lower costs. It's probably better to have a separate ARM and x86 machine rather than rely on some hybrid that may or may not have physical ARM cores that are swapable on demand.
>>54115015
>>54115327
>He was there for 5 years.
This is the incredible thing about this, he was 5 fucking years, his last time he was a single year
>>54119363
>rendering on a workstation
>professional
sure m8
I don't need to upgrade but I'm assblasted with Intel changing socket every fucking year making my mobo obsolete by the time I build my computer.
>>54115342
Apple has poached quite a lot of good engineers from everywhere, they might be unable to get another giant leap but they will keep up for quite a long time
>>54119540
Yeah it's a cock move but they get away with it just like crApple.
>>54115015
He left once the work was done and taped out. Zen is complete.
Hopefully you're a little less retarded now.
>>54119361
Well for one, ARM is a lot more energy efficient than X86 ever will be. ARM is also a lot more efficient at using its cores, is a better platform overall and it's compatible with a lot of software and hardware, unlike X86. The only thing really keeping ARM from coming to the desktop is that its performance just isn't quite there yet. However, ARM is better than X86 in every single way possible, except performance.
>>54117959
>That IPC increase doesn't mean it will apply to single-threaded performance. That could mostly apply to multi-threaded performance, which means that Zen might match the 6600k in 4-threaded performance only.
Except it does
>>54116381
>there's people out there who fell for the Pentium meme and can't even run games nowadays
I found that even the g3258+cheapestdGPU was more expensive than a Kaveri A8 where I live, the only accepatble quad core I could get was with AMD, the only Intel offerings at those price points where Atom's
>>54119606
Do some research on it, it wasn't fully completed, some Pajeet with 0 accomplishments was put in charge of it after Keller left, I guess that's the best they could do, hope you feel more retarded now
>>54119536
These machines are worth thousands of dollars and are used to design fiber optic gyroscopes nad wafer architectures. Our servers are no different, they don't use AMD either. What the fuck do you consider a professional .
>>54119620
What
The
Fuck
>>54114350
>meaningless buzzwords except for maybe DDR4
>>54119671
Do your research. X86 is outdated garbage, the only reason why it still exists on desktop is because there is no alternative yet.
>>54117223
He was a single year the last time he worked at AMD, we all know how insanely great things went
>>54118774
They're doing a third round, it kinda has finally worked out, in their fuckexpensive Iris Pro offerings
>>54118804
>intel has been using ddr4, pcie, usb 3.1 for years now while amd is still stuck in the past
Oh, I forgot you were stuck in 2004, when AMD only offered AGP
As for the rest it's been less than a year of DDR4 and USB 3.1
I really hope AMD delivers. We need higher IPC for teh emulatorzzz!!1!!11
>>54119668
Professionals don't render on workstations, they render on render farms
Any workstation is worth thousands of dollars, as any server is, unless you live of equipment that has been dumped by some company like /g/ does
>>54119678
Here's your (You)
>>54115096
I don't get it, how can one person be so much better than anyone else in the industry..? Was he born with Mozart-level cpu-design skills?
>>54119620
>Well for one, ARM is a lot more energy efficient than X86 ever will be
Only because Intel puts performance first, THEN works on efficiency. ARM is the other way around. An instruction set might have some impact on efficiency but to say x86 will "never" be as efficient is just a false assumption.
>>54119702
>because there is no alternative yet.
>what are SPARC, POWER, MIPS, Alpha, Itanium, and the numerous other home grown high-performance architectures that either failed or were relegated to niche markets because of incompatibility with existing software
There have always been alternatives, but in the end nobody gives a fuck about marginal performance gains when there's nothing to perform on.
>>54119764
>Professionals don't render on work stations
we do, and we are the best on in the market, every internet provider uses our tech, everyone uses our stuff
>>54119776
>Only because Intel puts performance first, THEN works on efficiency.
It hasn't been like this since Sandy Bridge
>>54119783
All of those bar Itanium where quite entrenched in the server market until commodity hardware started to become the norm
>>54119783
>Itanium
oh yeah, when intel failed to defeat intel
>>54119620
>ARM is a lot more energy efficient than X86 ever will be
>Double niggers actually believe this.
Haha. Oh wow.
>>54119811
>>Only because Intel puts performance first, THEN works on efficiency.
>It hasn't been like this since Sandy Bridge
Intel got performance even as far back as Nehalem, and then has been steadily increasing efficiency since.
I mean, the performance isn't great, but there's an 8 core atom with 12w TDP.
>>54119822
AMD, actually. IA64/itanium flopped, leading Intel to eventually adopt AMD64.
>>54115015
https://www.youtube.com/watch?v=eDmv0sDB1Ak
They were probably bug testing the CPU brah its hard..
>>54119859
>AMD, actually. IA64/itanium flopped, leading Intel to eventually adopt AMD64.
This is literally the main reason why x86 beat every other arch in the server market
>>54119764
>Professionals don't render on workstations, they render on render farms
not really, maybe huge jobs, but the whole idea of a workstation is to take load off of that infrastructure in the first place
>Here's your (you)
are you trying to tell me that anything in that OP was actually meaningful on its own? having moar cores and threads isn't going to matter for shit when they're weak as fuck and your applications aren't optimized or even suited for them
>>54119897
>are you trying to tell me that anything in that OP was actually meaningful on its own? having moar cores and threads isn't going to matter for shit when they're weak as fuck and your applications aren't optimized or even suited for them
>a jump to the latest node isn't anything meaningful
DDR4 is by far the least meaningful thing of all the things listed in the OP
why is this dude named Lisa? do you need a gender bender name to be CEO of a company in this SJW world?
>>54119811
>All of those bar Itanium where quite entrenched in the server market until commodity hardware started to become the norm
You are absolutely correct, even the Itanium was reasonably entrenched in supercomputing while Intel still cared about it, however the point is that even though these superior alternatives existed, they didn't trump x86 because performance isn't the end all be all issue that gamer faggots on /g/ think it is, compatibility is ultimately what matters in desktops and low-end servers.
>>54119929
>>a jump to the latest node isn't anything meaningful
Not when it's old news.
>DDR4 is by far the least meaningful thing of all the things listed in the OP
All the cores in the world don't matter for jack shit when they're gimped by slow storage.
>>54114350
I have been using AMD for 10 years and can definitely say that we've always sucked intelgods dicks. Recently I finally bought Intel based PC. The difference in performance is significant. But guys try to understand me, I was just poor and that's why I tried to fool myself and every time talked to everyone that you can buy 100500 core processor for cheap blah-blah-blah. Who I tried to deceive? I wanted just to be as normal as intelgods, I didn't want to be a fucking MAD-peasant. I wanted to be on the waves of their glory, the glory of people who have real computers with real veritable CPUs, not this slow, hot, stupid toy MAD shit. On behalf of all MADcucks I admire your superiority. Shame on me. I'm so fucking sorry. I'm sorry.
>>54120023
Not even DDR3 bandwidth starves current CPU's, there's literally no reason to get DDR4
>>54120036
Thank your pajeet, 10 rupees have been deposited to your account.
>>54120036
Lmao. Love how you can't have a single discussion in /g/ between two companies without it eventually being a shilling fest.
>>54120129
>>54120036542
If SB IPC is achieved, if the IHS is soldered and if good boards are released, with overenginereed VRM's and big heatsinks on them, i buy ten, twenty zenver1 chips no doubt. Unfortunately, AMD decided to join hardware backdoor trend with PSP, but fuck it.
If not, i'll stay on 1100t's, FX-8350's and i5 2500k's forever. They already do emerge -e @world in reasonable time and they all have no problem running SHOC (CoP and Lost alpha are slightly bottlenecked by [email protected] and [email protected]). Arma2 on the other hand, suffers even with [email protected], when critical amount of AI and scripts are active.
>AMD
>Hyperthreading
I just want AMD to be competitive again.
Their APUs are pretty good bang for the buck for Office builds.
>arm
>>54120213
Simultaneous multithreading is AMD's version. Hyperthreading is just a marketing buzzword, multithreading exists for quite a while.
>>54120051
This is wrong, though it's not about bandwidth, but latency.
Every generation of DDR after DDR1 has been worse and worse at latency until DDR4 3000Mhz. No other DDR has been good until now. At 3000Mhz+ you see big performance gains, and 3000Mhz is pretty standard for DDR4.
>>54120358
Someone tell me if this is true. I have some low speed DDR4, and have been thinking about getting something faster.
>>54120428
It's kinda true. Latency does decrease when you buy ddr4 memory around that clock speed, but it has nothing to do with the clock speed. It's just that good ram chips with very low latency are usually factory clocked at 3000mhz+.
>blaha Intel
>blaha AMD
Mfw I don't discriminate
>>54119647
>Pajeet with 0 accomplishments
is this why all their cpus are shit and have poo in loo names? God I hope China nukes the shit out of India. Worst country and worst humans.
>>54120358
Not quite. DDR2 was a laughably massive improvement over DDR.
It took DDR3 a while to beat higher end DDR2 kits in latency, but it still happened. CAS 10 and 11 2400mhz DDR3 has been a thing for a couple years now.
Recent DDR4 kits have been making serious headway though. CAS15 3600mhz DDR4 is pretty damn impressive.
Not many workloads directly benefit from latency improving 1-2ns, but many will improve from the higher throughput.
>>54119647
They have had some pretty good pajeet teams, the ones behind the cat cores was a team made entirely of pajeets, and all of those cores were extremely efficient and pretty competitive compared to the *dozers
Then Samsung poached the entire team
>>54120469
Is it worth the difference to spend $80-110 on new ram or is it insignificant outside of benchmarks? I'm just a filthy gaymer anyway.
>>54120485
If you nuke it, then AMD is over. If you don't know, the way AMD was conceived is a bunch of inidnas got together and decided to reverse engineer intel products and rebrand them as their own. Kinda fitting that they still uphold the company legacy of rebranding their hardware constantly.
>>54120514
Are their new APUs going to see huge improvements thanks to the DDR4 RAM?
>>54120485
>>54120533
AMD is american retards, and older than Intel
When will the shills stop?
>>54120532
Should mention mine run at 2133
>>54120533
damn now I feel dirty as fuck for buying AMD stuff
one rule we have in our business is to never ever shake hands with indian clients because even the elites have the subhuman habit of using their hands to wash their assholes
>>54120532
Literally so small, not worth the extra $. You'd be getting more of your money's worth if you just bought another video card to run in crossfire/sli or a better card model that can handle overclocking better. Hell, even a better cooler is a better option than spending money on high end memory. It's kinda like SSDs, sure it'll make your overall PC faster, but it won't make your games lightning fast or anything (at loading, caching, etc. yes, but no actual fps gain).
>>54120542
Carrizo/Bristol Ridge has the same memory controller that Kaveri has. It was purchased from a 3rd party instead of being developed internally, and it can't deal with high memory speeds. 2400mhz is about the absolute maximum without pushing baseclock higher, and if you do that it begins to effect other system components. Some of the highest memory overclocks with Kaveri are only around 2600mhz~
Bristol Ridge on the desktop won't be able to use any good DDR4 kits.
Everyone is hoping that AMD designed their own memory controller for their Zen based parts, because that would be a crippling downside if memory were to be so heavily limited again.
>>54120575
Alright good info, thanks.
>>54120590
Are the carrizo Athlons good for a budget gaming build? They are much cheaper than i3s and even Pentiums where I live.
>>54120619
They're not great.
>>54118929
Care to explain what you need those lanes for?
your GPUs aren't using them all by a longshot,
>>54120619
>Are the carrizo Athlons good for a budget gaming build?
$65 on a good sale for a true quad-core that can match a $120 i3-6100 in multithreaded applications?
Yes.
>>54120640
yeah but it's for cheap ass steam games that I'll play whenever I get time to kill at workl. Not even fucking MOBAs, just stuff like Don't Starve, Orcs Must Die! and whatever Trine comes out next.
>AMD can't win a price war with Intel either.
you vastly overestimate intel's fabs and understimate globalfoundries
>What other realistic options are left?
if the core count is the same and the zen chips perform less than 10% worse than kaby lake, zen will be at a small advantage given that all FX chips at all price points have unlocked multipliers, all features (AMD-V etc) built in yadda yadda
also, zen will have more chipsets components built-in than intel, and AM4 is designed to have a super long lifespan. another minor advantage in zen is being able to stick a brand new chip into a 5 year old motherboard with little to no penalty
that is, unless they pull another bulldozer and release 125w TDP chips that can't be used in older motherboards of the same socket, which was extremely stupid
>>54120645
SFF-8087-to-SATA RAID cards and a dual 10Gbps NIC card.
I want to put an PCIe-to-M.2 adapter card with a heatsink for both of my SM951s, but I have no room. For now, my M.2 port is occupied by a lowly XP941
Wait is it true that cpu industry is fucked? I wanted to work for intel or arm.
>>54120663
>>54115206
forgot to link
>>54120645
Also remember that PCIe storage is becoming more and more mainstream in enterprise servers, so if Zen wants to have a chance in the enterprise market, they need to up the PCIe lane capacity to allow for at least 9 PCIe 4x SSDs or 4 PCIe 8x SSDs and 5 PCIe 4x SSDs.
>>54120663
In the long run it usually was go with AMD if you want to save more money and get a good rig, or go with intel if you want to get the best tech at the moment and burn a hole in your pocket. Both companies fill their certain niches.
>>54119480
Somebody is building a consumer mainboard for power8 which is what google is moving towards for there servers but it'l hardly be competition because the only software that will run on it will be that with available source. Microsoft helped IBM fuck itself so hard and I don't think they will ever voluntarily venture into the pc market again.
>>54115639
>tfw make -j32
>tfw ffmpeg
>tfw more VMs
Fucking this. Yeah if you're a man child gaymer single core performance is all you care about. If you use your computer to do work your workflow can benefit from the additional cores. Gaymers like to point out that no software takes advantage of all those cores. They forget about the one program almost everybody runs: the operating system. One of the main jobs of the OS is to assign all of your processes time slices on a CPU. Moar coars means shorter queues for those slices.
>>54120967
>if you're a man child gaymer
I'm a man child gaymer too and care about multi cores. Mostly because I like to emulate shit instead of playing actual PC games. I don't care about the latest AAA title shitfest, they're all buggy, unoptimized, shitty cashgrabs.
>>54114350
If they get something with Intel IPC and 8 cores for around 400 € I buy it. Otherwise AMD gtfo.
>>54120967
>more VMs
I'm running a VM server that basically acts as the primary gateway to the internet. All of the browsing, downloading, VoIP, and streaming is done through those VM clusters. I'm not convinced that Zen will be able to beat my Xeon V3 processor, but it would be nice to see other non-Intel options if I need more than 8 threads.
>>54114350
You forgot
>2017
>>54115206
Intel has been doing nothing but 5% increments for the past six years so it is entirely possibly for AMD to catch up to them in performance.
I just hope they'll use a fucking LGA socket this time.
>>54121127
>I just hope they'll use a fucking LGA socket this time
nope
http://wccftech.com/amd-am4-%C2%B5opga-socket-1331/
it's still a PGA, but the pins are now even smaller.
>inb4 >currytech
It's been confirmed by shipping manifests of the prototypes. See the header in the article
>>54120358
Latency has remained about the same. The latency specifications given in clock cycles increase but the actual latency measured in time units has remained pretty steady. That is
latency in clock cycles/frequency.
>>54114446
I like my FX-6350
>>54121268
I just grabbed an i5 during black friday. fuck AMD.
>>54119363
>He is going to be fired soon too as it takes longer for him to render shit than others
Kek, you joking m80?
>>54121127
Well AMD said that they're going to use Zen for the next 5 years before they feel the need to create a new architecture (they'll make zen+ in the meantime, which is just a factory overclocked zen). They seem to have a ton of confidence in zen, I dunno if it's actually that good or not. I mean it's been in production for like what, 6 years now? It must be good, AMD has been spending a literal eternity developing this processor.
>>54114446
>DX12
>Vulkan
>programs that actually make benefit of many cores
If zen isnt a total flop and has a good price-performance ratio, shieet im sold
>>54121315
>production
shieet, meant development.
>>54119765
If zen succeeds then yes. He is the mozart of CPUs
>>54114350
I'm getting excited when they're on the market and benchmarked.
>>54119765
Kinda. He made the athlon 64 and saved AMD's face hard with AMD64, which to this day everyone uses. He's like the linus torvalds of the cpu industry, amazing at designing cpus, not kernels.
>>54114446
>15 unused cores
Shit, give this man a single core CPU with 6ghz, amirite?
>>54121301
I think post just cut me
>>54121268
>>54121301
how much did you each pay and what are your respective average FPS measurements for CSGO,? How long does it take each of you to unzip a 1gb file? What CPU utilization do you get running handbrake on a 4k transcode?
AMD is so retarded. Why didn't they just make a single core CPU with a really high clock rate?
>>54114350
If amd is so good, why is everyone using intel cpus then?
>>54121479
I hope you guys are right and he's some kind of wizard who wonders around designing miracles. I want my next machine AMD.
>>54121775
good reasons that you didn't want to hear but which I'm going to give you, pretending like you were serious and actually wanted to learn:
1. Intel was proved to use unfair business practices to keep AMD out of consumer's hands and it is extremely likely that these practices continue in some of the largest markets today.
2. Intel performs better than AMD in terms of raw performance, which people believe is what they need, though 90+ % of them would be better served by the superior cost/performance ratio AMD offers.
3. Intel manages its brand image better than AMD.
4. Intel started the x86 race ahead of AMD already and has consistently maintained a superior economic position.
5. There are legitimate reasons to use Intel instead of AMD because of the differences in design which exist between their products.
Except for #5, what it all boils down to is that Intel maintains brand image as "premium" compared to AMD - the same reason Applel beats out Android. People substitute blind brand loyalty for product knowledge because they find it too difficult to actually understand the product.
>>54121834
There are a handful of people who can be credited with shaping the computing world. And the rest of the world with it.
Jim Keller is one of those people.
>>54119668
Apart from your obvious bait. Im willing to wager a 9590 paired with some very beefy gpus will render pretty close to 5960x.... Seriously GPUs are the major factor even for rendering these days. And the 9590 has 8 "half cores" even if the ipc is shit, 5ghz lightens that somewhat.
But again i can tell your retarded by "WE USE INTELS HE USES AMD, INTEL GOOD AMD BAD" as if intel and amd are just single parts or "professional" workplaces just let you buiod your own computer with whatever you want. Then just fire you if they don't like it,instead of just replacing it like a normal company
Why hang around when you can hire pajeet to tweak the cpu for another 10 years?
>>54121946
Idk though, im not sure even jim keller can compete with intel when he is locked down to Global Foundries, they are thr main thing holding amd back really. Should of done anything in their power to keep their fabs open.
>>54121917
Thank you i was really curious. I heard that amd cpus might cause problems with some virtual machines (or maybe it was only hackintosh), so is it possible that amd cpus cause any problems or crashes that intel ones don't have or is it mostly the same?
>>54121834
I don't remember which one, but Keller helped Apple with the A* chips.
>>54121946
How even? All AMD did was capitalize on Intel's marketing strategy that preyed on "muh big numbers" mouthbreathers
>Not buying a 5 year old E5-2670 for $70
Shiggy
>>54121775
Because amd hasn't released a new mainstream part in 4 years.
And because >>54121917
>>54114350
So does it have 32 logic cores (or however they're called) or just 16?
>>54122206
Probably. But it's probably going to be the same way intel has 22 cores - Only on the chips you wont be buying.
>>54122206
32 logic 16 physical, since they implemented SMT in Zen
>>54115421
The only thing AMD dominated in 2000-2006 was the gamertard market and certain HP OEM systems
Anti-competitive practices aside, the Athlon MP was the sorriest excuse for an enterprise x86 chip ever and the Opteron was kickass for all of like five months before Intel rushed out their own 64-bit implementations that matched them or maybe even surpassed them in applications that used Intel's spacious L3 caches featured on Xeon MPs and later DPs
>>54122206
Dies up to 16 core with 32 core dual die Opterons (MCM, right?)
>>54120619
Get 860k. Overcock to 4.5ghz.
Or 845 for that excavator core.
>>54122224
Yeah they probably won't even fit in my 1150 socket anyways
>>54122240
32 cores would be awesome for blender rendering, since my gpu doesn't really seem to work.
>>54122122
AMD CPUs are quite stable. You can't always run hackintosh on them, but that's not AMD's fault, that's the fault of shitty Applel.
Some virtual machine softwares used to have trouble with AMD, but literally any virtual machine software worth running today will use both AMD and Intel hardware virtualization technology implementations equally.
>>54122120
>Should of done anything in their power to keep their fabs open.
Ironically enough.
GloFlo are technically AMD's fabs.
I believe it was Hector Ruiz who spun AMD's fabs off. And placed himself as CEO of that new company, while also remaining (For a short time) as CEO of AMD, he forced a long-term contract, locking AMD to GloFlo until 2020 or somewhere about there.
I believe he was arrested for Insider Trading.
Keller has been behind the majority of computing breakthroughs in the past two decades. Starting with AMD64. And continuing on to develop dual core processors thanks to the AMD64 instruction set.
His work in later years with Apple proved to the marketspace that ARM processors could indeed compete with x86, as the A7 and A8 chips both hit heavy blows to their Intel Mobile Core counterparts at the time. Intel were quick to respond and correct the situation however.
From all of his previous work, and his reputation in the industry and just how much companies want him, even for a few years, is sign enough that his being somewhere means to expect big things.
>>54122128
>All AMD did was capitalize on Intel's marketing strategy that preyed on "muh big numbers" mouthbreathers
Big numbers were all that mattered in the early 2000's. Technology had yet reached the point where architectural efficiency meant something. This was a time for innovation and invention in the computing world, power consumption was not a concern. Ghz were king. And Cores count meant the world to businesses, banks especially as they could expand their systems 10-fold with this newfound computing power.
>>54122256
if you plan on overclocking that high, don't get the 860K or the 845 (of course with the 845 you don't need to overclock as much).
The reason is that the 860K runs with the bad TIM between the heatspreader and the die, leading to poor thermal characteristics. You should get much better OCability by going just a step up, to 870K, which is derived from Godavari and therefore uses a soldered material between the heatspreader and the die.
Please note I'm not talking about the heatsink you put on top of the CPU's heatspreader+die package.
>>54122120
Stop being retard.
Gf knew they're shit and licensed samsung.
>>54121834
Athlon64 killed pentium 4, intel had to resort to bribery.
Apple bitch slapped quallcomm they had to housefire their chips.
