Core i7-6950X
Images are sometimes not shown due to bandwidth/network limitations. Refreshing the page usually helps.
You are currently reading a thread in /g/ - Technology
You are currently reading a thread in /g/ - Technology
>AMD shills on suicide watch
Damn this shit's expensive but 10 cores... See Intel can do it too
http://www.methodtech.co.uk/2016/04/intel-broadwell-e-lineup-confirmed-x99-support-i7-6950x/
>>54030905
That's impressive. What the fuck is anyone going to do with 10 Cores and 20 HyperThreads though?
>>54031056
It may not be super utilized now but in the future applications will become more and more threaded as cpu's like this become the norm
yay, I can finally run chrome at an acceptable performance
>>54031056
Maybe finally play GTA IV at 60FPS max settings.
>>54031056
Virtualization.
Seems kind of pointless. Who's going to spend $1500 on a consumer grade CPU?
>>54031106
LOL this is what I want. Better GTA frames. [email protected] and still only getting 50-80 average in the city. Drops down to 40 or so in certain areas or if explosions and stuff are happening around a lot.
>>54031098
That's what they said about the FX-8350
>>54031756
We just need more 6 and 8 core cpus in the casual market. little optimization was due to intel not having a mroe than 4 core on the mass market. optimization will happen soon since both manufacturers are due to have 8 cores on the normal market soon.
for gaymes any i7 performs the same if not even better. For muh performance xeons are much cheaper. This is some fancy toy for rich kids
>>54031756
And the FX-8350 works great. The more CPU demanding something is the more multi-threaded it is generally.
>>54031056
>>54031098
any good parallelized code will detect how many cores are available and spawn as many threads as needed
>>54030905
>10 cores
We don't need moar coars, 8 was fine, we need a lower TDP and higher frequency on the 8core variant.
Let me know when Intel ships a 4GHz 8core CPU.
>>54032013
I think you're getting hit pretty hard by Amdahl's law with 20 threads, though. Unless you're either doing something that's trivially parallel or multitasking heavily.
>>54032002
not in 99% of cpu intensive games.
>>54032251
Are games hard to write parallel? code for, or do devs just not do it?
>>54032326
>do devs just not do it?
pretty much this
>>54032326
development is hell
honestly how most studios work it's amazing they make things work at all
>>54031606
You can count on LinusShillTips.
>>54030905
why?
if amd pushes 8 core for mainstream, which they legitimately could, its going to completely shit on intels line up, then once price comes into play, shit on intels lineup again, and when the 16 core single chip for work stations comes along with some of the core locked for binning, shit on intels offerings yet again.
all we have to do is hope amd pushes the price down like rumors are speculating at.
remember, amd expects no growth at all in the consumer space, that may be just profit or could me market penetration, this means they could be selling the 8 core well below what they could.
>>54033742
>and when the 16 core single chip for work stations comes along
You do know Opterons are a thing right?
>>54033754
current ones suck dick though.
>>54033761
Current ones are on a 4 year old Arc, and an even older process. They've dropped in price significantly enough to make them a strong choice for workstations and servers where you might not be able to push for the budget you need for Xeons.
Opterons may not be as good as newer Xeon models, but they by no means 'suck'.
>more cores for muh video games xD
>>54033742
>its going to completely shit on intels line up
is it the early 2000s again or did I have a stroke?
What are you smoking anon, AMD hasn't been relevant for almost a decade in the CPU space.
>>54033784
>AMD hasn't been relevant for almost a decade in the CPU space.
Which is another way of saying
>There have been no notable advancements in CPU tech in a decade
Intel completely fucked the market with that stunt they pulled, slowing advancement towards more advanced CPU tech by possibly decades, if we're lucky. Could be more.
>>54033804
Blaming intel because AMD can't manage to come out with a decent architecture. This is how you know you've found an AMD shill.
Come on man, intel might be a bit lax on IPC improvements over the past few years, but only because they have no incentive to do so. How is it Intels fault AMD can't compete?
Your comments about AMD being able to shit all over the Intel line up are just weird, you must live in some fantasy land where AMD CPUs actually hold value.
>>54031098
That wont be the norm in this chips lifetime
>>54032251
You're living in the past, friend. Great strides have been made over the past couple of years to make games much more heavily-threaded. You can thank the shitty laptop-grade CPUs in the new consoles for that, with their eight cores and 2005 IPC. It's the reason why the G3258 looked a great buy at launch, yet was a complete joke that gaymes would refuse to even launch on six months later.
Even the can't-into-PC-development Japs are getting in on it. Dark Souls 3 runs great on my 1090T and makes good use of all six cores.
Has Intel gone too far?
>>54033823
>Blaming intel because AMD can't manage to come out with a decent architecture.
No. Blaming Intel because they deprived AMD of billions in income, forcing them into their current deprived state.
The mismanagement of AMD is an unrelated matter, but still one that has caused this decade of stagnation in which we now sit.
It's Intel's fault that AMD can't compete because Intel is the sole reason that AMD is unable to produce anything competitive.
>Your comments about AMD being able to shit all over the Intel line up are just weird
Different anon here m80.
AMD 'shitting' on Intel is a pipe dream, that will never happen again, Intel were smart/stupid enough to see to that matter in the early 2000's.
>you must live in some fantasy land where AMD CPUs actually hold value.
Everything has it's place.
I said this to an anon in a thread yesterday. Set yourself a budget of $600, go to PCpartpicker, and see what you can do.
Ultimately, the choice will be
>i5 and a 960
vs
>8350 and a 380x
The i5 has significantly higher single threaded performance, so while using the same GPU, it will outperform any AMD processor in a game. But the extra money spent on the i5 over the 8350 could have gone towards a notably more powerful GPU, which will make a world of difference in framerates vs a change in CPU.
I will not deny that Intel CPUs are superior. But to insist that AMD are useless in the market, that their products have no place, is just naive.
>>54033823
Completely missing what the anon said it's how you know this is either b8 or full blown underage
>How is it Intels fault AMD can't compete?
Bribing the shit out of OEM's and destroying the market when AMD pulled ahead, destroying the R&D budger AMD could had amassed with the Athlon 64 hit
>>54033844
>Great strides have been made over the past couple of years to make games much more heavily-threaded
Pretty much this, most stuff runs at multiple threads nowadays, thing it's that despite being able to harness SMP systems shit's still slower than it used to be, devs are completely incompetent
>>54033844
>Not 0-Indexed CPU count
REEEEEEEEEEEEEEEEEEEEEEEEEE
>>54033901
There's a bunch of edge cases where AMD is better for price/performance but that's about it. I really want to like AMD and buy their shit but it's difficult to find a build where they are the superior option.
>>54033901
Anyone buying AMD right now is a retard. They have no current platform and AM3+ boards are limping along using 3rd party controller chips for USB 3.0 and there is no native M.2 Support, etc.
Anyone buying an AMD platform today is god damn retarded with Zen dropping later this year and all the current Intel offerings.
AM3+ hasn't changed in what... 6 years? And that's the most recent platform AMD has to offer? FM2 socket is dead unless you're building an APU rig, which i woudlnt recommend to anyone except for very specific use cases. Gaming isn't particularly one of them.
Zen might be okay when it comes out, and a year ago buying an AM3+ board would have been okay for a budget build.
But right now in mid 2016, buying an AMD platform is just pants on fire retarded. AM3+ and FM2 are dead platforms with no future.
>>54033776
nah, cheapest 16 core amd i can find is in the 400-500$ range, and an even cheaper intel xeon in the 250$ gets close enough to nor warrant the cost for the amd. you may be able to find a 4 cpu amd board, and could possibly make up the cost difference if you spent it on 64 cores then getting a similarly performing intel setup... but still... for my money till zen i would go intel.
>>54033804
person who co authored x86_64 and lead designed amds best cpu is behind their newest architecture.
what we know is its at least as good as sandybridge at worst, and depending on the leaks/rumors, its putting it closer to current gen intel then older.
if amd goes for price competition, intel has nothing as value wise, who needs 10 cores on a consumer desktop? why not go a xeon if you need the power?
amd is tapping out a 16 core cpu,
and at their highest end, putting 2 16 core cpus in one socket for a 32 core cpu, so its possible to get disabled core 16 core chips cheaper and making up an entire enterprise lineup.
we have a reason to hope, and the best/worst thing is, intel sat on their dick so long that amd can catch up.
the first new architecture sense bulldozer as everything else was iterations.
>>54033844
amd will push almost every game to 60fps, the issue is the stronger single core of intel will get most games to 120+ fps unless there is a gpu bottlneck or the game is horribly programed.
dx12, mantle, vulcan fix some of amds problems and bright their "8" core to i7 levels, but there are still issues with them just not being strong enough.
as for the pentium, anyone stupid enough to go dual core deserved to get fucked, we told you for so long 4 core minimum and now some games wont run on anything less than 4 threads... sure they COULD run on the cpu, but the devs are to lazy to do it.
>>54033901
the new architecture is VERY promising.
>>54031106
>>54031725
>>54033844
You realise this is gpu related
yes? /g/ cant be this illiterate?
>>54033956
no, its a cpu bottleneck in many cases.
gta5 is an icp demanding mother fucker.
>>54033949
unless you need the power now, dont get a cpu
unless you build dies, dont get a cpu
if your build dies, you may want to get a stop gap instead of a cpu.
amd could drastically change what cpus cost.
>>54033956
>/g/
All I see in this thread is /v/
>>54031056
Currently not much but it comes to show that intel wil lstill be in the game if devs come to their senses and start doing async and multi core/thread in everything.
>>54031576
Xeons and opterons mate
>>54032036
Amdahl's law isn't as big of an issue as people make it to be
>>54033956
In GTA IV and V many areas are CPU intensive
>>54033950
>person who co authored x86_64 and lead designed amds best cpu is behind their newest architecture.
>the new architecture is VERY promising.
Oh I'm well aware that Keller came back to work on Zen. I didn't even mention Zen, as it would cause an unparalleled level of shitposting in this thread with WCCF links being thrown everywhere and rumour being equated to fact.
We don't know a damn thing about Zen except that it's going to have SMT and it will be on GloFlo's 14nm FinFet.
AMD are very good at keeping lips tight lately. Which is a good, and bad thing.
Very good in that it keeps the community interested and eyes are on them when we near product releases.
But very bad in that people just speculate without any ground to stand on. Look at the Fury lineup of GPUs. All we knew is that it would use HBM. And that's all we knew until it launched, not even it's name. But still people flaunted it as the "Be all end all, Titan killer supreme ultragigganigga x500000" and this led to undue hype, and ultimately disappointment for the people that believed any of the rumours that spread around Fury pre-launch.
While I don't think AMD should spill the beans and do a full paper launch of Zen months in advance, some mention of it's specification could go a long way to influence investor interest.
>>54031056
Vegas or Premiere would probably be able to render 1GB in about four and a half seconds with that many logical cores.
>>54034003
*tips fedora and makes a smug face*
>>54034067
Are there not already 12 core Xeons with hyperthreading?
This i7 is sort of null and void all things considered.
It's just Intel's equivalent of the 9590. A "Look at we can do! (Please no one actually buy it because it's not really that good for how much we charge!)"
>octacore is still 1000$
>2 hexacores, one of them 1.5x more expensive just because it has higher factory clocks
>>54034084
>It's just Intel's equivalent of the 9590. A "Look at we can do! (Please no one actually buy it because it's not really that good for how much we charge!)"
Intel has always launched some 1000 CPU that has terrible perf/$, it's for marketing purposes and to scam a few 'tards
They are known as Extreme Edition
>>54034059
the fury wasn't really a disappointment though, nvidia sacrificed their titan x to get a card to beat it.
underperforming sure, but i cant blame them on a bad design, they had a card built for 20nm, a card that need 20nm, and got fucked with their contracted fundry royally fucked them over.
>>54034084
2500~$ sure,
if you need the power, you need the power, and there's not a goddamn thing you can do.
yet again, intel refused to lower the price of their chips, while shrinking them, and not improving performance.
all these shrinks with no performance gain (or next to nothing) while charging the same price give me hope amd is ball park as good and just undercuts the fuck out of intel, just as a middle finger to their profit margin.
just imagine it, intel has FAR more overhead than amd does, amd could survive on the chips if the priced them in the 200$ range (the 8 core is 1/2 the size of their current 8350) but could intel survive if they priced the chips at the same level or a level or relative value?
>>54034112
like i said, of you absolutely need the performance, the fuck are you going to do?
>>54034106
Probably has more PCI lanes like they did with the 5820k vs 5930k.
5820k is 6 core 12 thread clocked at 3.3ghz and has 28 PCI lanes available for $390
5930k is also 6 core and 12 threads clocked at 3.5Ghz with 40 PCI lanes available for $580
So $200 for 200mhz and 12 PCI lanes.
>>54034084
Xeons have low clock speeds and cannot be overclocked.
>>54034124
>the fury wasn't really a disappointment though
To those that live and breath off of hype and rumours, it was.
To those whom are down to earth and base don't go out of their way to believe the unbelievable, the Fury line is pretty cool.
>>54034124
>like i said, of you absolutely need the performance, the fuck are you going to do?
If you need that sort of performance you're doing workstation/server shit, in which case you would be buying a Xeon and not some gamur chip
>>54033956
Except you're totally wrong. Many gaymes are heavily CPU-bound.
>>54034084
>Are there not already 12 core Xeons with hyperthreading?
yes but they're even more expensive and not clocked as high, even the 10 core Xeons (similarly clocked) are more expensive and probably more power hungry as well
>>54034141
There's Xeon's with pretty high clocks
>>54034138
And those PCI-E lanes are useless unless you have a quad Titan X setup, and if you do, you will probably get the most expensive one anyway.
>>54034144
nah, i was looking at the specs, and what we could get... also the price range, and in the performance, we didn't get 1:1 scale, in price, it came out on the lower end of what i thought it could be with water cooling.
not a bad card, but nvidia shit all over its launch with the 980ti.
>>54034153
except the xeon at that performance is usually as expensive, but a lower clock.
lets say you want to render video and play a game, the i7 would be the better option for someone who doesn't want to overclock, and alot of people don't want to do that, especially on expensive hardware.
>>54034168
The 3Ghz+ clocked 8, 10 and 12 core Xeons are stupid priced.
>>54034194
>except the xeon at that performance is usually as expensive, but a lower clock.
Except you actually need ECC ram if you actually need that kind of performance, unless you don't care about your time, in which case you wouldn't need that kind of performance
perf/$ render farms would be much cheaper
>>54034194
Fury X is a mess.
Fury is alright.
Fury Nano is fucking baller though.
>>54033823
>intel might be a bit lax on IPC
>a bit
https://www.pugetsystems.com/labs/articles/Haswell-vs-Skylake-S-i7-4790K-vs-i7-6700K-641/
>>54034272
Now compare to AMD and tell me what incentive there is to change.
>>54034221
if you genuinely needs ecc ram, then its more then likely someone else id footing the bill for the computer.
i can think of many applications where you would want a top end cpu, but if your shit crashes, you don't loose a whole hell of a lot.
sure a xeon is more stable, but at a FAR higher cost, which if you are doing a hobby, or low end work (as in a crash isn't going to set you back a few thousand dollars) its not a bad trade off.
hell, im on a phenom II right now with ddr2 ram, i use a lot of pro software for hobby use, and i haven't had something crash on me through use in... i want to sa 4 years, and during render, i had a lock up once, but rendering when im not at a computer means i don't really waste my time.
id love a faster cpu, and zen is the limit, either its good and i get one, or i cave and get a 6 core intel.
>>54034320
a significant improvement, means mass upgrades across the board.
>>54034320
AMD is irrelevant to my reply. I'm not comparing AMD with Intel, I'm trying to show you that Intel doesn't give two shots about IPC anymore. "A bit" isn't enough to describe the 4790k and the 6700k performing nearly the same.
>>54034349
>hobby
>20 threads
Sure senpai
>>54034360
Of course they dont, there is no incentive to have more IPC, so what you said has 100% everything to do with AMD, if AMD makes improvements in their IPC it will force intel to do the same.
Hurr duur how does competition work
>>54034370
if you enjoy 3d modeling or making videos, you can easily use that 20 threads, you may not use them all the time maxed, but when you need them they are there.
>>54034376
intel wont have shit for 1-2 years, if they even can which has been called into question in the past.
what we will likely see if amd is even close, is a price war, where amd is so much better value that intel has to reduce prices, and amd can comfortably go FAR lower then intel can.
why is intel making new broadwell cpus when they have skylake?
>>54034376
Are you retarded? I'm just saying that "a bit" isn't enough to describe intel's inexistent IPC gains. I'm aware that Intel doesn't need to improve anything due to the lack of competition and retarded underage gaymerz already buying whatever they release because of "newer is guaranteed to be beddur guise".
>>54034526
Because the HEDT platform is almost always an architecture behind. Haswell and Haswell-E only lined up because they had to delay Broadwell for a year and released Devil's Canyon instead. Skylake-E will be next year, at which point the mainstream platform will be gearing up for Cannonlake and 10nm (although it'll probably end up getting delayed into 2018).
who the fuck needs 10 cores
>>54034774
Video editors for one. Don't be so stupid when asking questions. Think before you ask.
>>54034774
Dx12 seems to be making use of them.
Besides I'm more than happy to take as many cores as possible when it comes to Photoshop or 3D modeling programs.
>>54031576
my almost 5 year old ivy bridge mobile i7 which is a hyper threaded dual core can run up to 3 VMs without a hitch.
I run out of RAM before I run out of CPU power
>>54034161
>many
>>54034161
>are
>>54034898
Yes, many. In fact, almost all when you start bringing multi-GPU setups into it. It's easily possible to amass so much GPU power that an overclocked 5960X curls up into a ball and cries, yet the same can't be said the other way around.
>>54032029
When AMD shits a 16 core that isnt a housefire
>in short never
i hope we get a 12 core with a nice tight little tdp and 4ghz+ one day for my 240hz vr porn
>>54034124
>2500~$ sure,
You can get a 14-core xeon for $1500, 12-core for $1200 and 10-core for less than 1k. See http://ark.intel.com/products/family/91287/Intel-Xeon-Processor-E5-v4-Family#@Server
The only problem is lower clocks and higher platform price.
>>54030905
AMD is working on 32 core APUs.
10 cores aren't impressive
>>54035109
CERN leaks say its a 16C APU and a 32c CPU.
>>54035232
Your mom says she's working on a device that'll let her suck 64 dicks at once. It should be ready to present at CES 2017.
>Just built a new build 2 years ago with a i5 4690
>look it up in charts
>gets the top of the chart in games/3dm
KEK i aint even mad bring on cheap 8 core i7's
high 4ghz+ when?
>>54032029
You can already overclock the 5960X to 4Ghz barely touching the voltage. The stock clocks on 2011 chips are extremely conservative.
Jesus Christ AMD fanboys are some of the most delusional people on the planet. No wonder Youtube is always flooded with pro-AMD comments shitting on everything else, their market is literally children and retards screaming about Zen.
>>54034013
>In GTA IV and V many areas are CPU intensive
they are but a lot of benchmarks have shown the 6700k @ 4.6ghz for example not only keeping up but coming out with a slight lead against a 4.6ghz 5820k in such games.
the reason being even though they are multi core titles, they are still very single threaded bound. they're not true parallelized titles. instead of having all 12 threads of a 5820k all working together on drawing, ai, and whatnot only one thread does ai, the other thread does drawing, extra.
virtually all current multi core games are like this. i saw one set where a 4.7ghz 6600k out performed a 4ghz 5820k in crysis 3.
as long as you have enough cores single thread performance is what matters the most. single thread performance is still the king.
some interesting benchmarks about it
>http://www.eurogamer.net/articles/digitalfoundry-2016-what-is-the-fastest-gaming-cpu
>https://youtu.be/ocwwaVGUFtk
>>54036193
wow they even ran sli benchmarks too and still showed the 8 x 8 skylake setup came out not only the same, but in some cases faster than the 16 x 8 and 16 x 16 haswell-e setups.
all those games known as cpu bound titles too.
impressive. moar cores truly don't mean moar performance.
>>54036193
wow gta 5 is a shit port
>>54030905
Do you think this could handle android?
>>54030905
>1500 USD
>>54036193
Problem is that without a newer API like DX12 that would truly allow all cores to communicate with the GPU(s), one core will always be overloaded with sending draw calls to the graphics, and all that the multi-threading amounts to is basic auxiliary processing for tasks like physics, ray casting, audio processing and AI on the rest of the cores, which historically is almost never as intensive as the work core 0 is doing. An exception would be for games like the Total War series where the AI takes a huge chuck of processing.
>>54035952
You're not proving to be much better.
put this in your UFO
![YesPerfect[1].png](https://i.imgur.com/W03eqKY.png "YesPerfect[1].png")
>>54030905
>Intel is now MORE CORES
>>54030905
lel
Why AMD's Zen will lose, even with lower pricing
>8 core Zen won't even get close to Broadwell's IPC
>it will have HT, but not anywhere near Skylake's HT efficiency
>it will have lower clock speeds than an equivalent i7 or Xeon to get the operating TDP to acceptable levels
>any high TDP Zen will suffer from thermal throttling and will likely not be overclockable like Piledriver, thus there will be a definite overclock "ceiling" below 5.0 GHz
>Zen needs to be overclocked in order to meet Broadwell's per core IPC, but again will be limited in how far it can overclock
>because of the advances in IPC and efficiency between Haswell and Kaby Lake, the top-end 8-core Zen will have the same multi-threaded capability as a 6-core Skylake-E
>but the 8-core Zen will cost more than the entry-level 6-core Broadwell-E, let alone a 6-core Skylake-E
>Intel's DMI has a much higher buswidth than AMD's FCH (or its successor in AM4 motherboards), which means it won't have any more PCIe lanes than an LGA1151 Kaby Lake board
>8 cores will almost certainly be the limit of Zen on consumer sockets, with 8+core Zens relegated to server sockets in order to properly compete with Intel on featuresets
>>54031056
Open 50 tabs in your web browser.
Can someone explain to me why their flagship $1500 processor is only 3.0 GHz, but their cheaper $550 processor is 3.6? What is the logic behind that?
>>54040905
They're constraining themselves to a 140w TDP, so with more cores, they need to run slower to put out the same level of heat.
TDP is just a marketing number, and doesn't really say how far you can actually push the processor assuming you have top of the line cooling, it's just that a 300w processor isn't as sellable as a 140w one on paper.
With Haswell-E, it's possible to get any of the processors to run at a solid 4Ghz for 24/7 operation, and all of them hit a max OC of about 4.6Ghz depending on chip lottery, even the 5960X.
Stock clocks are utterly meaningless in these chips.
>>54038928
Why your post was nothing but a waste of time
>wall of text no one will read
>horse shit by some retarded fuck who thinks he can tell the future
>NO ONE HERE READS RETARDED SHILLPOSTS LIKE YOURS
the end.
>>54041193
Alright, thanks. I was pretty confused.
>>54041219
Delicious AMDrone butthurt
I can smell it from here
