Intel is going to be caught off guard by Zen, aren't they.

>>57419061
>future proof
>amd
Nice try, Pajeet.

>>57419061
>twitch
>reddit
>instagram
>facebook
Leave

File: never_obsolete.jpg (143KB, 768x1024px) Image search: [Google]

143KB, 768x1024px

>>57419061

>>57419061
I hope AMD keeps overhyping. Hopefully Zen will be the final nail in their coffin. The last thing PC needs is poorfags.

>>57419061
This will be the generation where AMD finally catches up to Intel.

After this, we can expect actual competition. AMD will probably still be inferior, but they'll at least light a fire under Intel and probably be chepa

File: 1477435019330.jpg (112KB, 1024x279px) Image search: [Google]

112KB, 1024x279px

>>57419061
Not at all, though it will be difficult for them to do massive price cuts on i3/i5 processors because of how jewish they are which will hurt them in the long-run.

>>57419100
intel isn't future proof either. Enjoying that Nehalem i5-750 senpai?

>>57422021
https://youtu.be/_mPFfwVl1qA seems fine desu

File: react.gif (822KB, 600x366px) Image search: [Google]

822KB, 600x366px

>Future proofed hardware

>>57422050
>>57419100
>>57422021
>what is 2500k

File: 1459397639075.gif (2MB, 540x501px) Image search: [Google]

2MB, 540x501px

>>57422065

I remember when I bought a Q6600 because /g/ told me it was futureproof.

Now I have a 2500k.

>mfw waiting for Zen

File: 1477244967742.jpg (138KB, 653x726px) Image search: [Google]

138KB, 653x726px

>>57422083
dumb, poor animeposter

>>57422065
>Something released in the last decade
>It's future proof guys!
Kek

Why do you guys hate amd so much? Competition is good. Giving money to the Jews is bad.

File: modern_processors.png (86KB, 710x371px) Image search: [Google]

86KB, 710x371px

>>57422046
>>57422065
hmmmmmm

>>57419599
This used to be an acceptable pc. Fuck me, I'd have trouble getting some flavours of modern linux running on that glorified toaster. How did we survive those dark times?

>>57422109
it's just a handle of faggots from /v/ who come here to do nothing but shitpost anti-amd things.

>>57422065
2500K has proven to be the best CPU purchase ever, even better than the Q6600.

>>57422148
>1049£
hm

>>57422148
dat single thread rating doe

>>57422267
Even the 6100 is twice as powerful as your meme ass 2500K.

>>57422267
>not $
Please leave your irrelevant country out of this thread.

File: laughingsluts.jpg (2MB, 2560x1440px) Image search: [Google]

2MB, 2560x1440px

>>57419061
>future proof
>amd cpu
pick one

>>57422300
>smart
>animeposter
pick one

>>57422324
>clean streets
>you
pick one

>>57422341
>good memes
>you
pick one

File: okay.png (82KB, 710x371px) Image search: [Google]

82KB, 710x371px

>>57422288
To be fair it does for the most part, for now. Nobody turns to AMD when they need to do very CPU intensive tasks like 3D animations or 4K video encoding. Hopefully zen won't disappoint.

Though as an intel user I will say I'm fairly impressed by the X4 845 from AMD. It's a really neat budget solution for doing most things. You don't need an i7-6700K overclocked to 9001 THz if all you do is browse the web and do basic office work.

>>57422300
see >>57422148

File: 1477984435021.gif (1MB, 400x254px) Image search: [Google]

1MB, 400x254px

>instagram
>reddit
>/x/
>/grimes/
>AMD
I'm disappointed in you for not recognizing the obvious, /g/.

>>57421980
But there's always budget tier Intel options too that are pretty good. Like the i3 6100.

>>57419100
Well 2bh the fx series, and GCN, has aged pretty well, but yes futureproofing isn't always the smartest purchase factor to go with.

>>57422109
It's just a vocal minority. Believe it or not, most people don't care or just don't know any better.

>>57419061
every time I see that stupid round brush vector drawing I laugh.
It's like that time Sony used the Spiderman font for the PS3, it looked so fucking bad.

>>57422021
My Nehalem i7 runs modern stuff just fine

>b-but a brand new $400 i7 gets you 5 more FPS!

>>57422614
more like 50 FPS, lol.

>>57419061
>Parent Herald

just another generic news-aggregation website with rehashed articles every now and then.

>>57422619
I can run Witcher 3 on high at 60fps with a 7970GE, I don't see how shilling out for a new CPU and mobo will net me 50 more fps

I have better things to spend my money on than hardware to play games these days

>>57422083
There's still nothing wrong with a Q6600

>>57419061
>future proof
Written by an idiot.

File: 1477375730073.jpg (69KB, 960x544px) Image search: [Google]

69KB, 960x544px

>>57422640
Thank you.

>>57419061

>Intel is going to be caught off guard by Zen, aren't they.

AMD literally underclocked an Intel CPU to make a "comparison" between how well the Zen "performs". Just that in and of itself tells you how fucked they are.

File: 6NVC7.jpg (280KB, 2048x1536px) Image search: [Google]

280KB, 2048x1536px

>future proof
clickbait should be against the rules

>>57422673
>AMD literally underclocked an Intel CPU to make a "comparison" between how well the Zen "performs". Just that in and of itself tells you how fucked they are.
New leaks don't agree with you.

http://techfrag.com/2016/11/04/amd-zen-blender-benchmarks/

File: 1474727528628.png (119KB, 225x227px) Image search: [Google]

119KB, 225x227px

>>57422695
>http://techfrag.com/2016/11/04/amd-zen-blender-benchmarks/

>mfw i read this url as 'techfag'

>Future proof
>AMD

i7 2600k still going strong while Bulldozer cant even beat a 5 year old Intel processor.

Test

>>57422186

Man, I browsed 4Chan on a Windows 98 machine with a 350MHz Pentium II, 128MB of RAM, and an ATi 3D Rage Pro with 8MB SGRAM just for shits. It actually worked, it was just slowwwww.

>>57419061
>implying zen won't be a bust just like everything else amd

>>57422910
Did it work?

>>57419061
>there are literal children perusing this board who don't remember the days of AMD whooping Intel's pathetic ass

>>57423001
>muh athlon
former glory is irrelephant bruh

>>57423020
I am inclined to agree, but children such as yourself seem to deny any chance whatsoever of AMD once again surpassing Intel, even though it is a very real possibility.

>>57423034
>even though it's a very real possibility
possibilities are about as relevant as former glory, call me when it's an actuality

AMD has been announcing Zen every month for the past 3 years

>>57423035
>People supporting this big jewish company

>>57419599
well its still got the bd+dvd crap going for it

Good Luck to zen and its developers

>>57419599
It probably still works tho. Back in the days thay actually cared about optimization and the experience was comparable to what we have now.
The only difference is that now you need a quad core to run javascript analytics and malware

>>57422148
Rocking that 2500k baby yeah! I'll just overclock when I need more power

>>57423037
People were saying the exact same shit about Intel back in the day.
I'm not convinced Zen is going to be the magic bullet, but no company can be king forever.

>>57419061
I use Intel and I hope it does. Give them the push to make something better. Seriously.

Future proof is the worst tech buzzword

>future proof
They really want to meme this up.

I feel sorry for AMDrones, it was bad enough when the rx480 turned out to be a flop, I could only imagine the disappointment of Zen would lead to suicide

File: LinuxLoli.jpg (93KB, 791x427px) Image search: [Google]

93KB, 791x427px

>>57422324
>using animeposter as an insult
>on an anime website

Go back to >>>/v/

File: screen0003.png (204KB, 640x360px) Image search: [Google]

204KB, 640x360px

>>57423310

>>57422691
>unlimeted internet
>$19.95 per month
Wew lad, feels bad to be Australian

>>57422186
>How did we survive those dark times?
Web browsers were bare-bones, cybercrime was rare, and video games constrained their UI to the physics then available.

The only thing which was a challenge was streaming content, and in a world without streamed content, that also didn't seem too bad.

>>57421987
>This will be the generation where AMD finally catches up to Intel.

After this, we can expect actual competition.

AMD finally caught up to Intel when they finished the race years ago. Congrats.

But seriously, Haslel already hit the brick wall in performance. AMD barely managing to catch up when Intel's already trying to jump ship to new materials and technologies says a lot about their future.

File: georgeleadnew-619-386[1].png (503KB, 619x386px) Image search: [Google]

503KB, 619x386px

>>57422083

/g/ is wrong a lot, but man, when they're right, they're fucking right

File: 8808150-Portrait-of-laughing-senior-woman-working-at-computer-Stock-Photo[1].jpg (131KB, 1300x866px) Image search: [Google]

131KB, 1300x866px

>>57423035

> Charging $50 to enable hyperthreading

Hahahahahahahahahaha, fuck me I'm glad that shit never caught on. I guess even normies have a limit to how hard they can get jewed.

>>57422707
hur dur I am so nerdy

>>57419061
i7E increased to 10 cores because they needed to maintain a comfortable performance margin over Summit Ridge in all heavily threaded workloads.

>getting AMD, ever

All benchmarks point to Intel's superiority. Do you enjoy wasting money?

>>57423731
> they are wrong a lot of the time
> they are right when they are right

>>57419061
nah.
they'll just bribe all the oem not to use amd.

>>57421980
>being this much of a cuck

>>57422094
dumb pajeet

>>57419599
>upgrade to the fastest model on the market every 2 years for only $99
pretty good deal 2bh

>>57421980
>the last thing PC needs is competition
That's what you are saying and it is fucking retarded. Enjoy your dystopian future of consumer technology, cuck

>Intel is going to be caught off guard by Zen, aren't they.


No they won't.
Post with a trip so we can all laugh at you when Zen is going to be trashed by i3's in single core stuff.

>>57425962
$99 in 1974 (from when this "pretty good deal" was introduced) is $4599 nowadays.

>>57419061
I fucking hope AMD will not fuck up this time.
Intel went fucking nuts with the prices the past 3 or 4 years.

>>57425976
But i like to be cucked, i plan on buying my wife's son a i7 7700k when it comes out,

>>57426019
im pretty sure that computer is from the late 90's early 2000's

So.. AMD will let you swap out chipsets ?

Because that is where many features come from as well.

>>57426044
no but they wont change chipsets every 6 months like (((intel)))

>>57419061
>future proof
>CPU
Yeah ok.

I really want to think that Zen will be competitive to Intel's offerings, and I think they will, but I don't want to overhype it. That's ANDs problems, the fan base overhypes everything and when released its touted as a disappointment. The Polaris gpus are a good example. Perfectly good cards for what they are, but fans were hyping it to be some 1080 level shit.


You guys also have to think about the differences between AMD and Intel right now. AMD was a company on the brink, and as such they sold off everything that they didn't need, cut costs, and focused on making money. Right now AMD has low overhead, and has started making money again. Any small increase in margins and such is a huge increase in income. Intel has huge overhead, as evidenced by the layoffs in the last few years. Running fabs is expensive, not to mention the huge marketing dept they have. If AMD some out with something good intell will have to drop prices, and I think this will hurt them more than people think. Part of the reason Intel charges soo much is because they have to, they are a very large, bloated, company, and people need to get paid.

I hope Zen is good, Intel needs a fire under their ass.

>>57422021
intel would not price cut the i3-7 like this generation, they would do it gen after zen so it seems to more people like you aren't being fucked for price, possibly they would do it at the next die shrink.

>>57426256
intel doesn't need to cut any of their prices.

The only way AMD is staying competitive right now is by purposefully selling their products for barely any money.

Just take a look at AMD financial statements they've only recently (IIRC) gotten out of the red, last ~4 years they have lost money.
Why ? because if they raise their prices people will just buy intel instead.
AMD exists only in the sub $150 category right now.
They can't sell at a loss forever.

>i dont know how their gpu division is doing tho

>>57426290
no, remember a good part of the loss in sales comes from fucked wafers. that and amd gpu/semi custom business keeping up a architectural blunder cpu devision up.

with the zen, they will have a range that can go as low as 100$ per chip, as its half the size of a 8350, god knows it wont be under 200$, but point being they COULD go that low and still pull proffit, then you have the jump down to 14nm which takes the die size on the gpu and halves it for equal if not better performance depending on resolution and api.

amd is financially is a FAR better position then intel, because amd has pretty much trimmed all the fat that it can, especially if they regain any of the server market place, lets ask this, could intel cut their margins at all and go on business as usual?

>>57422627
witcher is a gpu bottleneck game if i remember right, any game that is cpu bottled though, 30-70fps is the range you can gain.

>>57422640
um... considering any dual core is more then any normal person needs, and damn near any quad core is all most gamers need, calling this future proof, when my quad core is 6-7 years old and barely showing normal computer use age, isnt to far fetched.

>>57422673
do you understand what clock for clock is? they downclocked to show the closest apples to apples comparison you can between 2 architectures, raw ipc difference. this is an ES and most of the time, these are clocked far lower than their final counterpart.

we won't know till amd says something if it clocks good or not.

>>57426518
Even shit tier Vishera performing well in newer games is telling. Eight very weak and high latency cores still do pretty well when the workload makes use of the available threads.
An 8 core/16 thread processor with significantly higher IPC will conceivably be relevant for 5 years easy.

>>57423037
intel sat on their ass so fucking long going after arm exclusively that they left the pc wide open with the only real changes being new instruction sets rather than architectural upgrades.

>>57419599
>writing drive maps on the bay
I remember old people doing this.

File: 1477970764749.jpg (85KB, 797x876px) Image search: [Google]

85KB, 797x876px

>>57423327
>muh weeaboo hugbox

>>57426562
not even that, unless there is a sea change, ipc WILL NOT bloat out like it use to, but this is assuming intel has been trying their damndest to improve it.

I mean look at it from a normal computing perspective. will programs start to suddenly only be programed to use this level of ipc? im betting money incompetency will take hold, but they wont be trying to push it.

whenever a cpu would fail at being able to uphold a certain level of performance, lets say in video decode, a gpu can do it without breaking a sweat and is far easier to upgrade that for 100-150$ if that's all you use it for

and in games, physics are the only thing that fuck a cpu over hard, if that gets an opengpu solution, then physics are off of the cpu and on the gpu for everything, not just physx.

in a professional setting, the most demanding shit you can ask a computer to do is art, as all the physics simulations have been offloaded to farms a long time ago, art on the other hand... lets use blender as an example, when you make a 3d scene its so fucking demanding that throwing processing power at it to see it in viewport cant happen, you have to set a visible limit and a render limit but even this aspect of computing is more on the gpu then cpu as well, with some programs like cirta being heavily driver dependant along with vram dependant.

now, I want you to think, what tasks could a cpu do 5 years from now that they cant do now?

Realistically its instruction sets right?

Im on a phenom II 955 with 16gb ddr2

There is VERY little if nothing that I outright cant do, with only things being done faster as a factor. I don't want to be a future predictor, but i'm almost certain that 8cores will be the last stopping point for consumer computers, you will have the prosumer higher core counts, but the next thing may be number of threads, I think, not sure, that currently consumer hardware has 2 per core, but server has up to 4 per core.

>>57423196
not even better, just cheaper.

File: amd.jpg (183KB, 1127x1231px) Image search: [Google]

183KB, 1127x1231px

>>57426442
>amd is financially is a FAR better position then intel, because amd has pretty much trimmed all the fat that it can,

How is that a better position ? is this AMD logic here..?

>>57426703
>but server has up to 4 per core
Depends on the architecture. The most recent Xeon Phis are 4 per core, and IBM's POWER can do up to 8 per core.

It begs the question though what the home user is going to use all those threads on though. I currently own an 8c/16t Xeon and aside from the occasional distributed computing or video transcoding (the latter of which is currently choked by disk bandwidth), my CPU rarely sees more than 3-4 threads reach full utilization.

There's also the fact that as things stand, there's physical limits as to how fast our shit can ultimately get. It takes time for a line of data to make it from a cache to the CPU, even if its just going from the L1 cache to the execution pipeline. Propagation delay means that no matter how highly clocked or how tightly packed your CPU is, if a block of data on-chip is on the physical other side of it, it has to transit that distance to make it to the core that needs it.

>>57426715
amd has no market share in the servers, and had their market share eroded fucking everywhere.

they regain even 10% of the server space, that's all their debt gone in a year, and amd having cut all the fat means they can survive on margins FAR below intel, not only that but a mcm cpu, which is likely what they are doing, will kill the risk of making a big monolithic chip due to yields, so amd is likely going to have amazing yields, along with a simple design that allows them to put out more, higher end chips for a lower cost, and if a chip is fucked, you can always sell it as a 4-8 core to consumers.

amd has low cost of production,
low costs to eat if something fails
low margins to need to compensate for

they could effectively out price intel on every chip they have, and on the chips they put out that go above intel's offerings, sell it at a premium.

meanwhile and intel, they are going to have to massively lay people off if they lose 10% of the server, if they have to cut prices to compete...

realistically the only way amd is going to get fucked is if intel does the monopoly shit again, but would they be able to? i'm not sure how closely they are monitored.

>>57426767
well my thoughts is this, currently my computer has 144 processes going just due to chrome alone, and on a fresh boot, has around 90, and after i close shit it has as low as 70, while windows 10 recently got rid of svchost, so the process bloat is real there.

now what if every process had its own thread and no need to share? it may not be ideal, but that has to have an effect, and could be a relatively low cost effect if you compared it to just increasing cores, or engineering a better architecture.

though i was not aware ibm power went to 8 threads, i thought they were 4.

but what you mentioned with latency, that will be handled not by architecture, but by either post silicon or post how we currently make cpus, which may still be a decade off, as 7nm is the limits of what we currently have, and both intel and everyone else isn't even at their real nm rating. there is ground to make up in the process, and if we stunt at 10nm thats what they will have to do for a few years.

>>57426787
>that's all their debt gone in a year, and amd having cut all the fat means they can survive on margins FAR below intel,

You do realize that if intel lowers its margins it would be the end of AMD , right ?

>so amd is likely going to have amazing yields, along with a simple design that allows them to put out more, higher end chips for a lower cost, and if a chip is fucked, you can always sell it as a 4-8 core to consumers.
jesus man, are you drowning in that kool-aid ?
You're just going to assume yeilds and then make it seem like AMD is the only one that bins their chips


>meanwhile and intel, they are going to have to massively lay people off if they lose 10% of the server, if they have to cut prices to compete...
Intel already has plans to lay off 12,000 workers over the next year or so, this was announced back in April or so.
layoffs tend to get headlines, but it never makes the news when intel hires people.

Honestly man, it sounds like you're betting the farm on a product that isn't even out yet that is competing against a behemoth of competitor.

>>57422236
to be fair, you oc the 6600 and its amazing, but the 2500k doubled ipc, if that was really needed or not is another matter entirely but the core quad you had was nice.

You guys are making it more complicated than what it should be.

>Is the new computer at least 2 times faster than your own?
>Yes: upgrading makes sense
>No: upgrading makes no sense

You don't future proof a computer, you make logical increments.

Does anybody know what the zen lineup is going to look like? I know its going to start as 8 core in early 2017, but when are the 6 core cpus meant to arrive?

>>57426884
At least two of each core count, and both the 6 and 8 core will be available at launch. 4 cores might be OEM only.

>>57426864
its not if intel lowers theirs, its can they?
they spend so much on research, get next to no where with it, set up new fabs, and compete at a loss with arm, they have so much shit that adds to the cost of their cpus, could they lower the margins?

lets say that zen is good, it is spitting distance from current intel to the point its margin of error, but amd puts the cpu out at mid range i7 prices. could intel answer that? sure they could lob off the gpu portion of an i7 and get the chips smaller, but they would loose the almost all in one solution they have going on for most people. what could they do server side to those prices?

as for yields, its the same process as the 480 correct? with an mcm design, they rely on smaller cpus and put them together for bigger ones, rather then one big silicon that has a far more likely to have some damage lets say instead of pinning hopes and dreams on 25 32 core chips, a batch, they are able to pool together 4 chips from a batch of 100 and its not like the 32 core full chip where you at best have 25 chips no mater how they fucked up.

and the intel laying people off, what caused them to do that was a drop in the bucket with goes further to illustrate they have no room for lowering margins, while amd is going to gain just form the fact they are half the size of a 8350 even if it was the exact same price, its doubling the margin.

>>57426854
>well my thoughts is this, currently my computer has 144 processes going just due to chrome alone, and on a fresh boot, has around 90, and after i close shit it has as low as 70, while windows 10 recently got rid of svchost, so the process bloat is real there.
>
>now what if every process had its own thread and no need to share? it may not be ideal, but that has to have an effect, and could be a relatively low cost effect if you compared it to just increasing cores, or engineering a better architecture.
Not really much of a point, because 90% of those processes are idling most of the time, and the ones that arent typically only use a couple percentage of a thread at a time.
Its easily possible to stuff the bulk of those idle processes onto a pair of threads (on a single core) and not adversely affect the rest of the system.

As for the IBM POWER 8-way multi-threading, it was literally POWER8's shtick: https://en.wikipedia.org/wiki/POWER8

The only way I really see anyone fixing physical distance latency is to start stacking transistors on die, 2.5D/3D construction and the like.

>>57426878
here is another point

the computer is 2 times faster on paper, do you actually notice a difference with that power?

and for almost everyone on the planet, no, it makes more sense to get that one big one now, especially when realistically, sense sandybridge ipc has stunted hard.

>>57422300
>pick one
bottom right

>>57426962
true, but how close are we to actually hitting physical distance being the main issue, and not just an issue keeping the chip cool?

>>57426929
cheers senpai

>>57427037
In a way we've already hit it. Going out to main memory and back takes quite a while, with part of it having to deal with the actual physical distance between the CPU and the memory ICs. Its why we have so many cache tiers these days.

>>57422148
>costs 4 times as much as 2500k did when it was new
>less than 3 times as powerful
Wow, such advancement.

>>57426019
You do realize this would be a supercomputer back in 1974, right?

>amd

>>57427067
Ok i misunderstood what you meant, I thought you meant going from cache on cpu to cores, not system memory... yea that is a bit of a sticking point.

probably moving on from ddr to something new would improve things, there is also the idea of using light for the data interconnects, not sure how much either of those would change things, but with ddr the timeings keep going up so the overall benefit is far more mild then it could be just increasing clock, honestly, how to fix this may be simple now that I gave it a minute.

what is stopping a cpu socket from being a bit bigger and having something like hbm socketed into it? maybe 1gb, maybe less, just something between the cpu and the ram, where it can hold a mass amount of data (at least compared to what the cpu can hold) while acting as closer to cpu storage area that can be fed new data at a higher latency, and feed data to the cpu at a far lower, the higher capacity is just so you feed that ram at a faster rate than the cpu can process.

with it being off cpu but still in the vicinity, it could be a take it or leave it add in chip.

>>57427121
you need the power, you need the power, there is no two ways around that issue.

>>57422148
>140W
>barely faster in single core
>~4x the price

Dunno man. I don't think I can argue with that logic.

Intel might just drop their prices a bit, their prices are probably the most off-putting thing for most consumers.

>>57427234
your not getting an 8 core cpu for single core, at least not a 1000$ 8 core.

>>57427226
A first gen Xeon E5 could offer the same number of cores and almost the same clock speed.

>>57427219
Going from cache on-die to the cores (SB-E L3 latency is about 7-8ns) is nowhere near as bad as going from memory to cores (>40ns), but even so 8ns is a dozen or so clock cycles the CPU is stalled waiting for data.

Changing the memory interface and type may help, but even if we went to optics there's still the memory ICs to optical to CPU IC conversion time, which is a non-zero value and puts a fairly hard limit on what an optical connection can do.

Putting something like HBM or HMC on-interposer/On-package is an option, but keep in mind that is an enormous number of connections for those technologies (on the order of thousands of connections), and routing that many connections is a gargantuan pain in the ass, which in turn drives up costs.
Intel has already tried and done this with their Crystalwell EDRAM for the consumer and HMC for Xeon Phi, but they use a relatively large amount of power, and for some tasks having an L4 cache has no benefit. Hell, many programs make minimal use of the L3 cache many chips have these days.

There's also a practical limit on how big the CPU package can get, from a package trace routing standpoint, a motherboard trace routing standpoint, and simple physical size. If you've seen Intel's latest Xeon sockets, they have over 3000 pins and are fucking enormous.

>>57422656
br has all that stuff but you usually get shot in the street

t. Brasilian

>>57427278
Nah, prices will go up. You see, if intel rises the price they're showing everyone just how much better their products are.

I mean, it'd be stupid to rise prices, right? So there's something to it. Better be safe and buy intel.

>>57422083
Who's this little angel?

nope

>>57427325
Have to ask, how in the fuck do you look into getting a xeon cpu? I have never been able to figure out how to look into the prices, the skus, or the mother boards... from the way it looks, I could get a Intel Xeon E5-1680 v3 for around 350$ and drop it into a 200$ motherboard and be happy?

Would I need ecc ram?
Can I overclock?
Will this in any way fuck me over?

This is a concern i have been having for awhile, I want and can use an 8 core cpu, my hobbies push my pII 955 to its limits, but I'm holding out for zen before I make a decision, the chance that its a simple no bullshit execution, comparable to intel, with me knowing exactly what is going to go into it when I get it, especially with being able to buy new parts and not third party parts off ebay, give me some ease of mind.

Able to chime in at all?

I have looked into dual cpu boards in the past because render times would be much better if i could just put more power into the boards, but price and age are a big concern.

>>57425976
>implying x86 is going to be around in the future

kek

>>57427490
>how in the fuck do you look into getting a xeon cpu?
>Would I need ecc ram?
>Can I overclock?
>Will this in any way fuck me over?
Simple: identify your needs.

>>57427525
I need to be able to hand video rendering, 3d rendering, multimedia and gaming, basicly every hobby I have and I need it to do it well, which a 3.2 8 core haswell, pretty fuckin sure it could do it... my main issue is not knowing what i need, Do I need ecc ram or can it manage with less expensive ram? Is there any oc ability? While I think 3.2 would satisfy me, if my phenom II is anything to go by, for the next 6-7 years easily, it seems diminishing returns comes into play at 4ghz, would love to be able to push that. the only two motherboards I can get new are gaming asrock from newegg, anything fucked with them?

On top of the whole, i'm getting a used part that may or may not work and it costs 350$ kind of risk.

Im willing to risk that if amd falls on its face hard, but till it comes out and I see its performance and price, its hard to justify, but knowing how to look up intel cpus from skus that are not meant for me at all would help immensely in the coming months.

>>57427298
I know, but many workloads today are largely one or two threads on desktops. Yeah it has its workloads but someone who actually needs 10 cores knows they need it, unlike people who just want it for the e-peen.

>>57426039
I can confirm, that was my uni PC I got in late '99. I had to put an ethernet card in it and asked about them installing it and some RAM, they wanted $40 to put the card in the slot and another $40 to pop the ram in the slot. When I opened it up and found how easy it was I still to this day am disgusted by Best Buy

They probably mean the socket and not the cpu itself.
Even though Am3+ was a heaping piece of shit it survived the entirety of ddr3

>>57427627
>Do I need ecc ram or can it manage with less expensive ram
How important are your rendering work? If you can not tolerate any delays due to faults of minor corruption, you need ECC. If you can spare a few minutes or hours on redoing corrupted renderings, then you don't need ECC.
>Is there any oc ability?
Absolutely not
> the only two motherboards I can get new are gaming asrock from newegg, anything fucked with them?
What chipset? ASSRock Rack have disappointed me in the past for cheaping out on controllers, but their consumer boards have pleasantly surprised me for having weird features that other manufacturers usually pass on (like preserving IDE support to Ivy Bridge and eSATA 6Gbps on their Z87/97 board).

>>57427761
but the argument is price for performance. single core is stunted to the point its back to clock speed again, at least with current architectures (discounting amds current till zen at least) multicoreis the metric you should measure these things by.

and even for games, people argue you don't need more then 4 cores, and at most 6, or that there is no difference between 6 and 8 cores, but when you take it out of a steril benchmark there clearly is a massive difference. both in how your system functions when you hit a load spike, to just what you are now able to do. an 8 core would change how I use a computer entirely, just because i can offload a render to 4-6 cores, get shit done, and still be able to use the computer myself for entertainment, something i currently cant and routinely scheduled for night when i sleep (cant bring myself to load a computer when im not near by, just on fire risk alone)

>>57427848
for me, currently all my rendering workloads are hobby, just dont know if the cpu could/can handle non ecc ram, as i know that is/was the case with skews a long time ago.

http://www.newegg.com/Product/ProductList.aspx?Submit=ENE&IsNodeId=1&N=100007627%20600533617%20601205171

keep in mind, I have no fucking clue about this at all, I just hit the i7/xeon option because if i buy something from there it will be with that option, i refuse to take the risk on buying things when i can't return a substantial component,

>>57422083
I was still running one a few months ago

>>57427914
Yeah, you don't need ECC then. The only reason why you'd need one is if you have a 24/7 file server that can not tolerate bit rot, a rendering farm master server, a network server that needs 100% uptime (like an SQL server or a business VPN server), or a compute supercluster carrying out the same workload on the scale of dozens, if not hundreds, of clusters where fault tolerance can mean the difference between restarting a week-long financial model simulation and throwing a few million dollars down the drain.

>>57422148
>cpumark

>>57427490
I bought a board off natex with two v1 E5-2670s. The mobo is the issue because of subpar PCI-E support but for 500 bucks, 128 gb of ram and 16 cores is decent. Don't expect to play AAA games on it as turbo on those is meh and, again, PCI-E is garbage. Still good for other applications (I do scientific computing sometimes for work or to make myself more marketable for work)

>>57427490
It is a bit complicated, the Xeon is marketed to businesses and serious working professorial that know their needs and want the advantages the Xeon offers.

First make sure your parts are compatible, ... before buying them.

Depends on what makes you happy?

No, ECC RAM is not required. But one reason for getting a Xeon is many versions of it can support that extra feature. Which can be vital for certain jobs. Basically are you willing to pay more to dramatically lower the odds a stray bit flip won't corrupt a file or crash your system? Most people never notice these errors as they are very rare, and often just result in a simple reboot with loss of the current save version. It gotten even less noticeable as newer journaling file systems often save the last version as a back up anyway. (double check this, but I fairly sure of it)

No*, Xeon are locked down even more then normal CPUs but are also optimized better at start. So over clocking it would be a huge task that would actually lower performance or more likely kill the chip.

*There are a few research edition Xeon chips which can be over clocked, but they are very rare and are only rented out to research labs under strict contract. So you can't legally own one and the cost to barrow one from Intel is insane.

Unlikely, in most cases Xeon chips are better in nearly everyway. However there can be some gaps when it comes to single thread computing and some hardware optimization issues, like those typically found in some video games. That said if you have such a question it would be worth it to do your research as Xeon chips cost considerably more for features you may not really use.

I haven't followed AMD or Zen much.

Depends, but some people I know swear a last gen power rig is a good deal as many new prefomace gains are in power consumption not processing power. Really depend on the money you got and your future predictions on CPUs and the energy prices on your application scale.

>>57428243
Using a Xeon cpu sounds more like an headache than anything.

Can you even game on that shit?

File: CPU-Z.png (28KB, 407x400px) Image search: [Google]

28KB, 407x400px

>>57428260
As the owner of one, you can, but performance wont be as great as on a CPU that prioritizes single thread performance.

Mine is about the same as a stock non-turboed 2500k in single thread performance. Useable, but not spectacular.

>>57428260
You can, it's just usually not worth it to get one JUST for gaming because xeons tend to be either expensive or clocked fairly conservatively. A quad-core E3 with hyperthreading used to be a cheap alternative for an i7, but with skylake Intel mostly killed that as you need a motherboard with workstation or server chipset despite them still using 1151 socket.

File: Alexis-Ren-14.png (2MB, 1074x1156px) Image search: [Google]

2MB, 1074x1156px

>>57419061

I remember how much hype there was before bulldozer's release. As it kept getting closer there was zero press from AMD about the performance figures and it was only apparent how fucked up the architecture was when AMD started shipping server products to enterprise customers (before they sent out samples to joe blow tech sites). I think Si-Soft got ahold of some samples for Sandra before release and talk about how poor its cache latency was and AMD drone fanboys went ape shit calling them liars and whatever conspiracy bullshit you can think of.

If they try to hide performance #'s the closer it gets to release then you know Zen will be shit. But that wont stop the shills from shilling till then.

>>57428396
This. There are hardly any word on Zen's clock speed other than the odd benchmarks every now and then which shows a maximum clock speed of under 3.0 GHz.
Not only that, but AMD had to downclock a competitor i7-6900K in order to show "clock-to-clock" comparison, only adding to the evidence showing how fucking slow Zen will be compared to even a low-TDP version of the Intel HEDTs.

>>57428421
>AMD had to downclock a competitor i7-6900K in order to show "clock-to-clock" comparison,
Source?

>>57428260
For normal use the differences are minor. So I don't see any headaches. Unless you heavily invested in a computer that can't do what you bought it for in which case you didn't research it enough. Even some reputable workstation companies only sell Xeon because they often are just better.

Yes, but many games don't use the Xeon's extra power so it is roughly the same as an i7 in many cases involving video games. Depends more on how the coded the games, with some games being better and other being worse, but over all seems about the same which makes it hard to justify the higher price of a Xeon for gaming.

And that gap has widened as companies try to separate the markets to make more money. Intel lost big one time back when people started to realized a low end Xeon was better and cost less then a i7, they have long since fixed this market cannibalization error. So don't expect more for less money.

>>57428431
http://wccftech.com/amd-zen-performance-demo/
>inb4 curryt-
Wait, that doesn't make any sense because the Pajeets at currytech are legit AMDrones, so of course they're going to make AMD look better than it actually is.

>Putting the pins on the cpu

dead on arrival

>>57428451
>wccftech
Nigger please

>>57422148
Relative to the Turbo clockspeed, the single thread performance is very close between all of those. Just overclock the i5s and they become as fast as the 6900k, not counting extremely multithreaded applications where the HT kicks in.

>>57428464
Read my disclaimer, fuckhead. Yeah it's currytech, but they always try to make AMD look better than they actually are. So read between the lines of their poo-in-loo propaganda and focus on the details
>had to downclock Intel's i7-6900K to make Zen look good
That is not a good sign, as it's clear that AMD knows that it has a clock speed disadvantage over Intel and had to compensate to get "even" benchmark results. From this we know that Zen will either be massively underclocked compared to Intel's HEDT or can not reach the same clock multipliers because of their lower TDP limitation.
>relying on Blender for benchmark comparisons
Blender can take advantage of other hardware within the PC (like memory size/speed) to speed up results. The fact that they only published the CPUs used in the benchmark and not the memory, storage, motherboard, etc. reeks of rigged benchmarks. So it's quite possible that Zen is underperforming Intel yet again and the Pajeets have to resort to shady benchmarks to lie to the public yet again.

No. Zen has been a long time coming; 'caught off guard' isn't a reasonable assessment.

It is going to be performance and price competitive, which isn't a situation Intel have found themselves in since the Excavator disaster.

They aren't able to die-shrink as they'd hoped; Cannonlake is sitting at 2018, maybe, if EUV pans out. They're not quite at process parity now (Intel still have a better process overall) but electromigration has proven to be a significant issue.

Kaby Lake is more of a chipset refresh than anything else; needed as USB 3.1 PHYs and more PCIe lanes are the order of the day now.

They may need to reconsider their on-die GPU strategy.

Zen is quite a bit smaller than Skylake or Kaby Lake. It scales to more cores more easily, with the right supporting bus. The new Opteron could put pressure on the Xeons they haven't had in years. Intel will be hoping 3D X-Point provides an advantage there, but in reality no mainstream PC OS is architected to really take advantage of it yet.

AMD have the chance to take some of the low end and midrange market back with Zen. Their Raven Ridge APUs with HBM2 deliver quite a powerful low-end punch.

Ultimately it comes down to if it'll clock well - don't know yet - and price - don't know yet. Intel may need to cut costs to keep their share, but with the huge cost of 10nm equipment and it going significantly over budget they don't want to do that. Time will tell. I have no idea which I'm likely to buy at this stage.

Either way I hope it puts the cat among the pigeons. Intel have had an effective monopoly on decent desktop and server CPUs for a decade now; the best thing Zen can do is break the stagnation.

>>57423046
That's cuz it's gonna be the best thing since indoor plumbing.

>>57425234
Source

>>57428243
what im looking at is Xeon E5-1680 v3, which is a haswell 8 core with .2 faster base clock. single core... im on a phenom II riding out the days to zen, my price limit on zen itself is 500$, that's the line where getting a xeon become a better price performance proposition, at least for me, Im assuming sandybridge ipc with the ability to keep up between intel 6 and 8 core, but it seems like they may be out performing them by a margin depending on where leaks and benchmarks are shown.

the cpu itself... im assuming was put under load, 24/7 and was never turned off from the start of its sold life till it came into my hands.

There are a few skews that I know of that have multi cpu boards, that come close to the performance of this single cpu, but that's more of a risk for me because i know even less about them then i do this.

knowing it works with non eec... I could build the total computer for around 1500$, and that's coming in significantly under budget, but in 4-5 months when zen lands, there is also going to be some big shifts in ssd, gpu prices, along with potentially cheaper ram so getting that 32/64 gb system wont be as costly.

>>57428325
350$ for a 8 core 3.2 haswell... fuck me the value.

>>57428628
do we know raven will have hbm yet? just looked into it, but is there official word on this? as this has MASSIVE potential to fuck with how laptops are.

also, for zen, It looks like it will be 158mm~ while skylake is 122, knoo will be a refinement, so it will likely be smaller, but they could add shit to make it bigger too.
unless someone has better numbers to go off of.

File: 1477838281777.jpg (128KB, 720x601px) Image search: [Google]

128KB, 720x601px

>>57428396
Zen is already going the same path as Bulldozer, we already have AMDrones trying to defend it with "Its meant for servers and enterprise n shiet!!!" just like they said for Bulldozer.

This shit is going to flop so hard that /g/ will be on meltdown

>>57422065

I can't believe I was naive enough to think I could get a cheap used 2500k by now.

If Zen was good AMD wouldnt be keeping it so secret. Its a defense tactic, they dont tell anyone how shit it actually is, and AMDrones overhype it for them and give them free PR

Same happened with Bulldozer and the RX480

>>57428925
Except the RX480 was never marketed as a true highend card and I suspect Zen will be the same. Not really highend but enough to hopefully shake up the midrange market.

If they make a half-decently clocked 8 core model thats cheaper than Intel offering then I'll definitely consider buying one.

File: 1473240801317.png (12KB, 880x131px) Image search: [Google]

12KB, 880x131px

>>57426642
Reminder to fuck off.

File: 1476351883308.jpg (90KB, 701x728px) Image search: [Google]

90KB, 701x728px

>>57428989
>Its gonna have GTX980 performance for only $200 guiz!!!

>Has only gtx 970 performance
>costs $250 almost $300 for a decent aftermarket version

The swift beatdown Nvidia gave AMD by making the 1060 have everything the 480 didnt only made it better

fuck, I can only imagine what happens with Zen

>>57419061

>future proof

I *hate* this idiotic meme. It's been around as a marketing term in tech since at least the mid 1990s. THERE IS NO SUCH THING AS FUTURE-PROOF. Technology is 100% deflationary. From the moment you buy a computer or component, it's being left behind.

You want to know how to future-proof your computer? Stop reading about tech as soon as you buy it. Enjoy the fucking thing and when you start getting frustrated with it, go buy a new one.

>>57429039
Here, have a few rubies and then fuck off, nvidia shill

>>57428989
>>57428925
with the gpus, we already see how massive the difference is from binned to what we get is, we also see how wildly the gpu power can change, with stock being 160 and some after markets only hitting 100, and these are the un binned ones.

that was about the only disappointment with the 480 we seen, I refuse to believe the people who massively inflated the gpus were people who liked amd and instead wanted to make fun of hype.

zen is likely going to have the same binning issue, where there are going to be some that oc for shit, and some that seem like a god crafted the chip in secret. Im willing to bet a lower stock clock all around compared to intel, with the silicon lottery being real, with a stock clock of around 3.2-3.6 being the most you get from factory, with what you can oc it to being anywhere from 3.9 to 4.5, possibly being able to set a 1-2 cores up to 4.8 but due to wild variance on a new process, the are going to hold the top clock lower.

>>57429081
i dunno, its been about 5 years and the 2500k/2600k is still one of the best CPUs you can have
>but muh 10% performance increase with Skylake

>>57429039
its almost like benchmarks of new games don't exist, and new apis becoming increasingly used.

>>57429095
"Everybody I dont agree with is a Shill" The AMDrone's guide to shitposting

>>57429039
>>Has only gtx 970 performance
>check a year later
>980 and 970 gimped
>480 still stronk

Enjoy your nVidia telemetry.

>>57426019
1999*

That's $110

>>57423103
>Back in the days thay actually cared about optimization

Are you kidding? That was the age of RAD, not the age of carefully-crafted C.

> and the experience was comparable to what we have now.

Take off your nostalgia shades. The experience was complete shit compared to what we have now.

* Shitty CRT monitor with 0.35 dot pitch, inconsistent whites that were splotchy yellow.

* Slow as fuck hard drive that filled up with porn as soon as you got an Internet connection.

* Internet accessed billed by the hour, assuming you could even get it.

* Could only buy from a handful of web sites, often requiring that you complete the transaction by phone.

* Literally no 3D accelerated graphics until about 1996 when the Matrox Mystique came out.

* Programmers created godawful UI abominations. It got better when Microsoft released their development and UI guidelines, but it was still the wild west of UI awareness.

* If you didn't have the right FIFO buffer on your serial port, you'd corrupt your data connection when you switched to a different window.

The computing experience is vastly, astonishingly better today than it was. And yes, great software was written with more care than most software today, but most software wasn't great software back then.

File: 1391007137-ruby.jpg (172KB, 1920x1080px) Image search: [Google]

172KB, 1920x1080px

>>57429095
>a few rubies
Do you mean the precious stones or a few of pic related?

>>57429099
m8 I'd be happy with a ~3.5ghz stock clock and 8 cores as long as it's more affordable than what Intel offers.

>>57429081
>>57429111
future proof as a term is something that needs to be defined.

I define it as the point from when you buy the computer to the point the computer no longer functions for the task you want.

I had a pentium 4, a pre build alienware that was a gift from my parents as we had a pentium 2 that barely functioned anymore. this computer did everything I asked it to do up until codec shifts in youtube caused 100% processor use at sd, and the 720p shows I downloaded could not play, and hand breaking 1 episode took 16 hours.

my p4 computer was 'future proof' for a grand total of 4-5 years. ph phenom II 955 was future proof for upwards 7 years and still going, just every task takes longer then it could take otherwise.

there is a chance with the way tech is going, and if zen is good enough, it could be the last cpu I buy for a power need till a new material is in fabrication.

I also remember dual cores are all you need for gaming when i got the phenom II but a few years ago marked the beginning of games requiring 4 threads. my 4 core was more future proof then the dual cores of the same era.

>>57429219
>m8 I'd be happy with a ~3.5ghz stock clock and 8 cores as long as it's more affordable than what Intel offers.

define more affordable, because 1000$ is more affordable then their 1050$ cpu.

like i said above, 500$ is my limit on what i'm willing to pay and the used xeon looks nice enough to roll the dice on.

>>57429230
>I define it as the point from when you buy the computer to the point the computer no longer functions for the task you want.

That's an idiotic definition.

By that definition, literally anything is "future proof". Oh, it doesn't function the next day? Well, it meets anon's definition of "future proof".

What you appear to intend is that the term refers to the above definition being a LONGER THAN EXPECTED period. That's a critical bit to leave out.

>my p4 computer was 'future proof' for a grand total of 4-5 years.

No, you just weren't bothered by the relatively poor performance.

> ph phenom II 955 was future proof for upwards 7 years and still going,

Only if you keep running the same old programs. I had a 955. Now I have an 8350. It's still working acceptably, but that's because I've upgraded everything else around it (SSD, GPU).

> just every task takes longer then it could take otherwise.

WELL THEN WHAT'S THE FUCKING POINT OF EVEN DISCUSSING IT? Your definition comes down to "how much shit I'm willing to put up with".

The reason that you guys think your stuff is "future proof" is because the rate of development has slowed tremendously. If you have an SSD, even an old CPU will still be tolerable for most common stuff.

But if you're doing something CPU, GPU, or disk-intensive, using software that's continually evolving, then NOTHING is "future-proof".

>>57429261
I'm not going to wait for long because Intel locking down the current xeons to workstation boards only will probably make older xeons even pricier.

I might wait 6 months or so and if Zen still isn't fucking there then I'll find myself a nice used workstation with a xeon or two in it.

>>57429286
nah, putting 200$ into the cpu instead of 150 has allowed me to go strong for a few years longer then i otherwise would have been able to deal with.

got an ssd and a gpu, still didnt need to change the core of the computer yet, I value that quite a bit as the reinstall of everything I use is a pain in the dick.

as for the pentium 4, when that was a slide show, that was it, cant deal with it anymore, thats the point where it was no longer useful to me and the computer was at deaths doors. up to that point, what was going to give me better performance? I used a computer to watch media and read,

video editing, recording, where cost prohibitive back then due to the cost of space, the cost of hardware to do it painlessly, and just the limits of processing power in general, not to mention 3d applications demanded a pro gpu to not crash and render, along with sculpting being in its infancy.

as for tasks taking longer, I don't mean opening a web page, watching a video, downloading, nothing that would effect the day to day use, I also dont mean in programs like adobe, maya, vegas, 3dcoat or zbrush, none of these feel sluggish to use or painful, i want to render something out, shit sucks but i'm not using the computer for an hour or so, and this final rendering step is the only part that would really be improved if i got a new cpu, a few games i don't own or like also demand a higher power cpu then want i have, but it doesn't effect me yet.

Im not being paid to do this faster, and im not really out anything by fucking off for an hour or so to watch a movie.

with my pentium 2, the internet became a slide show.
with my pentium 4, small tasks I aksed of it no longer worked

neither of these is the case.

>>57429301
same, but I honestly doubt its going to shift the price, these old xeons were sold in such high volume and were replaced, that's why I can find a 1000-2000$ cpu for under 400$ and most people are not going to have any fucking clue this is possible... I may not know as much as I should but that's beside the point.

could also roll the dice on engineering samples.

>>57429191
Most of the stuff you pointed out is
>durrrrr le superduper specs XDXD 4k VR ready gaming amirite :DDD
I have a laptop with windows 2000 that has like 128Mb of ram and offers a comparable experience to any other pc when browsing 4chan or using Office.

>>57427823
>Even though Am3+ was a heaping piece of shit
How so?

File: justaddmoarcores.png (39KB, 930x414px) Image search: [Google]

39KB, 930x414px

>>57422751
wew lad

>>57429500
>botnet 10

>>57429518
muh gaymes etc

i mean what the fuck else am i going to do with a fucking 4.4 ghz housefire

>>57429417
>allowed me to go strong for a few years longer then i otherwise would have been able to deal with.

And we're right back to what I already said. It's a matter of how long you can keep your head in the sand.

>with my pentium 2, the internet became a slide show.
>with my pentium 4, small tasks I aksed of it no longer worked

And *again*, we're right back to the same fucking thing:

>>57429286
>The reason that you guys think your stuff is "future proof" is because the rate of development has slowed tremendously.

>>57428852
I hate to give more advice as there is so much that goes into it. But really for me it boils down to three options
1. good laptop (less power but it stays mobile)
2. go big with workstation (do anything machine, try future proofing)
3. decent consumer desktop (admit software changes are going to break my hardware every few years, so plan to keep replacing them with something that's cheap and just good enough)

I know this was for someone else but I got to say
I actually I want a chip to get bigger, as more spacing offers better thermal dissipation, but then people as why pay for so much empty space on an expensive die.
I say better cooling, clocking and life time stability, but doubt people would go for that.

>>57429590
>>The reason that you guys think your stuff is "future proof" is because the rate of development has slowed tremendously.

tell me, what the fuck could be done with more power for, I don't know, a web browser? an image viewer? I would say video but thats offloaded to a gpu.

tell me, what advancement would the average person need more power?

I want it because of hobbies, but i sure as fuck don't need it.

>>57429618
I would argue for a better thermal interface, a bigger chip means that there would be more latency, but also potentially more robust connections so failure in the process happen less.

>>57429655
>tell me, what the fuck could be done with more power for, I don't know, a web browser? an image viewer? I would say video but thats offloaded to a gpu.
>tell me, what advancement would the average person need more power?
>I want it because of hobbies, but i sure as fuck don't need it.

Yeah, so... Exactly my point.

>>57429547
Compile Linux

File: Screen Shot 2016-11-07 at 1.10.54 PM.png (935KB, 2046x1578px) Image search: [Google]

935KB, 2046x1578px

>>57426854
>while windows 10 recently got rid of svchost
no it didnt

>>57429655
It doesn't exactly work that way given the proximity and modes of thermal transfer.
Kind of like the how the best cooler is significantly effected by bad thermal paste. (still wouldn't turn down a good cooler as it helps)

You can use a diamond substrate to cool things down a lot more (great stuff, why we don't do this is beyond me) but you still have limits to the interface geometry. Empty space doped with thermal conductive additives make it behave more like a 3D surface, thus more functional surface area interface.

IBM is working on liquid cooling inside the chips itself for even more cooling gains, but not sure how much I trust micro size pipe works, also phase change issues (boil, freezing).

File: s-l1000.jpg (130KB, 800x647px) Image search: [Google]

130KB, 800x647px

Im really curious about the performance.
I would like to upgrade my i5 [email protected] but then again im also skeptical the zen will beat it.
>What i want
like 6700k level performance
>What i expect
i5 performance
I really hope this isnt an i5 equvalent cpu with poor single thread.

Whats your expectations?

>>57430085
it wont be near either because its another processor with 8 shitty cores instead of 4 good ones like bulldozer
theres a reason why AMD compares it with a 6900k instead of a 6700k, they dont want anyone to know its bulldozer 2.0

File: lol.png (51KB, 512x512px) Image search: [Google]

51KB, 512x512px

>>57422148
>yfw not everyone can casually afford $1000 silicon in $3000 pc's
>yfw you suckle Intels cock

>>57423310
i feel sorry for shills who suck the dick of a company because they throw them a bone and give them free shit. especially because all 3 major companies offer up solutions for almost every problem

>>57426164
Intel charges more namely because they hold that share of the market by the nuts, they can charge $1000 for an 8c/16t chip because theyre the only ones who offer that kind of power

>>57428260
>can you even game on that shit?
looks like someone thinks computers are only for games

>>57428319
what do you use your Xeon for? I have a 2670 that i want to run calculation intensive Fortran programs on

>>57430085
what are you going to be using your computer for? if game is the most youll do (or light content creation), you can stick with your i5. I went from an i5-3450 to i7-3770K and do some content creation and am more than glad I got it (for free, nonetheless)

>>57430437
>what do you use your Xeon for? I have a 2670 that i want to run calculation intensive Fortran programs on
I use it as my daily driver, as it is a significant step up, both in single and multi-thread, from my overclocked Q9400. I would use it for some video transcoding to shave down 1080p60 recordings to a more sane size, but lack of drive speed puts a cap on how much of a speedup i can get out of that.

>>57430534
do you use ECC? if so have you noticed an advantage over normal DDR3?

>>57430572
You don't 'notice' ECC or not, you have it or you don't. Read up on what it does.

>>57430597
i know what it offers, just wanting to know if there are any discrepencies in time taken to accomplish Task X
i want to get a 4x4gb kit of ECC over my 2x4gb kit, only $55

>>57430629
https://www.pugetsystems.com/labs/articles/ECC-and-REG-ECC-Memory-Performance-560/

>>57430572
ecc just minimizes the write/read fails and corrects them in your ram
its most important for raid configurations that runs 24/7 where the extremely low % of an error actually becomes a threat

>>57430572
Nope, I'm using ordinary ram. Even if i was normal ECC ram should not have any noticeable performance impact.

>>57430629
It depends on what sort of DIMM you're using. If you're using Unbuffered ECC DIMMs, the impact on memory performance is slight to nonexistent depending on the application. RDIMMs have a significant delay in latency due to the extra controller built into the DIMM itself.

File: Speccy.png (44KB, 660x558px) Image search: [Google]

44KB, 660x558px

>>57430437
>yfw not everyone can casually afford $1000 silicon
pic related

>>57430597
Yes you do, I certainly have when I looked at my SMBIOS log and saw a DIMM was failing due to excessive errors but my computer wasnt crashing. And since most ECC ram are RDIMMs they have higher latency due to the buffers.

>>57419061
that's like being caught off guard by a zamboni, you see that shit coming and have plenty of time to react.

we've been talking about zen for fucking years. I think Intel has a contingency plan by now. they're either going to price cut kaby lake or divert r&d budget back into the x86 desktop. or both.

the fiercer the competition, the better. if neither Intel nor AMD win then we do.

>>57430664
http://www.newegg.com/Product/Product.aspx?Item=9SIA5753MW0382

this is what i want

>>57430673
>hurr you see it in the logs

Well of fucking course you see it in the logs. My point is you don't notice it during day to day activities, wether that is shitposting or rendering your moms enormous arsehole.

>>57430726
>you dont notice uptime

>>57430708
Those are RDIMMs, so they're more expensive and slightly slower overall. Why do you want that particular kit? Just go for a 4x4GB kit of Unbuffered DIMMs if they cost less. RDIMMs are for capacity above all else, and that shit doesn't make sense unless you're running eight of them on a dual socket board for double the total memory capacity.

>>57430759
How do you know if at any given moment your machine will crash or not?

That's like saying you notice your car insurance from day to day life. You fucking don't, you only notice when shit hits the fan and even then ECC doesnt stop all bad things from happening.

>>57429805
I thought a bit ago one of their updates killed them giveing every service its own process.

>>57430196
thats because its a real 8 core this time, not the mostly core, look at intels 6700 and 6900, in an application where single core matters more, the 6700 is going to beat it just because it will clock higher, but in an application where multicore does, it gives a 6900 a run for its money.

>>57430085
If you are just going to game, fuck the zen for now, wait till zen+ as intel should respond to amd by then either in price, or pulling something they have been holding back, who knows, if you are doing more then just gaming, zen will probably be good for you, just remember, 8 core zen will clock like shit, intel's 8 cores do too, but they make up for it in getting shit done and load balancing.

>>57430673
does dual cpu really have no effect on gaming or is there atleast some performance improvement?

File: Screen Shot 2016-11-07 at 2.35.18 PM.png (199KB, 3286x562px) Image search: [Google]

199KB, 3286x562px

>>57430779
>you dont notice that you have car insurance
>even though you are billed monthly

>b-b-but ECC ram doesnt protect against hard drive failures

>>57430815
>I thought a bit ago one of their updates killed them giveing every service its own process.
Thats a Win 10 Enterprise build 1607 VM with the latest patches.

>>57430822
>does dual cpu really have no effect on gaming
pretty much, i use it for more than gayming; pic related

>>57430822
Why are you asking so many questions?
No, dual CPUs won't help at all with almost every single game on earth because there is a limit to how many threads they can scale up to (due to lazy developers). The most is between 6-8 thread utilizations. You only need two or more processors if you need more than a certain number of active threads at once, like with large-scale virtualization.

>>57430822
How many games make good use of 4 threads? Now try finding one that uses 8+, nevermind a fucking dual xeon with 16 cores and 32 threads.

Only good thing is you can do your gaymen while running VMs, converting videos while not giving a fuck.

Also some of the xeons with a higher core count have a lower base frequency and turbo frequency thus making them arguably a bad choice for gaming since you would be better off with 4 really high clocked cores than 16 low clocked ones.

>>57430863
>No, dual CPUs won't help at all
That isnt true, it will help prevent caches from getting flushed due to inactive threads which occasionally become active. So to say at all would be incorrect.

>>57430845
ok, you either need more then 3.5gb of ram for it to do that, or they didn't roll the update out to everyone, not sure which.

>>57430845
>pretty much, i use it for more than gayming; pic related
accesing web clients?

>>57430897
>ok, you either need more then 3.5gb of ram for it to do that,
That VM has 16GB allocated to it

>>57430917
then not rolled out to everyone, apparently insider got the update in october.

>>57430845
>CAPACITY: 43.98GHZ

Erect

File: Screen Shot 2016-11-07 at 2.42.07 PM.png (271KB, 774x1654px) Image search: [Google]

271KB, 774x1654px

>>57430910
>accesing web clients?
no

>>57430890
>it will help prevent caches from getting flushed due to inactive threads which occasionally become active
Tell me, how much of a benefit do you get out of your gaming experience solely because of this?

Exactly. It's negligible. You're right if you're talking about virtualization or running a significant number of parallel tasks, but not so much for gaming. All that extra money you spent on a second 8-core Xeon and a dual socket board is wasted when a simple GPU upgrade or clock multiplier adjustments can net you more performance increases.

>>57422109
AMD is for poorfaggot NEETs

>>57431001
>Exactly. It's negligible. You're right if you're talking about virtualization or running a significant number of parallel tasks, but not so much for gaming. All that extra money you spent on a second 8-core Xeon and a dual socket board is wasted when a simple GPU upgrade or clock multiplier adjustments can net you more performance increases.
Mine are 10 core chips (E5 2660 v2) and i'm the fag running ESXi as the host OS. The anon with the 8 core looks like it is a single socket as the popup button to select CPUs is greyed out.

>>57426005
Is that like a joke. Most i3s are clocked very high and competitive in single core performance with the i7s.

File: speccy 8.png (45KB, 835x634px) Image search: [Google]

45KB, 835x634px

>>57431037
If you're talking about me, yes, this is in fact a single socket system.

Its only because this chip is literally top of the line for Sandy Bridge E that its useable as a daily driver.

>>57431069
>Its only because this chip is literally top of the line for Sandy Bridge E that its useable as a daily driver.
The E5-2687W is better as far as "daily driver"ness goes.

>>57431069
>8GB RAM
>55GB SSD
>Xeon E5-2690
you got some weird priorities

>>57431107
The difference is fairly minimal.

>>57431130
Ram and CPU were freebies along with the board, and I've had this shitty SSD for a while now.

>>57422083
I bought a 4770k 3.5 years ago for $300 and it's still within 10% performance of the new i7. I doubt I'll be upgrading my CPU for another 4 years at the rate Intel is innovating. Getting almost 8 years out of a CPU will be pretty rad.

File: RAID.png (153KB, 1171x1165px) Image search: [Google]

153KB, 1171x1165px

>>57431169
Why do you even have a 2690 if the rest of that box is so shitty?

>>57422300
My waifu is in that picture and you're calling her a slut. Delete this image immediately.

>>57431199
Because my previous CPU was a Q9400, and I knew someone in the industry who was willing to freebie this shit to me in the first place.

The CPU is good enough that I dont really have to worry about upgrading for a good long while.

>>57431253
>I knew someone in the industry who was willing to freebie this shit to me in the first place.
I just looked and saw it was an engineering sample. What a fag, bug filled CPUs dont count.

>>57431286
Its the release stepping, literally the only difference is the damn S-spec number.

>>57426019
What the fuck no
$99 in 1974 is equivalent to $507 in 2016 tyol

>>57419061
>tron 3 news
The fuck?

>>57419061
Please be good, i dont want to give my money to the jew

>>57431326
>I dont have an errata sheet
>but seriously its just as good

File: 6ae.jpg (3KB, 165x115px) Image search: [Google]

3KB, 165x115px

>>57431253
>>57431286
>>57431326
RRRRRRRRRRRRRRRRRRRRRRRRReeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee

fuck you fucking insiders and your shit. Where do I get my rare engineering sample tier hardware???


t normie

>>57431413
ebay where they are dirt cheap because they were bug filled cpus which may have been heavily abused for purposes such as accelerated aging in an oven.

>>57426715
25% profit seems good?

>>57419061
>future-proof
LMAO

I want to see zen speccy threads don't let me down /g/.

>>57426884
>>57426929
I doubt there will be 6 cores at launch unless they have a shit load of failed 8 core cpus. Zen is 4cores per module, not 2 like the excavator cores. So the only 6 cores will be rebinned 8 cores with deactivated cores. I honestly think you'd be more likely to see an 8 core without smt before 6 cores.

>>57427490
Any xeon will function in a compatible motherboard. Just know the socket, then check the cpu support list supplied by the manufacturer. An e5 v3 or v4 will work on x99 boards, v1 and v2 on x79 motherboards. The only dual socket motherboard that would support overclocking is the evga sr-2, which x58 is the last platform you could overclock the base clock of a xeon.

E5-1600s have unlocked multipliers, do not expect to go get a cheap 2670 and overclock it. You will be sorely surprised by the 103mhz base clock maximum. Golden chips might hit 105mhz.

Ecc memory is an option, but not all consumer boards will support it. It is also not required. 1200's and 1400's may also have unlocked variants, but I have never found any information confirming that.

https://en.m.wikipedia.org/wiki/List_of_Intel_Xeon_microprocessors

Handy list of pretty much any xeon you could ever encounter, along with all specifications including turbo steps. I think the only thing they're missing is the tjmax. Also, they will accept memory beyond specification just like any other chip. Some are more responsive to faster memory than others.

Source: my 4400mhz e5-1650 v2 with 2400mhz memory.

>>57419061
Does nobody question the site this is on?
>PREGNANCY
>BABY
>TODDLER
>BIG KID
>TEEN
>EDUCATION
>SPED KIDS
>WELLNESS
>FAMILY LIFE
>CELEBRITY MOMS
How the actual fuck did you end up on that site?

>>57422186
Because everything is bloat now

>>57430845
>Thats a Win 10 Enterprise build 1607 VM with the latest patches.

>>57430897
>>57430917
You're both fucking retarded.
The anon that mentioned Windows 10 getting rid of svchost isn't wrong. But it happened in a recent insider build, and hasn't hit the Pro/Home releases yet outside of insider previews.

Mentioned here: https://blogs.windows.com/windowsexperience/2016/10/07/announcing-windows-10-insider-preview-build-14942-for-pc/#6W1Wb4tMYH8VO3uK.97

>>57432980
when looking it up to see if the patch was out, I found the 3.5gb of ram or more is needed otherwise it stays with svchost largely due to the plethora of win 10 tablets they sold. so i, >>57430897, was right.

did not no there were 2 different patch tiers for win 10, thought that shit just patched at the least opportune time to fuck with you the most.

>>57432864
>The only dual socket motherboard that would support overclocking is the evga sr-2, which x58 is the last platform you could overclock the base clock of a xeon.
No, SuperMicro makes dual socket Xeon E5 boards which can be over clocked via BCLK. They call it HyperSpeed.

>>57432980
>Mentioned here: https://blogs.windows.com/windowsexperience/2016/10/07/announcing-windows-10-insider-preview-build-14942-for-pc/#6W1Wb4tMYH8VO3uK.97
No, you're retarded, it clearly says svchost.exe still exists

>Note that critical system services (services whose recovery require system restarts), as well as a couple of select service hosts, will remain grouped.

>>57433017
>did not no there were 2 different patch tiers for win 10, thought that shit just patched at the least opportune time to fuck with you the most.
Case in point, you're retarded.

>>57422614
It's more like 20% less fps.
My 2600k/gtx1080 Screws about 14% lower than a 6700k/1080

>>57433042
or i didn't get win 10 and never update windows anyway.

need to figure out what ip ranges to block before next year though, really want to dual boot 10 but the moment windows decided, mid render to shut down for upgrades ill murder someone.

>>57433064
>need to figure out what ip ranges to block before next year though
http://www.arin.net and search for Microsoft

>>57433064
Or just install the Enterprise edition and run your own WSUS or SCCM server.

>>57433041
>HyperSpeed
That doesn't work with ALL Xeon-E5s for some reason. I think the model needs to have a W suffix for that to work.

>>57433121
I don't trust microsoft to let me keep my os that I pay for patched the way I want. I plan to make a 'me' port on the router, split it, and have my computer filter all microsoft ip's, and if I ever need an update from them, switch ports.

>>57423103
is malware even noticeable on high end pc's?

File: Screen Shot 2016-11-07 at 5.36.13 PM.png (519KB, 3360x2100px) Image search: [Google]

519KB, 3360x2100px

>>57433125
Most of the prebuilt systems they sell dont use W suffix processors
https://www.supermicro.com/products/nfo/Hyper-Speed.cfm

>>57433149
>I don't trust microsoft to let me keep my os that I pay for patched the way I want.
Thats the entire reason behind SCCM/WSUS.

>and if I ever need an update from them, switch ports.
And you'll get all the patches rather than only the ones you want

>>57433064
>the moment windows decided, mid render to shut down for upgrades
It's something that doesn't happen unless people are retarded.
Windows 10, just like previous versions, gives you a notification that you should get on restarting for updates, You can either postpone it or just reboot. Postpone it for long enough, and Windows will force a reboot.
Know which other versions of Windows do this too ? All of them since XP. Literally. Unless you disabled updates entirely.
Yeah, Windows 10 is a tad more aggressive in not allowing several weeks to go by before forcing a reboot, but you still have more than enough time to finish whatever the fuck you're doing and reboot before you start your render.

That's not to say that Windows 10 doesn't handle updates in an utterly retarded way, though. It does. But "muh forced reboots" is seriously the least you have to worry about.
Worry about how every upgrade that isn't just a security patch essentially pulling a full reinstall of a whole new build of your OS instead.
Or how it'll fucking reset a lot of your settings to default upon doing so.

>>57419061
Intel will run more TV ads to make their processors run faster.

>>57433251
never once had xp force a reboot for updates, thought win 7 was, but found out its a bug once i go over 150 processes the system randomly shuts down.

and thanks, didn't know there was a way to make windows 10 seem less appealing but you found a way.

File: e5-1650v2.png (42KB, 801x373px) Image search: [Google]

42KB, 801x373px

>>57433041
>Hyperspeed
While I`ll admit being wrong about dual socket overclocking, it maintains the 105mhz base clock maximum. And that`s an absolute maximum, most xeons will not run higher than 103mhz, I suppose if you really need that 3%, it`s an option.

>>57433125
It`s an issue with stage 3 being 103.9mhz. Disable that for your exceptional 102.6mhz base clock, and it`ll likely work with any xeon.

>>57433294
>never once had xp force a reboot for updates
Then you didn't postpone them long enough.
Once important updates rolled out, you had (if I recall correctly) about 3 weeks before XP, Vista, 7, 8, and 8.1 would all just give you a 15 minutes countdown until a forced reboot.

This was/is handled in a more graceful way than Windows 10 does, but you'd still be left with no other option than to reboot. The main difference here being that Windows 10 doesn't allow more than a few days, and once the forced reboot does hit, it does so in the form of a notification banner taking up half of your screen that 1) cannot be dismissed, and 2) prevents you from focusing any other window, so you don't get to save your work.
I'm not going to defend this way of causing a reboot. It's retarded, and they should have stuck to the same deal as earlier versions of Windows. But it's the user's own fault if they get there.

>>57423782
yeah..

I wonder whats the difference between i5 and i7 again, for a 100$.

>>57433480
I think I always had updates off or at my discretion, on windows 7 i just have the service shut down because it likes to use 100% of a cpu core and never stop.

>>57433493
the difference is if you can use the threading or not, generally with normal programs, its 5-10% more performance, when a program can use it, it can get close to doubling.

File: Wow+bane+is+now+truly+a+big+guy+good+job+_96632e384e3ded58d624fb70daf9589d.jpg (10KB, 210x325px) Image search: [Google]

10KB, 210x325px

>>57419061
DAILY REMINDER THAT ZEN AND KABY LAKE ARE Windows 10 EXCLUSIVE

>>57433717
nope

>>57434270
Yep
Zen and Kaby lake are Windows 10 exclusive.

Any other OS will not work.
MS trying to Cuck everyone.

>>57434430
>Any other OS
No, just any other version of Windows.

>>57434476
Anything that's not win10 will not work.

MS wants that spyware win10 on every device

>>57419061
I'm using a Pentium 4 right now. Being "future proof" is not impressive.

File: Intel-P-States.jpg (730KB, 3000x1537px) Image search: [Google]

730KB, 3000x1537px

>>57433717
>>57434430
>>57434476
>>57434543
This isn't even remotely true.
Any OS will run on Summit Ridge and Kaby Lake, any version of Windows too.

This tech illiterate retard is regurgitating clickbait headlines.
What Microsoft said is that legacy versions of their OS would not receive updates to support new hardware features. They were referring specifically to intel's new pstate scheme present in Skylake and newer chips. AMD's equivalent is already present in Carrizo.

You can run Win XP or any other OS on these chips. You just don't benefit from their newer pstate switching. Thats it.

File: 1dAi2af.png (111KB, 660x1010px) Image search: [Google]

111KB, 660x1010px

>>57427067
>We're hitting distance limits
>Going out to main memory and back takes quite a while

DRAM access times have been ~10ns for the longest. Moving the memory controller in to the CPU didn't change this. 10ns is enough time for light to travel 9.8 feet. Unless motherboards are now putting RAM on the opposite side as the CPU, it's not a distance problem.

DRAM is cheap and fills the gap between cache and persistent storage nicely. That's why we have high RAM access latency. People are happy just upping the bandwidth.

My question is when are we're going to see an increase in L1 sizes? We've been riding the 64kB meme for 10 years now.

>>57419061
>neverbegameover
>he's one of THOSE autists

>>57434661
I guarantee MS is gonna use this to block any use of win 8 or 7 on new CPUs.

They will f ind a way to fuck us

But but but Muh headlines

>>57433251
>A rapist will only rape you if you keep saying no.
>don't worry about the forced sex it only happens to retards
Kek

>>57435338
a rapist will only rape you if you dont carry a gun. windows will only update automatically if you dont configure a WSUS/SCCM server

>>57425341
do you not understand basic language?

"frequently, usually, they are wrong in a prediction, however, when it happens that they are correct in their prediction, they are typically spot on and very right in their prediction."

File: 1469620373836.png (431KB, 800x584px) Image search: [Google]

431KB, 800x584px

>>57435359
>this rapist will only rape you if you don't employ its father
all of my keks

File: tmp_6930-1445871807546917288051.gif (3MB, 245x263px) Image search: [Google]

3MB, 245x263px

>>57422083
>mfw waiting for Zen

>>57435428
>implying niggers want jobs

>>57419061
nope

>>57422186
>This used to be an acceptable pc.
>Celeron
Nope

>>57429159
this

>>57430777
so might i be best off getting a normal 4x4gb DDR3 non ECC kit?

>>57435472
there was a time when they weren't complete shit.

>>57435793
Yeah, for 2 years.

>>57435776
ECC > non-ECC. Also you do realize that you can buy used 8gb rdimms dimms for cheap.

>>57435860
i kinda want to get new ones, however theres literally 2 new parts in this pc (the Nova 650G and 212 Evo)
so if I want to be super autistic about speed just get ECC, but if im not an autist i can get Registered ECC

>>57435886
>so if I want to be super autistic about speed just get ECC, but if im not an autist i can get Registered ECC
Yes but unregistered ECC memory is uncommon and thus more expensive than registered memory usually. If you look in to 16GB sticks, the price difference is absurd due to economies of scale.

>>57435903
what would be a site to browse/kit/sticks to get?

File: blackgirls.jpg (15KB, 480x409px) Image search: [Google]

15KB, 480x409px

>>57435451
Oh don't worry his father doesn't do anything. You still get raped. He's an affirmative action hire. So you get to feel culturally enriched while the the throbbing botnet cock squirts its last drop inside you.

>>57435886
Also depending on your workload the latency difference doesnt really matter as much due to hyper threading. Hyper threading exists for when cores stall due to waiting for memory to load.

>>57435918
ebay, or servethehome forums

>Bulldozer 2.0
wow its nothing

>>57435929
thanks, if worst comes to worst ill just get the ones listed in the earlier post
>>57430708

>>57435963
For $16 more you get twice the ram and it is low voltage (wont matter too much for only 4 dimms but still) and it is 1600 vs 1333. Also it has free shipping vs whatever jewegg charges.

http://www.ebay.com/itm/SAMSUNG-8GB-PC3L-12800R-DDR3-1600-ECC-Registered-1-35V-LV-RDIMM-M393B1K70DH0-YK0-/371749640654?hash=item568e0159ce:g:Hp0AAOSw8w1X7ao9

>>57436025
thanks anon
i think i might do that, ill get 32gb to start. hopefully i wont need more
I cant conceive of a situation in which i would need more

>>57436101
>I cant conceive of a situation in which i would need more
you can never have enough - >>57430845
>>57430937

>>57423138
>still with 2500k
>hasn't overclocked
You stupid nigger.

>>57436119
i mean, my computational physics professor is having me put together a system for him and he says 32GB per system is a minimum
i lurched when he said
>32gb minimum

>>57422300
Lain or Ritsu.

>>57436166
32gb doesnt go far. For instance my 2 file server VMs take up 24GB just for dedupe. 2 small SQL server nodes take up another 32GB. 12GB is used just for the web interface to VMware ESXi.

>>57436186
i kinda want to try to max out memory in my personal i7 system with 32GB or the Xeon system I have once i get that 32GB, other than Intel Burn at max load
I am interested to see how my CS GO map renders on it

>>57435776
No, RDIMMs only make financial sense if you need to exceed the 16GB per DIMM limit on DDR4 (8GB for DDR3). Otherwise, it's cheaper to go with unbuffered ECC RAM.

>>57436232
an anon earlier suggested i get http://www.ebay.com/itm/371749640654
>>57436025

>>57436219
>max out
>a Xeon
LR-DIMMs are expensive. My dual E5-2660 v2s max out at 1.5TB, but my motherboard is limited to 1TB. It would cost ~$7k to max out.

>>57436232
lol no, this is so full of wrong. first off there are 16gb unbuffered ddr3 ecc dimms, go look at atom servers. secondly rdimms are cheap as shit due to economies of scale, and how every server which is being retired is full of them.

>>57436267
when i said max out i meant fill up all the memory the system has available to use.

Currently my max is about 11GB in attempting to replicate my friends situation where he says he needs 64gb (2x32GB) ddr4 to browse the internet and play games (with a skylake i3 accompanying). I was running Steam, KF2, 78 tabs of chrome, discord, Origin, and Battle.net

>>57436267
>first off there are 16gb unbuffered ddr3 ecc dimms, go look at atom servers
I have an Atom C2750 server. It won't take single sticks of 16GB DIMMs unless it's registered. I don't know where you got your info from
>secondly rdimms are cheap as shit due to economies of scale
Compare the MRSP of a Crucial DDR4 unbuff ECC 4x16GB 64GB set and the same set in RDIMMs.

>>57436344
>I don't know where you got your info from
SuperMicro, the C2750
https://www.supermicro.com/products/motherboard/Atom/X10/A1SA7-2750F.cfm
>Memory Capacity
> 4x 240-pin DDR3 UDIMM socket
> Supports up to 64GB DDR3 ECC or Non-ECC memory
which clearly states udimm, go show me something which says they support rdimms.

>ddr4
were talking about ddr3 anon

>>57436382
Well shit. More reasons never to buy ASSRock Rack

>>57419061
6 months later Intel is going to release something 10% faster than kaby, at 10% lower clocks using 20% less power
And AMD will be dead again.