SANDY BRIDGE FAGS ON SUICIDE WATCH

>>58090100

Jokes on you faggot, I'm already dead.

File: 1471465419966.png (205KB, 505x431px) Image search: [Google]

205KB, 505x431px

>>58090100
>i3 K
N O N S E N S E
O
N
S
E
N
S
E

>>58090100
>an all new i3 overclocked to fucking hell and back is slightly more powerful than a 4yo i5 2500k at stock
no shit

>>58090100
how come i3 has less single core IPC if it's same cores as i5/i7? doesn't it also have almost same clock?

>>58090100
Woo, well my 4790k is fine so who gives a fuck anon.

>>58090406
Probably the lower cache or some shit

>>58090100
no. its still good enough for me.

THANK YOU BASED INTEL

>>58090100
why is single core performance so terrible on cpus now?

>>58090100
>i5 2500k, Q1'11
>i3 7350k, Q1'17
>?????

>>58090478
because enterprise doesn't care about single core performance
they care about cache size, latency and multi threaded applications

>>58090484
oh fuck, here >>58090392 I wrote 4yo when it's actually 6.

INTEL PERMANENTLY BTFO'd and BANKRUPT

ZEN will save us.

AYYMD IS FINISHED & BANKRUPT

>>58090560
my fucking sides.

>>58090560
Nice MaY MaY

File: 1481647560449.png (166KB, 673x599px) Image search: [Google]

166KB, 673x599px

>>58090573
>>58090576
>>58090578
Out in force.

File: 1482177860004.png (157KB, 1008x854px) Image search: [Google]

157KB, 1008x854px

>>58090598

>>58090100
But I have 2600k.

>>58090598
nah just the truth, itll fall flat like all other AMD shit. Theats the problem they lie so badly they make whats coming out sound amazing then its a bucket of stale cum

>he didn't buy a 2500k
wew lad, literally no reason to switch (until I get the need for gpu-passthrough)

>>58090100
that's some pretty good single core performance desu

t.amdfag

> have 2500k and 1070
> still can play the Witcher 3 at ultra settings at 1080p at high framerates
> all the compute tasks I do for work are now on the GPU

I see no reason to upgrade.

>>58090654
big fucking deal

i have an a10 5800k and a rx480 and I can do the same

you stupid, stupid jew

>>58090608
>>58090628

Giving that all the """ rumors ''’'' are constant, and the stuff AMD reveled, even if it does flop(it won't) it will still be better than shelling out 500$ for a sub par processor and a 5-10 % increase in power every subsequent generation.

>>58090654
A-am I wrong in thinking that most programs really aren't designed for parallelization and so real-world CPU performance hasn't fucking budged in years?

>>58090662
Lol no you can't

>>58090697
yeah sure, stay in your bubble jewcuck

>>58090654
>have 2500k and 1070
same, it's amazing how well this thing still runs after 5 years

File: ayy.png (62KB, 657x282px) Image search: [Google]

62KB, 657x282px

>>58090636
>>he didn't buy a 2500k
But anon I did.

>>58090732
>5.9
jesus christ that poor thing

>tfw I still didn't clock mine
too jew to buy a new cooler

>>58090748
>>5.9
>jesus christ that poor thing
I- hang on a second I swear I didn't have it set that high.

>>58090560

I really wish it would but unfortunately I really doubt AMD has that kind of magic in them. I miss the days when AMD made quality processors that were competitive with Intel processors at about half the cost. Drove both companies to be better.

Jokes on you, im still rocking an E8400 Wolfdale

I have a few 2500k and 4690k cpus still. My GCN GPU's are still going strong due to AMD drivers. Feels fukin gud that even my old five or six year machines are still relevant and plenty for gaming today.

>>58090774
Have faith young one, Keller will save us.

File: 1478149531975.jpg (17KB, 250x255px) Image search: [Google]

17KB, 250x255px

>mfw I still have an i5 2500k from 2011

>>58090774
>yfw that was 12 years ago

>>58090732
>5.9ghz

You CPU is moving faster than light? How is your PC not on fire or travelling back in time?

>6 years later
>5 generations later
>i3 has higher stock clocks

wow it's fucking nothing

>>58090811
I think it was adding the current clock speed of all the cores actually

File: 1480818516033.png (37KB, 769x540px) Image search: [Google]

37KB, 769x540px

Still rocking my 2500k
Any real life situation that justify the upgrade ? It may be slower in bench but it still feel like a beast when gaming! How much IPC improvement justify new mobo/cpu/ram ?

>>58090809

Fuck... it was. I think 2005ish was about the last year that AMD wasn't a complete fuckup. When they started jamming MOAR CORES on all of their products and stopped focusing on the actual performance of the cores they fell behind and still haven't been able to catch up. Like I said I'm hopeful for Zen but I was hopeful for Bulldozer too and that turned out to be a fucking mess.

>>58090839
I'd upgrade for kaby.
IPC difference is now visible because of so many +5% generations and high speed DDR4 does affect CPU bound games framerates.

Sure 2500K is still good but if you're not broke af it's a good time to upgrade.

>>58090839
It's not worth the upgrade. Upgrade your GPU and you're good. I have a few 2500k's a couple 4690k's and a new memelake i7 and the difference is I had to switch a bios setting on to install winbaby7.

File: ayy2.png (78KB, 626x352px) Image search: [Google]

78KB, 626x352px

>>58090811
Hmm I ran stress to max out the CPU and it still says 5.9ghz even though it's maxed out at 4.4.
What gives?

>>58090839
No idea senpai, all I know is that buying into the thinkmeme made me realize how much all the new shit is irrelevant. Sure, new GPUs can justify their existence when their whole purpose (for consumers) is vidya, and said vidya ARE running poorly on older harder (2500k vs 7500k or whatever equivalent would be like a GTX 580 vs GTX 1080, which is like 3x faster), but as far as the CPUs are concerned I can't tell a difference between my mobile i5 and my 4th gen desktop i7 when doing anything other than video encoding.

>>58090880
I should have said "preforms 3 times better".

Lots of talk about the 2500k, but how long until my i5-4670k is obselete? The i7-7700k only has like 20% better single core performance.

Am I ever going to see something worth giving a shit about?

>>58090880
This is especially true provided you're using an SSD for your OS. I might encode some vids here and there but either way I'm making a sandwich or doing something else regardless of the CPU for awhile.

>>58090100

Let's try that again with my OCed i5 2500K at 4.8Ghrz.

Also I'm pushing 1.4v on my i5 for like 3 years now and my CPU is literally IMMUNE. Try pushing 1.4 without water cooling on this Shitty Lake.

>>58090899
No, unless you need precious seconds for some professional reason. Xeon would likely be best if you're one of those people.

File: 52547425.png (253KB, 480x480px) Image search: [Google]

253KB, 480x480px

>>58090920
That's a shame.

Seeing faster and faster shit come out was something that excited me growing up. Getting on a more modern system and clearly experiencing the difference.

Is the fun over? Just moar cores?

>>58090662
>>58090654

Both full of shit, witcher 3 is cpu heavy too.
Then again you say high fps you mean over 30 obviously. It cant do a constant 65 with youre configs

>>58090100
I wouldn't know about that, my 2600K still seems to be in the upper range.

>>58090940
Why would you specify a constant 65 as opposed to 60? Is that a thing people do now, care about a constant 65 versus 60?

>>58090938
That ride ended in the mid 00's.

>>58090955
no I look for a lot higher than that.

>>58090971
Yeah okay sure but why did you say 65 instead of 60? Sorry to be so autistic about this, but it's just a common thing to hear people talk about 30, 60, and then 144. I've never heard someone go "a constant 65" as a target.

>>58090964
Ah. Well I had a C2D as a teenager, either the BTX mobo or PSU died so I shared a PC with mid-range AMD quad, and then this. Guess I made a decent purchase. Should be fine for another 4 years, haven't even OC'd the thing yet.

>>58090980
I just picked a figure out. like you say 60 is overmentioned so 65 seemed an appropriate close number hehe.

File: 1f2ee2fdcvhnshw3.png (45KB, 806x601px) Image search: [Google]

45KB, 806x601px

>>58090839
>Any real life situation that justify the upgrade ?
Hardware failure.
I feel it coming, i've got an asrock motherboard
Boys, I think i'll be checking out early.

>>58090991
Ah, okay thank you.

>>58090985
I have an old C2D machine that had vista on it. I can feel the difference between that and a 2500k or newer for sure. Added an SSD to it to speed it up some. It's not that bad. I watch the cpu utilization spike for a bit and then it levels off. Still good imo.

>>58090100
This makes me sad. I have a i7-3770K, and I would love it if after all these years and years of waiting, a new i3 could blow me the fuck away with its single-threaded performance.

But alas, it's only a modest improvement -- 140 to 177, running at stock speed. Do I have to wait another decade to get enough single-threaded improvement to make it worthwhile to upgrade?

I'm just about on suicide watch, but for the exact opposite reason OP is >implying. This is so painful for me to see every new Intel generation now barely benching better than the previous one.

>>58091027
Look on the bright side, if you just collected all the loose change you accumulated over the time you've had your cpu you'd probably have enough to buy the next actually good CPU.

>>58090392
>12% overclock
>to hell and back

>>58091019
It seems an SSD is probably the biggest upgrade you can make these days. Threw one in a family member's laptop and the difference is night and day. She doesn't want a new one anymore because $100 made it better than it was new.

Fingers crossed something better is out by the time it packs up.

My only concern now is how much OCing it might reduce its lifespan, even with a poofan meme cooler.

>>58091085
Just because it's stock speed is high doesn't mean it isn't overclocked to hell and back.
Unless the thing has as much OC headroom as the thing you're comparing it to, it's a dishonest comparison.

>>58090938
>Just moar cores?
Core counts for consumer CPUs double like every decade.

>>58090811
retarded question: what is actual physical clock limit?

>>58091133
Power consumption and heat.

>>58091133
>retarded question: what is actual physical clock limit?
Would there be one if you had a superconducting CPU?

File: 1457235108670.jpg (88KB, 1175x650px) Image search: [Google]

88KB, 1175x650px

>>58091148
there is only so much faster you can make electrons run, i'm asking how much at current node

no idea how would photonics work, they'd still have delays in non photonic parts
not sure if it's possible to make purely photonic chip

>>58091204
>there is only so much faster you can make electrons run
Physical electrons don't actually move very fast through a circuit (think inches per hour)

Also check out this:
http://www.lps.umd.edu/Superconducting%20Computing/SuperconductingComputerIndex.html

>>58091204
I think would probably be worthless.
Electrical signals already propagate at speeds relative to C slowing down based on resistance, light travelling through fiber optic glass is actually slower than electricity in high quality copper. So using light in a processor would likely introduce latency unless it was free air or vacuum based
There's more to a cpu though like whatever clock rate is, I dunno, i'm not very smart. All I know is electricity is real fucking fast so I don't see the point in moving over to light, especially light in some dense medium.

nice bait op. the whole video is hilarius. kaby lake confirmed to bean overclocked skylake.

i really liked how the 7350k that everyone said would be a hit with gaymers turns out to be total shit at running games too.

what a fucking disaster. only buy this garbage if you can't find a good deal on a 6th gen part.

>no 3570k

Forgotten again

>>58091001
I have the same board, been running my 2500k @4.6 for 5 years.

>>58091356
>been running my 2500k @4.6 for 5 years.
Tfw you go out of business because you make a product so good nobody needs to buy another one.

>>58091334
>I don't see the point in moving over to light
capacitance, inductance, resistance to name a couple reasons.

>>58091118
this

the core architecture is reaching its limits and all Intel can do is try to improve power efficiency so they can increment stock frequencies every year

>>58090938
>Is the fun over? Just moar cores?

Looks like it. Starting about 2003, two things hit the CPU industry at the same time: (1) the top CPUs started requiring so much power that there wasn't a whole lot of room left in a desktop case for the heatsink+fan plus enough room for airflow to manage all that heat; and (2) the rise of the mobile market, which had an absolute demand for low heat and low power solutions.

So they decided to top out the desktop CPU power at about 125 W, which pretty much brought an end to increasing the clock speed to get more power. (The power draw is proportional to the square of the clock speed.) Then they turned the development of their product lines to focus mainly on the needs of the rising mobile market, where they knew most of their future revenues would eventually come from -- and where low power is a mandatory requirement, so higher clock speeds are simply not an option.

About the only thing they have left to improve the speed is architectural changes, like trying to squeeze another 2% efficiency out of the pipeline architecture, or try new design tricks like speculative execution and branch-prediction techniques to maybe get another 2% increase.

>>58091375
I don't know what inductance is but light based processors will have limits similar to capacitance and resistance (at least if i'm remembering what barrier capacitance presents to CPUs, but as said i'm pretty stupid so i might be wrong).
Anything you make to emit the light will have a cool down, with LEDs is something like millionths of a second but it's still not infinite and there will always be energy lost to heat no matter what you do, the material you pass the light through will absorb some of it, the detector would likely lose some of it, i can't see it being any more efficient than a photovoltaic cell so unless some weird shit starts happening with photovoltaics at the nanometer scale it will probably lose over half of the energy here.
As a general rule of thumb i don't get hopeful about a technology that requires the invention of a much more efficient or entirely new technology to be feasible, but i don't know enough about this to rule it as impossible so i'm just sharing my thoughts.

>>58091352
we're just too good

>>58091204
http://www.extremetech.com/extreme/195322-us-intel-agency-is-developing-a-superconducting-exascale-computer-and-cryogenic-memory

>>58090100
>tfw your 2500k doesn't take kindly to overclocking

Or maybe I've just been doing it wrong. And by "wrong" I mean "auto"

>>58091583
Yeah you need to manually adjust voltages to get good numbers, i usually just use voltage offsets and trial and error because anything else with my board loses voltage step down when it's idle.
You can usually find a list of what clocks and voltages people were able to achieve with their cpu, pick a low one and add a couple millivolts to that and you've got a stable overclock most of the time, then stress it to see if your cooling can handle it.

>>58091583
Some just don't, it's a lottery. I had one that wouldn't budge past 4.2 no matter what.

>>58090100
meanwhile the 6700k gets BTFO in cinebench by an almost 5-year old, $70 sandy bridge xeon

>>58090875
>What gives?
Screenfetch is shit

>>58090100
Nvidiots on suicide watch.

Guys apparently superconducting CPUs could reach >500ghz. There is still hope.

>>58091133
>>58090811
I was there to watch CPU speeds hurtle from 33mhz to 4000mhz, slam into a brick wall at the end of the Pentium 4 era and slowly roll back to where most of them are between 2000 and 3000mhz now.

I know it's almost meaningless due to IPC but still there's no rush to make new ultra-powerful 33mhz processors, you know?

>>58091630
But you can't game on a xeon xd
Those xeons would be two times more badass if motherboards weren't so fucking expensive

I'm still on Westmere and my CPU gets almost double that score.

>>58091334


Here's a photonic dual core CPU they made a year ago.
Apparently it has fucktons of bandwidth and consumes jack shit in terms of power.
They've actually been pumping a lot of money into photonics for some time now and that's most likely what they're going for next.
Silicon-germanium photonics, rather than graphene.

http://news.berkeley.edu/2015/12/23/electronic-photonic-microprocessor-chip/

> new chip touting a density of 300 gigabits per second per square millimeter. That’s 10 to 50 times better than traditional electrical microprocessors.
> the chip is also energy-efficient, using only 1.3 picojoules per bit, equivalent to consuming 1.3 watts of power to transmit a terabit of data per second.

File: nochuckle.gif (2MB, 245x246px) Image search: [Google]

2MB, 245x246px

i'm gonna need to replace my motherboard due to obsolescence before I need something faster than my 2500k

File: 1481984851447.gif (338KB, 538x572px) Image search: [Google]

338KB, 538x572px

Socket change a year

>>58090636
I got a 2600k instead. I'm sad Kabey lake isn't an improvement I was actually thinking of upgrading this year.

>>58092149
why not? I do it all the time.

File: fakelol.jpg (239KB, 1537x723px) Image search: [Google]

239KB, 1537x723px

>>58090560

File: cpu.png (10KB, 461x219px) Image search: [Google]

10KB, 461x219px

>>58090636
but i did. also it was only $100 when i got it.

>>58093239
My motherboard is probably going to die before I need to change the CPU, unless I fuck up on getting rid of ME.

File: 1460438266851.png (30KB, 406x414px) Image search: [Google]

30KB, 406x414px

>>58090100

4790k vs 6700k is the relevant bench duel imho

>haifa designed lake backdoors

File: dXtuCne.jpg (186KB, 586x880px) Image search: [Google]

186KB, 586x880px

BREAKING NEWS. THIS JUST IN.

>BENCHMARK REVIEW SITE SHOWS A 4.8GHZ DUAL CORE WITH HYPER THREADING MATCHING, OR SLIGHTLY EXCEEDING, A 3.3GHZ QUAD CORE NO HYPER THREADING FROM FIVE YEARS AGO. A 1.5GHZ GAP BETWEEN THE TWO.
MORE NEWS AT 11.

>>58094715
What the fuck are you on about?
>>58091630
Wow a $350 cpu isn't quite as good as a $1300 cpu?

File: fixed.jpg (219KB, 1920x1080px) Image search: [Google]

219KB, 1920x1080px

>>58090100
i fixed it for you.
need to show value for monies.

File: 1546879879.jpg (67KB, 920x767px) Image search: [Google]

67KB, 920x767px

>>58094853
BREAKING NEWS. THIS JUST IN.
>THE KABY LAKE I3, KNOWN AS THE 7350K, COMES STOCK AT 4GHZ VS THE 3.3GHZ 2500K. A 700MHZ GAP ALONG WITH THE +15-20% IN SINGLE THREADED PERFORMANCE BROUGHT BY SKYLAKE SINCE KABY BROUGHT VIRTUALLY NONE OVER SKYLAKE.
THIS IS ABSOLUTELY MIND BLOWING. MORE NEWS AT 11

>>58094867
>Wow a $350 cpu isn't quite as good as a $1300 cpu?
>$1500
you know you can get sandy bridge 8-core xeons for like $70 right?

File: e5a393ac65cd4a7f938268d0d679d1d5.png (168KB, 1080x791px) Image search: [Google]

168KB, 1080x791px

>>58094853
>>58094908
BREAKING, BREAKING, BREAKING. THIS IS KTLA10 BRINGING YOU THE LATEST 11 O-CLOCK NEWS!

THANK YOU TOM FOR YOUR BREAKING NEWS COVERAGE. THIS IS ABSOLUTELY INSANE. HOW WILL 2500K USER'S WHO PICKED UP THEIR QUAD CORES FOR THE AVERAGE STREET PRICE OF $210 BACK IN 2011, FIVE YEARS AGO, EVER RECOVER?

>>58094906
+ 500 mobo + ecc ram + 250 watts + 2x coolers
Fuck off with your non Ex/Ep v1
You fags act like you discovered El Dorado with your used parts
>>58094911
Not fucking new you can't

>>58094911

The Skylake Xeons are gonna be kickass imho

>>58095138
They already have SL xeons. Maybe SL-ex is what you mean.

File: 1461156770321.jpg (74KB, 739x600px) Image search: [Google]

74KB, 739x600px

>>58095119
THANK JOHN! THIS IS TOM AGAIN. I DON'T KNOW HOW 2500K USER'S WILL EVER RECOVER. CONSIDERING THEIR 2500KS RETAILED FOR ABOUT $210, THE 7350K AT AN MSRP OF $180 AVERAGE FIVE YEARS, GOING ON SIX YEARS, IS SIMPLY INCREDIBLY. ITS REMINISCENCE OF STEVE JOBS. IT TAKES COURAGE OF INTEL TO PULL OFF SUCH A FEET.

BACK TO TO YOU JOHN

>>58094908
This image induces the worst type of rage

>>58090100
DELET THIS MOTHER FUCCKER

KABY LAKE? MORE LIKE SHITTY LAKE

I think the Sandyfags are a little delusional, they look at the 5-10% improvement every year and go "LOL, looks like my Sandy is still good"

But don't realise 4 generations have passed, so that 5% has turned into 20-30% between the new chip and sandy.

File: hyeassGWfIqC1rv3SFlN874y8yYQDszL9qxmGOe75Cw.jpg (110KB, 548x768px) Image search: [Google]

110KB, 548x768px

>>58095187
YOU SAID IT BEST TOM. 7350K WITH AN MSRP OF $180 MATCHING A 2500K FROM NEARLY SIX YEARS AGO RETAILING FOR $210 BACK THEN IS NOTHING SORT OF COURAGE FROM INTEL. JUST REALLY INSPIRING AND SHOWS OFF THE PURE INNOVATION THAT INTEL IS ABLE TO IMAGINE AND CREATE.

File: Piriform Speccy.png (33KB, 758x554px) Image search: [Google]

33KB, 758x554px

>falling for the Sandy Bridge meme

>>58095267
>Falling for the windows 10 meme

my 3220 is fine tho

>6 years newer
>barely faster
>barely cheaper
Intel = joke

My i7 920 PC is nearly 8 years old and it's still fine.

File: Speccy64_2016-12-20_21-07-51.png (3KB, 310x58px) Image search: [Google]

3KB, 310x58px

>Not using based 2550K

>>58091583
I had such an easy time ocing my 2500k I couldn't believe it and thought I was doing something wrong and that the system was lying to me. Went into bios and up the multiplier until I hit 4.8ghz and left everything else on auto/untouched. Motherboard is an asus P8P67 Pro.

Strangely, it was unstable like that at 4.6ghz and wouldn't boot properly sometimes but on 4.8ghz, not a single startup problem.

Enjoy your Intel ME, retards.

File: JUST.png (9KB, 800x195px) Image search: [Google]

9KB, 800x195px

>>58090100
>Everyone arguing about their space age tech.
>Then me.

>>58096142
>1090T masterrace

still running that 4790, feels good

>>58091583
I'd check it out myself. I've never seen a 2500k that cant run a safe 4.5ghz. SB/e is a champion overclocker, almost as good as the nehalem chips.

File: 1462275001676.jpg (27KB, 850x460px) Image search: [Google]

27KB, 850x460px

sandy/ivy bridge is to icore cpus how the P4 was to pentiums

What's even the point of new shit i7s? I bought a EVGA x58 Mobo, i7 920, and 4gb ripsaws for $140. And after my cooler gets here I will oc the shit out of it, but with that being said it is great even at stock speeds

>>58096322
>overclocking on older boards
I hope you enjoy the smell of burnt plastics in the morning

>>58090100
how is this anything new you dumb shitposter

>>58095394
Yeah it sucks but I need it for teh gaymes brah

>>58096320
Pretty sure they're not hot as fuck and shit mate.

>>58096489
>P4 shit
le sheldorn face

File: Screenshot_20161220-164453.png (332KB, 1080x1920px) Image search: [Google]

332KB, 1080x1920px

>>58096280
Some will need some crazy voltage , some won't , they all do overclock still it doesn't mean it overclock well!

>>58096381
Is fine. Is only oc. Plus this board seems prime. Has potential, and if anything happens, I'll buy a new one for cheap

>>58096684
>I'll buy a new one for cheap
Good Luck

>>58090478
Because multithreading is far superior in most circumstances

>>58090478
Because single core performance only matters for muh gayming.

>>58091133
not sure about physical limits but for silicon chips there's diminishing returns past 3.5GHz or so for heat/performance, iirc. that explains why most of intel's non-low power chips are clocked between 3.4-4.0 GHz and less than 95W. i'm sure any chip at 5GHz will be using 200W if not more, and if it could sustain 6GHz without melting it'd easily use twice that. so even if you could reach the limit of what's physically possible, it'd be completely impractical to do. clock speed, voltage, heat, number of transistors all contribute to power consumption.

>>58096684
you should have gotten one of those 32nm 6-core LGA1366 xeons that can clock as high as sandy and run cooler than the 45nm i7's.

>>58090100
Sandy Bridge, Sandy Hook. Think about it.

>>58090478
almost everything can benefit from multi core, and the few things that do not have been good enough sense sandybridge that it's hard to bitch.

>>58090478
improving single threaded performance is difficult and expensive without increasing clock speed, and they can't significantly increase clock speed. They've hit a wall where increasing the number of instructions per clock cycle or improving branch prediction is getting more and more difficult.

>>58090100

There is no conceivable situation where you would ever need more than 452 of whatever that graph is measuring

>>58090100
Trolling /g is like pissing in an ocean of piss.

>>58097048
not even then, only old games rely on single core, and any old game even a 8350 can drive to playable.

>Zen

soon all of this Intel faggotry will be irrelevant.

>>58090100
Fuck off skype. My core quad is doing just fine. Never buy new.

>>58102307
I would argue that if you play any shitty F2P MMO, AMD CPUs are just asking for 10 FPS in crowded towns.

>>58090100
A seven year old technology, while still fairly powerful, is slowly becoming obsolete as more powerful and efficient models and technologies replace it.

I have never seen such a thing happen in my life.

So I have x5650, 3570k, 4590 and 4790 in my homelab. Also for lower wattage server I have e3-1220Lv2 and dual e5-2670s for one lease machine which I use for client's stuff.

I really dont see a reason to upgrade any machine in my lab. At least till zen hits and we get real life test results for it.

>getting a 7350k instead of a 6400

You'd have to be completely brain-dead, unless you're doing absolutely nothing other than using dolphin emulator.