POLL: How much are you willing to pay for 8c16t Ryzen

>>58024220
Nothing because i am a poorfag and whatever it will cost it will be too much.
More interested in their 4(8) CPUs.

>>58024220
>tfw poor neet.
i'll wait for the $99 4 core 8 threads.

These are too high end for me. Looking forward to seeing something compete with i5 6500 performance

>>58024220
I want 8core 8t for 250-300$

i5 6600K go for around $250 So less than $240 for an AMD CPU.

>>58024220

max i would pay would be equal to the 6700k at $330usd

>>58024269
I doubt they will have models with SMT turned off, as it counts for a huge amount of the performance.

You can buy a 8 core Xeon E5-2667 v4 for $332.

http://www.ebay.com/itm/Intel-Xeon-E5-2667-v4-ES-LGA2011-3-2-9GHz-Compatible-with-X99-i7-6900K-6950X-/152336234507

You would be a big fool to pay more than this for the AyyMD equivalent.

>>58024220
Why would I pay for the flagship when I can conceivably just buy the "slower" 8 core and overclock it?

If I were in the market for an 8c/16t class CPU, I'd give 'bout 400€.

I sincerely hope their mid-range Ryzen APUs with 4c/8t go for no more than 150€.

File: 1481671240637.png (418KB, 627x627px) Image search: [Google]

418KB, 627x627px

>>58024220
I'm not looking for the high end but any ZEN that going to fall in the $350 range is what i will be upgrading to.

>>58024220
>I said $600 - 700
but I olny know NZD so....

>>58024313
>you would be a big fool to pay more for a better chip

so tired of sandy bridge hipsters

>>58024313
no thanks, pleb

>>58024404
>better chip
We are talking about AMD chinkshit here.

poorfag so I'm gonna wait for the $100 Zen chips. It'll have to be bought with a new motherboard and ddr4 ram. And on top of that, I'll be getting some other parts to put my old CPU in as a home server.

>>58024416
meanwhile the latest and greatest intel 8 core is getting btfo by 'chinkshit', let alone a five year old piece of trash

>>58024419
>not having made a couple grand off of AMD stocks

>>58024419
I'm very hopeful that AM4 will have longevity as a platform, and with Zen it'll actually be worthwhile.

>>58024220
The price of its Intel counterpart so around $150.

>>58024440
Just put in an order for 45 shares, I'm looking to ride the rise after CES.

File: HC28.21.121-Next-Gen-Mem-GPU-Kim-SAMSUNG-v02-t1-3-07.png (155KB, 2000x1125px) Image search: [Google]

155KB, 2000x1125px

>>58024441
DDR5 is already coming in 2018 so either AM4 gets replaced soon or it will be as outdated as AM3+ is now.

i think $350 is the ideal price for their cpu. otherwise ill probably go with the intel 6700k counterpart once they drop in price

>>58024475
>DDR5 is already coming
In development. There is no planned market penetration from any DRAM producer. DDR4 sales are still ramping up.

>>58024475

when does HBM memory arrive?

>>58024475
DDR3 stayed around for 7 years
DDR4 is expected to have higher longevity and has only been around for 2.

>>58024475

Gddr5 has been used for years in amd graphics cards

>>58024524
Its out, just not cheap enough.

>>58024524
HBM goes directly on the chip. It's not like other kinds of memory.

$0. I won't buy sandy bridge level IPC in 2016, and I certainly wouldn't buy from a company who reduces costs by firing people and replacing them with H1B.

$299

>>58024568
Anything to back that up? I'll sell my fury x right now and give the money to nvidia

>>58024568
so i guess you'll be buying from VIA, then?

>>58024470
Terrible timing. You should've bought at ~3 like I said in the Polaris prediction threads. This is just stock gambling 101, but NEVER buy on the peaks. You're just asking to lose your money.

>>58024568
>I certainly wouldn't buy from a company who reduces costs by firing people and replacing them with H1B.
So you're never going to buy any computer product ever again?

since its probably going to be the regular, unimpressive and usual let down from amd i won't pay anything more than my i5 4690k which was about £150.

IF by some miracle it actually performs nearly on par with 6900k, then 500-600$ max.

The price tag of 1100$ on 6900k is utter bullshit, especially when compared to the sub 400$ price on 6800k.
I'd be willing to pay 500-600$ for the 6900k and same goes for any equivalent CPU.

>>58024220
personally, I believe it will be close to a 6700k in price, but am willing up to 500$ for it.

>>58024220
$450 is my limit

>>58024389
you were in the 400-500$ range...
google is your friend for currency conversion.

File: USD-NZD.png (10KB, 518x166px) Image search: [Google]

10KB, 518x166px

>>58024676
Right.
well I did as you said.
its roughly in the ball park you said.

that is my max on PC spending for a single component.
pic related.

what die size are 6900k mm?
if we compare it to GPU silicon of same size, which is in not less complicated, would it be same price?
if it's less i'd say AMD gonna price it in same range as GPU
they will go for bare minimum margins to stay afloat this year

also how they did GPU got serious conference
and CPU got gaymer one indicates pricing is under $500

>>58024780
amd need money and they'll charge whatever they think they can get. CPU margins have traditionally been much larger than GPUs as well.

>>58024779
depends entirely on performance and if it seems preformance isn't getting better.

but yea, 500 this time around is my top as ill probably have the computer for a good 6-10 years before an upgrade MAY make sense.

>>58024780
no 6900 size, but haswell 8 core was somewhere around 330-360mm
zen is half a 8350 so around 160mm

>>58024220
300-350usd
3.4ghz base sounds about right

If the benchmarks come out and it actually can hang with a i7 6900 i'd be fine paying that but if it cant beat my old i5 4690 or 6 core i7 ill just keep waiting

>>58024819
>zen is half a 8350 so around 160mm
for 8 cores? seriously?
my point is they sue same fabs for gpu and cpu the prices depend on size as far as I know on fab side
and if they can get decent margins from 200mm 480s they sure can get a ton from 160mm

>>58024816
remember consumer gpu is as of now a write off for amd, they are looking to make money in enterprise.

if am4 is a several generation long socket, right there is where you have your returning customers. If you can either give us better cpus or more cpu cores, that is where the upgrades will come from.

get someone with a low end quad core, if they feel the need, get them again with the next gen quad core or a higher core count cpu. Its not like people are locked into an intel platform as they change cocked all the fucking time.

>>58024849
remember, what adds a fuck ton of space to a die is the fpu to my understanding, at least a fuck ton of complex space, intel has a 256bit fpu and amd has a 128bit where it splits a 256 up in two to run it. this makes higher fpu applications an intel win, but in a home user base, how many applications use 256 or 512 bit fpu hammers?

in the i7 area, you have a 4 core i7 that takes up a crap ton of space because of the gpu.

the amd cpu, as I can tell, has no fat to it at all, its a sweet spot hitting almost everything.
they said it in a earlier conference saying their cpus will be on equal or better ground for 80% of all workloads.

and to the yields, if they really do make a 4 core cpu, they have half a cpu they could fully disable worth of errors.

>>58024819
>depends entirely on performance and if it seems preformance isn't getting better.

You mean major improvements?
not likely AMD would need to be more than compeditive but RYZEN (Zen) will bring compeditiveness back for sure.

Given all the data i can't belive sandybrige trolls
because many of them have not had a AMD CPU and they work differently and excel in different areas

lets forget Bulldozer as that is not SMT it is CMT and totally not suited to consumer users
no body here has had a rea first gen bulldozer you all talk about versha which is not the orginal bulldozer

I had the original bulldozer and I don't think /g/ really knows outside of meme's
I'm now conviced of this now

my mid range FX6100 held its ground right up to 2015
but as soon as the amd 300 series came out
it was clear it was never going to surive as a cpu.

Zen based CPU's feel more on the lines of broadwell-e era
thge kaby lake and so forth don't really match and i fear for the APU market because it really feels underpowered at this very moment

i want to see how RYZEN (Zen) holds on multitask duties.
IPC is kinda nosence as its only really used in older games DX11 and older.
not to say its not useful its but a farcry from main stream useage of the futurei my eyes.

No more than 250 euros

250-350 burger bucks

>>58024313
>http://www.ebay.com/itm/Intel-Xeon-E5-2667-v4-ES-LGA2011-3-2-9GHz-Compatible-with-X99-i7-6900K-6950X-/152336234507

>Engineering Sample
y-yeah, no thank you

>RYZEN (Zen)
Stop this.

>apple adopts ryzen apu with hbm2 for system and video ram.
>hsa is now real life.

>>58025320
>RYZEN (Zen)
U mad?

Product name (CPU Name)

I made this to make it easier to understand.
and no i will not stop this your not my boss.

>>58025607
>your
No, but you're a cuck.

>>58025292
If I knew 100% that it would work, I would, but im not rolling dice on 350$

>>58024220
$350 max

They'd have to pay me to you that garbage.
>b-b-but the marketing says it's good
Fuck off.

>>58024220
three fiddy

I guess if the 6c/12t goes for around 250€ it will be a nice upgrade to my 3570K

>>58024220
Let's just say if it's more than $600 I'll buy Intel, a proven brand

max probably $500, hoping it'll cost less than that

>>58024220
considering that we expect a 4c/8t and a 4c clean id say
8c/16 no less than 500
4c/8t 350 or about that area..
4c clean 200-250

>unironically voting for high prices

>>58024220
If the leaks are true. I'll pick the $350 version 8c/16t even it is not overclockable.

>>58025821
4c/8t is minimum, they aren't selling cpus without smt

>>58025945
they will otherwise there is no point selling mid to high stuff and leave the low market where they shine without nothing

>>58025957
they could just sell 2 core 4 threads for the low end.

>>58026092
>2c4t unlocked multiplier, oc all the way to 5GHz
might buy.
it's over i5 is finished.

Sandy Bridge octocores are around $70, so the IPC is going to need to be drastically better to make Zen worth buying.

>>58024568
>sandy bridge level
Zen is above Haswell-level performance

>>58024568
So you're saying you wouldn't buy Skylake.
That's great, but this thread is about the AMD Zen processors.

File: 79ab0f4b2caf3d35c0135e4c66699f84.png (201KB, 799x574px) Image search: [Google]

201KB, 799x574px

>>58024220
>$300-$400 is in the lead

You poorfags are in for a serious wake-up call. Intel might overprice their CPUs because they're jewey as fuck, but AMD needs to recoup their massive investment. It will be at least $700. Maybe even $100 less than Intel.

>>58026841
That's great, but when they go out of their way to compare it to an i7 6700k in their pr blurb there's no reason to think they're going to price anything that high.

There is no market in the 'enthusiast' sector that needs to churn out blender renders.

There's no market in the enthusiast segment that has a user too stupid to enable hardware encoding off their video card for streaming.

The intel hex and octocores aren't the enthusiast segment, they're the retard bait segment.

If AMD tries to set the price of their cpu at 700 they're done. The mobos won't even have any future proof features.

AMD doesn't go after the .1% markets either, even for server skus.

>>58026198
this, although 2011 motherboards are expensive due to demand.

>>58026921
Cheapest 2600k I can find on fleabay is $180-$200. No clue where you're finding these dirt cheap ones, unless you're talking about low end i7's. And yeah the motherboards for them are stupid overpriced now.

>>58024220

Eh I'm happy to wait for the 4c version.

Less than 150 bucks.

>>58024251
>that will happen

50 less than the Intel equivalent and no more

>>58024540
GDDR5 = DDR3 but with higher data output at the cost of higher latency
Its not DDR5

>>58024607
>not shorting on a peak
Study more little one.not saying this is actually peak. Watch it double again. Then sell and short. Ride that wave baby.

>>58027133
we are talking about 8-core xeons, i.e the $70 E5-2670, not i7's

i will buy a intel's cpu with lower price if amd is that good

>tfw recently bought AM2+ motherboard

sorry AMD, I'll have to wait a few years

>>58024313
That's a huge risk you are taking with that ES.
A shitload of mobo does not support it (e.g. gigabyte) and the ones that currently does might dump support in future BIOS releases.
Also, sometimes early ES miss certain CPU features.
If you have that kind of money to spend buy something else.

>>58024656
This.
Intel went bonkers with the prices, 1000 USD for the top of the line desktop CPU I can understand, Fucking 1100 USD for a mid/high range CPU and fucking 1800 USD for the best one is crazy.
If, and that's a big if, AMD ryzen 8c16t manage to get decent multi thread performance (let's say better than 6800k) at a decent price (~400/450 USD) I will get one of their CPU just to say a big fuck you to Intel.
If performance are on par or higher than 6900K then 600 USD is what I would spend for ryzen 8c16t.

>>58024220
i don't need 8 cores/16 threads

i mean, i won't mind if it'll be retardedly cheap, but we all know it won't

>>58024220
I'll pay as much as I need too, I need to support AMD no matter what

-Sent from my FX 8350

Will the 4 core zens have better/equal perfromance compared to a skylake i5?

>>58028899
Comparable performance and wattage at ~75% of the Intel price point is expected.

>>58028899
We have no idea
AMD says yes but they've said yes many times before
Remember when they said two 480s only running at 50% were better than a 1080

>>58028911
With 2/3 IPC compared to intel

>>58028911
75% off the average price of an i5 you mean?

>>58028911
>>58028899
>>58028927
If it sounds too good to be true with AMD, it is

It's like people learned nothing from Bulldozer marketing and Polaris marketing

>>58028934
Of whatever Intel CPU with a similar performance to their offerings, if they want to play it clever. Not too cheap so they don't look too good to be true, not too expensive to avoid people going Intel for just $10-$15 more.

>>58028946
this
bulldozer was supposed to clean our dishes and mop our floors the way it was marketed, everyone here freaked out at how amazing it was just like they are now and then it turned out to be fucking nothing

>>58024220
I am willing to pay the leak prices

eg $350 USD for the 3.4ghz Ryzen chip

anything more will be difficult for AMD to justify

>>58024313
>http://www.ebay.com/itm/Intel-Xeon-E5-2667-v4-ES-LGA2011-3-2-9GHz-Compatible-with-X99-i7-6900K-6950X-/152336234507


>falling for the eng sample meme
>buy this exact listing
>end up with a 2.6 ghz 8 core

dafuq

>>58025957
They'll probably target the low-end market with APUs and Athlon-equivalents. But they're releasing the high-end desktop parts first, this time.

>>58024244
get a job magill

>>58024313
Where can I find used Zen cpus?

>>58029152
Trip fag kill yourself

$300-400
It's only dual channel. If they want more money sell unlocked opterons.

>>58024313
How much is the motherboard?

For me it all depends on what kinds of clockspeeds end being offered/attainable on the 4c or 6c variants. If they lock the good single thread perf to the 8c parts I will have to strongly consider my options instead of just going with AMD

>>58029390
>If they lock the good single thread perf to the 8c parts

why on earth would they

>>58024220
250

>>58024220
I want to build a budget pc next year. I assume that ryzen will be far too expensive? Or are they going to release a cheaper version?

if i can get 4c/8t one for ~100€ might consider upgrading my 965BE

>>58024251
>>58024269
>>58024372
>>58029860
This. I would get a 4c/8t if it was a decent price.

>>58028119
The motherboards are still expensive, aren't they?

>>58029977
>The motherboards are still expensive, aren't they?
At at least $200 per board, yep. Same issue with LGA 1366. The boards arent made anymore, and then the server farms dump their Xeons on ebay for cheap, which then further drives up demand for remaining units.

I'm going balls out with an enthusiast build next year (got to show my e-peen and all) and if zen undercuts x99 by a decent amount in price i'll buy it and not even blink. I just hope ram prices come down a bit as as much i'd like 32gb its hard to justify that for my usage needs but eh, 2x8gb is enough as I can always simply get another kit later on.

When is that AMD showcase. Wasn't it today or tomorrow?

depends how well the single core performance is because I want to play Total War games at 60FPS yes I know I'm trash

>>58029390
This makes no sense at all. It's much easier to achieve high clock speeds on lower core counts, because for one your TDP is much lower clock to clock.

>>58030218

Tuesday. As in, 4 days ago

>>58024220
$200 or less
There's no performance justification for me to spend more than that anymore.

>>58024220
I'm hoping it's not more than $349 for the base 8c/16t variant, but considering it's competing with an $1100 CPU performance wise that might be asking for a bit too much.

>>58030097
Are 1366 mobos really that expensive?
I have a P6TD deluxe in a box somewhere, with ram and i7 920

>>58030097
LGA1366 motherboards are more affordable, but still expensive considering they're old tech.

used LGA2011 boards are so expensive that I just bought a new server motherboard with warranty and paired it with a E5-2670 and ECC RAM.

>>58030373
They can be depending on the board.
An unusual example (because of its configuration) is the SR-2 Classified. They run between $500-$600.

File: 1322332065482.jpg (408KB, 800x1199px) Image search: [Google]

408KB, 800x1199px

I honestly couldn't give two shits, as I moved to 1440p there is no way my 2600k at 4.4ghz will be a bottleneck anyway. I'm much better off spending the extra cash on upgrading my GPU when the time comes

File: 1480901409567.jpg (73KB, 405x430px) Image search: [Google]

73KB, 405x430px

http://cpuboss.com/cpus/Intel-Core-i7-6700-vs-AMD-A12-7th-Gen-A12-9800

I'm trying to decide which one would be better.

>>58026463

no, it isn't. if AMD's marketing team claims it's haslel performance that means it will be 20-30% slower.

>>58030373
Yep, even the cheapest boards that are no-name chink brand shit are still just shy of $100. High quality boards are easily 2-3x as much.

Its why I got enormously lucky in the purchase of this DX58SO board for $30 practically new in box. Decent overclocker too, I've had it up to 215mhz bclock with no stability issues.

>>58024220
If there 3.4 gh, 8 core processor isnt sub 300 then most people would have no reason to upgrade to it.

File: 1396693462394.jpg (117KB, 330x319px) Image search: [Google]

117KB, 330x319px

>AMD poorfags
>willing to pay 400-500 bucks for a CPU

>>58030479
Well, if you plan on doing anything that requires performance, the i7 will be the best option.

>>58030625
The amd seems to have high numbers but doesn't have l3 cache and the extentions on the amd and i7 would suggest i7 can do a ton more work but idk what those extentions do and if are unlocked with a good board.

>>58024220
All you people here hoping it would go for 400 bucks or less,yet it has performance greater or equal to intels 1000$ CPU. Real poorfags here

File: 1481561773160.jpg (78KB, 717x729px) Image search: [Google]

78KB, 717x729px

>>58030705

i have a 4790k, if it isn't a significant improvement, then 0

>>58024504
>>penetration
nofap. Don't trigger me.

>>58030494
>beats Broadwell-E clock-for-clock
>"hurr 30% slower xd"
I think there's a term for inventing shit as a coping mechanism

denial? delusion maybe?

>>58030705
Intel's $1000 processor isn't all that impressive. The advancement of processors has slowed to the point that each generation is only a negligible improvement over the previous. That's why Intel bought Altera, the most value will come out of being able to retool the CPU depending on workload.

>>58030786
>inventing

real independent benchmarks done on zen engineering samples showed performance closer to westmere and sandy vagina levels. AMD's marketing team (history of lying and exaggeration) says haslel. pretty obvious that it will be somewhere in between, which is 20-30% slower.

File: 1419206769559.jpg (328KB, 810x587px) Image search: [Google]

328KB, 810x587px

>>58030854

>real independent benchmarks done on zen engineering samples showed performance closer to westmere and sandy vagina levels.

>ES
>unknown clocks/very low clocks
>unknown featureset
>unkown what they were testing specifically
>AoTS website, known for being easily faked
>all legit guise I swear!

>>58024568
>reduces costs by releasing the exact same chip but using a better binning process after realizing overclockers were getting absurds amount of performance from the same chip
>charge over 5x for the better binned chip that's rebranded
>keep the price of the worse binned chips the same
fuck amd.

>>58030786
>falling for the Chinese Jew
People have run that Blender benchmark on i7 6900K at stock speed and have gotten times a few seconds faster than the Zen sample (between 42-50 seconds). AMD gimped their 6900K to make their CPU look better, as usual.

>>58030854
>chooses to believe leaks over real, verifiable demonstrations
kys

>>58025320
What's wrong with it
Ryzen is the name of this cpu series.... and zen is the name of the architecture

>>58030854
you realize vishera is already sandypoo single core?
you realize its multithreaded performance still competes with $1000 intel chips?

>>58030887

wow, you're deflecting pretty hard. are you really going to try to claim anything that doesn't come from AMD's ministry of truth is faked? this is shillary clinton levels of delusion. maybe next you'll tell me russian hackers are forcing AMD to design shit products.

>>58030902
Also, someone with an i7 5960X at stock speed managed to get a time of 52 seconds

>>58030910
>taking AMD's word after being deceived every single time over the last few years

you are both idiots
how about wait for real independent reviews to see how it actually performs

>>58030902

Alright - link to some of these runs.

Even mr "fuck AMD" Kyle Bennett over at hardocp got to within margin of error times running at 150 samples.

http://www.hardocp.com/article/2016/12/15/amds_new_ryzen_cpu_smt_ipc

>>58030902
other accounts from 6900K owners have shown it to be even slower than AMD's own results. there are factors other than the CPUs themselves that could influence rendering time. all that's necessary is for the two test systems to be identical.

there's no reason to assume those same people wouldn't also beat AMD's results with the zen chip as well. but you choose to believe so because you're a dumb fanboy cunt.

>>58030938
>you realize vishera is already sandypoo single core?

no, it isn't. poodozer and vishera are only roughly on par with C2D in terms of IPC.

>you realize its multithreaded performance still competes with $1000 intel chips?

no, it doesn't. an 8 'core' vishera chip competes with sandy and ivy bridge i7's in heavily threaded integer workloads, it doesn't come anywhere near the performance of the 6 and 8 core chips in intel's -e lineups.

>>58030947
if that's the guy from /g/ a few threads ago, he later admitted CUDA rendering was enabled

>>58030960
>>58030977
>http://www.hardocp.com/article/2016/12/15/amds_new_ryzen_cpu_smt_ipc
>dual channel only
This is the Jewish tactics we should expect from AMD. All of the tests that BTFO out Zen were using quad-channel memory, not this deliberately gimped bullshit. Why AMD thought they could get away with this is beyond me. Zen is a failure of a chip that can only support dual channel memory. It will always lose to Intel's quad and now hex-channel memory.

>>58030938
>you realize vishera is already sandypoo single core?
Maybe if you overclock it to 5GHZ it could probably keep up with a 3GHZ sandy bridge....

>>58031024
>Zen is a failure of a chip that can only support dual channel memory.
>I'm an idiot and I have no idea what I'm talking about

OK.

>>58031024
Operton Zen will support quad channel memory or more ryzen/summit ridge will not

>>58031024
>toggling back and forth between dual and quad channel memories at cited 2400MHz seems to make no difference at all in this particular test, so if you see someone complaining about that, their point is moot for this discussion.

Nothing...I'm still using my fx and I'm happy with it. Get back to me in a year or 2 when my wife is ready for a new rig.

>>58031104
That's bullshit. People's results show that quad channel memory does impact the time on Blender benchmark. That must have been one big check that AMD sent Kyle to shut his fucking trap up about it.

>>58024475
JUST WAIT 2 MORE YEARS GOY

File: IMG_0065.jpg (58KB, 932x576px) Image search: [Google]

58KB, 932x576px

>Buy 6/12t core Reisen
>overclock the fuck out of it because its only 65w TDP
>mfw I get Skylake maybe beyond Kaby Lake IPC but with 2 more cores for muh rendering

Most of you guys arent realizing what the best deal is

>>58030978
>on par with C2D in terms of IPC.
LOL
>download cpu z
>benchmark tab
>see reference 4.7ghz vishera (these things usually go up to 5ghz btw)
>1376 single threaded perf
>view 2500k single core perf 1317
not to mention the multi threaded perf which beats even 6700ks
then view some benchmarks on 4ghz-4.5ghz 8350s for ingame performance
realize it's a competitor for even todays i5s
realize it still shits on intel multi threaded performance
realize it's half the fucking price
kill yourself after realizing how much money you wasted on the jew.

>>58031113

> People's results show that quad channel memory does impact the time on Blender benchmark.

Then cite some fucking sources. We have one source saying it doesn't matter.

>>58031137
JESUS CHRIST

File: cpu-z bench.png (46KB, 807x403px) Image search: [Google]

46KB, 807x403px

>>58031137

>>58031151
...Have you never used Blender before?

>>58031113
now you bring the validity of your own source in question

fucking hysterical

>>58030835
Yes, but hoping that the competitor would be more than 50% cheaper is absurd

>>58031174
Just show those results

8C/16T or 6C/12T Ryzen for $200 with motherboards at $65 for it to be any competitive.

I5-6500+Asus H110M-K costs $260 USD here
Which is damn good for a budget gaming build.

>>58031197
50% might be pretty steep but they'll need significant discounts if they want to rapidly grow market share

>>58031229
>8C/16T or 6C/12T Ryzen for $200 with motherboards at $65 for it to be any competitive.
DEFINITELY not on launch

but AMD parts actually do usually get cheaper over time.

>>58031197
it's not absurd. Intel will cut prices in response.

then we'll see just how much they've been jewing us lately.

>>58031292
Intel is the kind of company that will sell for loses to flood a market, literally chinese dumping tier....

>>58031305
no, they'd rather dig into their pockets and pay OEMs directly rather than let consumers have it.

>>58031325

Place your bets on how many illegal deals they will place to stop zen gaining a foothold. Then place your bets how much more they have to spend to pay off the EU and uncle sam to look the other way.

>>58031278
By that time Intel would already have another strangle hold on the market.

AMD needs to come to the table thrashing.

If they can't bring i5-6500 performance + motherboard down to less than $200

Then it won't be taking any marketshare.

File: cpu-z_bench.png (20KB, 403x402px) Image search: [Google]

20KB, 403x402px

>>58030938
>you realize its multithreaded performance still competes with $1000 intel chips?
Retarded statement. It performs on par in some workloads (operative word SOME) with an i7 6700k at best, which is not a 1000 dollar chip. It costs somewhere around 350 dollars.

This is what a 1000 dollar chip looks like in CPU-Z benchmark. Oh and the TDP is less than half of the AMD offering.

>>58031408
The 4-core Zen can't even match an i5-6400. It's dead on launch.

>>58031386
honestly I don't think they'll pull that shit again. most likely they'll divert funds back in to x86 and pull something out of their ass.

hopefully zen sells enough while they have their pants down so that AMD can keep up the arms race on the long term.

>>58031450
at least show us a picture of your hairy swamp ass so you can say you supplied a source

>>58024220
8C16T is a bit too high end for my tasks so I'll just stick to the 4C8T variant when it comes out. I'm planning on replacing my i7 950, which will be almost 7 years old by the time this thing comes out, and I expect to pay about $250 for it without a motherboard.

It's probably going to get a tad expensive when I have to replace the motherboard and RAM at the same time. Might as well replace the case with something less shoddily made than the Bitfenix Neos I'm currently using along with switching out the PSU for something modular to reduce the mess.

>>58031450
lol the i5-6400 is a $170 USD chip here
$235 usd including a H110 motherboard

AMD needs to bring what ever is faster to the $200 mark
Or what ever is near that performance to sub $150

>>58031518
Unless that 4C8T is faster than the Skylake i5-6600K at $229 on newegg, which is doubtful, $250 is way too expensive to be competitive, not to mention Kabylake will be in that same price bracket.

>>58031577
>Unless that 4C8T is faster than the Skylake i5-6600K at $229 on newegg, which is doubtful,
At this point I'd call it probable that it'll be significantly faster than 6600k in tasks that can use more than 4 threads effectively. And Kaby Lake is just Skylake with slightly higher clocks and worse thermals.

>>58028922
Can i get a source? I remember hearing two 480's being comparable to a 1080 for cheaper, but not at half the power.

>>58031624
lol no
Zen is clocked way too low for it to even touch a 6600K. Even if they came in comparable stock speeds, the 6600K has the advantage because it doesn't have hyperthreading, which makes each core work more efficiently since they don't have to arbitrarily split and share resources for virtualized cores.

>>58031651

>Zen is clocked way too low for it to even touch a 6600K

3.4ghz and up yo.

>6600K has the advantage because it doesn't have hyperthreading, which makes each core work more efficiently since they don't have to arbitrarily split and share resources for virtualized cores.

You're retarded.

>>58031651
>>>/v/
The only thing we know about Zen's clockrates is that the 8-core part will come at 3.4+ GHz. Parts will lower core count will almost certainly come clocked higher. And you clearly have no idea how SMT to works - in practice, if it's implemented correctly there's no performance penalty at all for single-threaded tasks. And even when it's done badly(like Sandy Bridge), the performance penalty is insignificant.

>>58031624
I think you're giving SMT way too much credit.
40% IPC increase and SMT will not make Zen overtake Skylake IPC performance, i7-6700 is at $315 on newegg and dropping, by the time Kaby comes out i7-6700 should be around or sub $300

the i7-6700 hits 4ghz turbo and the i5-6600 hits 3.9ghz turbo

Does Zen even turbo up to 3.9~4ghz stock? if it doesn't it would have to have better IPC than skylake to beat it, which we know that it does not.

>>58031702
Buttblasted AMDrone detected. Enjoy your housefires.

>>58031725
>if it's implemented correctly there's no performance penalty at all for single-threaded tasks.

>operating system scheduler moves thread to fake core instead of real core leading to worse performance

>>58031727
>intel
>dropping prices
citation needed??!?

$0 because single core is a shit and the benchmarks are embellished.

>>58031735
>>58031754


No you just don't understand how SMT works. SMT is essentially just a smart scheduler and will only give negative scaling in very specific scenarios. Even ignoring that the extra cache a 4c/8t then is likely to have (half the 8c is a good guess) gives it an advantage.

>>58031727
If 8-core part with SMT is performing on par with 6900k, a 4-core part with SMT will certainly beat 6600k soundly in multithreaded performance unless the clocks are significantly lower.

The single-threaded performance is still in question, since we have no idea what AMD's SMT implementation is like. It could well be that AMD is making up for the worse single-threaded performance with superior SMT implementation.

>>58031524
205€ for i5-6400, come andd gibe some capitalism

File: 1479423682450.jpg (29KB, 366x268px) Image search: [Google]

29KB, 366x268px

>>58031702
>yo

>>58031429
>caring about thermal design point
are you some sort of fuggit who can't into cooling?

My vote goes for the missing
>$0 because poorfag
option.

>>58031784

http://www.eecs.harvard.edu/~fedorova/papers/wiosca06.pdf

Not sure if operating system schedulers have been updated since 2006 but doesn't this paper literally prove we're correct

>>58031806

>with superior SMT implementation.

It makes things rather uncomfortable for the intel defence force - either Zen's single core is superior to Intel's or their SMT (which Intel has been working on for best part of a decade) is superior.

AMD has shown two bennchmarks - one integer heavy and one floating point heavy and at this point people are grasping at straws to find legitimate reasons to say why zen isn't competitive. Note I said competitive, not that it is inherently superior.

>>58031806
The i7 6900K is a broadwell chip and has worse IPC than Haswell and Skylake

It's also a 3.2~3.7ghz stock clocked.

A 6600 vanilla will be faster than 6900K at stock for anything 4 threads or less.

Also your 6900K benchmark was clocked at only 3.0ghz on both the zen and the 6900K

Which means they had the 6900K turbo disabled and UNDERCLOCKED.
Which tells you something about Zen's clocking capabilities.

>>58031831
Can you explain what the advantage of using double the electricity and having twice the heat to dissipate for about 60% of performance is?

What is the advantage of doing this?

If it competes with the 6900k soundly I might pay $300-$400, but that is a very big if. How good the new chipset is also really important factor, will we have more than 2channel ddr4? This is a bit of a huge flaw with the current intel chipsets, but their upcoming server chipset is rumored to have 6channel memory so I might just wait for that instead.

I'd pay whatever would be competitive with contemporary Intel products.

>>58031876

> broadwell chip and has worse IPC than Haswell

No, it doesn't.

>Skylake

Not all the time - the fat L4 cache goes a long way for broadwell. Remember: IPC isn't static.

>>58024220
Nothing, just going to stick with my i7 6800k because it's assured to perform well.

>>58031725
My source said that
Zen can't overclock past 3.8ghz

The same source that told me the furyX was gonna have 4gb if hbm1 not 8.

Makes sense because why would amd not run it at 3.7ghz to match that 6900

>>58031925
They didn't even run it at 3.2ghz to match the stock non turbo clock of the 6900K, they clocked them both at 3.0ghz.

>>58031876
>The i7 6900K is a broadwell chip and has worse IPC than Haswell and Skylake
Broadwell has slightly superior IPC compared to Haswell, and slightly worse compared to Skylake, but in practice all three are identical in 99% of workloads.

>Also your 6900K benchmark was clocked at only 3.0ghz on both the zen and the 6900K
Apparently AMD redid the benchmark in yesterday's stream with 3.4GHz Ryzen and 6900k at stock(including the ability to use boost clocks). Ryzen still performed practically identically compared to 6900k.

>>58031941

That was months ago you retard. The event on tuesday had zyzen locked to 3.4ghz and the 6900k operating "out of the box" (i.e turboing as it normally does).

>>58031941
Oh, well that supports what I said even more.

Zen will probably be bad to overclocking

>>58031882
housefire memes

>>58024220
Planning to get the 4c / 8t, or whatever is their cheapest CPU over that with super high boost clocks

>>58031916
Broadwell-E doesn't have any L4$, that was only for the 5765C and 5775C.

File: 1467474391835.gif (2MB, 320x180px) Image search: [Google]

2MB, 320x180px

>Mfw the 300-400 is leading.
From what we can gather, that's if we blindly believe the evidence AMD has presented, the Ryzen is on par with 6900k which costs 1100 fucking freedoms.
That's 1100$ and you guys seriously expect to get the same performance for 300-400$
No way in hell they're selling this for under 650$ and if they are, they're massive idiots.
AMD isn't a charity and they're not out there to bring you a cheap product just for the sake of it.
They're going to make as much money from this as they humanely can, while trying to dominate the market.
650$ would still be a ~50% difference in price for the same performance and even that's an absurdly low price point to sell the product.

>>58032159

Thats because /g/ won't buy anything from AMD unless its better than Intel for half the cost (as in, a coimpletely unsustainable position for AMD). For the most part /g/ actively encourages getting fucked by Intel.

>>58032159
It's about the phrasing of the question, anon. The question isn't "how much is the CPU worth" it's "how much are you willing to pay for it." /g/ is full of poorfags and people who don't really have use for 8-core CPU beyond bragging rights, not to mention that most people already have CPUs suitable for their needs, so anything new would need to offer a significant advantage over what they currently have at a price they're willing to pay, and I don't think you'll find many people willing to spend $500+ on a CPU for shits and giggles.

All of that is before you even take into account that 6900k is overpriced for what it offers and the only reason it's going for $1100 is that intel hasn't had any competition except their own older CPUs in that market segment for years.

>>58032159
it's absolutely guaranteed that AMD will undercut Intel on this

let's be real here, the 8 core Intel chips are overpriced because they probably have shit yields on those and make a buttload less than they make their other SKUs

and to price Zen anywhere near the same is going to be suicide considering it will very likely fall short of skylake in single core perf

You guys really think it's like 300 dollars for that thing?

>>58032159
The 6900k is just a overclocked server chip, same with ever "extreme" processor intel has sold in the past 10 years. They price it as such as well. The top of the line non extreme chips offer almost 90% of the performance at less than half the price. Zen will need to compete with stuff like the 6700k, not the 6900k, they just put that shit in the press releases because that is intel's 'best' consumer chip, even though nobody actually buys that crap.

>>58032434
Their best consumer chip is the i7-6950X

File: 1444286872755.jpg (88KB, 594x395px) Image search: [Google]

88KB, 594x395px

>>58032434

Personally I have a Xeon E3 1231 V3, hopped on to this about an year ago from a Core 2 Quad Q9300 @ 3.0ghz.

I honestly don't see myself upgrading anytime soon, but if I do I'd probably have a budget of about 450~500 max for the CPU and Motherboard combined and it well have to be a hell of a lot faster than my current setup.

Going from a Q9300 stock to an E3 1231 V3 was double to triple the performance.

I don't believe that the 8C/16T Ryzen will be as cheap as people are hyping it to be.

>>58032483
I definitely think it will be less than 500

>>58025957
150-200$ is what the zen will bottom out at if the leaked prices are right.

zen is not their low end shit though, that is their apus which won't be out for a few months after zen.

>>58032598
Zen is the name of the core architecture, not just the high-end chips, anon. It covers everything from low-end APUs to 32-core server chips.

>>58031902
Am4 is only dual channel. This is standard as it affects so little, especially with non enterprise workloads and low core counts.

Umbrella up coming 6channel sockets are coming in above their 2066 4 channel sockets, they'll be priced way above even hedp and I'll be surprised if they're even offered on boards with less than four sockets.

>>58032647
I'm not typing ryzen, I have enough issues when it comes to spelling, i'm not going to sit there and think of how its fucking spelled every time I want to refer to a zen cpu.

>>58032598
>that is their apus which won't be out for a few months after zen
No, their APUs will be more expensive than the four-core Ryzen because of the iGPU. Look at the price difference between the FX-4300 and the A10-6900K.

>>58026902
not retarded, you get a higher quality at a lower bitrate with cpu, if you have crappy uploads like most people do, it matters.

>>58032747

>it matters

NOT IF INTEL IS LOSING IT DOESN'T

>>58032743
I don't believe amd is going to sell only a 4 core apu though.

we got a spread of 4 6 and 8 on the non apu because of presumed bad parts, and im assuming the same 2 3 and 4 core spread will be applied to the apu's too, with special skus that disable the apu part and just sell very low end cpus for damn near nothing.

>>58032702
Hmm that's a shame. Memory bandwidth and latency is pretty much the biggest factor in compiling speed so I wanted something with at least 4ch ddr4.

>>58031628
it was a bad slide they are referring to.

what amd showed was to polaris beating a 1080 in aots and then having it look like it was 50% utilized, really, polaris was 87% utilized.

>>58031292
intel wont lower prices in the same generation, however it would be hilarious if they push what we know as an i7 down to ballpark i3 prices in one gen time and tell us 'we made a huge breakthough'

>>58032789
AMD could release low cost opterons to compete with intels "enthusiast" line

but I really doubt that will happen

>>58033150
Not sure how well this rumor will pan out but supposedly naples will have 8 channel memory

https://www.extremetech.com/extreme/222921-amd-is-supposedly-planning-a-32-core-cpu-with-an-eight-channel-ddr4-interface

>>58024220
Personally I think it should go for about 800$

>>58033189
They wouldn't release the first zen chip to compete with enthusiast Intel range.... would they?

>>58024220
350$ tops. If not i already have z170 mobo that will probably be able to use 6c12t cofee lake cpu. Its price point really. What i really want is the 6c12t model for 200$ or less.

>>58024220
I just hope that amds first cpu will be decent and competitive and not too expensive. They have to increase their visibility again before they can increase the prices like intel does.

>>58033332
Will a 6c12t model even exist?

>>58024568
>posted from his samsung phone with facebook jew app on it

File: 5jl9x5sed02y.jpg (25KB, 480x443px) Image search: [Google]

25KB, 480x443px

>tfw unlocking 6 extra cores on my dual core zen

watching new horizon right now

ayyymmddd is our goy?

>>58033180
Thats not a rumor.
The Zeppelin die has two memory channels.
The Snowy Owl SKUs are two die MCMs with quad channel memory.
Naples is an MCM comprised of four Zeppelin dies, having eight memory channels.
This is what Papermaster was referring to when he promised disruptive memory bandwidth in enterprise. 153.6GB/s and higher. Any highly memory bound workloads will love AMD's new Opterons.

The IO on them is massive all around.

>>58033180
That's a 10 month old article.
4-chip Naples MCMs clearly have the potential for 8-channel DDR4, but AMD's been extremely tight lipped about what the platform sockets/mobos will support, and supporting absolutely all the IO, especially in 2-socket platforms, will absolutely require 4000+ pins.

> 8*DDR4 channels
> 128 bidirectional PCIe lanes.
> fat GMI (HT replacement) interconnect link(s?)
> potentially a bunch of SATA lanes, maybe 10 GbE too
> ...

>>58033370
Why wouldn't it?

>>58033417
"A Pioneer in the chip world"
come on goy pls jim keller
"a great great person"
jim motherfucking keller!!!!!!
"doctor lisa su"
wow... what a fucking faggot chink ceo

>>58033438
>> fat GMI (HT replacement) interconnect link(s?)
I believe it's called Infinity Fabric™ now

>>58033438
Now the question is: is AMD crazy enough to go 4P with this?

>>58033563
I believe its limited to 2p. An Anandtech article from a conference where they first displayed Naples said as much.
2p, 64c/128t, shitloads of RAM.

>>58033370
4 models will launch.
8c16t that is higher binned. Think about i like black edition. Higher base clock, higher turbo, more OC potential.
8c16t base model that will have lower clocks, just shittier version of the other, probably will have minimum oc potential outside turbo frequency.
6c12t like above but with 2 of the core defective - they have some problems with manufacturing and like before will disable defective cores. Bad news - they will be laser disabled, so no unlocking.
4c8t - this is another chip, it is not defective part that got locked. Crystal is not the same. This will cost ~150, will be like 4770, probably will not go above 4-4.2 they will use waffles with imperfections for smaller chips.

>>58033607
what will the 8 and 6 core models cost?

>>58024475

Kek I've hand modded a DDR3 Stick to run GDDR5 memory and its not that much better.

>>58033563
>>58033594
Is anything known about how coherency will be managed this time, given that each sock is gonna have somithing like 8 NUMA islands?

Just lots of snoop tables, or something more?

File: 1446667091224.png (8KB, 196x171px) Image search: [Google]

8KB, 196x171px

>>58033667

>>58033654
No one knows, they will launch in late feb or march, the budget model will come one month later. Pricing was leaked only for the cheapest model, they probably probing now the market.

>>58033674
The interconnect fabric they're using is brand new IP. Not a ton is known about it, but they have said everything on die is coherent. Socket to socket scaling is supposed to be pretty high as well.

>>58033674
Naples is expected to be quad-die dual socket, whereas G34 was dual-die quad socket. Different arrangement, but ultimately the same number of NUMA nodes.

And i doubt they'll treat each 8c die as a pair of nodes, thats just insanity right there.

>>58033702

It's easy if you know how to do shit like it.

>>58033607
>4c8t - this is another chip, it is not defective part that got locked
Nope. It's the same exact piece of silicon as the 6 and 8 cores. And there will be a lot of them since AMD is so bad at quality control.

>>58033856
AMD doesn't even package their chips anymore I think

>>58033888
They're so incompetent that they can't even make their own chips
You can't make this shit up

>>58033856
That guy was entirely talking out of his ass. Pay him no mind.
Also you have literally no idea what "quality control" even means in this regard. AMD has no control over how GloFo yields with a given process. Not that this matters since 14nm LPP is a mature node.

>>58033888
They sold their stake in the company that packed their chips, though this firm is still being used. They never did it themselves.

>>58032159
/g/ is full of /v/ermin overspill who don't have $100 to their name. If people really think AMD will sell a chip that is equal or better to a $1100 one (even if it is grossly overpriced) at 35% cost then their in cloud cuckoo land. They've spent 4 years of investment on this thing. I expect $600-$700 at absolute best, with $500 being a fucking miracle. Intel have also been purposely costing along with the 5% improvements due to lack of demand and competition. If this causes a dent to their profits then they'll release whatever they've been sitting on for years.

1) will zen support smt?
2) ecc?
3) amd-v?
4) vt-d

If yes to all, 2-300e for 8c. 4-500e for 16c.

Depends on pci-e lanes and igpu, too.

>>58033994
> I expect $600-$700 at absolute best
AMD doesn't have the brand strength to sell a consumer CPU with a pricetag that high. Things are not priced what *you* think they're worth. They are priced at the equilibrium between sales volume and margin.

> Intel have also been purposely costing along with the 5% improvements due to lack of demand and competition
No, no they have not. Intel has been increasing R&D expenditures, their fab costs have been increasing, core arch development funding has been steadily rising. Its called diminishing returns.

>If this causes a dent to their profits then they'll release whatever they've been sitting on for years.
Thinking that they have a radically higher IPC design ready to implement just sitting around in a vault is the epitome of the childish gamer nonsense you yourself try to denounce.

>>58033790
>And i doubt they'll treat each 8c die as a pair of nodes, thats just insanity right there.
but they're advertising flat L3 latency, which implies lines striped across 4 slices connected in a crossbar with 4c clusters maintaining redundant copies if necessary.

>>58034037
Yes to the first 3

>vt-d
I think it's called AMD-Vi in AMD land

>>58034037
>Depends on pci-e lanes and igpu, too.
32 and none FYI for the first Summit Ridge chips.
Not sure about the Zen-Vega Raven Ridge APUs coming mid-late 2017 though.

>>58034108
They touted the ability of any core to access any chunk of L3 on the same die with the same *average* latency. This implies a buffer.
They haven't mentioned how this will scale, nor the specifics of their new fabric IP.

32 lanes of PCI-e is probable

>>58034179
>same *average* latency. This implies a buffer.

I took this to mean that access time within a slice was location dependent and that access times differences to slices as a whole were comparatively negligible.

Buffers of some sort are a guarantee given that caches have supported multiple in-flight requests for ages though.

File: HailSatan.jpg (58KB, 500x502px) Image search: [Google]

58KB, 500x502px

I'm willing to drop $500 on the top end Zen, so long as the MB is <$300.

I would do this to support AMD just as much as I have the need for some serious VT.

>>58024524
>High Bandwidth Memory Memory

>>58034591

Excuse him, Nvidia fans aren't used to being at the bleeding edge.

>>58024220

It's too much CPU for my usage. I could envision using 8 cores in a low-power server environment but I would want a version with roughly 2GHz clocks and a third of the TDP or less.

Outside of running VMs on a desktop or video encoding, I don't see the purpose of the horsepower this provides for everyday use other than bragging rights. Still, if AMD isn't being deceptive about the Ryzen's performance, it is an impressive CPU.

>>58034809
I think they're targeting the prosumer/PC gaming streamer market with these chips and the branding if I had to take a guess. I think AMD's first priority is just winning their good name back among CPU enthusiasts and getting something positive associated with their brand name so that their cheaper silicon, which is what they actually live and die on, will have its reputation enhanced.

The ironic thing is AMD just finally released the software suite to enable some pretty good hardware encoding on their GPUs in the ReLive update, though software encoding is still superior for a number of reasons and of course will benefit from beefy CPUs.

>>58034990

> though software encoding is still superior for a number of reasons and of course will benefit from beefy CPUs.

You'll never fit in on anandtech with that attitude.

>>58024220

I scored an RX480 on sale and eagerly will spend $250 for my 8350 to be retired.

So if it launches for much more, I'm waiting

>>58035065
kek, who would try to argue that hardware encoding is better

>>58035154

People who must defend fairest maiden Intel's honour. The "in" thing after the ryzen demo is either 1) Nobody streams at 1080p because its retarded (ignoring the fact top streamers often have a 2nd machine dedicated to encoding or 2) just use your gpu - its waaaaay beeter than cpu encoding because its faster!

/g/ actually has been one of the more civil places to discuss ryzen believe it or not.

of course this is ignoring one of Intel's top dudes (dude with a french name) shit talking on twitter of course.

>>58035154
It is better if you don't have a good enough CPU for software enc.

>>58035249
Yeah I mean you'd have to be a poorfag or a non-streamer to make that argument. A 720p/60fps stream with a hardware encoder like NVENC, Quicksync, or ReLive will look like shit at 3500 kbps on Twitch or Youtube so there really is no argument for it.

>>58035288
It's only better in the sense that they don't have another option for recording. At least it gives them an option for recording gameplay without really hurting their performance which is cool, but it's completely inferior for streaming at this shit upload bandwidth allowed for non-partnered streamers.

>>58031136
fuk ye can encode mah animoo faster

File: wpI2YSD.jpg (38KB, 960x398px) Image search: [Google]

38KB, 960x398px

>>58031735
tfw nvidia is only one who's ever caused house fire

>>58031806
I know what it's like.. Jim Keller did it.
so It's better than intel's implementation.

>>58031965
this
on the new horizon video the fucking AMD chip had turbo boosting disabled and intel chip was stock out of the box with turbo boost of 3.7 and stock of 3.2

>>58024220
$400 is the most I'd spend on any CPU, ever.

>>58036203

All core turbo for the 6900k is 3.5ghz.

File: 1458923848059.png (162KB, 284x281px) Image search: [Google]

162KB, 284x281px

>>58024220
$500
It maintains parity with a $1050 Intel CPU but any more and it would be out of my budget.

>>58036288
damn that's garbage
so zen has at least better ipc than the 6900k all cores go

>>58036320
That or drastically better smt scaling

Do you all have Windows 10?

File: Screenshot 2016-12-16 21.41.08.png (288KB, 1920x1080px) Image search: [Google]

288KB, 1920x1080px

>>58036631
Yes

>>58036631
how else would i play dx12 games??

>>58032747
>Higher quality
>Matters when you stream to twitch/youtube/memebox

Doesn't matter at all. If you think you're going to archive your sick frags for all eternity then think again.

Even if you did care you sure as fuck wouldn't do pc encode off your gaming pc. You'd just get a high end capture card and a second streaming pc.

>Streaming 1080p with poverty internet
>Streaming 60fps with poverty net
Well that just completely negates your streaming quality argument altogether. It's 2016+1 not even friends want to see your pixelated mess at 540x360

>>58037308
You're capped at 3500 kpbs anyway so you need the highest quality possible. For streaming the limiting factor isn't what your PC can handle or what looks best at a massive bitrate, it's what looks best at the shitfuck low bitrate that twitch allows you to use if you aren't shilling their stupid products.

The stream will look like shit at 3500 kbps with any hardware encoded method. Has nothing to do with poverty net.

>>58031651
>>58029390
>>58031051

Has no one on this fucking board learned from Pentium 4?

>>58037421
It looks like shit with software too. I actually tested HW vs software yesterday, and while hardware was horrible software was pretty bad too.

>>58037546
>>58037308
>>58037421
been looking into streaming for a while. Mostly just as a what if people want to watch me kind of deal.

my options are a hardware encoder that is crap, a software encoder that is better but demanding, or spend around 2 grand on a separate computer to render and upload.

>>58037546
What's your setup? 3500 is going to look like shit no matter what, no doubt, but you want it to look as non-shit as possible.

>>58037591
Yeah so you're in that uncomfortable middle ground. Definitely don't need a separate encoding PC if you're willing to invest in just upgrading to X99, especially if you're just getting started. My 5820k doesn't break a sweat in recent Battlefield games which are pretty CPU heavy.

>>58037638
What do you mean setup, like settings? I tried various. The only thing that looked okay was something that took 60% of my CPU. With low CPU usage I got pretty crap results. It's just not enough bitrate for anything.

>>58037679
What's your CPU and yes what encoding settings are you using? The quality options in OBS do make a difference but they are very CPU intensive. It can look passable but I'm not denying that it takes a powerful CPU.

>>58037728
I have a 12 core 24 thread Xeon. It can do fast very fast and faster presets, but going further than that is not great. I don't remember what settings I used that took 60% CPU since it was custom arguments, but the slow(er) presets will probably get you to the exact same spot.

>>58037862
Yeah I've had the best luck at superfast. Not a huge difference but it could just be that you're holding it to a higher standard than I do.

>>58037531
kek
obv not, moar clock = moar guuder

jim keller is going to do it again.

>>58037955
Well, probably. I probably haven't even watched streamers who don't have over 3.5mbits, because to me that's borderline unacceptable even with fast preset.

File: 1481682303614.jpg (29KB, 399x385px) Image search: [Google]

29KB, 399x385px

>>58035065
>mfw juan's multi-page long spergout

>>58038001
>I was only off 14% here
>I was only off 20% there
>30% of the time I'm 50% right
>stop making fun of me
>I-I know thing!

Fucking Brazilians.

>>58024475
When DDR5 goes mainstream expect AM4+

I'm running a 4.3GHz 3770k I'm perfectly happy with it but it is getting a little long in the tooth


I'll honestly give it a look if:

A: motherboards aren't shit
B: at least competitive with the kabylake ipc

>>58038085
That's not unrealistic when you think about how long AM3/AM3+ has been around.

>>58028833
a 2520m is almost equivalent to your shitty processor lel

>>58038193
>intlel

embrace the zen kid

File: Screenshot 2016-12-17 00.48.41.png (279KB, 1920x1080px) Image search: [Google]

279KB, 1920x1080px

>>58030938
>you realize its multithreaded performance still competes with $1000 intel chips?

Zen was already shown with above broadwell performance but if you think vishera manages $1000 intel CPU multicore performance you are delusional.

>>58037987
drop it down to 720 30 or 60

someone else said it too, most people aren't watching full screen, the drop in resolution makes a fuck load of artifacts go away.

Its by no means perfect, but it should get you by

also, if you are recording for editing later, dxtory with lagarith lossless codec is probably the best in terms of space taken up vs quality.

>>58038755
It _kind of_ can. You can get quad socket G34 vishera level opteron system for less then 6900k cpu alone these days.

>>58038755
>>58039439


bulldozer has come a long way but fuck me, they never competed with intel's higher than 4 core offerings, if a program was coded bad enough that intel was held back or was one of the 5% of things that amd did fucking amazing count yourself lucky.

>>58024220
However much I spent on my 4790k is how much I'm willing to spend on any processor, so around $300 IIRC? Never bought anything AMD before. Besides the AMD Sempron 3100+ the family computer had back in like 2001, anyway. I went from that to a P4 to this 4790k. But I like AMD's business practices and support of OSS so I want to support them in return. I'd probably be willing to pay a little extra.