Is threadripper /our cpu/? AMD is finally bringing competition

File: 1.jpg (271KB, 720x540px) Image search: [Google]

271KB, 720x540px

>>60734642
Its definitely industry changing.

i hope to see cloud servers / vps get a shit load cheaper for core count. companies should (possible) be able to offer 2 or 3 times as many cores for the same with ebyc

Infinity Fabric changed the game

if single core performance beats ryzen, that'd be really great

>>60734762
It's just several ryzen dies slapped on interposer, retard.

>>60734762
That's what Zen 2 will be. Ryzen is hilariously unrefined, but most of it's issues should be fixable. If everything goes right, the jump from the R7 1800X to the R7 2800X will be as big as the jump from an i7 880 to an i7 2600K.

>>60734642
Why is everyone pretending they're going to buy the $1800 + $600 mobo halo amd product

>>60734797
>>60734897
The rumor is that TR is going to use the 2:1 IF frequency that Epyc uses, which should boost it's single-threaded performance compared to Ryzen.

File: lisa_su_smug.png (123KB, 240x286px) Image search: [Google]

123KB, 240x286px

>>60734974
>WAAAAH! Stop caring about market segments in which Intel is getting BTFO'd the fuck out!

>>60734984
If most consumers can't afford it might as well not exist

>>60735037

No one bought X99. Don't worry.

>>60735037
>If most consumers

Not him but this is targeting enterprise, not consumers you dumbass

>>60735055
No one bought or is using amd, not even you or the next guy reading this.

>>60735037
>>60735037
>>60735082

You're overwhelmingly underestimating enterprises.

Not surprising for a basement dwelling neet that never experienced working at all.

File: smug_lisa_su_face.png (355KB, 777x656px) Image search: [Google]

355KB, 777x656px

>>60735037
FOR YOU

>amd
Yeahhhhhh, no.

>>60735082
No one buys a katana, yet you have a collection of it.

>Is threadripper /our cpu/?
About 1% of /g/ users will actually purchase it.

>>60735342
Yeah, everyone on /g buys thinkpads, retro ibm keyboard, korean monitors, OnePlus and chink stuffs all the time!

>>60734897
>Q1 2018
>then there's Zen 3

fuck
I should probably wait for them to release stronger mobos then

>>60735342
>technology board
>you cannot care about the techonology unless you buy one

>>60735342
I might, if it's cheap enough.

>>60734642
yep

>>60735431
Huh when's Zen 2 coming out then? I want existing issues ironed out before upgrading.

>>60737887
somewhere around first half of 2018, it will come in a 7nm process and have both architecture improvements and a 50% bump to core count because of 6 core CCX against current 4 core

>>60734642

I love competition. Finally intel hardware will be cheaper.

>>60739022
Should I wait?

>>60739065
the motherboards/sockets will be the same, so no, you can buy stuff right now and upgrade later when you can/feel like doing it

>>60739085
hmm but what about those hardware faults that were reported?

>>60735431
>I should probably wait for them to release stronger mobos then
According to AMD the utilization of very fast ram is not only up to their microcode, but also requires refined circuit design from MOBO partners

>>60739097
as of now there's nothing there that can't be fixed with microcode updates, they even announced recently that there will be support for both high speed ram (3600~4000MHz), and also fixes to VM issues

>>60739097
They may not even exist on the CPU, and if they do, who knows if you'll run into them in a tangible way.

>>60734897
I doubt it will be like the jump from Nehalem to Sandy. I'm expecting more like 10% higher IPC.

>>60739022
GloFo's 7nm DUV (absolute ass for mass production due to 80masks and quad patterning) is Q3 2018 at best. Zen2 on 7nm is Q1 2019 based on GloFo's roadmap stating that 7nm EUV is q1 2019.

>>60739143
well, the 7nm process will both allow for higher efficiency and clockspeeds, but I agree, the jump won't be that big

>>60739154
DUV might be for CPUs, EUV might be for GPUs

Or the opposite, anyhow CPUs usually have smaller die sizes than GPUs so DUV would be a better choice for CPU

>>60739164
EUV is always going to be a better choice due to fucking 7nm DUV requiring quad patterning.

>>60739134
>>60739133
Will AMD release any revisions of existing models?

>>60739172
Not if you have to wait another year for it and you have a roadmap to uphold

>>60739173
Zen+ (Pinnacle Ridge) is Q1 2018.
>>60739185
>another year
It's like 4-6 month between DUV and EUV on GloFo's roadmaps.

>>60739190
Then there would be no point in getting DUV equipment working for GloFo

>>60739173
As of now it's unknown, but they probably won't until (next?) year, seeing how the dies with higher steppings will be kept for HEDT and server CPUs

>>60739205
There is, since risk production for 7nm DUV is Q2 2018.

>>60739221
And EUV mass production is in late 2018? What's the point of DUV?

>>60739133
>as of now there's nothing there that can't be fixed with microcode updates, they even announced recently that there will be support for both high speed ram

It will largely depend on the quality of your motherboard's cirquitry and fixes to the hardware stuff.

So it might be a very good idea to wait for the next generation of the boards that will shit with that latest AGESA outta the box.

Not a disaster if you buy right now though, but you probably wont get your ram way higher than 3200

>>60739238
EUV is q1 2019 at best. 7nm DUV is a halo node for some small dies.

>>60734642
Threadripper is just 2 Zen CPUs stapled together creating a huge hot steamy piece of shit, I don't know what people are pumped for?

File: Screenshot from 23-05-2017 14:52:28.png (43KB, 747x562px) Image search: [Google]

43KB, 747x562px

>>60739253
There's only a quarter difference between the two.

>>60739259
this was explained a number of times, so you're just making low quality bait

tl:dr: fuck off

File: 14961377582910.png (402KB, 617x463px) Image search: [Google]

402KB, 617x463px

>>60739259
Hello Brian.

File: Screenshot from 18-04-2017 01:44:09.png (186KB, 1585x891px) Image search: [Google]

186KB, 1585x891px

>>60739259

>>60739266
Anyway, ask GloFo why are they ever doing 7nm DUV since even Samsung skips straight to EUV for their 7nm node.

>>60739294
Because Glofo is using IBM's process, not Samsungs for 7nm

Also that means GloFo will be fabbing POWER9+/POWER10, which means crazy clocks.

>>60739306
Yes, I know that, but why are they not going straight for the EUV for 7nm is beyond me.

>>60739324
Could be capacity issues, there's no doubt one AMD product will at least be DUV while the others will be EUV, just which one?

>>60739340
Probably laptop APUs. Having node advantage in mobile space is everything.

>>60739340
>>60739361

Could be just some pipe cleaner product. Something like a vega version of a rx550/rx560, Fills out their product portfolio and might be a laptop dgpu etc.

>>60739395
I'd say it's going to be some ULV laptop APUs.

>>60739395
Vega is designed for 14nm, so unlikely.
Vega replacements for Polaris will be late this year anyway.

>>60739404
I wonder how smaller Vega will look like, since Vega's front-end is MUCH beefier.

Cryptocucks be damned I swear I'll murder every cryptocuck i see if they are going to buy Vegas en masse.

How much is the cheapest Threadripper?

>>60739494
We don't know. But it'll be cheap since 80% yields.

>>60739494
the 16 core with 1700 clockspeeds is rumored to go out at $850, that may give you an idea

>>60739536
>$850
Shoah.

>>60739536
>850

Still expensive 4 me

I'll take the SKU above the 1800X

>>60739553
at least they all have 64 pci lanes

File: hedtlol2.png (67KB, 1427x575px) Image search: [Google]

67KB, 1427x575px

>>60735082
lmao.

>>60735074
It's targeting "prosumers", enthusiasts, etc.

Epyc is targeting enterprise.

File: intel ceo JUST.jpg (119KB, 1000x1000px) Image search: [Google]

119KB, 1000x1000px

>>60735253
Go away, Brian.

>>60734642

I rather still buy intel. Their shit just works.

>>60739667
>intel literally became the apple of processors
kek

>>60739667
Hi Brian.

>>60739667
It just explodes.

http://www.legitreviews.com/intel-x99-motherboard-goes-up-in-smoke-for-reasons-unknown_150008

It just bricks itself.

https://www.extremetech.com/computing/244074-intel-atom-c2000-bug-killing-products-multiple-manufacturers

>>60734974
>$1800 + $600 mobo halo amd product

> AMD charging Hewbrew prices

It's like you've entirely missed the point of why Ryzen/Threadripper is a popular topic.

>>60734974
>$1800 + $600 mobo
That'd be Intel's HEDT option.

>>60740095
If threadripper was $100 cheaper. I'd call it affordable.
Can still be if the mobo's aren't stupid high.

Not that I have any use for it right now. But I'm sure there's plenty of people running dual Xeons that are desperate for an upgrade.

>>60740146
the non X 16 core will cost (rumored) $850 anon, this puts the X 16 core at around $950~$1150, this is way cheaper than intel's one, not just $100

>>60740199
I meant for home non profit application, many of our Anon's are on secondhand duel Xeons. Because of price/perf
But yes, it's still stupidly cheap.

>>60740140
I could see TR4 mobos being in the $400-$600 range mostly just out of historical inertial (AM4/TR4 don't even need big chipset, PCIe switches, etc., just more layers for DDR4 and PCIe), but I'd be rather surprised if the top Threadripper exceeds $1200, and I think it actually has a good chance of coming in at or under $1000 a few months after release with discounts, etc.

The $895 intro price for non-X 16c model feels plausible, but I trust it about as much as any rumor this far ahead of release.

>>60740223
fair enough, but it's still a really good long time investment concerning socket compatibility

>>60740228
why the hell people are thinking x399 boards will be so expensive?

File: TB2f4vIvrJmpuFjSZFwXXaE4VXa_!!2341477237.jpg (413KB, 1060x1569px) Image search: [Google]

413KB, 1060x1569px

Remember to not choke from laughter.

File: 1496328708889.png (745KB, 1260x673px) Image search: [Google]

745KB, 1260x673px

>>60740370

>>60740260
Same reason why you nee a k series cpu to overclock. The Jews find a way.

>>60740370
Oh look, Mr. Delay displaying the reason why it's so fashionably late.

>>60740435
>jews
>amd

>>60740488
I meant fuckary behind the scenes. Intel bent assrock? over with the overclock bypass. So it shows power to make them do what will.

File: donald-trump-shrug-1024x551[1].jpg (101KB, 1024x551px) Image search: [Google]

101KB, 1024x551px

>>60734642
Already bought a 7700k. I don't need that many cores, I'll wait for Zen 2 and see if they boosted their single core performance. I doubt it though, Jim Keller left AMD.

>>60740577
>single core performance

You don't belong on /g/. Go back to >>>/v/.

>>60740260
It's got a fuckload of features and 64 PCIe lanes. Of course it's going to be expensive. More expensive than X299? Maybe. We'll see.

>>60740577
Their IPC will definitely improve with Zen 2. The lead architect of Zen isn't even Keller. He was just the project manager.

>>60740611
PCIe lanes are on the CPU, not on the motherboard, they'll probably be the same price, AMD has a bigger socket with more pins, but Intel's X299 chipset costs $50 alone while AMD is using the cheappo x370 one since their CPU has all the I/O the mobo needs.

>>60740488
you forget that mobos are all made by various flavors of chinks, aka yellow jews.

if Threadripper >>>> i9 but priced better, mobo makers can charge higher by a degree less than the CPU price to extract more of the profit for themselves.

>>60734642
VMware HCL EPYC when?

>>60740577
Zen was designed by Mike Clark you fucking retard. Keller was managing K12 and Zen team.

>>60739613
>ryzen, much cheaper, hot

>>60739281
So Intel HC CPU are simply 2 low core cramped connected, only unlike Ryzen it scales like shit?
EBYN :DDD

File: AMD-EPYC-CPU-740x416.jpg (26KB, 740x416px) Image search: [Google]

26KB, 740x416px

>>60741490
Zen with more than 8 cores are on seperate dies, which still scales way better than ringbus crap.

>>60741490
Yes. Bingbus was never designed to scale anyway. See their original bingbus topology.

>>60734642
>Is threadripper /our cpu/?

Threadripper 12c/24t will be /mycpu/ in a month or two, followed by an upgrade to the full 24c in a few year for "7" nm Zen2/3/++/whatever.

>>60734975
>The rumor is that TR is going to use the 2:1 IF frequency that Epyc uses

is this actually being touted on real tech "journalism" sites or is it just forum speculation?

The ~1.2-1.6 GHz, ~40-50 GB/s for Ryzen IF is clearly too low for 4 channel DDR4 systems to do proper interleaving, and there should really be headroom to run the crossbars at 2.4-3+ GHz, but AMD might keep the frequency down just to avoid housefire scenarios.

>>60741744
Even fucking debug boards for Zen has 2:1 crossbar ratio, anon.

>>60741744
and in various websites they were saying TR would have "full speed" IF at 100GB/s

>>60734642
Yes it is, i'll be moving from x99 to thread ripper

File: vr.jpg (216KB, 1620x599px) Image search: [Google]

216KB, 1620x599px

>single thread performance still 2 years behind Intel
>everyone excited about more cores that they won't use

File: this kills the SL-X.png (3MB, 1651x1475px) Image search: [Google]

3MB, 1651x1475px

>>60741744
>>60741762
>>60741781
Multi-die Zen absolutely requires faster than 1:2 clocked Infinity Fabric.

Each crossbar port on Ryzen runs at exactly 2*DDR4 channel bandwidth (1/2 nominal DDR clock * 32B = DDR clock * 64b/clock/channel * 2 channels), so even 4ch Threadripper will be fucked without more internal bandwidth.

>>60734642
Yup.

File: muhsinglecore.png (453KB, 1920x1080px) Image search: [Google]

453KB, 1920x1080px

>>60741927
Kill yourself, tripfag faggot.

>>60741927
And intel's answer to that is releasing a cpu with more cores that don't scale well, full-on bingbus mode, with lower frequencies than those available on threadripper
And they don't even have their precious ipc lead over amd anymore

>>60742146
real gaymers stick to their 4c/4t cpu for maximum single thread awesomeness

>>60742251
i5's are trash for that now, i7-7700K for max FPS or 1600X for max value.

>>60742321
joking aside, would it make sense to wait for kaby lake x to come out and get a i7 7700k for about 150-200$? Surely it's the best poorman's solution?

>>60742340
Wait and see what Coffee Lake S looks like. If the new i5 is a 4 core 8 thread it will become the new gaymur CPU.

>>60742340
unreliable, you don't know you'll get anything cheap

>>60742350
they gonna be expensive as hell like the current kaby lake lineup and have the usual 5-10% speed improvement over their predecessor.

Only reason to wait is for the price to come down even more

>>60742391
If i7-7700K still has value as a gaming CPU you won't see prices drop on that either.

>>60742428
b-but i wanna be one of the cool kids D:

File: 1486856944571.gif (1MB, 235x177px) Image search: [Google]

1MB, 235x177px

>>60740397

>1.7 volts

>>60734642
How big will the coolers have to be for threadripper? I assume because of it's physical size something like a H100i won't work

File: DA-Iz0kUAAADyE7.jpg (66KB, 1024x576px) Image search: [Google]

66KB, 1024x576px

>>60742596
They are pretty fucking big. But that's a good thing, because it leads to better heat dissipation.

>>60742649
that's bloody massive. Will TR get AIO support or will it be stuck with either air or custom water cooling?

>>60742715
Don't know. Presumably some AIO coolers will be made, it just needs a bigger contact plate.

>>60742596
>>60742649
>>60742715
Apprently most coolers that fit socket 2011/2066 will also work fine on TR4, because the dies are both right in the centre of the package. Coolers with larger base plates might provide slightly better temperatures though.

File: amd-zen-core-scaling.jpg (33KB, 597x430px) Image search: [Google]

33KB, 597x430px

>>60739259
Cry yourself to sleep shekelchaser, because 1.99 scaling Threadripper is coming to dumpster literally every single socket Xeon.

File: 1485115684236.gif (913KB, 460x351px) Image search: [Google]

913KB, 460x351px

I hope Vega makes nVidia squirm as hard as TR is doing to Intel

>>60743089
>ebin Pajeet of nvidia rushing Fermi to respond to R700/R800
Probably.

>>60743089
It won't, since Vega isn't scaleable, unlike TR.
You should wait™ for NaVi.
Plus, NVidia has been doing at least something instead of sitting on their asses like Intel.
P.S. Even if Nvidia will shit themselves and release Fermi 2.0 as an answer to AMD architecture, they'll still hold vast majority of market share

When are OEMS going to sell workstations based on Ryzen?
It's hard to convince enterprise to build their own desktops.

>>60743216
Wait for Epyc(tm).

>>60743216
Intel workstations are 99% Xeons, so if there are OEM workstations they will be EPYC or threadripper, since both support ECC and more than 2 channel memory.

>ryzen makes corelets cry
>kikeripper makes overcompensating Intel dead on arrival
>ryzen 3 soon to stomp I3 and bentium
Holy fuck, it's happening. And just needs to up their game a little on uber cards and cheapo entry level ones, and Intel/nvidia jewness is done for.

>>60743299
Nvidia is not half as bad as Intel.
They're only scummy for their proprietary and vendor lock in practices but they make good GPUs

>>60743147
true. nVida has been making actual progress with their R&D, but releasing two titans for one generation is bullshit. It is nice that they're not turbojews like Intel though
>>60743216
If they do it would probably be low end workstations. What AMD is essentially telling us now is that Ryzen is for consumers and Threadripper/EPYC is for HEDT/workstations

>>60739143
even if they only achieve 5% over current IPC and can get the clocks of zen+ up to 4.5-4.6Ghz on OC It will still probably be a good chip, plus I know if I buy an AM4 or LGA 4096 motherboard it'll be compatable with Zen2/+ because AMD isn't shit

>>60739239

There's already an AGESA update supporting 4Ghz if you're okay with a beta BIOS

>>60743299
you forget about AMD Finewine™. In a few more months ryzen will BTFO everything jewtel has to offer in every category

>>60743425
>poor optimization is now fine wine

Can Threadripper compete with the i9 outside of the possible pricing advantage?

>>60743495
It's two 1800X, you do the math.

>>60743506
So it will inherit the low single core performance?

>>60743495
>More lanes
>More Ghz
>More RAM
>Lower TDP
>Lower power draw
>Theoretically higher IPC
>Will be released this year rather than Q2 2018 like skymemeX is delayed to

>>60743529
Broadwell is now low single core performance?

>>60743319
They do make good hardware. But the whole "optimised for Nvidia, buy nvidia, goiym" shit is scummy as fuck. Vulkan and dx12 can't come in mass soon enough.
Also, g-sync is not free. Wut

>>60743495
It will absolutely fuck Kabylake-X up, as in demolish it at almost every level, as intel has decided to only allow 14 PCI-E lanes on their X299 platform. They went completely fucking retarded.

Skylake-X will perform just slightly better but won't be capable of the clockspeeds the consumer version has because there are 10-18 cores on one die so all those transistors packed together not only make it impossible to reach skylake speeds, but also increase cost to intel by an insane amount...


Intel will be barely making money on these even if they're twice as expensive as the AMD counterpart, and IPC wise they will only be 1-5% faster, while they are completely and utterly gimped when it comes to PCI-E lanes and features. Intel lost big on the HEDT front this time around.

>>60743552
Maybe he thinks he'll overclock these 10 cores to 5.0 or something? Wouldn't put it pass idiots.

>>60743583

Forgot to mention, X299 boards will have features disabled from purchase unless you buy a USB key to unlock lets say ECC and extra PCI-E lanes so they are trying to close down to the platform so that it doesn't interfere with their Xeon platform. Basically intel completely screwed up, Even Linus thought what they were doing is pants-on-head monkey-retarded

Sorry for being an idiot.

Threadripper runs on an own socket right? Did AMD announced how long it will get used?

>>60743638
https://www.youtube.com/watch?v=TWFzWRoVNnE

>>60741329
>goalpost moving simulator

>>60743640
Considering its size, a good 3-4 years.

>>60743216
If it's a HIGH end workstation, I think it will use single socket Epyc for that sweet octa channel RAM and 64 MB of cache

>>60743640
The socket will probably be around as long as AM4 is being used

>>60739266
So TSMC literally confirms nvidia volta cards are coming in early 2018.

Seriously thinking about getting on the Threadripper train.

I hope AMD can also significantly improve the single core performance over Ryzen 7.

>>60743719
>early 2018
Can mean January or May

>>60740228
maybe highend x399 will be 600, but cheaper boards should be around 250 at launch and around 220 after some time.

>>60743719

They won't be that good, someone calculated the performance of the MASSIVE 815mm die size GV100 and found that it's only 9% better than GP100. That means that Volta GPU's will have much higher memory bandwidth thanks to GDDR6 but will only be slightly faster than current pascal. If you need a fast videocard you should just buy what you can now.

File: LRM_EXPORT_20170411_145818.jpg (1MB, 1293x1730px) Image search: [Google]

1MB, 1293x1730px

>>60741927
Kill yourself you perma-BTFO'd tripfaggot

>>60735342
Only 1% of /g/ has an actual need/reason for such a CPU.

>>60743766
it have like +5TF?
How the fuck does that translate into +9%?
It the same fucking retarded speculation ppl did with Pascal, claiming it wont actually improve performance and how wrong they were...

>>60743719
No, that's just when mass production starts, I'm still calling Q2 2018 for Volta and it's still going to be 10% faster than Pascal.

>>60743723

Zen+.

Even more IPC, more instruction sets.

>>60743810


It's 9% because GP100 was 600m and GV100 is 815mm. GPU's don't work the way you think they do. If we get a Volta 2080 Ti it will be almost 600mm for (BEST CASE SCENARIO) 25% of the current 1080 Ti. Except in 4K (because of the increased memory bandwidth where you'll get 30%)

It won't be like the jump from Maxwell to Pascal where the 1070 was as powerful as the 980 Ti. The 1080 Ti will probably match the 2080.

>>60743810
He means it has like 9% better perf/mm^2. Volta is oversized hot poo of Huang trying to move away from consumer GPUs while forgetting that GPGPUs are irrelevant to 80% of datacenter market.

>>60743851
>>60743845
always new goalpost metrics?
All that matters is final product performance, nobody gives a shit if threardryper is 4glued cpus, nobody will give a shit in nvidia card is massive if it bring in better performance, not to mention nvidia cards already much more efficient then amd..

>>60743886
The bigger the die, the higher the price assuming Huang will keep his margins, anon. Volta is oversized poo. They desperately need another Maxwell.

>>60743886
You're not selling a 800mm^2 die to the consumer market you twat, unless you want gamers to pay $12000 for them.

>>60743900
>>60743912
Ok i get it you are amd fan, what you think then 2060 gtx performance and price will be?

>>60743928
Assuming it'll have 10-15% bigger die - 10-20% with 10% higher price.

>>60743928
Somewhere around the 1070

>>60743944
Ok i get it, you are a fucking retard

>>60735342
I might pick one up and then move my ryzen 1700x over to the micro server I'm building. Just need a mini-itx board and a sata card.

>>60743809
And I'm part of the 99% and will stay buy one.

>>60743960
t. Jen Hsun Huang

I can feel NVIDIAs desperation flying in the air at every keynote. Literally ML memes every fucking time.

>>60744041
Nvidia is fine, Intel is the one shitting themselves.

File: file.png (708KB, 640x640px) Image search: [Google]

708KB, 640x640px

>>60735037
>AMD is bad because nobody can afford HEDT platforms anyways
>AMD is bad because Nvidia's highest end halo products are faster
Really makes me think

>>60744073
NVIDIA desperately wants to get into datacenters space. Literally every fucking keynote is ML memes and no one cares about ML memes besides big dudes who can simply develop ASICs. What was the fucking point of Computex keynote? At least Intel was honest enough to cancel IDF.

>>60743340
>nvidia aren't turbojews

>Hey goy, buy our "founder edition" card, only 100 shekels more! It comes with a shitty blower cooler!
>Make sure you create an account so we can track when you update your drivers!
k

>>60744236
Nothing compared to jews who want $200 per 18mm^2 of silicon

>>60744245
Blue Jews had no competition. Green Jews just abuse mindshare for the past 8 years.

>>60744236
nVidia is definitely a bunch of jews, but not turbojews Intel. Intel is in a league of its own

>>60744103
I thought NVIDA wants into the automotive sector.

>>60744500
She wants literally anywhere away from extremely volatile gaymen gGPU market.

>>60744500
They want the AI meme sector.

>>60734689
CCX killed Ryzen 5 though.
I hoped for low-mid end changes in the market, but Intel continues to reign there with Pentium/i3 and i5. Sadly.

>>60744245
>>60744275
>>60744339
There are reasons both companies are jews, but I legitimately think nvidia is worse.

Intel's sins recently are:
-Having the better product / no competition, allowing raise their prices to take advantage of the market.
-Stupid feature-gating which basically all big vendors in a market do to try and extract more money.

Nvida's recent sins are:
-Make an account or you get no drivers.
-Making their products run worse through driver updates (buy our new cards goy!)
-Gameworks in general, along with honest to god anti-competitive behavior like adding extra things under the ground to fuck with AMD's tessellation
-Charging more for a "founders edition". They're charging more for the same thing. They're literally managing to jew the market against their board partners. "Ey, babe, I'm gonna release this early and get some crazy margins. You can have what's left".
-Gsync vs. Free-sync. They are trying to get royalties out of a new technological improvement that has a good open standard and fuck AMD over in the process.

Intels products, ignoring price, until Zen were better and they weren't actively trying to fuck AMD. AMD just wasn't as good. IMO actively working to fuck over AMD like Nvidia has been doing is way worse than "we're legitimately the best, lets charge $$$$$".

>>60744604
>Intel wasn't fucking with AMD
They've yet to pay 1.2b fine.

>>60744604
>Intel's sins recently are:
You forgot
>Blackmailing/bribing OEMs into not using AMD products

>>60744603
todays high end is tomorrows low end.

>>60734642
Is that a woman? Dropped if it is.

>>60744603
Uh, what? There's no reason to buy i3's due to poorfag pentium, and no reason to buy i5's because they CPU bottleneck way too easily and 1600X is just flat out a better CPU at that price point. i3 and i5 are both dead already.

>>60744603
>CCX killed Ryzen 5
Nigga what? Half of that lineout is amazing with the 1600. If you want something that competes with lower end, then wait for the APUs. Those will use only one CCX.

>>60744644
Yeah, even the poorfag pentium will meet it's end when the APUs come out. And R3's would kill i3's if they weren't already dead, killed by Intel themselves.

>>60744627
But that's the dumb part, like you know how progression works with CPUs. Like... this will represent a very small portion of sales, and a very small segment.

Intel could have thrown up their hands and said "Wow we had this all along with Xeon", I mean you walk into Dell and say you need a serious machine for serious work, you get a Xeon. And it's got 2 CPUs, a fuckton of threads, and can handle anything.

>>60744642
>>60744644
Even the cheapest Ryzen is like 150$ more than the i3-7100 for example.
If I go up to the Ryzen price range, I can buy i5s are better in single core performance.

I agree, the top-tier Ryzen chips are great for multithreaded load. But I am a poorfag.

>>60744626
I thought this stopped awhile ago? After the court case in the P4/K9 era? I specifically said "recently" because of this... I guess you could argue they would've but just didn't have to.

>>60744621
Oh?

>>60744678
btw Pentium costs the same as an i3, so... yeah, why would you buy the Pentium? I know they are very similar, but if they cost the same...

>>60744681
>recently
Jew shit under the IHS is worse than any of nVidia software memes.

>>60744678
>Even the cheapest Ryzen is like 150$ more than the i3-7100 for example.
Like I said, the Ryzen 3 series aren't out yet. They will be priced as low as $100-$120, and be true 4 core CPUs. Not 2 cores w/ HT.

>>60735476
What do you personally consider as "cheap" when considering a cpu? Serious question, because it's a different answer for most people.

>>60739133
Latency issues can't be completely fixed, but workarounds can happen.

>>60739259
No one is really "pumped", but amd marketing needs to fabricate the hype for reasons.

I have no use for Threadripper but I'm gonna buy one anyway and just showcase it like a trophy cause T H I C C

>>60744845
this
>$5000 PC with all the bells and whistles
>only use it to shitpost on a Tibetian basket weaving board and watch animu
It's just planning ahead for when japan finally starts making 1080p anime.

>>60744843
There's fucking tons of bad press for Intel right now. "Don't overclock", "Jizz TIM", "Fragmented platform", "Hardware dongles" .. AMD could just quietly release Threadripper and still be ahead.

>>60744880
Oh and I forgot, "14-18 cores delayed" lol

>tfw ryzen 1500x and rx560 until new mid-range radeon is released
You'll never take me alive, injews /jewdia

>>60744604
>Nvida's recent sins are:
>-Make an account or you get no drivers.
You can easily download newest drivers from nvidia website
>-Making their products run worse through driver updates (buy our new cards goy!)
They dont, if you believe they do provide the benchmarks
>-Gameworks in general, along with honest to god anti-competitive behavior like adding extra things under the ground to fuck with AMD's tessellation
That was Crysis dev's, not nvidia fault but i somewhat agree gameworks run terrible on weaker hardware
>-Charging more for a "founders edition". They're charging more for the same thing. They're literally managing to jew the market against their board partners. "Ey, babe, I'm gonna release this early and get some crazy margins. You can have what's left".
FE is literally early adopters tax, dont see problem with that
>-Gsync vs. Free-sync. They are trying to get royalties out of a new technological improvement that has a good open standard and fuck AMD over in the process.
Gsync works differntly than Free-sync, it need extra hardware installed on monitor, also nvidia is careful enough not to allow anyone to use gsync. Gsync monitors are quality controlled unlike Free-sync, which is basically Free-for-all fielday

>>60744914
>goysync works differetly
It's the same.

>>60744914
He forgot on-by-default telemetry.

>>60744880
>There's fucking tons of bad press for Intel right now. "Don't overclock", "Jizz TIM", "Fragmented platform", "Hardware dongles" ..

And your point?

>AMD could just quietly release Threadripper and still be ahead.

AMD would be ahead if their products were actually constantly better than intels, but AMD never lives up to the hype they fabricate on image boards.

>>60744914
About gsync. It does the same thing as freesync, so nvidia shat the bed and their cost expenses are their problem. Quality control is such a fucking jewy fairytale justification. Nobody said that freesync monitors don't work because muh lack of quality control. And they do work

>>60744989
You gonna say that rx and ryzen didn't deliver?

>>60744989
Intel's product line has been reduced to two CPUs, and one of those is going to get creamed by APUs later this year.

TR 16 core is two 1800X on one CPU. With near linear scaling. It's going to beat Intel's 18 core, for half the price, guaranteed.

>>60745018
They do work, but when some shitty 150$ monitor comes with Free-sync you just know it a piece of shit

File: 1492489342710.jpg (119KB, 1065x1300px) Image search: [Google]

119KB, 1065x1300px

>>60744914
fuck off I'm forced to buy a gsync monitor that will lock me in the Nvidea ecosystem for years

>not joos

>>60745072
you are not forced to buy shit

>>60745041
>You gonna say that rx and ryzen didn't deliver?

I skipped Rx, but ryzen wasn't as advertised when released. I ended up returning 2 to the retailer myself.

>>60745046
>implying that's amd fault that chink monitor is shit
>implying that 150$ monitor with g-sync won't be shit
>the quality for no less than 500 $ goiym meme

>>60745125
there are g-sync monitors for 380euros in my country, fullhd 144hz.

>>60745113
Why? It was cheaper then Intel competition, good core utilisation potential, overall great performance and oc.
Rx 460-580 are great cards

>>60745151
or even dell 165hz 1440p g-sync for 470

>>60745042
>Intel's product line has been reduced to two CPUs, and one of those is going to get creamed by APUs later this year.

This is what redtwam tries to force people to believe. This time I'll actually wait until after the magical apus release before believing any claims.


>TR 16 core is two 1800X on one CPU. With near linear scaling. It's going to beat Intel's 18 core, for half the price, guaranteed.

Again I'll wait until aftee release to see where exactly the CPUs stands.

>>60745151
That's my point. 150$ monitor is shit regardless of Tech in it. And decent things cost a bit more, maybe not 500$ but you get my point.

>>60745113
>but ryzen wasn't as advertised when released
And how exactly, was it advertised?

I seem to recall AMD comparing it to the 6900k in a professional environment and seeing Ryzen come out on top 80% of the time, and even when it didn't it was within spitting distance.

Or do you mean to say that it wasn't how /g/ said it would be?

>>60745042
There's no way they can beat Intel with even cores or less.

File: help.png (6KB, 350x314px) Image search: [Google]

6KB, 350x314px

>>60745203
>3.5 vs 2.7

>>60745190
It does come pretty close or beat the 6900K in most workloads. http://imgur.com/a/wMm1C

I think most of the butthurt comes from when AMD showed some GAEMS, especially streaming while gaming, and people assumed it would demolish Intel in gaming as well. It's still better at streaming while gaming on the same PC, but not gaming at 1080p 144hz while not GPU bottlenecked.

>>60745190

Only games matter. It is irrelevant that AMD pitched ryzen 7 against the 6900k is their massive event earlier this year and all other events after that.

>>60745203

Agner disagrees.

>>60745203
see
>>60740370
Take the 1800X cinebench score and multiply by 2. That's likely where it will end up.

>>60745228
>Only games matter.
>>>>>>>>>>>/v/

>>60745164
>Why? It was cheaper then Intel competition,

It wasn't really cheaper though, and they had irritating issues up until the last update (can't confirm myself, but some issues including mine were supposed to be fixed).

> good core utilisation potential,

I didn't witness any of the good parts when I owned them.

>overall great performance and oc.

It had good multicore performance, but OC wasn't good.

>Rx 460-580 are great cards

Maybe, but I wasn't interested in them myself at the time. (Already had gpus).

>>60745228
I almost hope you're wrong. I run a 1600x. It they signifigantly step up single core performance from where it is now, so soon, it's kind of going to suck for me as an early adopter.

>>60745219
Yeh, I'm not sure people realise the advantage to having, simply, more cores.
While streaming/gaming performance may be comparable to the 7700k, the difference lies in that the 1800x can do twice as much before it starts to struggle, before it starts to stutter, to hang the system up.
Hell, 60% of streamers on Twitch run a separate PC just to offload the stress of stream encoding just so their systems don't hang up on them and effect the quality of the stream as a whole.

>>60745291
They aren't abandoning AM4 or anything. Just upgrade CPU and sell your old one when Zen 2 comes out.

>>60745250

Your sarcasm detector is broken.

>>60745291

Ryzen already can handle more instructions per clock than kabylake can - it is just a matter of the software catching up to the hardware.

>>60745311
Poe's law.

File: Overclocked 1700 boinc.png (219KB, 1541x889px) Image search: [Google]

219KB, 1541x889px

>>60745301

I'd like to see a 7700k keep up with this (note there are only 12 threads running - I don't use boinc for the remaining 4 so my typical usage is not impacted).

>>60745354
>b-but that's not a vidya gaem ..
>s-stop cherrypicking .. ;_;

File: 6ee.jpg (55KB, 600x600px) Image search: [Google]

55KB, 600x600px

>>60744989

>>60745190
>And how exactly, was it advertised?

It was advertised as a competitor to the 6900k, but it ended being more of a competitor to the 6800k a cpu that I had no use for.

>I seem to recall AMD comparing it to the 6900k

This is true, but there were many ommisions to the comparisons.

>in a professional environment and seeing Ryzen come out on top 80% of the time,

Sometimes that 20% that they ommit carries 100% of the deciding factor when choosing a best cpu from a mainstream.

>and even when it didn't it was within spitting distance.

If ryzen had 40+ PCI lanes, and more memory capacity then they would have been the best choice at the time.

Or do you mean to say that it wasn't how /g/ said it would be?

File: 1481744513695.png (349KB, 716x2588px) Image search: [Google]

349KB, 716x2588px

>>60745113
>6900k performance for less than half the price
>Not delivering
Pick one and only one.

File: hedtlol.png (67KB, 1585x559px) Image search: [Google]

67KB, 1585x559px

>>60745415
>It was advertised as a competitor to the 6900k, but it ended being more of a competitor to the 6800k a cpu that I had no use for.
What. Did you not see this right here? Multiple sites running multiple tests against the 1800X and 6900K. http://imgur.com/a/wMm1C

Yeah the 6900K was on a HEDT platform, but it was also twice as expensive, and nobody fucking bought it.

>>60739536
Insane but the important thing is how far it can overclock and if it's possible to disable some cores and push it past 4ghz for gayming

>>60745219
1080p is poo tier anyway

>>60745439
24 PCI lanes = Didn't deliver.

Remember everyone's use case is different, and ryzen didn't meet my needs. Therefore I had to choose more fitting products.

>>60745495
That's cool.

>>60745450
See
>>60745495

Different use cases poojet.

>>60745495
AMD never advertised AM4 platform as having more PCIe lanes, though. So your initial point about AMD Ryzen being "not as advertised" is bunk.

>>60745518

Watch this space - in a few weeks 44 pcie lanes will be enough for anybody and 64 a waste of time and/or overkill.

>>60745291

I can see a 1650X and an 1850X being released before 2600-2800 etc.

I would imagine that the only real improvement would be small tweaks to allow it to get to 4.4Ghz boost.

Honestly, Ryzen 5/7 looks really good but there's this tiny bit of me that says It's a side-grade because I'd be losing 10% clockspeed over my 4790k even though IPC is essentially identical.

I do feel massively bottlenecked by my 4c/8t CPU in some workloads though.

>>60742649
Can't wait for noctua to make a cooler with three 140mm fans and 3 towers.

>>60745518
AMD advertised it as being a cheaper alternative to the 6900k, but it was really just a competitor to intel mainstream CPUs.

To understand that you would need to understand why some people buy a 6900k over a 6800k, or 7700k though.

>>60745550
It's already happening, along with "Jizz TIM is superior to solder" and "Intel 18 cores will clock to 4.8 GHz".
>>60745554
>10% clockspeed
Literally nothing.

>>60745591
Keep grasping at those straws. You'll grab one of them someday.

>>60745550
More PCI lanes the better imo.

>>60745603
No straws grasped though.

File: bd97de060f[1].png (14KB, 392x391px) Image search: [Google]

14KB, 392x391px

>>60745594


Keep in mind that my 4790k is overclocked so single-core performance is more like 15%, and that single-core performance is enough to push some emulators into playable territory. I'm not saying the 1800X isn't better for multithreaded workloads, it absolutely destroys any mainstream i7 in that regard. My issue is, is it really worth it to upgrade? That's why I'm looking at Threadripper. If there's a 10-16 core model with 4Ghz clockspeed I'll buy it. Especially because the rumors say 16 cores start at 850$.

CAN I PLAY ARMA WITH THREADRIPPER?

>>60745724
Please don't use CPU-Z, it is really a crap benchmark. Worse than pissmark, even. http://www.guru3d.com/news-story/new-cpu-z-upgrade-lowers-ryzen-performance.html

>>60745724
Where your 4790K really stands up against an 1800X is more like 90% faster multicore and 10% slower single core. Even after accounting for overclocking.

>>60745752
They lowered the performance because AMD "Samsung-ed" that benchmark.

>Please don't use CPU-Z, it is really a crap benchmark.

I remember when these benches where once amongts pojeets favorite to post.

>>60745806
>AMD built its hardware around CPU-z
You must be mentally retarded.

>>60745591
Uh, you are aware that Threadripper is going to bring 64 PCIe lanes and that the 10-core version at 3.6/4.0ghz will still be significantly cheaper than the 6900k, yes?

>>60743780
Nice

>>60741927
Faget

Will Adobe ever update their shitty software package to make more than 8 cores somehow useful?

>>60746105
No.

>>60746105
No, that's the advantages of having a monopoly. That's like asking MS to port Office to Leenux.

>>60746128
How is not using more cores a result of monopoly?

Especially if Intel is now also pushing more cores.

>>60746137
Because they can allow themselves to do fuck all to multithread their software suite since people will use it anyway.

>>60739143
Kaby lake is only about 10-15% faster IPC.... a lot of that can be attributed to the nearly 1ghz clock advantage too. (7700k vs 1700x)

>>60746186
SKL is 7% better IPC for Intel-optimised code.

>>60734897
>been waiting for literal years to build a new rig
>hey AMD apparently released the best CPU ever, time to jump on
>this post

Back to waiting, I guess.

>>60746186
Kaby Lake and Ryzen are within 10% of eachother in IPC. IPC isn't really clockspeed, though. Intel's 4 cores just clock higher. Anything more than 4 cores and that advantage evaporates pretty quickly, though.

>>60746211
Not really, because they're not switching the sockets every two seconds like Intel likes to do.

>>60743780
lmao

>>60744603
Hahahaha what?

The 1600 basically made the i5 series obsolete. It's the best CPU out for that price, even for games.

>>60746271
Yeah fuck the socket bullshit. The office had a pile of AM3 machines we built dirt-cheap, the Athlon X4s did precisely what we wanted at the time. The athlons were showing their age at the advent of the Bulldozer, so we looked into just swapping the CPUs, but no.

My speculation on an improved 14nm Ryzen would be sticking in a full 256-bit AVX-muncher instead of the current 128-bit. For the 7nm chip I'd just stick in more L2 and L3 caches to compensate for recycling the socket and any RAM speed limitations.

>>60746271
Even so, replacing a $300 CPU after a year is awful. I guess I could try to sell it a month or two before the next generation releases.

>>60745823
Damage control post.

Let's forget what AMD tried to do, and lets give them a free pass after getting caught.

>>60743299
Do we have any info or rumours about Ryzen 3 yet? I'm interested in the R3-1100, all I need is a cheap Zen quad core to run TF2 at 144hz, which my i7-920 currently fails to do even at 4ghz.

>>60746494
Confirmed, mentally retarded.

>>60746541
Q3 2017. $100-$120. That's all we got.

>>60734642
SIXTY PERCENT MORE I/O I AM FUCKING SOLD
INTEL KEKS GTFOOOOOOOOOOOOOOOOO
why do you need all this power?

>>60741564
omg its beautiful
thanks for posting that image

>>60746583
>server
Dunno.

>>60744685
Think percentage change instead of linear, also new pentiums are i3s, just lower clock speeds

>>60735206
> DELID_THIS.jpg

For you.

>>60746558
At the end of the day we'll both be left wondering why AMD IPC in benchmarks doesn't reflect real world usage, but deep down we'll both know why without even having to say anything.

Threadripper? More like Threadnigger.

I will remove myself.

>>60746721
Yeah, because deep down inside we know you're a massive idiot without a ounce of hardware knowledge to argue about this subject.
And calling you out on it is too pitiful.

>>60746541
Ryzen III with integrated GPU would have been neat.

>>60734642
Yes. Built a Ryzen 5 system for a friend, am totally on board with AMD now.

I won't be buying until Zen+ comes out though. Already have a 6700k that does what I need it to.

File: 1492292585330.jpg (56KB, 500x500px) Image search: [Google]

56KB, 500x500px

>>60746801
enjoy that 10-15% lower maximum fps in some gayms then IDIOT

my 120fps>your 108fps TOPKEK

>>60746886
The human eyes can only see 24fps anyway

>>60746801
Some shitty friend you are, I'd never build a Ryshit system for anyone, they would never want to hear from me again and they'd think i'm tech retarded

>>60734642
intel can't even compete with performance per watt which is a big deal to the market.

Doesnt AMD have an ARM CPU that is in a removable socket on an ATX style motherboard?

How is that going? Id love to build an ARM desktop into a regular ATX case without having to do custom shit

>>60747403
You talking about this? I don't know if it's available for consumers. Seems like server tech. https://www.extremetech.com/extreme/221282-amds-first-arm-based-processor-the-opteron-a1100-is-finally-here

>>60746784
Ryzen APU for less than 200 would have been perfect.

>>60747425
yep thats the one

>>60746402
>after a year
Is there any indication of Ryzen 2 being here after an year?

>>60739667
good goy

I DON'T CARE
RELEASE VEGA YOU FUCKS
AND MAKE IT GOOD

>>60747938
AMD says early 2018. http://www.trustedreviews.com/news/amd-zen-2-specs-technology-release-date-price

>>60734689
>Its definitely industry changing
Its definitely industry changing what?

>>60749938
Core. And the way they scale their designs. It's literally LEGO CPUs.

>>60749960
Its definitely industry changing core what?

File: The_Best_I_Can_Do_2_i9_Boogaloo.png (2MB, 1920x1200px) Image search: [Google]

2MB, 1920x1200px

will a pinch magnetic field fusion core be enough ro power my i9 when it comes in? i need to plan out my HEDT build.

>>60750043
The throughput and perf/watt is absolutely bonkers, anon. And I'm not even talking about IF.

>>60750070
I think you're gonna need at least two for that, anon.

>>60740397
Fake and gay
>1.7v
>Clocked at 4.5GHz but barely faster than 6950X@3GHz

>>60750230
Its throttling.

>>60750086
You're not gonna give up, are you?

File: dynomak-31.jpg (661KB, 1417x1031px) Image search: [Google]

661KB, 1417x1031px

>>60750217
nigga deuterium aint cheap

>>60750324
Can't help it if you're a poorfag, anon. Don't you want the best possible computing experience known to mankind?

>>60750324
Forever 20 years away.

>>60735055
Funny because I did. I bought two 5820k CPUs.

>>60751832
Funny because you are actually retarded bought those two

>>60751832
>>60751855
Hey, the 5820k was the best buy for a while. Every dumb shill and shitposter was parroting i7 quad cores, but the 5820k had 50% more cores and could OC very well. It's a shame the 6800k turned out shit moreover now when it's all moot due to the 1600.

File: DEEP WEB 1.jpg (159KB, 721x481px) Image search: [Google]

159KB, 721x481px

where can I learn more about CPUs and how they work

>>60751973
They are serving me well. I run them on two esxi servers and were worth the money if the last me a few years