Quantcast
[ 3 / biz / cgl / ck / diy / fa / g / ic / jp / lit / sci / tg / vr / vt ] [ index / top / reports / report a bug ] [ 4plebs / archived.moe / rbt ]

Due to resource constraints, /g/ and /tg/ will no longer be archived or available. Other archivers continue to archive these boards.Become a Patron!

/g/ - Technology


View post   

[ Toggle deleted replies ]
File: 1.15 MB, 534x752, Ayy.png [View same] [iqdb] [saucenao] [google] [report]
62478560 No.62478560 [Reply] [Original] [archived.moe] [rbt]

http://wccftech.com/amd-ryzen-mobile-apu-benchmarks-leaked-90-faster-bristol-ridge

>~90% increase in overall performance than AMD's last generation of APUs
>This literally kills and incinerates Inturd's Jewish attempts at kikening out a piece of integrated graphics that will ever be able to compete with AMD's offerings

JUST.

>> No.62478641

>>62478560
The niggerfucking OEMs will still gimp it with single channel memory.
I hope AMD is not skimping on L2$ for DSBR.

>> No.62478709

Come te fuck on AMD

Just release this bad boy so I can pick up a new laptop ASAP, can't stand this piece of shit anhmore

>> No.62478755

>>62478709
They can't; the software for consumer Vega is not ready.
They've recently pushed DSBR into Radeonsi specifically tuned for RR.
It's coming Q4 2017.

>> No.62478768

>>62478641
It was already confirmed that no gimping will be allowed and will be directly supervised by AMD, considering the long-lasting big Zen contracts made with Dell, HP, OriginPC, iBuyPower, Alienware, Lenovo, OLDI, and Maingear. All of these make notebooks, not just custom pre-builts, and AMD sits heavily on their asses right now as to not allow for any of such shit to happen. The QA on AMD's side is insane, even when it comes down to watching over third parties.

>> No.62478774

>wccf

>> No.62478812

>>62478774
This is not a meme, /g/oy, they've became quite accurate lately. And from time to time they also simply just repost what's already on other outlets, like in this case.

>> No.62478856
File: 387 KB, 1000x1000, JK.png [View same] [iqdb] [saucenao] [google] [report]
62478856

Say his name.
Say it and say it LOUD.
You KNOW you WANT TO.

>> No.62478862

>>62478768
HOLY FUCK AT LAST.
A275/A475 refresh can't come soon enough.

>> No.62479033

Will motherboards come out with many sata ports? I want to build a nas with this.

>> No.62479062

>poobench

>> No.62479094

>>62478862
Bristol Ridge was FX garbage anyway, and all I always personally cared for in integrated graphics is only the ability to run something like PCSX2/Dolphin or at the very least ePSXe & PPSSPP while everything maxed out with stable 60 FPS and not turning my notebook into a fucking holocaust oven within ~10 minutes of play. No integrated graphics is currently able to fully accomplish this on all fields, not even 5775C's Iris Pro which up to now was considered the most powerful Intel's integrated video out there. Zen's APUs will be the very first ones which won't be only very cool and power efficient, but will also perform greatly enough as to allow 100% comfort /g/oyming in 1920x1080 or even 2560x1440 resolution. I cannot fucking wait to try out the most top gen1 Zen's APU offering by playing either Witcher 2 maxed out with and without Super Sampling while riding a train, OR by running modern /g/oyming Emulators with MAXIMUM GRAPHEN on it. I have almost no doubts that top gen1 Zen APU will be able to run all of this with relative ease while staying cool and efficient. Even if it'll just be a 1920x1080 notebook at 15" and 60Hz with IPS, this will be more than enough for me personally. AMD APU's are definitely "shit's so cash" right now.

>> No.62479111
File: 353 KB, 1534x1966, IMG_4197.jpg [View same] [iqdb] [saucenao] [google] [report]
62479111

>>62478560
> pic related is a 3 year old chip
> lowest end Crystal well
> Iris pro graphics 5200 (128MB L4)

>> No.62479120

will it be a botnet tho

>> No.62479132

>>62479111
>HQ
We're talking 15watts here.

>> No.62479141

>>62479094
Autism too hard there buddy

>> No.62479144

>>62479033
What's wrong with pre-built NAS from third parties?
Not all of them use Infail's garbage Atom's and other inept POS in them, you know? Even right now you can find decent third party NAS with AMD processors in them, and when first Zen APUs start coming out it'll be just a matter of months before first third party NAS solutions based on Zen will get on sale.

>> No.62479173

>geekbench
>ignoring that there are 2500U results with 25% more performance than that

>> No.62479197

>>62479111
Look at the frequency of that Zen APU, you dumb down syndrome-induced retard. It's an engineering sample, you stupid faggot. This shit will be at 2.8~3GHz by the time it goes gold and you can imagine what scores it'll get when that happens. Learn to fucking count, Brian.

>> No.62479218

>>62479173
It's not even it's final form, though. Just an IS.

>> No.62479233

I wonder when will AMD get bored enough of no competition to integrate HBM2 into the package.

>> No.62479244
File: 77 KB, 1265x638, 1505464018375.jpg [View same] [iqdb] [saucenao] [google] [report]
62479244

Who cares, the CPU will be close to Intel, but the GPU will demolish Intel.

This is Vega 8, not Vega 10 even.

>> No.62479269

>>62479244
I wonder about the size of L2$ on these little boys.

>> No.62479282
File: 104 KB, 500x354, Intel AMD Hybrid.jpg [View same] [iqdb] [saucenao] [google] [report]
62479282

>>62479233
>I wonder when will AMD integrate HBM2 into the package
Zen 2. They're already finishing it up by the end of this year and it's slated for release somewhere next summer or autumn. Absolute holocaust.

>> No.62479320

>only 4MB of L3
But why.
Though AMD uses 512KB of L2 (and it's 8-way) so the hit ratio must be decent.

>> No.62479449

>>62479233
Never. Low-cost HBM is more likely, but maybe not even that because of the size of the package will be disruptive for laptop motherboard form factors. Laptop motherboards are very dense so having a large package will piss people off.

>> No.62479471

>>62479449
>Never.
Oh i wouldn't say that.
>the size of the package will be disruptive for laptop motherboard form factors
And yet no one complains about HUGE slab of eDRAM that is Crystalwell.

>> No.62479518

>>62479244
>the integrated video of gen1 Zen's APU will demolish Intel
That was the point to begin with, though. Just like OP-post says - the "~90%" performance increase is overall, meaning on ALL fields, not just computation or just video, but BOTH. I presume with a pretty big portion of confidence that using top version of gen1 Zen's APU will be something like putting a full-sized i7 2600K (after all, 8 core RyZen without SMT is LITERALLY "two 2600Ks" by sheer PPC, and gen 1 Zen's APUs will have only 4 computing cores at the start) in your notebook + new instruction sets + less power consumption and lower thermals, and then adding a GTX 1050 Ti/RX 560-level card over that to the boot. And all of this just from the CPU alone. This is the most realistic course of action in my personal opinion. But I actually think it might turn out to be even better than that.

>> No.62479526

>>62479471
>And yet no one complains about HUGE slab of eDRAM that is Crystalwell.
Because Intel still subsidizes their chips and development of products. AMD doesn't do that and can't afford to do that, so they don't have the power to tell OEMs to shut their filthy peasant mouths and know their place.

>> No.62479537

>>62478560
This is tested with a single memory channel.
Check the AES and Memory bw scores.

>> No.62479538

>>62479320
Engineering sample. A literal "proof of concept" stone.

>> No.62479545

>>62479526
>they don't have the power to tell OEMs to shut their filthy peasant mouths and know their place.
They did exactly that with EBYN.
OEMs were eager to comply.

>> No.62479560

>>62479537
Sandbagging? Nice.

>> No.62479564

>>62479538
Or the R5 is only 4MB and R7 is 8MB to differentiate a bit more between them.

>> No.62479592

>>62479564
Both are still 512KB [email protected] Zen lives and dies on bretty large, low latency and fast L2. It's quite the opposite of faildozer (L2 killed it).

>> No.62479598

>>62479526
>they don't have the power to tell OEMs to shut their filthy peasant mouths and know their place
Except that this is EXACTLY what they're doing right now by heavily supervising OEMs for not gimping out pre-built systems and notebooks based on Zen APUs via single-channeling. Long-term contracts, bitches.

>> No.62479616

>>62479592
L3 is one thing, but cutting L2 would be too destructive to the performance.
I don't think it's even possible to cut L2, it's not made into subslices like L3 is

>> No.62479649

>>62479598
Ye, long-term partnership with chinknovo is that.

>> No.62479743

>>62479592
>L2 killed it faildozer
Wrong.
What Keller'd FX is FPUALU-trash architecture and artificially inflated corn pseudo-Hz. FX was doomed from the very start simply because by sheer IPC and PPC there was a formula in place which literally translated as "3 FX cores = 1 Sandy core", and this was due to FPUALU approach. They advertised FX APUs as 4 core solutions while IN REALITY there were only 2 actual cores, same with desktop - FX's pseudo-8 cores were actually 4 in reality and pseudo-6 core stones were only having 3. And now compare Sandy's (for example) 4 TRUE ACTUAL cores to FX's 4 "pseudo-8" cores...this battle was lost even before it began, because FX's IPC and PPC were absolute garbage shit and just 1 (ONE) Intel's Sandy core LITERALLY performed like 3 FX cores.

>> No.62479768

>>62479743
No you fucking retard.
4-wide decode per 2 sets of ALUs and horrendous cache latencies killed Bulldozer.
Stop pretending you know anything.

>> No.62479799

>>62479649
Not just Lenovo. There are a shittion of big OEM/third party players contracted with AMD on Zen right now, see here >>62478768.

>> No.62479861

>>62479768
>Only 4 out of """""(((8 supposed)))""""" cores were working at one time, with other turned off
>FX could never turn all cores on simultaneously, only working on a "flickering" methodology
>WAAAAAH, U LIE! THERE WERE 8 REAL COARZ I SAID!!!!

Amusing.

>> No.62479889

>>62479861
>Only 4 out of """""(((8 supposed)))""""" cores were working at one time, with other turned off
>FX could never turn all cores on simultaneously, only working on a "flickering" methodology
Bulldozer is not a barrel processor.
You've just described Sun's Niagara.
Please stop shitting in this thread you tech illiterate moron.

>> No.62479947
File: 57 KB, 1000x667, 1486872762378.jpg [View same] [iqdb] [saucenao] [google] [report]
62479947

>>62479120
If you want some delicious RF ID chips alongside with your CREAMY EQUINE MAYONNAISE under the lid - AMD is not the fast-food joint you should go to.

>> No.62480005

>>62479889
Faildozer was EXACTLY working like that, you DUMB fuck. Entire FX family works this way. There were only 4 real cores at one time, because ONLY 4 cores were working at any one given time, while others disabled themselves automatically. This shit was so bad it was literal "monitor panel PWM"-equivalent of CPUs. FX flickered out of it's goddamn ass ALL the time, because it was BUILT that way.

>> No.62480023

>>62480005
No you fucking retard it's not a barrel processor.
Stop shitting this thread you tech illiterate moron.

>> No.62480051

>>62480023
Alright, fine, I admit your defeat. Go and educate yourself, you reality-denying inexperienced imbecile.

>> No.62480065

>>62480051
>abloo abloo Bulldozer is a barrel processor!
Please stop.

>> No.62480102

>>62478641
fucking this, ishygddt
just make a small machine with dual channel, i want a new laptop

INTEL BANKRUPT
INTEL FAGS ON SUICIDE WATCH

>> No.62480157

>>62480102
>just make a small machine with dual channel, i want a new laptop
SOON, just wait

>> No.62480161

>>62478856
THANK YOU BASED JIM

>> No.62480179

APU'S are CUTE!

>> No.62480186

>>62480005
>fx flickering
Dude stop.
You're being too technical to understand you.

>> No.62480191

Well not that Xeon-SP and KBL-R are dead it AMD has CFL-S to deal with.
I think they'll a round of price cuts the day Coffin Lake launches.

>> No.62480204

>>62478560
>taking wccftech seriously
LUL

>> No.62480224

i want to succ lisa su
>>62478856
THANK YOU BASED JIM

>> No.62480238

>>62480179
A-CUTE

>> No.62480242
File: 85 KB, 835x703, doitfgt.jpg [View same] [iqdb] [saucenao] [google] [report]
62480242

>>62479111
>it's the stupid faggot who thinks Intel's $400+ spaceheater mobile CPUs are the same shit they put in $500 OEM craptops

>> No.62480263
File: 944 KB, 1228x1502, 5kl46879udy67ojykdjhgdflyujo96uo7.png [View same] [iqdb] [saucenao] [google] [report]
62480263

>>62480191

>> No.62480275

https://www.servethehome.com/dual-amd-epyc-7251-linux-benchmarks-least-expensive-2p-epyc/
Even the cheapest of EBYNs is doing fine.

>> No.62480282

>>62479144
What are these third party nas with AMD processors? Really intrested here. I don't like giving my money to those kikes at this point arm is better.

>> No.62480300

>>62480263
And you know what's the worst part of all the shitfest?
Intel damage controlled so fucking hard instead of staying fucking silent.
>>62480282
There's QNAP ones with full fat Ryzen 7's.

>> No.62480318

>>62479545

To be fair EPYC BTFOs Xeons in 80% of use cases. They see the dollar signs. While Raven Ridge is very good, so far it doesn't look so much better that AMD can slap people in the face with their dick.

>> No.62480338

>>62480318
>To be fair EPYC BTFOs Xeons in 80% of use cases
So does RR.
iGPU alone makes it ages ahead of anything Intel.

>> No.62480368
File: 155 KB, 804x978, Truth.jpg [View same] [iqdb] [saucenao] [google] [report]
62480368

>>62480191

>> No.62480421

>>62480368
>over FIFTY
>SKUs
>from THREE dies
HOW

>> No.62480422
File: 71 KB, 1183x820, Screenshot_17-09-2017__185636.png [View same] [iqdb] [saucenao] [google] [report]
62480422

https://browser.geekbench.com/v4/cpu/3852138
https://browser.geekbench.com/v4/cpu/3988647

>> No.62480423

>>62480282
http://www.anandtech.com/show/11523/qnap-launches-ts1277-nas-with-amd-ryzen-cpus, for example. But they're not the only ones. There's also Synology, D-Link, LaCie, Thecus.

>> No.62480429

Meanwhile Ice Lake will have 8c/16t a year from now, and you know what that means.

>> No.62480444

>>62480422
Just to show that the R7 2700U is alive, though a INVALID

https://browser.geekbench.com/v4/cpu/3770816

>> No.62480449

>>62480421
>From THREE dies
I have very bad news for you - http://i.imgur.com/K92QwvT.gif

>> No.62480456

>>62480449
No-no, i mean Xeon-SP.
It's THREE DIES.
AND OVER FIFTY FUCKING SKUs.
THEY ARE SEGMENTING EVERYTHING EVEN MY FUCKING ANUS.
>>62480444
>that multicore score
?

>> No.62480457
File: 140 KB, 350x263, TambourineAttacks.png [View same] [iqdb] [saucenao] [google] [report]
62480457

>>62480422
BASED
AMD

>> No.62480458

>>62478856
JIM KELLER
JEW KILLER

>> No.62480477

>>62480456
>>that multicore score
>?
Bugged obviously, it's an ES

>> No.62480479

>>62480456
>?
ENGINEERING SAMPLE™
JUST WAIT™ WHEN IT GOES GOLD

>> No.62480521

>>62480422
Why is it so much faster on x86 android?

>> No.62480524

>>62478641
They would be stupid to cut their products performance in half when Raven Ridge will actually be good.

Still have no idea why they had single channel APU products before though.

>> No.62480525
File: 192 KB, 512x512, 4nJ16bfQ.png [View same] [iqdb] [saucenao] [google] [report]
62480525

>>62480456
>No-no, I mean Xeon
>He doesn't know that Inturd's "desktop" and "XXTREME ENTHUSIASSD" processors are LITERALLY cut Xeons, LITERALLY same shit just BACKWARDS

>> No.62480542

>>62480525
No, i mean actual Scalable Family Xeons.
OVER FIFTY
50
SKUs
FROM THREE DIES
THE SEGMENTATION DOESNT MAKE ANY SENSE.
>>62480524
>Still have no idea why they had single channel APU products before though.
*smiles in shekels*

>> No.62480558

>>62480542
Why do you write like that?

>> No.62480567

>>62480558
Nothing stops me from doing so.

>> No.62480591
File: 308 KB, 800x445, The Know.png [View same] [iqdb] [saucenao] [google] [report]
62480591

>>62480282
>>62480423
>>62480449
>>62480525
Oy vey

>> No.62480647
File: 321 KB, 681x438, 1504904148876.png [View same] [iqdb] [saucenao] [google] [report]
62480647

>>62480429
>8 core Intel mainstream
>you know what that means
Of course I do, Brian. We all definitely do:

>> No.62480804

>>62480429
INTEL MOBILE HOUSEFIRES

>> No.62480964

I want my GPUless ITX box and I WANT IT NOW

>> No.62481192

>>62480804
>Mobile 8 cores
>Intel
Pick one.

>> No.62481205

>>62481192
45w SKUs.

>> No.62481227
File: 971 KB, 484x682, Inturd.png [View same] [iqdb] [saucenao] [google] [report]
62481227

>>62481205
>Meanwhile at AMD
>~15W

>> No.62481238

>>62481227
>ha ha we have a node advantage
And then 7LP comes in and tears Intel a new assholes.

>> No.62481273

>>62481238
I have no idea how 7nm LP will do for low power since it's an IBM node.

>> No.62481322

>>62481273
It's >60% lower power as ISO frequency or >40% more performance as ISO power.
It's as good as nodes can get.

>> No.62481323
File: 4.00 MB, 2588x1618, IACH.png [View same] [iqdb] [saucenao] [google] [report]
62481323

>>62481238

>> No.62481343

>>62481322
That means 60% lower power at 3.0GHz? (no zen2/+ improvements)

>> No.62481351

>>62481322
>It's as good as nodes can get
Except that they're already working on going 5nm and 4nm. If it'd had too extremely diminishing returns or was unfeasible, everything would've stopped at 7nm, but no they didn't.

>> No.62481384

>>62481343
Exactly, the stats are compared to 14LPP.
And it's >60%.
Probably ~70% lower power for non-EUV 7LP.
>>62481351
I mean as good as single node shrink can get.
Denser than Intel's 10, with better performance, perf/watt and yields (7LP has lower mask count compared to Intel's 10) to boot.

>> No.62481390

>>62481322
Except that's typically at target frequencies, which is 5GHz or something for the 7 leading performance node. Scaling down usually doesn't turn negative but it can certainly be less than 1:1, which is already mediocre considering scaling down on Zen becomes >1:1 after a certain point

>> No.62481404

>>62481390
At ISO frequency, anon.
It's >60% lower power at 3ghz, which is 14LPP's sweet spot.

>> No.62481426

>>62481404
But they plan to seriously put Zen 2 at 5GHz as a "mainstream" frequency.

>> No.62481445

>>62481426
You do understand that mobile offerings clock lower?
Intel's 14+ can achieve nuclear clockspeeds yet mobile chips don't do that.

>> No.62481485

>>62481426
Mobile is not technically mainstream. Mainstream is a non-enthusiast desktop.

>> No.62481487

>>62481404
Yea except you don't get the same scaling at every frequency. ISO frequency is usually at some best case frequency for scaling and not an average across the board.

>> No.62481510

>>62481487
>ISO frequency
I mean the measurements they take at an ISO frequency are usually taken at some best case frequency.

>> No.62481516

>>62481487
ISO freq is ISO freq.
You are saving 60% power at ISO freq or getting 40% more perf at the same power.

>> No.62481522

Is Glofo 7nm EUV an improvement in perf/power or just density/cost?

>> No.62481534

>>62481487
But Zen's yields are extremely good, so binning is of super-high quality. Literally the only thing that restricts gen1 Zen's Overclocking right now is 14LPP-based techprocess which they took from smartphones. There's nothing wrong with silicon quality.

>> No.62481537

>>62481522
http://www.anandtech.com/show/11558/globalfoundries-details-7-nm-plans-three-generations-700-mm-hvm-in-2018
First stage EUV is yields and costs, second stage is performance and density.

>> No.62481577

>>62481522
Naturally both, because 7nm is much smaller than 14nm (more CPUs on same platter - lower costs) and there's also massive gains PPC-wise when same frequency as gen1 Zen.

>> No.62481640

>>62481577
I meant compared to DUV

>>62481537

>> No.62481650

>>62481577
>massive gains PPC-wise when same frequency as gen1 Zen
That depends on uarch improvements.
>platter
It's called a wafer.

>> No.62481722

>>62481650
uArch improves naturally alongside with the shrinkage, otherwise making things smaller wouldn't have had any reasonable incentive.

>> No.62481748

>>62481722
>otherwise making things smaller wouldn't have had any reasonable incentive
Except performance, cost and power.
Fucking nothing i swear!

>> No.62481929
File: 112 KB, 691x771, 1499963016439.jpg [View same] [iqdb] [saucenao] [google] [report]
62481929

>> No.62481961

>>62478856
JIM THE PUSSY DESTROYER KELLER

THANK YOU

>> No.62481971

>>62481961
>THANK YOU
/v/ pls go.

>> No.62482015
File: 141 KB, 750x709, 60272054_p0.jpg [View same] [iqdb] [saucenao] [google] [report]
62482015

JIM THE "I BOUGHT MORE THAN 28000$ WORTH IN AMD STOCK WHEN IT WAS JUST 2$ PER SHARE, THANK YOU ETERNALLY FOR NOW I AM RICH" KILLER

>> No.62482024

Hi guys! I like Intel processors ! Well, that's all I have to say so have a good day!

>> No.62482039
File: 971 KB, 484x682, JewFail.png [View same] [iqdb] [saucenao] [google] [report]
62482039

>>62482024

>> No.62482054

what matters is, will lenovo's next gen laptops have injew or gay-M-D processor?

>> No.62482062

>>62482054
https://twitter.com/LisaSu/status/905828511520313345
Have a wild guess.

>> No.62482076

>>62482054
>Trying this hard

>> No.62482092
File: 120 KB, 1920x1080, tluafedserxam.jpg [View same] [iqdb] [saucenao] [google] [report]
62482092

>>62482062

>> No.62482110

>>62482092
And these are Bristol shitboxes.
Chinknovo will totally refresh them with Ryzen Mobile Pro somewhere Q1 2018 for ultimate holohoax.

>> No.62482112
File: 422 KB, 672x794, 052.png [View same] [iqdb] [saucenao] [google] [report]
62482112

>>62482062
I FUCKING HATE FUCKING AMD SO GODDAMN FUCKING MUCH!

>> No.62482129

>>62481971
>/v/ pls go.
/b/ pls go.

>> No.62482133

>>62482112
We all do, Joao.
We all do.

>> No.62482140

>>62482129
>/b/ pls go.
/a/ pls go.

>> No.62482141

>>62482110
What "Bristol shitboxes"? You've posted a tweet with RyZen Pro announced in Lenovo notebooks. There's nothing about no Bristol Ridge.

>> No.62482150

>>62482140
>/a/ pls go.
/x/ pls go.

>> No.62482158

>>62482133
>Joao
But Brazilians are pro-AMD as fuck. SOPA DE MACAQUE, UMA DELICIA.

>> No.62482159

>>62482141
It's AMD PRO (i.e. it's Bristol for now) powered ThinkPads.
Ryzen Mobile is yet to get to it's Pro stage. which is Q1 2018 according to FAD roadmap.

>> No.62482186

Did i forget to say that 2 chink hyperscalers are adopting EBYN?
I guess fucking chinks over Early Purley was a bad, bad idea, Intel.

>> No.62482189

I just want Apple to switch to AMD fully. Thankfully Intel has already relinquished its grasp on Thunderbolt and soon AMD can incorporate it into their products. Then there will be zero reason for anyone to use any Intel product again.

>> No.62482194

>>62482159
How is AMD Pro not RyZen Pro? And how is RyZen Pro - Bristol Ridge? Are you saying there's two different AMD Pro segments?

>> No.62482210

>>62482194
Ryzen Pro =/= Ryzen Mobile Pro.
AMD PRO includes Bristol Ridge.

>> No.62482219

>>62482189
>Then there will be zero reason for anyone to use any Intel product again.
Except XBAWKS HUEG volume discounts.
You see, Intel already lost EVERYONE relevant, AMD grabbed OEMs by the balls.
They will do everything to at least keep Apple on tight leash.

>> No.62482224
File: 59 KB, 519x644, 1488686979361.jpg [View same] [iqdb] [saucenao] [google] [report]
62482224

>>62482189
>MFW even FireWire is better than Thunderbolt

>> No.62482232

>>62482224
TB has its uses.
Still relatively niche.

>> No.62482259
File: 64 KB, 600x800, AD.jpg [View same] [iqdb] [saucenao] [google] [report]
62482259

>>62482210

>> No.62482263

>>62482219
>Intel
>discounts

Excuse me? Did you not see the rushed platforms and CPUs they released when they realized they were losing to AMD a few months ago? They were worse, priced higher and included cute little tricks like fucking RAID keys. This is how Intel acts when they're losing. They won't discount shit to anyone, much less Apple.

>> No.62482284
File: 215 KB, 1920x1080, [pls] Hibike Euphonium S2 - 11 (BD 1920x1080 x264 10bit AAC)_0001_00:03:56_[2017_07_29_13].jpg [View same] [iqdb] [saucenao] [google] [report]
62482284

>>62478560
>gaymen
>laptop
>mfw you can pick two now

>> No.62482297

>>62482263
>I don't know how volume discounts work
Whoa there, friend.
No one big buys mobile chips or Xeons at recommended prices.
>>62482259
Don't worry, chinknovo will gladly refresh them with Ryzen Mobile Pro.

>> No.62482321

>>62482284
Dumb kumikoposter.

>> No.62482330
File: 91 KB, 660x708, 1497563503011.jpg [View same] [iqdb] [saucenao] [google] [report]
62482330

>>62482232
>Thunderbolt has it's uses
If you're sissy gay and a coprofag...oh wait
>I just want Apple to
I see...

>> No.62482349

>>62482189
>wanting Thunderbolt
that shit need to die already, it is an actual housefire (just like most intel products)
https://www.youtube.com/watch?v=KV_WfuIlAmo

>> No.62482359

>>62478560
>performs less than a dual core i5
kek

>> No.62482370

>>62482189
Apple already grabbed AMD by the balls. The best polaris chips go to macbook pros.

>> No.62482387
File: 2.64 MB, 400x225, 1502011199203.gif [View same] [iqdb] [saucenao] [google] [report]
62482387

>>62482349
>MFW Thunderbolt's name at the beginning was "Lightning" (the LITERAL epitome for "overvolting that burns down trees and causes housefires")

>> No.62482392
File: 120 KB, 480x521, 1504275173842.png [View same] [iqdb] [saucenao] [google] [report]
62482392

>>62482359
15W chips BTFO by 95W chip. WOW AMD IS FINESHED.

>> No.62482404

>>62482370
It's actually somewhat backwards. No one really like to work with nVidia.
>The best polaris chips go to macbook pros.
And the best Vega10's go into iMac Pro's.
Full [email protected] at 220w?
You'll get one.

>> No.62482436
File: 15 KB, 438x403, Intbecile Inside.gif [View same] [iqdb] [saucenao] [google] [report]
62482436

>>62482359
>dual core
>i5
This is the state of Inlel and Inlel's shills in 2017, folks.

>> No.62482457

>>62480368
>someone posted my image

A warm feeling I have.

>> No.62482472
File: 49 KB, 565x486, 1498759795881.jpg [View same] [iqdb] [saucenao] [google] [report]
62482472

>>62482392
>AMD IS FINEWINE
>AMD IS FINESHIT

>> No.62482508

>>62482472
UMA?

>> No.62482564

>>62480458
that acronym, wtf is up with this timeline?

>> No.62482567
File: 827 KB, 854x480, not just trees.png [View same] [iqdb] [saucenao] [google] [report]
62482567

>>62482387
it certainly did a number on that pcb
imagine having that churning away on top of your bed sheets

>> No.62482581

>>62482297
>Don't worry, chinknovo will gladly refresh them with Ryzen Mobile Pro
Well, it's not like I'll be getting a Zen APU-based notebook from Lenovo in particular, though. Because I don't want no bloatware and SUPERFISH™ in my mobile system. I'll highly likely will be buying either an Asus or a Dell's Zen APU-based notebook for myself. I'm not against ThinkPads, I just don't trust Lenovo after the SUPERFISH™ fiasco, neither on the notebooks nor on smartphones.

>> No.62482589

>>62482349
Apple's fault obviously

>> No.62482602

>>62482581
>Asus
Mama's boys.
>Dell
Literally sold their asses to AMD.
So you'll totally get your RR-powered XPS of choice.

>> No.62482624

>>62482589
yeah, also those bricked rangleys are Cisco's fault

>> No.62482642
File: 275 KB, 600x335, 14120615701910.png [View same] [iqdb] [saucenao] [google] [report]
62482642

>>62482602
>Literally sold their asses to AMD
>Implying that's a bad thing, given the latest circumstances and also that Dell makes one of the best calibrated monitor panels in the world currently

Also - XPSes are bretty gud, ahckchyually.

>> No.62482643
File: 128 KB, 1680x1050, laptop.png [View same] [iqdb] [saucenao] [google] [report]
62482643

>>62482392
>>62482436
Isn't AMD amazing?

>> No.62482670

>>62482624
inb4 inteldrones
>but, muh errata. its ciscos fault

>> No.62482679

>>62478856
Jim "I barely had anything to do with Zen" Keller

>> No.62482693

>>62482643
Engineering sample, you Intbecile.

>> No.62482717

>>62482679
Except for the fact that he literally created it and that he produced amounts of ideas on how to develop platform further which will be enough for at least three full generations more. Keller is LITERAL ZEN.

>> No.62482731

>>62482643
I too love 30 seconds boosts.
>>62482717
He has nothing to do with Zen.
At all.
Zen was created by Mark Clark (and was named by him too).
Kelled designed K12.

>> No.62482745

>>62482581
wasnt superfish on ideapads trash only?

>> No.62482756

>>62482643
https://browser.geekbench.com/v4/cpu/3988647
Nice cherry picking.

>> No.62482763

>>62482745
It was everywhere. In notebooks, smartphones, tablets.

>> No.62482795

>>62482731
It's a 2GHz engineering sample. Final version will be 3GHz stock at the very least, that's not counting in any Turbo. Intel's mobile, however, already cannot do more than ~2.8GHz tops.

>> No.62482820

>>62482756
Can I play?
https://browser.geekbench.com/v4/cpu/3557230

>> No.62482849

>>62482820
That's retail sample.

>> No.62482850

>>62482820
oh, wait. You don't want a engineering sample, do you?
https://browser.geekbench.com/v4/cpu/4002635

>> No.62482853

>>62482820
No. Go kills macacos for soup.

>> No.62482870
File: 156 KB, 689x1024, 14847432956760.jpg [View same] [iqdb] [saucenao] [google] [report]
62482870

>>62482820
>Can I play?
No, you can play games only with AMD APU, because Intel's integrated graphics are absolutely inept garbage.

>> No.62482889
File: 10 KB, 279x181, 1481930704359.jpg [View same] [iqdb] [saucenao] [google] [report]
62482889

>AMD is shit
>b-but you are a nigger monkey eater

>> No.62482895

Are we going to get quad cores in 13 inch laptops with Zen or are they going to have to gimp the clock speed to accommodate it?

>> No.62482907
File: 112 KB, 255x231, 1430236296498.gif [View same] [iqdb] [saucenao] [google] [report]
62482907

>>62482870

>> No.62482920

>>62482889
>AMD is THE shit
At least you understand this, /g/oy. Don't tell your kiker anal overlords, though, or they'll take away your shekels.

>> No.62482928
File: 10 KB, 276x270, 1493246348565.jpg [View same] [iqdb] [saucenao] [google] [report]
62482928

>>62482870
>buying gaymen laptops
This is the state of /g/

>> No.62482937

>>62478768
I'll believe it when I see it, OEM's have been putting AMD chips in the shittiest laptops possible, AMD's reputation in mobile is in the gutter.

Also why did AMD even bother releasing Bristol Ridge? It's getting massacred in reviews and only hurts AMD's reputation.

>> No.62482955

>>62482937
Fill the gap? Probably.

>> No.62482956
File: 1.25 MB, 458x250, WTHIWWYP.gif [View same] [iqdb] [saucenao] [google] [report]
62482956

>>62482895
>Not getting a 15~17" top tier Zen APU notebook that's thin and light as fuck while also being power efficient and great for mobile /g/oyming

>> No.62482959

>>62478560
Any news as to how the GPU portion is going to perform?

>Jewish attempts at kikening out a piece of integrated graphics
Please consider suicide.

>> No.62482972

>>62482959
800-1200 gflops.

>> No.62482973

>>62482928
But with new APUs they'll actually be feasible and stop being a meme finally. So why not?

>> No.62482987

>>62482937
>OEMs have been putting AMD chips in the shittiest laptops possible
How is it there, in 2012~2014?

>> No.62482992

>>62482972
There's more to GPU than simply shader math.

>> No.62483008

>>62482992
It's little vega. We don't know.

>> No.62483017

>>62483008
Vega is yet to work properly.

>> No.62483039

>>62482959
>Any news as to how the GPU portion is going to perform?
>Literally says ~90% better performance than Bristol on ALL fronts (CPU and GPU altogether)
>The ~90% is engineering sample clocked at 2GHz, retail version will have at least 3GHz on stock
>He doesn't read

>> No.62483044

>>62479282
>its only holocaust when AMD does it

>> No.62483059

>>62483044
I don't think the jews would holocaust themselves.

>> No.62483061

>>62482928
>Integrated graphics
>Gaming laptops
Did you have a stroke while typing this?

>> No.62483063
File: 194 KB, 443x390, Intel Nazi Swastika.png [View same] [iqdb] [saucenao] [google] [report]
62483063

>>62483044
Well, there was this one thing in the past, but...

>> No.62483092

>>62481971
>>62482129
>>62482140
>>62482150
Fuck all of you! I tried.

>> No.62483095

>>62482987
https://www.extremetech.com/computing/222662-poor-oem-designs-single-channel-ram-sabotage-amds-carrizo
http://www.anandtech.com/show/10000/who-controls-user-experience-amd-carrizo-thoroughly-tested
https://www.notebookcheck.net/Ryzen-is-great-so-why-are-most-AMD-notebooks-embarrassingly-bad.237106.0.html

More like 2016.

>> No.62483143

>>62483039
The article says nothing about the GPU performance other than mentioning AMD's claims of a "40% improvement" you tard. Geekbench is a CPU benchmark.

>> No.62483242

>>62482643
those intel multi core scoreson the low power chips

LOL.
#rekt.

>> No.62483345

>>62483242
>2 cores
>4.5W
pajeet pls

>> No.62484766

>>62479244
> demolish Intel
> 15.5fps vs 12.4fps
> "demolished"
> Not even Iris 650
> not even the 128MB L4 Iris
Sure it will beat those low end i7's dual core i7's, but won't compete against Intels high end quad core i7's with Iris Plus graphics

>> No.62484802

>>62484766
These are ES stats, friend.
And if Intel really makes AMD angry it'll get APUs with HBM intergration (which will still be cheaper than HUEG eDRAM slab btw).

>> No.62484882

>>62484802
> it'll get APUs with HBM
not so soon anon...

>> No.62484906

>>62484882
IF makes PHY integration a trivial task.
They can intergrate one HBM PHY direcly into iGPU portion of the die.

>> No.62485024

>>62484906
> trivial
sure thing bro...

>> No.62485476

>>62479768
when they doubled decode width on Steamroller not much happened

>> No.62485533

>>62485476
The cache latencies were still horrific.

>> No.62487733

>>62478560
Geekbench is Trash as benchmark.

https://browser.geekbench.com/v4/cpu/compare/3989386?baseline=3770816

R7 2700U appeared too, 2.2 GHz base. Can't tell turbo, the results are too rubbish.

>> No.62487988

>>62482015
They probably had to pay him in stock

>> No.62488154
File: 46 KB, 150x262, lisa.su.amd.png [View same] [iqdb] [saucenao] [google] [report]
62488154

RAVEN RIDGE

I NEED MUH LAPTOP DAMNITTTT

>> No.62488191

>>62487733
>2500U
>Beating 2700U
>High end model beating the midrange model, in the same lineup

Sure thing, man

>> No.62488229

>>62479111
>comparing a 3.4Ghz turbo 35w part with an unknown turbo 15w part
Bitch nigga please

>> No.62488399
File: 73 KB, 633x758, 1498703482745.jpg [View same] [iqdb] [saucenao] [google] [report]
62488399

>>62482643

DELET!

>> No.62488451
File: 54 KB, 1000x1000, jpdwin.jpg [View same] [iqdb] [saucenao] [google] [report]
62488451

>>62478560
So, can i have one of those with the Raven Ridge?

>> No.62488480

>>62482643
This looks fine and dandy until you toss in 3D performance.
It will be a quite full oven.

>> No.62488487

>>62488451
If you want your hands on fire

>> No.62488501

>>62488487
You always can underclock and undervolt it until it gets to a decent temp.

>> No.62488526

>>62488501
Good luck with undervolting that to 2W

>> No.62488561

>>62481322
I'm hopeful but dont believe for a second that Glofo 7nm would allow a proper shrink of Zen 1800x(not Zen2 itself) to reach 5Ghz at ~110 watts torture loaded.
Would imply the process allows Zen to be stable at 5.8Ghz using 220 watts.

It's much more likely that the >Glofo marketing slide< is showing both the optimum HP-design voltage target, OR, the max reliable design frequency attainable. Stress on reliable

>> No.62488566

>>62488526
Well, you obviously will have to cut on the clock, and maybe even disable a couple of cores.

>> No.62489543

>>62478560
>was considering building a new machine with an FX-8350 but ended up not doing it
>it turns out this new processor gets ~86% of the multi core performance and ~130% the single core performance of the FX-8350, and it's designed with a 15 watt TDP vs the FX-8350's 125 watt TDP
>this new processor also has comparable single core performance to my current Phenom 9550's multi core performance, and my Phenom 9550 has a 95 watt TDP
>7nm hasn't even been released yet and will make these gaps even wider
I wonder if I'll just end up using an SBC when I finally get around to upgrading. Specs these days are getting completely ridiculous.

>> No.62489566

>lelnopo didn't wait for raven and released bristol ridge laptops.
Why?

>> No.62489653

>>62478856
THANK YOU BASED JIM

>> No.62489698

>>62478560
Give us a Zen laptop with an 13"-14" IPS screen and a long battery life, maybe even extendable battery like in Thinkpads and a metal body and it will sell like crazy. Don't make the mistake of making them too thin so they overheat like Apple trash. Only the weakest of the weak would absolutely require paper thin laptops. Anything thinner than Razer laptops would be a terrible mistake.

>> No.62489961

>>62489698
that would be gold.

>> No.62491096
File: 129 KB, 354x504, Delid dis goy.png [View same] [iqdb] [saucenao] [google] [report]
62491096

>>62483063
You're not funny

>> No.62491105

>>62484766
>he doesn't know what engineering sample sandbagging means

>> No.62491106

>>62489566
those excavator laptops are on the level of the flagship broadwell t450 but are a lot cheaper

>> No.62491108

>>62485476
>doubled decode
You meant to say "flickered the code the fuck out".

>> No.62491121

>>62487988
Nah.
First of all - he left before AMD gone 2$ per share.
Secondly - he got HUEG royalties for all his work in AMD through the entirety of the years, basically a literal golden parachute. Those royalties were bigger than what he's getting as an yearly wage at Tesla right now.

>> No.62491152
File: 400 KB, 816x1020, 1496330997182.png [View same] [iqdb] [saucenao] [google] [report]
62491152

>>62488229
There was no Turbo. It was bare engineering sample with stepping 0 (even says so in specs), a literal "a mere proof of concept right from the lab" unit, at 2GHz with almost everything disabled. There was no Turbo and also no SMT - just bare 4 cores at 2GHz. AMD is sandbagging SO GODDAMN HARD right now it's downright GLORIOUS.

P.S.
http://www.youtube.com/watch?v=ZVaJH8pWnWk

>> No.62491179
File: 3.38 MB, 480x270, 45854cfe0bf05cc9c54bd72823aa4da2.gif [View same] [iqdb] [saucenao] [google] [report]
62491179

>>62489543
>End of 2017
>He has Phenom 9550
>Not the GODLIKE Phenom II X4 950

>> No.62491194
File: 348 KB, 688x720, 1494102602792.png [View same] [iqdb] [saucenao] [google] [report]
62491194

>>62489698
>13"-14" IPS screen

--> >>62482956.

>> No.62491267
File: 952 KB, 500x685, Dudeness.gif [View same] [iqdb] [saucenao] [google] [report]
62491267

>mfw it's late 2018~early 2019 and I still absolutely seriously and completely unironically think of building a desktop system on Zen APU instead of just buying a notebook

>> No.62491313

>>62483095
this is first to me.

>> No.62491434

>>62491267
You can't go wrong with an R7 1700

>> No.62491511

>>62491434
I said Zen APU.

>> No.62491613
File: 3.95 MB, 307x325, 1478936303408.gif [View same] [iqdb] [saucenao] [google] [report]
62491613

>>62482062
>http://news.lenovo.com/news-releases/lenovo-reveals-new-thinkpad-series-powered-by-amd-pro.htm
>12.5 inch
>14 inch

Fucking dropped. I hate that there are so many manlets in tech. I want a 17.3" or nothing at all, everything is too damn small. I'm well over six foot though so there's that...

>> No.62491741
File: 565 KB, 367x265, 1494945059634.gif [View same] [iqdb] [saucenao] [google] [report]
62491741

>>62491613
Good to know that I'm not the only here who thinks that any notebook screen smaller than 15" is for sissy boy fags and brainless bimbo loli.

>> No.62491767

>>62479094
I wouldn't expect anything higher than a RX 550/460 level for these new APU's iGPUs at best, and if you think you'll be able to get 60 FPS at 1080p on Ultra in The Witcher 2 with that, you're in for a huge, huge dissapointment.

>> No.62491823

>>62491767
You really never used AMD APU ever before, huh.
Go back to your Inturd Integrated Graphics 2000, kike.

>> No.62491824

>>62491767
Duh, it's only 704 shaders in a very bandwith limited scenario.

>> No.62491856

>>62491767
Even Richland 6800K's video can already get 30~35 average FPS in Witcher 2 on 1920x1080 with everything maxed and if fast RAM is installed. Zen's iGPU is at least 90% more powerful than Richland's even on default slow RAM, so I hope that you're able to count 2+2.

>> No.62491877

Boy DSBR will be a killer feature for APUs.

>> No.62491911
File: 1.89 MB, 484x682, JUST.png [View same] [iqdb] [saucenao] [google] [report]
62491911

>>62491767
You forgot to enter your tripcode and log in, Brian.

>> No.62491918

>>62478755
the driver will be ready with the next iteration of the crimson suite
late sep early oct probably

>> No.62491921

>>62491877
>A Keller feature

>> No.62491964

>>62491918
Yes. They've already teased Afterburner-like overlay from Crimson Vega edition.

>> No.62491970

>>62491964
post da link ma nigga didnt saw it at all

>> No.62491980

>>62491767
>I wouldn't expect anything higher than a RX 550/460 level for these new APUs
>if you think you'll be able to get 60 FPS at 1080p on Ultra in The Witcher 2 with that, you're in for a huge, huge disappointment
Video below is 20~50 FPS 1080p everything maxed out with RX 460 on a FUCKING ATHLON:
http://www.youtube.com/watch?v=i0ncrf_dv-M
Are you so utterly retarded as to openly proclaim to us that Athlon is more powerful/better than Zen? And Zen's integrated graphics are not of RX 460-level, but of GTX 1050 Ti/RX 560 level.

>> No.62491986

>>62491970
Guguru it, sempai.
But it had two modes: classic and a fancy one.

>> No.62491997

>>62491980
It's actually slightly below 1050 according to SiSoft stats.

>> No.62492011

>>62491997
Engineering sample

>> No.62492019

>>62491997
ES without optimized drivers, go figure.

>> No.62492021

>>62492019
They may actually ran it with 17.320 batch.

>> No.62492023

>>62491986
sorry cant find anything related to it only tons of amd current drivers
am i missing something?

>> No.62492032

>>62484766
Its 4c8t i7, those have 4Ghz turbo.

>> No.62492039

>>62492021
Even if so (which I personally highly doubt simply due to how heavily AMD is sandbagging right at this current time), this still doesn't cancel the fact that this is a raw as fuck engineering sample working on 2GHz with Turbo and SMT disabled. And highly likely on gimped RAM too (and we all know how AMD's APUs love and greatly benefit from faster memory).

>> No.62492050

>>62492023
RTG dude teased the overlay like a day after RX Vega was launched.
But everything related to WiP drivers are under the NDA now as a part of Vanguard program.
>>62492039
The RAM for geekbench results was single channel afaik.

>> No.62492070
File: 2.02 MB, 1021x1068, BRHDRPC.png [View same] [iqdb] [saucenao] [google] [report]
62492070

>>62492050
>The RAM for geekbench results was single channel
Holy shit. This sandbagging is SO GODDAMN HARD it's downright getting RIDICULOUS now. Y U DO DIS TO US, LISA? This is beyond "smart trolling", this is just burning/gassing kikes in batches of hundreds of thousands now.

>> No.62492081

>>62492070
No one knew about SEV and SME before the very launch of EBYN.
AMD is very careful with NDAs and information on general.

>> No.62492088

And you all do understand that AiB Vega's are delayed artificially?

>> No.62492101

>>62492088
What's the reason, though? It's not like they're waiting for noVideo's nearest release because Huang already shitted out he possibly could so far. Are they fucking seriously and absolutely unironically waiting for the elusive "1070 Ti"?

>> No.62492110

>>62492101
No, the drivers.
They want a favourable second round of reviews.
Because if ASUS of all OEMs finalized the designs then Sapphire and XFX did it ages ago.

>> No.62492113

>>62492088
the sole reason they are on hold is because of the driver

>> No.62492115

>>62492101
>already shitted out everything he possibly could
Selffix

>> No.62492151

>>62492110
>>62492113
But WHY are they holding out the DSBR and PSC/PTC-enabling drivers, though? Isn't this EXACTLY because they foresaw 1070 Ti and knew that fully enabled /Ve/Ga/ can utterly demolish 1060, 1070, 1070 Ti and 1080 altogether in one step just by enabling those features? But even if so, that still won't fix prices and power envelope issues.

>> No.62492152

>>62492101
we wont see any cut down of the V's anyway not unless they figure out a way for their hardware SC to not eat away the power of the east coast
and since amd hasnt found a way to do it pretty sure nvidia cant hell they even officially annc that they have an emulator for async on the cpu
plus the warp sc that runs on the cpu
this will be a fucking indirect housefire where the gpu will heat up the cpu to do its job

>> No.62492168

>>62492152
>we won't see any cut down of the V
But Raven Ridge has an integrated Vega inside of it that was massively downclocked, no?

>> No.62492175

>>62492151
well you must understand vega was being built when amd has a bit more money than greece..
we already know they focused more on the workstation segment since the driver IS working there the culling is literally shitting on p6000
add to that that lisa got all the money from rtg to funnel it to ryzen
its no wonder why they are delaying it

also there is the fact that traditionally amd factory cards have shit coolers(which this time around they didnt for some reason..) so it was better to have a shitty driver and wait for the good aib to come with the proper driver
moving the goalpost the amd way i agree but still is far better

>> No.62492180

>>62492088
>>62492101
>>62492110
>>62492113
>>62492151
>meme driver
i thought it was already enabled on the Frontier Edition?

>> No.62492183

>>62492151
They are not holding out the drivers.
They are still working on them.

>> No.62492184

>>62492168
i meant we wont see any cut down volta not vega

>> No.62492186

>>62492180
DSBR does fuckall when there's a front-end bottleneck.
And it's still lacking deferred rendering step.

>> No.62492187

>>62492175
So what you're saying is that they basically OUTSOURCED development of good drivers to third party OEMs? What the actual fuck?

>> No.62492198

>>62492187
No you fucking retard.
They are delaying AiB cards because the software is not ready.

>> No.62492200

>>62492187
who the fuck told you that?
i said they developed the workstation drivers FIRST cause they didnt had MONEY to actively develop for both segments
plus the server and the ssg that needs special drivers also

aib's cant fucking get their shit together most of the time and you think amd will outsource the driver development to them?

>> No.62492215

>>62492183
If they'd didn't, then they'd release them together with Vega during it's launch, and no one would've bitched about Vega's /g/oyming performance in comparison to 1070/1080/1080 Ti then. But here we are, still waiting™ for proper fully enabled drivers, and I bet they had them on their hands since day-fucking-one. I just don't believe that they're not sandbagging in this particular case, considering how strongly they're sandbagging right now with RyZen, APUs, and Epyc.

>> No.62492228

>>62492215
Both Vega FE and current RX Vega exist to fellate shareholders.
Everyone does this.
Both P100 and V100 were "launched" with zero availability simply to fellate shareholders.

>> No.62492237

>>62492215
>they're sandbagging right now with RyZen
you mean my ryzen 1700 have some hidden power?

>> No.62492244

>>62492237
Threadripper is RyZen too. I was talking about Threadripper, naturally.

>> No.62492250

people seems not being able to understand what amd is doing...

they are getting back all the work a dev can do with the shaders and try to do automaticly
the primitive shaders normally needs to be exposed to the devs but amd doesnt because lets be real here last time they gave a new tech to the devs (tessellation) they fucked up amd big time

>> No.62492260

>>62492250
Ahh, this years-old "AMD has problems with tessellation" FUD of a fairy tale again. Haven't seen that in a while.

>> No.62492263

>>62492250
It's not exactly that.
They are nuking ye olde hardcoded shaders in favour of flexible new ones.
Basically poojeet got angry about the state of current APIs being warmed up DX9.

>> No.62492272

>>62492260
i didnt say they had any problem with tess i said that certain devs were overly using tess which has been proven already

>> No.62492286

>>62492260
It's much weaker in tesselation compared to anything resembling Fermi (i.e. anything that employs Polymorph Engines).

>> No.62492295
File: 239 KB, 620x497, true-water-mesh-620.jpg [View same] [iqdb] [saucenao] [google] [report]
62492295

>>62492272
But that was never "Radeon's shortcoming", though.

>> No.62492303
File: 200 KB, 602x483, debris-water-mesh-620.jpg [View same] [iqdb] [saucenao] [google] [report]
62492303

>>62492250
>>62492272
>>62492286

>> No.62492314
File: 1.73 MB, 1239x495, crysis2-tesselated-water-hidden.png [View same] [iqdb] [saucenao] [google] [report]
62492314

>>62492272
>>62492286

>> No.62492320

>>62492263
they made the already flexible ACE/HWS now being called NCU even more flexible by interconnecting them all together thus having a pool on holding what each engine is doing what and finally prioritizing better the workflow
(plus given what i saw on the videos of the deve talks they literally have ace on every place possible )
>>62492295
it is since the old idiotic ceo scrapped the plans of a hardware rasterization engine till polaris came along
nvidia gained a massive perfomance cause even if you had x8 tess or x64 the impact would have been the same due to the culling
but on amd an x64 tess was literally a tsar bomba hitting the card add to that that physx stopped running on amd cards and went into the cpu and you get an idea

>> No.62492333

Mind you, ALL of this was done DIRECTLY under noVideo's orders (a collusion, even) to PURPOSELY gimp Radeon. So given that, how "tessellation was Radeon's problem" any ever? I mean, it sure was, but NOT because of hardware itself.

>> No.62492364

>>62492333
You gotta play dirty if you want to win.

>> No.62492374

>>62492364
Logic like this is the exact reason why we can't have nice things these days anymore, unless they're Zen.

>> No.62492388

>>62492374
well amd had to play the role of the open source guy cause there wasnt any other way

remember when ati started to gain market share with the 4xxx? the next gen they bumped +150 on the price on every bracket
just a hint of victory was enough to do this and what for?

>> No.62492400

>>62479320
A single ccx has 4MB$, ryzen has 2 CCX and apu's have 1

>> No.62492439

>>62492400
we have seen a r7 apu also
https://compubench.com/device.jsp?benchmark=compu20d&os=Windows&api=cl&D=AMD+Ryzen+7+2700U+with+Radeon+Vega+Graphics&testgroup=info

>> No.62492440

>>62492400
Single CCX in Zeppelin has 8MB of L3$, dummy.
>>62492388
No, Evergreen was bigger dies in newer node for slightly higher prices.
$350 for 5870 was a steal.

>> No.62492446

>>62492400
Was it confirmed that Raven Ridge is one fully enabled 4 core CCX and not two CCXes with 2/0+2/0 core configuration? I thought that Zen's entire developing process makes it much easier and cheaper for them to just disable particular cores on full-sized CCXes and ship them like lesser core configuration, than to make fully enabled single-piece CCX separately, no?

>> No.62492460

>>62492446
CCX is 4 cores + cachez.
One Zeppelin die has 2 CCX, one RR die has one CCX.

>> No.62492470

>>62485476
They also gimped cache even harder destroying most decode gains

>> No.62492474

>>62492460
So it's actually just one fully enabled 4 core unit for the APU. Damn, that's actually pretty good considering that it'll literally obliterate any possible latency increase.

>> No.62492487

>>62478768
>, even when it comes down to watching over third parties.
and yet third parties are selling the vega chips at 600+$ before it even touches retailers.

Like others said, ill believe it when I see it but I believe amd has a potential winner

>> No.62492498

>>62492487
The supply chain for GPUs is very long.
Stop comparing incomparable.

>> No.62492506

>>62492498
amd doesnt crack down on that hard, why would they crack down oems hard?

>> No.62492514

>>62492487
>Like others said, ill believe it when I see it but I believe amd has a potential winner

the winner for amd is that they literally got the renderfarms and the independed ssg users from nvidia..
now an ssg costs 7k grand and you need a simple pc to run it given its massive cache
for nvidia you need a p6000 7k grand a dpg 1000 2k grand 1tb of ecc ram almost 1k grand
its a no brainer

>> No.62492515

>>62492506
You are comparing incomparable.

>> No.62492518

>>62492487
There are no third party-made Vega, aside from Asus' teasers of new Ares, all of the so-called "third party" offerings are pure references right now. These are basically "FE", so it's natural that they inflate prices for them to stuff buttcoin miners.
AMD seriously just sits on it's ass and waits for 1070 Ti to come out so that Radeon Group can release fully enabled features via drivers and only AFTER that 100% non-reference solutions begin rolling in. Just wait™.

>> No.62492524

>>62492514
1TB worth of 64gig LRDIMMs costs much more than one grand assuming $900 per DIMM.

>> No.62492525

>>62492515
i'm comparing amd directly handing shit over to someone, and doing nothing to stop them from being cunts.

>> No.62492532

>>62492525
They can't influence suppliers or retailers.
Doing that is illegal.

>> No.62492541

>>62492518
no, further up the supply chain before it gets to the companies we knows the names of are selling the chips at 600-800 usd meaning by the time asus gets it the chip is already 600-800$ on top of what they charge for profit.

>> No.62492551

>>62492532
hence why the shit gibbo started

he knew people eventually will find out his lies and he probably sold a lot of inflated cards because of it

>> No.62492554

>>62492532
then you don't work with them, you don't supply with them, where the actual fuck does everything think cracking the whip on people fucking you is illegal I have no idea.

>> No.62492557

>>62479743
>They advertised FX APUs as 4 core solutions
nope

>> No.62492571

>>62492487
By "third parties" I've meant custom pre-built system selling and notebook manufacturing OEMs & partners, not video card OEMs & partners. AMD is currently sitting hard on Dell's, HP's, OriginPC's, OLDI's, iBuyPower's, Lenovo's, Alienware's and Maingear's asses, directly supervising them all so that they don't combine Zen APU with a single channel memory and thus gimp it the fuck out. I was talking about this kind of a QA in particular. Considering that all of the aforementioned third party manufacturers made long-term partnership contracts with AMD on Zen, the fuck-up with memory is extremely unlikely now. AMD won't allow for any of that shit to come through.

>> No.62492609

>>62492557
>Bulldozer/Piledriver was never advertised as 8 core processor
>FX APU was never sold anywhere with a "4 cores" tag and FX 6300 was never sold as a "6 core" unit
Puh-effing-lease, you dumb fuck.

>> No.62492610

>>62492571
i doubt amd is directly supervising them cause they wouldnt let dell to release that alienware with threadripper and that fucking motherboard and memory with shit timings and the worst cooling solution ever..

>> No.62492618

>>62492610
Anything Alienware is shit, anon.

>> No.62492634

>>62492618
hence why amd probably isnt oversighting anything

which is bad but..

>> No.62492639

>>62492634
No. I mean anything Alienware is shit.
Their i9 prebuilt is just as shitty.

>> No.62492646

>>62492610
Again - they have long-term contracts established. AMD direct supervision is purely a natural course of action now. Because if any of these third party players fuck up with Zen in any way, and ESPECIALLY with Zen APU in the notebooks, they'll suffer great consequences due to breach of contract. I wouldn't be surprised one slightest goddamn bit if these contracts they made with AMD directly state in them that these third party manufacturers are simply not allowed to gimp Zen APUs by implementing single channel memory or in any other ways whatsoever at all.

>> No.62492729

>>62478560

Redpill me on Ryzen vs Coffee Lake. I need a new CPU for rendering.

>> No.62492753
File: 1.17 MB, 940x433, Legitimate Intel Fire Group.png [View same] [iqdb] [saucenao] [google] [report]
62492753

>>62492729
>Redpill me on Covfefe Lake
Here you go:

>> No.62492757

>>62492295
>>62492303
>>62492314
noobs still lying about Crysis 2 tessellation years later ...

>> No.62492766
File: 84 KB, 640x429, 1504720274478.jpg [View same] [iqdb] [saucenao] [google] [report]
62492766

>>62492729
>A new CPU for rendering
>Intel

>> No.62492772

>>62492729
they manage to put fx9xxx series into shame

that should be enough evidence

>> No.62492788

>>62492609
AMD is not the one selling them

>> No.62492791

>>62492753
>>62492766
>>62492772

What's the AMD alternative to the (((Coffee Lake))) then?

>> No.62492792
File: 198 KB, 400x300, 14794152882150.png [View same] [iqdb] [saucenao] [google] [report]
62492792

>>62492757
You forgot to enter your tripcode and log in, Jensen.

>> No.62492805

>>62492792
you forgot to learn something about their renderer

>> No.62492812

>>62492791
Threadripper. Even cheap 1920X is more than sufficient for serious work and already obliterates 7900X in 8 tasks out of 10.

>> No.62492819

>>62492791
for prof render and stuff TR4 is a no brainer
for the rest a 1700 is the best atm you can render play watch movies do whatever the hell you want with it and still you wont be able to saturate the damn thing

>> No.62492833

>>62492819
>for the rest a 1600X is the best

>> No.62492857

>>62492833
well depends
im a farmer on lineage2
i had a 6900k the damn thing was struggling after the 10th instance plus with 10 bots i still dont know why
a friend of mine had a built with 1600 and he was able to open 22 instances with 22 bots on low res obviously
i managed to open 39 instances with my 1700 and still the cpu was on 70%
for that kind of usage its far better having moar coarz

>> No.62492858

How the situation with Linux support on new amd GPU's? I remember they were incompatible back in the days, especially on laptops, so I didn't even consider buying one.

>> No.62492873

>>62492857
>End of 2017
>Lineage II
KYS, onegai kudasai.

>> No.62492899
File: 29 KB, 640x424, Torvalds.jpg [View same] [iqdb] [saucenao] [google] [report]
62492899

>>62492858
>How's the situation with Linux support on nGreedia GPUs?

>> No.62492919

>>62492812
>>62492819
>Even cheap 1920X is more than sufficient for serious work and already obliterates 7900X in 8 tasks out of 10.

Wew, looks like AMD and not Shlomo will be getting my money this time. I was doubtful about AMD because I've only had bad experiences with their GPUs.

>> No.62492922

>>62492873
bitch i get 2k per month usually from idiots that buy armors
let alone selling adena on various servers

its far from what it was 5 years ago that NA farmers had a monthly income of 20-40k by selling epics but still its going hard

>> No.62492938

>>62492899
Yeah, but I heard things changed after that performance. And indeed there are linux Nvidia drivers, even thought it takes an evening to spend to install them (doing this https://www.if-not-true-then-false.com/2015/fedora-nvidia-guide/)

>> No.62492999

>>62492858
Actually?
Really fucking good.

>> No.62493014

>>62492919
AMD's processors were bad only during dark ages of FX era, before that (Athlon 64, Opteron, Phenom II) and after (Zen) they're super great. Their video cards, however, were good during HD 6xxx, were meh during HD 7xxx, were good during R 2xx, were bad during R 3xx, were meh during RX 4xx, are "decent" during RX 5xx and are "just wait for proper fully enabled drivers™" currently with /Ve/Ga/.

>> No.62493039

>>62493014
>GCN1
>meh
I'll shank you.

>> No.62493075

>>62493039
The power envelope was bad and Crossfire was utter shit (THE most worst Crossfire Radeon EVER had) on HD 7xxx, so no, fuck YOU.

>> No.62493109

>>62493075
>Crossfire
Gay.

>> No.62493163
File: 45 KB, 1140x647, Intel advantage over AMD.jpg [View same] [iqdb] [saucenao] [google] [report]
62493163

>>62492919
7900X currently costs almost ~1000$ for the CPU itself and a decent motherboard for it will also cost you no less than 400$, also add RF ID™ and DRY HORSE CUM™ under the lid alongside with RAID KEYS™ and LESS LANES™ to that, with a Fukushima-tier power envelope and temps on top as a shit-smeared cherry. Meanwhile, 1920X costs ~800$ (can get as cheap as 720$ at some places during sell-off holiday days), has more cores and lanes, no RF ID spying, no RAID KEY restrictions, top tier quality direct solder, motherboards are decently priced, and it outperforms 7900X in 8 test out of 10. There is literally no choice between the two, because everything is very clear and apparent.

>> No.62493180

>>62493109
Not when it's done right (read - cordless XDMA with very low FPS latencies/discrepancies, which R 3xx and RX 4xx with 5xx have, for example).

>> No.62493294
File: 162 KB, 633x900, ffffffvey.png [View same] [iqdb] [saucenao] [google] [report]
62493294

>> No.62493322

>>62492858
Open source is ranging from "really great" to "requires tons of config to get working", where "really great" typically applies to Polaris-based chips, and the later referring to Vega (though this will probably change in time).

>> No.62493379

>>62493322
Those are the newest right? So all videocards released this year and later should work without any problems?

>> No.62493413

>>62493379
I can't say for R 2xx (most of those are HD 7xxx rebrands and HD 7xxx had shitty Linux support), but R 3xx should work just fine too.

>> No.62493417

>>62492857
That's because intel has shitty stuttery cores.

>> No.62493442

>>62493379
RX 5xx is a slightly updated and higher clocked RX 4xx with more memory on board, so not just from this year (RX 4xx is older than that).

>> No.62493465

>>62493379
RX400 and RX500 series are Polaris.

RX Vega is Vega.

Older stuff works to varying degrees.
>>62493413
R9 390's got some really broken DPM support in both the radeon and amdgpu kernel modules. I assume the same applies to the other hawaii-based cards. Anyone buying a card today should probably avoid those; Best bet is either Polaris or wait(tm) for better Vega support. Vega atm requires a lot of fiddling, what with custom kernels, and there are a few quality of life issues with actually using Vega on Linux as a daily driver.

>> No.62493512
File: 35 KB, 599x388, .jpg [View same] [iqdb] [saucenao] [google] [report]
62493512

>>62493442
>>62493465
I think I got it

>> No.62493522

>>62493465
Wasn't R9 390 in particular a "Tonga update"-based Hawaii? R9 285 was the very first Tonga (an architecture which wasn't both a mere HD 7xxx rebrand nor a full Hawaii), and 3xx had a refresh of that so wasn't it R9 390? Or was it 380X?

>> No.62493526

>>62492729
>CPU for rendering
>intel
>ever

>> No.62493529

>>62493522
>Or was it 380X?
This.
390/390X are rebadged Hawaii chips.

>> No.62493532
File: 68 KB, 750x418, Fire.jpg [View same] [iqdb] [saucenao] [google] [report]
62493532

>>62493526

>> No.62493561

>>62493529
Not really "just rebadged", because they had more memory than 290/290X had. With R9 390/390X 8GB pretty much became mainstream, while 290/290X had 4GB by default and 8GB 290X was a rare sight.

>> No.62493570

>>62493561
Well, just rebadged.

>> No.62493582

>>62493570
Adding more memory is not "just rebadging". Because units are being physically altered.

>> No.62493586

https://www.servethehome.com/intel-xeon-silver-4114-linux-benchmarks-and-review/
>The “elephant in the room” is AMD EPYC. At around $700, you are in price ranges of 16 core EPYC parts (e.g. the EPYC 7351P and EPYC 7281.) You can make an argument that Intel has some per-core performance advantages, but the advantage Intel has is not 60% per core. EPYC has the larger platform (8 DDR4 channels per socket and 128 PCIe lanes) but uses more power. The strange part here is that Intel is essentially increasing pricing just as AMD entered as legitimate competition.
Ayy.

>> No.62493610

>>62493586
I don't get it - is he complimenting Epyc or trying to trash-talk it? This shit is written too vaguely.

>> No.62493621

>>62493610
He's questioning Intel's sanity.

>> No.62493649
File: 2.91 MB, 896x1136, Shoah.png [View same] [iqdb] [saucenao] [google] [report]
62493649

>>62493586
>>62493621
I find it rather hilarious how he "forgot" to mention about RF ID™, HORSE CUM™, RAID KEYS™, and FUKUSHIMA POWER/TEMPS™.

>> No.62493662

>>62493649
That small Xeon is really cool.
It just happens to compete with EPYCs that have performance out of it's league.

>> No.62493757

>>62493662
>4114 literally loses 8 out of 10 tests to cheapest Epyc
>Out of those said 8 lost tests, gets literally OBLITERATED by the cheapest Epyc in 5 out of 8 tests
>Happens to compete
This is the state of Inturd jew's delusional denial of the reality in the end of the year 2017, people.

>> No.62493770

>>62493757
I didn't say it's a good competition to EPYC.
It simply is one.

>> No.62493779

>>62491767
I used to have a decent APU, it could run Skyrim on medium settings at 1080p at probably like 30fps or something.

>> No.62493783
File: 1.23 MB, 712x604, YUNOBUY.png [View same] [iqdb] [saucenao] [google] [report]
62493783

>>62493770

>> No.62493793

>>62493783
Don't worry, i'll probably get 7351p box.

>> No.62493818

>>62493770
If it's not a good competition, then there is no competition.

>> No.62493828

>>62493818
You still have a token choice if you're hellbent on buying Intel.

>> No.62493830
File: 234 KB, 800x612, 1486816070397.png [View same] [iqdb] [saucenao] [google] [report]
62493830

>>62493770
>>62493793

>> No.62493862
File: 329 KB, 800x533, 1497955599640.jpg [View same] [iqdb] [saucenao] [google] [report]
62493862

>>62493828
>Literally deliberately gimping yourself with less lanes and less cores, become cucked on higher price and worse performance, just in spite, "by principle"
Intel fanboys in a nutshell, ladies and gentlemen.

>> No.62493919

>>62493779
>I used to have a decent APU that could run Skyrim, but then I took an arrow in the knee

>>
Name (leave empty)
Comment (leave empty)
Name
E-mail
Subject
Comment
Password [?]Password used for file deletion.
Captcha
Action