[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / r / s / t / u / v / vg / vr / w / wg] [i / ic] [r9k] [s4s] [vip] [cm / hm / lgbt / y] [3 / aco / adv / an / asp / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / qst / sci / soc / sp / tg / toy / trv / tv / vp / wsg / wsr / x] [Settings] [Home]
Board
Settings Home
/g/ - Technology



Thread archived.
You cannot reply anymore.



File: unnamed (5).gif (3.91 MB, 500x281)
3.91 MB
3.91 MB GIF
https://www.tweaktown.com/news/59104/rtg-boss-takes-sabbatical-lisa-su-over/index.html

What did they mean by this
>>
Probably fake news, but It would make sense that Raja it's trowing a tantrum since almost every resource it's going into EPYC and Ryzen
>>
>>62401858
Could AMD win sufficient capital to not fuck Navi when Zen2 is ready?
>>
>>62401809
So he's back once Navi passes the tapeout?
Anyway Lisa needs to clean RTG up, their communication was seriously lacking.
It took AT's main editor poking them with a stick to get a comment on Vega's primitive shaders.
>>
>>62401937
what happened to them features anyway
>>
>>62401967
Straight out disabled on ANY public drivers. Both deferred rendering and primshaders.
>>
File: 11512356626.jpg (1.08 MB, 2344x2832)
1.08 MB
1.08 MB JPG
>>62401967
>>62402070
That's because AMD is playing the long game and sandbagging vega until after volta is released. Once volta is released, the full working driver for vega will be released which will enable all features, improve efficiency, and increase performance by as much as 60%. Doing this prevents nvidia from altering volta, and AMD will take the performance crown.

You read it here first!
>>
>>62401809
It meant that they're soft firing him when the press is distracted with the Apple releases. It's 4D chess
>>
>>62402159
t. (now fired) Raja Koduri
>>
>>62402159
That's dumb, because Vega's improved scheduler is going to be catching up to Maxwell at best. Meanwhile nvidia will have independent thread scheduling instead of warp scheduling.
>>
>>62402159
the poo fueled delusion is real
>>
>>62402182
That's compute.
>>
so you have to say "we heard" and everyone will believe you? amazing times we live in, can't wait for AMD response.
>>
>>62402182
You literally described async compute, a tech that legally belongs to AMD.
>>
>>62402182
when you pretend to know what you talk about try to be less specific, it won't show as much
>>
>>62402070
why would they do that
>>
>>62401809
kek, hope he fucking kills himself
>>
>>62402377
It's very very angry kids versus very angry shareholders type of situation.
>>
>>62402256
Async compute refers to multiple schedulers working in tandem. ACE for any compute work (shaders), DMA for memory tasks, graphics command processing for graphics pipeline to reduce waiting, which is the point of any asynchronous system. Wavefronts are specifically for issuing work to the compute units. If you have poor fill then you're doing work inefficiently. GPUs implement SIMD processing, if you have a boat big enough to fill 64 people headed to the same place but only fill 12 then you have extreme waste. Which is GCN's fundamental problem and has been since GCN1.

Nvidia is already good at filling warps, now they're claiming they can group threads that diverge during tasks to reduce the effects of divergence. Ex. normally you have multiple threads of execution but one takes a different code path and performs a different operation and the entire warp must wait for that one thread to complete and now the threads are out of sync, so this behavior becomes a repeating pattern. Now imagine of multiple take different paths, your warp will wait for each thread again and again. Independent thread scheduling will try to make sure the threads that diverged sync up with other threads when possible so that the wait doesn't occur the next time.
>>
>>62401809

Based Su.
>>
>>62402565
GCN has saturation problems because of overkill shaders:tris/clock ratio. NVIDIA faces no such problems since Fermi, Polymorph Engines are bretty flexible.
>>
>>62402159
I know that this post is bait, but I've seen so many delusional faggots repeat the same thing with 100% seriousness it's hilarious

vega is trash for gaming, wake up already and stop lying to yourself
>>
>>62402614
>Vega is trash for gayming
>every feature besides HBCC is related to gaymen
?
>>
I want to fuck lisa su. seriously. I don't even care that she is sort of masculine looking I want to please her for running AMD so well.
>>
LMAO is that gif real, what a nice party Raja
>>
>>62402624
it can have whatever feature it wants related to gaming, but as long as performance is crap in games - it's crap in games
>>
>>62402667
It's crap in gayms because it's not working properly.
>>
>>62402683
oh yeah I can't wait for those magic drivers to improve performance by 50%
just you wait until AMD releases em!
delusional fanboys, every fucking time
>>
>>62401875
this is not 1990
Navi has to be set in stone already
>>
>>62402697
>d-delusion
>i-i swear!
Whoa there.
>>
>>62402609
I will believe it if NGG actually improves performance
>>
>>62402719
what's your point exactly? vega launch was an absolute disappointment, both price-wise and performance-wise
now you're telling me that once AMD pushes out those super duper mega giga epic drivers it's gonna perform sooo much better?
you're either a shill or learned nothing after polaris
>>
>>62402720
We've yet to see it.
>>62402741
Enabling uarch features are exactly the thing you'd expect from super mega ebin epik drivers.
>>
>>62402790
ok, see you in 2 years when navi gets released and you parrot the exact same thing
>>
>>62402805
>l-lies
Okay buddy.
>>
>>62402741
>both price-wise and performance-wise
I haven't notice doubts about price/performance of Vega 56
>>
>>62402814
you forgot to call me an nvidia shill
>>
https://www.pcper.com/news/Graphics-Cards/AMD-Radeon-Technologies-Group-Raja-Koduri-Goes-Sabbatical

Pajeet is out, he got fired
>>
>>62402822
yeah, vega 56 would be a good card if it were sold for it's msrp

where I live vega 56 costs 800 USD, which is more than a 1080 ti
>>
>>62402831
No, it's all just according to keikaku.

Raj will come back when volta launches, and he will release vega driver that enable all features, resulting in double performance.

Just you wait!
>>
>>62402831
>goes on vacation
>is fired
?
Did you forget 2011 GPU division layoffs?
>>
>>62401809
just like the last sabbatical that turned out to be his vacations?
>>
>>62402159
>That's because AMD is playing the long game and sandbagging vega until after volta is released
>people actually believe this horseshit
>>
https://twitter.com/bitsandchipseng/status/887736810625019910

>People actually believe this horseshit
>>
Vega 56 for $430?
>>
File: ebyn.jpg (1.49 MB, 2048x1536)
1.49 MB
1.49 MB JPG
>>62401809
alright, best news for RTG so far. fuck this streetshitter, Mommy gonna make RTG great again!
>>
>>62403668
He'll be back in December, pal.
>>
File: 1504696113665.png (293 KB, 750x1000)
293 KB
293 KB PNG
>>62403682
so 2 1/2 months without shitting on RTG? great news!
>>
>>62403682
To a totally different RTG.
Raja isn't fired, but he's not competent at management or structure either - he's an engineer.

When he took the last Sabbatical they didn't replace his position with Lisa Su temporarily...
Su is coming in the clean house whilst Raja is out of the way and can't object.
>>
>>62403734
The best thing she can do it hire more code monkeys.
That and better marketing.
>>
>>62401858
It's real: https://twitter.com/ryanshrout/status/907812489672392704
>>
>>62403734
She's making changes to Navi
>>
>>62403793
Navi is one the way to tapeout.
>>
>>62402159
>completely mess up
>fanboys on the internet will come up with the most insane theories to justify it
>>
>>62403741
I'd imagine Hyderabad is in for some bad news though.
>>62403793
Unlikely, the chips are fine - Vega was sampling to AIBs November last year.
>>
>>62402697
>just you wait until AMD releases em!
NVIDIA WILL BE DESTROYED

JUST

WAIT

(TM)(C)(COPYRIGHT: AMD, CALCUTTA)
>>
>>62403839
>Hyderabad
Who?
>>
>>62402159
>deliberately losing $100+ per card so you can sandbag nvidia
>after marketing it as a gaming card
>and its slower than almost 2 year old 1080 and 1 year old ti
Your high on baked thermal paste mate.

RTG fucked up so bad i went with a 1600x and a 1080 oc for my new build and i couldnt be more happy, low temps across the board tiny power usage

AMD Vega is dogshit and you know it, a new process will help it but Raja's Vega is Kaka
>>
>>62403846
ATi already nearly murdered nVidia twice.
Nothing (besides time and money) stops them from doing that again.
>>
>>62403862
Wrong W, you want the third one.
>>
>>62403880
>losing $100 per card
Your dad works at AMD?
>>
>>62402159
I said this right after Vega came out. Literally weeks ago.
>>
File: Lisa.jpg (222 KB, 665x782)
222 KB
222 KB JPG
What's her full name /g/?
>>
>>62403991
Lisa "Poos in the Loo" Su
>>
>>62403991
The kike slayer.
>>
>mommy finally takes over the RTG to snap Jensen and his company in half and show him who's the true king of taiwaneese
>>
My RX Vega 56 flashed to 64. Ignore the clock speeds.
>>
>>62404435
>Vega
Let's call it a big Polaris.
Because it is.
>>
>>62404435
Isn't it 58 though? Or you can unlock the full gpu?
>>
>>62404470
There are no unlocked extra's apart from HBM being 1.25v instead of 1.25v and ability to use more power for overclocking. It was a GPU-Z bug showing extra TMU's.

I undervolt and overclock and for the same amount of power usage and end up with a faster card. The biggest downside is the power usage and heat output. AiB partner cards should be much better on the heat side of things (less noisy too).
I am considering modding it with a Morpheus II Core with 120mm fans for some fun tinkering.
>>
>>62404555
>HBM being 1.35v instead of 1.25v
fixed.
>>
>>62403780
A tweet doesn't make something real
>>
>>62404442
A Polaris that size probably would have been faster. Vega is about 2.2x the size.
>>
https://www.pcper.com/news/Graphics-Cards/AMD-Radeon-Technologies-Group-Raja-Koduri-Goes-Sabbatical

>As we enter 2018, I will be shifting my focus more toward architecting and realizing this vision and rebalancing my operational responsibilities.

Sounds like he's being bumped from VP and Chief Architect to just Chief Architect
>>
>>62402570
This. Can't wait for her to unfuck RTG.
>>
>>62406281
>bumped
he'd be just VPing if it was a bump
good thing really, they should've hired proper manager for the task not an architect
>>
File: radeon.jpg (161 KB, 610x320)
161 KB
161 KB JPG
>>62406281
It's where he excels.

His presentations at capsaicin and others are amateurish and too off the cuff, his industry partnerships are one sided or so out of touch with reality that his biggest "get" is pic related.

His gaming technology basically shifted any and all development from the graphics vendor and forced the industry to learn intricate and complex graphics coding with half baked, and inconsistent tools.

Why this took as long as it did is astounding.
>>
>>62406919
red, epic, bethesda
is pretty huge deal for AMD, they got big win this year at least on that front
>>
>>62403914
They are not losing money on vega cards. Those muh 100 calculations were done by some seriously retarded "journalists". AMD's o mnly problem is their inability to increase supply to meet demand. They AR leaving so much money on the table and nvidia is just walking buy and picking it up by having a supply of 13 month old cards just laying around.
>>
File: rofl.jpg (16 KB, 469x338)
16 KB
16 KB JPG
>>62401809
Praise Jesus. Get rid of the pajeet




Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.