AMD WON
Images are sometimes not shown due to bandwidth/network limitations. Refreshing the page usually helps.
You are currently reading a thread in /g/ - Technology
You are currently reading a thread in /g/ - Technology
IT'S OVER
>>54893678
they look pretty good for budget systems, but wow was that vendor shillfest cringy.
But AMD laptops still only come with single channel RAM.
http://store.hp.com/us/en/pdp/Laptops/hp-envy-x360---15z-touch%C2%A0laptop-t5y05av-1
>A12
Well..This has potential to fail in marketting. Unless it offers notably superior graphics performance.
>>54893678
>one th is different from all the others
Rrrrrrreeeeeeeeeeeeeeeeee
Why are they using the fx name on APUs? That seems confusing
>>54893733
also the A9
THEY RUINED THE EVEN NUMBERS
>>54893782
look for P suffix for POO IN LOO
>>54893782
They are APU now
>>54893807
Didn't even notice the A9.
If it comes with HBM then Intel and nvidia are fucked
>>54893808
only for mobile
Desktop FX won't have it from what we've been told (based on current desktop FX and Zen rumored info)
>>54893808
*APOO
>>54893830
When APUs get HBM I wouldn't be surprised if AMD's stocks jump up above $10. That would really open up OEM opportunities for them. And they would forever be crowned the indisputable king of mobile graphics.
In short. These APUs do not feature HBM.
>>54893857
Zen APUs were rumored to have HBM onboard
>>54893749
You mofo, I noticed that because of you.
>>54893869
These are not Zen APUs.
>>54893678
>still having any sort of hype over Dozer derivatives
My soul was crushed when Bulldozer came out. It is literally the main reason AMD is so far behind right now. Zen better save their ass, I'd buy their 8 core in a heartbeat.
My favorite part about AMD cucks as they continue to wait is that they also think Intel will somehow just do nothing.
>>54893883
I mean for the future.
>>54893678
>Bristol Ridge for desktop apparently canceled as there was no mention of it anywhere. Even Stoney Ridge which is a brand new die is being spoken of as mobile only.
>winning
I mean its probably a good thing since the laughably shit DDR4 performance on desktop would have been chewed up by every single reviewer on the web, but still it means we aren't going to see AM4 early.
The end of the year is going to be fucking harsh.
>>54893733
A12 has been a think since mobile Kaveri.
>>54894025
>A12 has been a think since mobile Kaveri.
Really? First I've seen it.
>>54894036
There were some A10 Pro SKUs for business machines, and one or two variants called FX Pro and A12 Pro. Though they were MIA. Mobile Kaveri had some major issues behind the scenes since even their mainstream 35w SKUs disappeared and they stopped advertising them on their own website.
AMD is still a bit of a mess internally. Carrizo was also supposed to ship 630,000,000 units, and its looking like they're going to fall incredibly short on that.
>>54893678
Not really. The 'dozers kind of suck.
Now a Zen + HBM2 APU, then we'd be talking.
>>54894135
I'd be insane for these to get in mainstream desktops.
>>54894211
I'd be happy with 6 cores and 2048 shaders with no TDP limit.
All unlocked.
>>54894135
Keep your fingers crossed that Raven Ridge actually has HBM on package.
>>54894280
If it has HBM, it will be on Package, that's the only way to implement HBM in a way that makes it beneficial to use over system memory.
>>54894289
You missed the point entirely. Good job.
Raven Ridge isn't confirmed to have HBM, its just a rumor.
How does HBM compare to Intel's EDRAM?
Can't possibly be more power hungry, right?
>>54894325
A first gen HBM module draws 3.5w~ for 128GB/s in 1GB density.
Intel's Crystalwell structure was 50GB/s bidirectional, and 128MB in density. Power consumption is harder to narrow down though all the Iris Pro parts can suck down a fuck load of power vs the non Iris Pro counterparts.
>>54894429
If Crystallwell can be used as a victim cache/L4 then its latency must be notably smaller than HBM
>>54894637
Yes that is true.
>>54894238
you looking at the next die shrink then, buddy
Most you'll see is 4 cores operating at mobile i7 4th gen levels with around 1200 SPUs- about 370x performance.
I would buy a laptop with that APU, and if they don't have it by mid-2017 AMD is finished
>>54893678
I honestly want them to succeed this time. It's not brand loyalty or anything, I just want them to get into competitive scene at least in cheap builds with decent performance. PLEASE FUCKING PLEASE.
>>54894709
>tfw blue balls for 5 years now
hahaha their stocks tanked because investors knew they are selling their new GPU's at a massive fucking loss.
>>54894787
it tanked because they showed off Zen, which will gimp FX chip sales until it launches
>>54894787
The chips are tiny, there's no way they'd be selling them at a loss.
>>54893996
they are about to finish the nvidia licence they have for their igpu and they dont really have any tech to compete with amd
naturally they will ask for licence from amd about it giving another shit dump on nvidia
Doesn't help, cause there's almost no laptops available with AMD chips
Atleast in Yurop, only HPs
they really should have fucking released dual Fury at the same time just to shut everyone who wants "muh maximum powah" up.
I am strongly considering moving off my GTX 760 SLI setup to the 480, just on pricing alone.
>>54895326
They had tons of vendors lined up, but they all end up backtracking. Lenovo was part of an AMD presentation for mobile Kaveri and they promised in the middle of it all to provide a bunch of models featuring the new chips. They ended up releasing like 1 shitheap.
A ton of fuckery was going on with Carrizo too. ASUS and some other vendors openly said they weren't going to sell certain models in the US. I think intel put pressure on them to keep the AMD systems out of the market. They do shit like offer a free Atom for every i5 system sold, or give vendors free SSDs if they sell a certain volume of chips. It'd be super easy to say
>Don't offer an AMD system that competes against this or the free shit will dry up
Theres no other explanation for it.
I just won a nVidia pin in their order of 10 sweepstakes. Will this skyrocket in value as a collector's item when they go bankrupt?
>>54894876
By my math they're selling between $10k and 15k of product per wafer, which could be anywhere from slightly less than half to three quarters margin.
At 6.5 million units thats between 500-750M cash.
>>54893749
TRIGGERED
what is this and why should I care about it?
weren't we hyping zen?
>>54894965
>naturally they will ask for licence from amd
They've been in talks for a few weeks now I believe.
ZEN WHEN?
>>54896376
Summit Ridge at end of the year. They're holding a HotChips presentation on the architecture in a few months.
>>54894018
This is not the future
>>54896396
POWER9 and Zen on the same day.
August 23.
Start counting the days.
>>54897076
80 days until August 23
>>54897076
Desktop and server competition for Intel on the same day?
a good excuse for them to push forward.
>>54893678
>35W
>3.0 to 3.7 GHz
>>54897211
It doesn't maintain those clocks, they won't even maintain their base clock when the IGP is being used. Every single Carrizo chip has the CPU drop down to 1.6ghz when the IGP is being heavily utilized. Even if the cTDP is set higher it still throttles down.
Desktop Kaveri did it too and you could only stop it by using a patch made by a guy on the Anandtech forums. 95w Kaveri with a 3.7ghz base clock was dropping down to 3ghz in normal gaming benches.
>>54897251
Ah that's disappointing. Still, any competition for Intel is a good thing.
>>54893919
AMD would have been fine if there were more multi-threaded stuff at the time they released bulldozer, or at least Vershia
>>54897937
It doesn't change the fact that it's shit. I'm impressed that they've somewhat salvaged that abomination with rehashes, but it's still their fault.
>>54898001
Nothing is unsaveable with enough tweaking. bulldozer is fantastic in theory, and if they had the r&d, could have probably made it competitive with intel and in low end systems, id rather have a decent amd apu then intel, if only because im not doing processor heavy shit on a laptop, but anything that could be processor heavy can get offloaded to the gpu, which amd is stronger then intel when intel does not dedicate a fuck load of die space to ram.
>>54894025
those are still doser cpus, personally i think its smart for amd to make it only mobile, as there is a good chance average people, if they put out a doser based next to a zen, or even same half as a zen, may confuse the two.
that and desktop zen is hitting first with apu zen coming later, something to fill the gap if its better would be nice.
>>54893883
nobody said they were
dumbass
>>54898728
Whatever happened to HSA? I saw some APU years ago beating Intel chips at some tasks.
>>54899177
The spec just got updated again, they're at v1.1. Every programmable piece of logic in a system is addressable now, everything from audio DSPs to FPGAs, and everything in between.
Theres a decent amount of software that makes use of it, and for web backbone stuff JAVA has fully incorporated HSA.
>>54893678
Literally no one will ever own an AMD fattop
>>54895546
Isn't that illegal?
>>54900310
Something is immoral in someones opinion.
>Thats illegal. They cant do that!
Giving the right to vote to non-landowners was a mistake.
where the fuck is muh zen
>>54900310
No, but if there were any contracts signed saying "put this chip in X notebook lines," there's grounds for a civil suit
There probably weren't
>>54900120
I was gifted an HP laptop with an AMD CPU about 7-8 years ago. It started to struggle playing 10bit BD rips on Windows with MPC, so it was time for an upgrade. I wouldn't mind a laptop with an APU nowadays since all of those chips can playback video fine. Pic related, my brother completely destroyed it. RIP ;_;
>>54900715
A series would be okay in an ultrabook form factor but why would anyone pay almost a thousand for a fat laptop with only slightly better graphics than the Intel equivalent.
>>54897251
Die shrink makes a lot of things possible
>>54900880
Bristol Ridge is still cheap 28nm bulk, it is not a shrink of Carrizo.
Pay attention.
>>54900715
> Odd sized system ram.
It hurts to look at.
>>54901048
At least it's not an odd number of cores.
>>54901048
If it's not a power of two, then my autism kicks in.
>>54901217
So, 24GB because it's the maximum amount of memory on an old workstation is bad?
>>54901235
Autism would still kick in, but having more RAM is never bad.
>>54900715
>45 nm technology
jesus man you're just now upgrading? that shit is bigger than some people's dicks
>>54901261
I seem to remember that 2006 MacBook Pro had a maximum of 3GB.
>>54901284
Nah, I upgraded soon after. Like two or three years after.
