There's just no way this won't be yet another mild

I hope it'll be okay.

If it's priced right it could do really well.
Given AMD's track record though- there will be like a i5-7600k (or whatever) competitor at $200, but they only make 12 of them so you can't find it for less than $500

>>58805525
Performance increases of Intel processors have stalled, meaning AMD will eventually get very close even if it takes them a while.

File: 1480386643099.gif (471KB, 512x288px) Image search: [Google]

471KB, 512x288px

>>58805596
Yeah pricing is now all that matters. Here's hope they deliver.

>>58805525
I'm interested in what they've got in the low end. Now that fucking Pentiums have HT and cost $60 it's going to be super hard to compete in that segment.

>>58805636
It never struck you that they've stalled because they have no competition, not because they've hit a wall?

>>58805670
They've been competing with their previous line of products for at least a decade.

If the problem was no competition, intel chips would have built-in self destruct mechanisms to force you to buy the new generation.
Since they don't, and intel CPUs can actually last decades (they're really solid), there's no reason to upgrade for 98% of people for years and years.

You can see this in their attempts to add extra functionality, hardware decoding, even going for the iGPU enchancements. They're looking desperately for improvements that aren't measured in pure x86_64 performance.

>>58805670
It's not the lack of competition, there's too much money to be made in the ultra high-end CPUs. It's physics and engineering. We're at 14 nm anon, shit is just crazy hard and absurdly expensive.

>>58805525
AMD should just focus more on APUs, they are good at that

/̶g̶/̶ will judge it solely by how fast proprietary video games for /v/intendo run on it.

>>58805525
>CPUs isn't a feelgood sports movie
you might say that the RND budget is 1/10th
but this isnt the 80's. Intel and AMD share most major technical feature patents and differentiate them in the package.

for me the real reason to consider amd is the fact that i would enjoy having more cores and the early integration into the kernel which Intel hasnt completed yet even for kalby lake

>>58805525
1) AMD beat Intel before on a fraction of their budget, they could do it again
2) reading news or hype is your own fault.

>>58805752
>early integration into the kernel
What do you exactly mean by this? Is this Linux thing or other else?

>>58805698
>intel chips would have built-in self destruct mechanisms
If they degrade their CPUs the they would have competition retard.

>>58805715
Doesn't matter if you're ahead by 20% or 2%. If the price is roughly the same as your competitor you'll still get 80% marketshare

>>58805777
>AMD beat Intel before on a fraction of their budget
lol
>reading news or hype is your own fault
>just posted hype himself
LOL

>>58805739
>CPUs are optimized for video games. Video games are a huge market---a
market where people pay close attention to performance and adjust their purchasing decisions accordingly. Yes, yes, some buyers pay attention to CPU performance for weather
prediction, or movie rendering, or various other important applications. But most of these applications have the same basic bottlenecks as video games: most importantly, low-precision multiplications and additions.


>https://moderncrypto.org/mail-archive/noise/2016/000699.html

File: certified shit wrecker.png (105KB, 255x345px) Image search: [Google]

105KB, 255x345px

>>58805525
>betting against Keller
Pity the fool

yeah man, I hear you. there's no way in hell that the mythical AMD K7 is going to beat the Pentium 3, let alone the Pentium 4. Intel just has so much more R&D.

>>58805807
you seriously are to young to remember K7 or K8?

>kids

>>58805829
This is the only reason why i have high hopes for Zen.

The Keller doesn't fuck around.

>>58805945
What did you expect from neo-/g/?

>>58805945
That's exactly the point retard. This is technology. Anything that happened almost 20 years ago is irrelevant. Most people who worked on that architecture probably aren't even with AMD anymore.

You might as well say that an American car is going to win Le Mans this year

>>58806027
Guy who designed those chips worked on Ryzen..

That's kind of why we're interested.

>>58806047
>one guy makes a CPU
>still ignores that it was 20 years ago
Just no.

Also, if anything, your example should tell you that AMD has only succeeded with 1 out of 10 launches. So there's a 90% chance Zen will be the same as always
>buh-buh muh hype
no

>>58806027
>Histroy doesn't matter, it doesn't ever repeat itself

You know why K7/K8 were so successful, fuck even why Cyrix was so successful?

Because Intel was stuck with same inefficient core design and was trying to improve it even when it was past it's time.
Same now is happening with i7 core.

>>58806094
Yeah, except he left AMD to go do other things. He hasn't worked on anything for AMD since those days, but has been working on microarchitecture designs for other usage.

Even if he gets AMD to Haswell like IPC, that's honestly good enough given they're looking to do serious 8core chips.

I'm not even hyping this, I just want to see what lands. What irks me is people making absolutist comments.

Intel have hit a wall, everyone is going to catch them in time. That's always been the way this was going to work out. It's no like Intel are still charging ahead with the designs. Even top of the line ARM chips are starting to compete with lower end X86 designs.

>>58805791
>What do you exactly mean by this? Is this Linux thing or other else?

not or other else but yes integration into the linux kernel like the one you find on kernel.org

are you a flint bathwater taste inspector or whats wrong with you sentence?

>>58805715
>We're at 14 nm anon, shit is just crazy hard and absurdly expensive.
Greyhawk is due in 2019 7nm.

I really am tempted to go with ryzen, mostly on the hope that i can get at least two CPU generations out of it.

>intel hit the wall in x86
>amd playing catch-up
>people keep forgetting that the battle is in RISC-based mobile SoC since 2011
>area where samsung, apple, and qualcomm dominate, and intel struggles upon

Mmm

If it's cheap and can deviler i5 2500k levels of performance then I'll have no reason to use an intel cpu until they come up with something that's actually better and not some meme 1% improvement

>>58806196
>giving a shit about mobile at all

>>58806196
Yea, battle to catch-up with x86.

>>58806196
Intel really does struggle with that due to the complexities in their x86 implementations.

The benefit they have is that x86 is the defacto PC standard. Even if they keep being shit, they have the benefit of being what everything has been designed for.

If ARM or whatever RISC type chips you want to mention are going to take over, it's going to take a really long time.

>>58806213
>not giving shit about mobile
>failing to see how mobile is eating into intel's and amd's profit margins and market share a couple of years now
>the server market, that are intel and amd's major clients, are moving onto low-power ARM from x86

>>58806196
remember surface? the one with risk-socs? failed hard and forgotten. as long as nobody can produce a competent risc desktop replacement = something you can build software on without feeling like a retard x86 will be around and your statement is the same as saying facebook games will overtake the gaming market because they have volume of consumers

>>58806257
>the server market
lmao

>>58806122
see >>58806094 again

>>58805525
did kevin sorbo every failed on creating any cpu/gpu?

NO never did

the fact that intel is already sampling new cpus is enough indication that ryzen is going to be good shit

>>58806281
Saw. Replied.

The assertion that 90% of launches were a fail is bullshit. AMD were absolutely competitive before the P4 days at which point they got out marketed and outplayed by Intel.

I am still not hyping this. I'm just waiting to see. I'm just saying that the assertion that it's going to be a failure because "my hypes" is fucking moronic.

I'll just wait and see.

>>58806309
>Replied.
You just repeated your previous shit about this one miracle guy being able to turn AMD around by himself.

>>58805829
based pussy destroyer

>>58806269
>desktop replacement

Everyone's using their phones and tablets for basically anything right now. Compared to power users, normies will only use a computer for social media, watching videos, listening to music, playing games, and do basic productivity, which a phone or tablet can sufficiently provide.

Give a normie a tablet or a phone, and they won't ask for a traditional desktop. Power users are few compared to the normie market, which sadly dominate the consumer computing market.

>>58806364
Reading. It's hard.

I didn't even say he was going to turn it around, merely that that was the only reason it was interesting.

The point I was making was that his releases to date have been the 10% (in reality more like 30%) worth using. Could well pull something out, may not. Either way, worth waiting to see.

>>58805525
>1. Retards forget that making CPUs isn't a feelgood sports movie where the underdog comes back in the end. It's about science and enginering and if your RND budget is 1/10th then you're not going to compete in all areas.
True. The only time AMD were really beating Intel was when Intel were trying to polish the turd that was the Pentium 4.
K6 and the first Athlons were mostly more bang for your buck CPUs.
>2. The same people have hyped it up well beyond realistic goals. They yet again think that it's going to BTFO intel and make AMD market leader overnight
Also true. People want it to be real so hard and they'll be disappointed yet again. Intel and Nvidia will continue to dominate the high end.
Still, I can't wait to replace my FX6300, and I'm hoping for a good bang for the buck 4C/8T CPU. Fuck Intel I'm not paying a ton of money for a fucking dual core CPU in 2017.

disappoint is only possible if you're hyped, ie are not a true /g/ent but a /v/ag.

sure, P4 made K8 great, but K7 was han honest all out win, even though it was not much better than P3.

the last can certainly happen again. ryzen doesnt need to beat current intel, it just needs to match it again, as done with K7. all the signs point to that going to happen, which is great. ofc, benchmarks will provide proof of one or the other.

>>58806094
>There's a 90% chance Zen will be the same as always

Just like the chance that this post wouldn't have dubbs

>>58806417
>1 guy is the difference between an architecture being interesting or not
see >>58806094

>>58805797
If no one had proof that they self destructed would they still have competition?

>>58806580
>just no
is just not an argument

>>58806602
I'm not saying "just no". I keep repeating the same shit because that's what you're doing

>>58806590
>computer stops working completely
>no evidence ("proof" is wrong)
nigga wat?

>>58805670
>It never struck you that they've stalled because they have no competition, not because they've hit a wall?
Their 10nm roadmap got delayed THREE times now, so they're totally holding back until AMD catches up :^)

>>58806415
>Everyone's using their phones and tablets for basically anything right now. Compared to power users, normies will only use a computer for social media, watching videos, listening to music, playing games, and do basic productivity, which a phone or tablet can sufficiently provide.
That may change as VR and AI begin to really ramp up and produce content that consumers will salivate over. You're not going to be able to have a holodeck or strong artificial intelligence agent with a shitty mobile cpu.

>>58806309
AMD was definitely competitive in the P4 days. AMD only really have had trouble with the K10 family. They did the same mistake as Intel did with the P4, they made the pipeline too long, trying to make really high frequency chips. I bet they learned the same lessons as Intel from that.

>>58806094
>every project keller has worked on is a success but it doesn't matter because I say so

Getting hyped because of this reason isn't that logical but it's good news regardless of your preemptive damage control.

>>58805525
Your arguments are invalid because of Keller. AMD will wreck Intel so hard that they will go bankrupt and finally AMD will buy Intel.

File: 27639.jpg (11KB, 184x154px) Image search: [Google]

11KB, 184x154px

>>58807203
Why would anyone want AMD to buy Intel?

>>58805525
1. AMD will not get the performance crown, they hope to give a real budget option, something like 75%~90% of the performance at 60%~80% of the price.
2. Yes the r&d budget of intel is huge but they also started to jew since the i7-2600K era, meaning with every gen you dont get 10% performance you get 2% with steadily increases in price.
3. I dont really know why you fagots love the idea of intel dictating prices and shit, it is baffling to see how much you want to get jewed over and over again...

File: 2QGTn3g.gif (162KB, 404x416px) Image search: [Google]

162KB, 404x416px

>>58807203
>AMD will wreck Intel so hard that they will go bankrupt and finally AMD will buy Intel.
Yeah, right after the modern day Joan-of-Arc Sarah Silverman leads the military coup overthrowing Trump.

>>58805525
For every previous AMD disappointment, we got benchmarks a few months before release that showed how bad they really are.

Zen only got benchmarks so far that shows that it is actually good.

And even so, it was already delayed about a year, so it's not all roses.

>>58807292
>it was already delayed about a year
What the fuck, no.
The earliest that was ever officially claimed was "limited availability by the end of 2016".
Limited as in samples available to OEMs and other partners, which was actually somewhat correct.

>>58807087
>AMD was definitely competitive in the P4 days.
If there ever was an understatement, this is it.

Intel shit got wrecked so bad after that nobody remembers the ViiV and the netburst

>>58805596
>i5-7600k (or whatever) competitor at $200

except it will be at least 10% slower with 2x cores, and overclocking may be really slim vs 4.6-5ghz on i5 7600k.

>>58807475
>IPC shown to be on average almost on par with Intel
>B-B-B-B-B-BUUUT MOOOM IT WILL BE SLOWER WITH TWICE THE COOOOOORES!!!!!

File: rx 480 cf vs gtx 1080.jpg (104KB, 1600x905px) Image search: [Google]

104KB, 1600x905px

>>58807531
>>IPC shown to be on average almost on par with Intel
>ALMOST ON PAR WITH INTEL
>ALMOST ON PAR
>ALMOST

>>58805525
Rumors point to 700$ MSRP for the top one. AMDfags might deny it but it's DOA for sure as even if it sells , intel is just going to drop skylake X , especially considering AMDfags were expecting 500 or less for the top one, which makes it out of reach for all the dreaming idiots who expected anything less than 700-750. It will most likely sell , but I wonder if it will generate lack of trust for future releases. Though if you ask me, I won't be buying it and I believe you also shouldn't.

>>58807203
I sometimes think that Intel is purposely slowing down to let AMD catch up otherwise Intel will be the only desktop CPU maker and get slammed for being a monopoly

>>58806951

They don't need to have a powerful system. The end goal is to stream all of that VR shit to the consumer.

>>58807548
The fuck are you on about?
>a fucking midrange GPU for 1/3 the price you can crossfire together for double the performance
>is now the same as a CPU
There are people pathetically grasping at straws, and then there are you.

>>58807589
>Rumors point to 700$ MSRP for the top one. AMDfags might deny it
Well yea, its a rumor that has no basis, who would believe it?

>>58807400
still competitive with phenom ii.

File: 1486028322785.png (4MB, 1543x1615px) Image search: [Google]

4MB, 1543x1615px

>>58807660
>>a fucking midrange GPU for 1/3 the price you can crossfire together for double the performance

only in single game it has 100% scaling, in 99.99% of games it has barely 60-70% scaling

>is now the same as a CPU

Amd has bad track of exaggering and just lying to customers - rx 480 bullshit crossfire 100% scaling, "Plentiful supplies of AMD Radeon RX 480 cards are reported" and many more with bulldozer/phenom/athlon. So it may be the same case for zen. IPC improvement may be just bunch of made up lies. But you are too naive to spot that.

>>58807647
I don't think you realize how much bandwidth that shit is going to use.

>>58806027
The demented people such as yourself that come forward screeching and and calling everyone retards, are the grand retards yourselves. Fool.

CPUs are basically the same thing for a LONG time now, only basically incremented and optimized.

The x86 architecture is very old as are many instruction sets, and they're still here, only improved.

If you think that the cpus of some generations ago have nothing to do the ones now, that is, its a entirely new game, you are the drooling retard here, talking about shit you know nothing about with a sense of self righteousness. You are beyond saving.

Kill yourself.

>>58807760
>only in single game it has 100% scaling, in 99.99% of games it has barely 60-70% scaling
Depends, obviously. That should be clear to everyone using two GPUs, be it SLI or CF.

>Amd has bad track of exaggering and just lying to customers - rx 480 bullshit crossfire 100% scaling
Exaggerating yes, I give you that. Not that this is this AMD exclusive, though.
I can't really remember a single instance where they flat out lied about the performance.

>Plentiful supplies of AMD Radeon RX 480
They actually had a lot better availability than most of their previous launches.

>bulldozer
They showed benchmarks of programs the architecture was good at. Again, nothing wrong.
I can't remember their power consumption claims and whether they were correct or not.

>phenom
The TLB bug is the only thing that comes to mind with Phenom. That's neither really lying nor exaggerating.

>athlon
Can't really remember any false claims, assuming you mean the original Athlon.

>>58806951
>vr
>ai
All memes. You're a meme master.

>>58805638
it's going to be like >>58805596
said, AMD retail price will be lower than Intel but you won't be able to find it for that price

I'm sitting on a shitty old i5 4670 and I will be upgrading this year. I would love to live in a world where I though AMD was not going to shit the bend and I could finally buy a good cpu from them for a fraction of what intel is asking.
I live in the real world however and I know by the end of the year I'll probably have a i5 7600k

>>58807760
>in 99.99% of games it has barely 60-70% scaling

You're being too generous. There are many games that don't support SLI/Crossfire at all, or have much worse scaling than that.
Multi-GPU support among developers is terrible and getting worse.

Because the outside is cold so kids stay indoor playing with PC.

Or a more serious note, they have less red tape. People are more open to talk about piracy and hacking.

nvm the above post, wrong tab.

>>58805525
amd is going to fuck up with the prices being too expensive

screencap this

File: 1480858901324.jpg (74KB, 627x627px) Image search: [Google]

74KB, 627x627px

>>58805829

>anon doesn't follow CanardPC's twitter

Always bet on Keller.

>>58808718
wrong thread dipshit

File: Zen team.jpg (75KB, 715x573px) Image search: [Google]

75KB, 715x573px

>Its a "butthurt fanboy concern trolls and tries to spread FUD" thread
Oh boy, haven't seen plenty of those before.

The top end 8c/16t Ryzen SKUs performs comparably to the i7 6900k.
Deal with it.

>>58807980
While I agree to a point (performance for example) there have been significant changes regarding power saving and energy use in general. Also other baked in features like intel management stuff. Alot of new stuff. Better use of hardware crypto junk and cache configurations.
To say it's a whole new game isn't entirely correct, but to deny it also shows your ignorance. You are probably just a deaktop user that doesn't care about those features and that's fine, But age of the instruction set (which has been modified, and improved upon quite a bit over its existence)doesn't mean the game hasn't changed.

>>58805970
has he designed something bad yet?

>>58807292
the only indication we ever had was it could have come mid 2016 based on tap out times and jims departure, on something amd said.

>>58811637

No.

>>58807760
that chart was made with 87% scaling,

>>58805525
Considering my firsthand experience with how broken Intel is as an organization (moar meetings!) I would not be surprised if a company with 1/10th the resources could build a competitive product.

>>58805829
He looks like he could be Kevin Sorbo's fat nerdy brother

>>58805525
Seriously, Intel, stop it. Your bullshit shilling is going to drive people away. Intel's shit is just rehashes and has been for YEARS now. Kaby Lake's big new feature is that it can authorize media for you. The last couple rehashes they had before that offered what, a 1% performance increase over their previous efforts. They're using consumers to subsidize lower-powered server chips.

Intel is also botnet as fuck.

>>58806112
This. Intel abandoned the P4 architecture and went back to rehashing the P3. Today's Intels are just steroidally-enhanced P3s.

They are literally working with a 20 year old CPU core at this point.

>>58806122
>IPC
The important thing will be the spec benchmarks. As Intel shills get more and more desperate, expect them to avoid talking about overall performance and just focus on one bullet point.

>>58808800
GET

>>58806172
Same year the next Duke Nukem comes out.

>>58806227
>it's going to take a really long time
Not true, when Apple releases their first ARM laptop it will be a done deal and within a year or two Intel will be relegated to server room tasks and Windows gaymen.

>>58806269
>competent
It was fine, what killed it was the fact that 99.99999% of Windows software would never run on it and most companies couldn't be assed to compile their software for a platform that Microsoft barely paid attention to. It's dead now because of MS, not because non-Intel architectures are bad.

Intel's only good at x86, remember the i860? What a shitshow that was, Intel has been cowardly ever since about true innovation.

They even wanted you to buy into another dead architecture to do 64 bit computing, the Itanic, but that failed too. AMD fucked Intel in the ass with 64 bit.

>>58809878
where the shitwrecker at???? :(((((

But seriously, how involved with Zen was Keller actually?
I heard many times that his (main) focus was actually on K12, their ARM "Zen" core that has been put on hiatus right now.

>>58808800
>8088
>00
Neis

>>58805525
Intel shills have been sweating and shaking a lot recently.

Kek, Ryzen is making them piss their pants before it's even out.

>>58805777
with the same engineer, too.

>Oh look, Intel is freaking out about Ryzen and is sampling new CPUs in panic. We'll tell you more about it Monday. (Sample incoming)

>But yes... I saw internal presentations dating last fall, where Intel was explaining that Ryzen was a joke

W E W
E
W

>>58815466
It's really funny to see them make the same threads over and over, they're triggered for sure.

>>58815440
Jim Keller was AMD's head of arch development. He was in charge of the teams developing both Zen and K12. He was the big boss running the show.

>>58815323
If I have a correct understanding, intel has 14-16 pipeline cpu, and amd has a 10, this should mean that amd can get shit done faster with future revisions to boot.

>>58817739
It's all so much more complicated than that these days that I'm waiting until the Spec benchmarks are done. My guess is, though, that they will probably cream Intel.

/g/ will be disappointed, yes, but the general public won't

Ryzen won't be a match to Intel's CPUs but it will give customers a much better price per performance ratio that what we're used to right now
It'll be a success regardless but /g/ will see it as a failure forever, there's no chance it will be seen as a success by anyone here

i think it'll finally bring their cpus to the level of their gpus. almost as good as intel for a little less. almost as good as nvidia for a little less. the end.

if they can get a decent price performance ration that would be good for most pc users. having the best intel cpu just leaves you with a bunch of unused potential and wasted money. bang for the buck is where it's at.

>>58817863
>you don't need a car that does more than eighty because speeding is illegal and you would never do it

>>58817791
its both more and still just as complicated depending on how abstract you want to go with it.

>>58805636
>Performance increases of Intel processors have stalled
>"I'm not fast enough because my competitor always far behind me!"

Whoa man i can't explain how much brainfart did you insert into my head.

>>58805970
>>58805829

But jimmy left after design the foundation for zen, you'd think the other AMD heads have enough brain to continue for the next 5 years ?
Foundation is there but it's up to the others to build what is right on top.

>>58818192
High level arch for Zen was finished before 2015. In may 2015 the work on Zen+ was already under way.

>>58805525
You know how celebrity deaths always come in 3s? Well so do comebacks.

1. Trump
2. Patriots
3. AMD

Screen cap this post

>>58805829
Is that Alex Jones's brother?

File: 1480601274569.jpg (78KB, 676x693px) Image search: [Google]

78KB, 676x693px

>>58818666
>patriots
>not Cubs
Damn new englanders

>>58805525
The reason they lost last time is because they bet on the donkey (more cores) but people wanted the horse (fast cores). Now they bet on both.

>>58815323
Is this like the Gundam Wing meme where the old mechs are actually superior?

>>58806227
I wouldn't be suprised if ARM will die in the next decade because when people switch it doesn't matter whether they switch to ARM or some other new ISA. The amount of effort is the same.

>>58815401
The problem with the first Surface was that they managed to sell 1 million devices but produced 2 million devices.

>>58805525
cpus will be fine. but the gpu will be a massive dissapointment.

1080 performance a year too late.

>>58815555
> It's really funny to see them make the same threads over and over, they're triggered for sure.
qft - quads for truth

File: 902025-iolaus.jpg (26KB, 278x420px) Image search: [Google]

26KB, 278x420px

>>58813557
>tfw no Iolaus

File: 1362180638885.gif (433KB, 500x275px) Image search: [Google]

433KB, 500x275px

>>58820809
Doesn't matter if it's significantly cheaper. I want a card that will allow me to fully utilise my 1440p 144 Hz FreeSync monitor and right now there isn't really a great card for that. GTX 1070 and 1080 don't have FreeSync (thanks for your £200 GSync tax, nVidia) and all of AMD's cards are a bit underwhelming. The 390X and 480 aren't quite fast enough to be future proof and the Fury lacks VRAM.

I'm hoping for a £300 card that is at GTX 1070 level performance with 8 GB of HBM2. Probably too much to ask for though.

>>58823771
>a £300 card that is at GTX 1070 level performance with 8 GB of HBM2
http://wccftech.com/amd-vega-10-20-slides-double-precision-performance-1500-mhz-vega-10-x2-2017/
Maybe Vega 10 can help you out there, just not sure whether it will cost you just £300 or below.

>>58807548
silly AMD this is not how crossfire/sli works...

File: lisa-su.jpg (37KB, 887x634px) Image search: [Google]

37KB, 887x634px

I am glad you guys are finally seeing the progressive light and supporting companies that is openly against Trump and his fascist views. Stop supporting Intel and their oppression of the Palestinian people! Stalman would be proud!

>>58825432
Intel donates to Feminist Frequency.

File: 1484357505875.jpg (291KB, 863x752px) Image search: [Google]

291KB, 863x752px

>>58825432
And AMD is supported by terrorists you false flag operative.

Companies shouldnt play politics

Despite Google trying to spy on me at every-step, I hadn't chose to change search engines and email providers until there latest stoop to enforcing their political will over others.

>>58825432
She wants to keep her Pajeet wage slaves for a little longer. Too bad they're gonna go back eventually thanks to this
https://www.law.cornell.edu/uscode/text/8/1182
That lawsuit is largely symbolic and even the 9th Circuit Court of Appeals knows that they're only stalling for time against his EO.

>>58805525
>yet another mild disappointment from AMD

By my count, they've knocked CPUs out of the fucking park a few times, most notably the Athlon and the Athlon64. My impression is they're doing their best to get back to those days, and erase the mistakes of that fuckbag Rory Read.

>>58805525
Well shilled.

>>58806580
are you implying that there aren't any engineers who are a cut above, say, an average PhD? surely there must be an average among even Ivy-League PhD's right? is it just impossible for someone to be exceptionally above that? is it because he only holds a BS from Penn State? what exactly is the basis for your argument?

>>58808163
do you realize how much money is being put into VR development by massive corporations? on a very subjective level, i think it's fair to say this could be similar to the advent of motion pictures: people love to pay for spectacles that feel like real experiences, no pun intended.

>>58831786

Decent VR needs as much $$$ as a 1440p gayman rig. The prices are going down, but it will be a while until it's mainstream. Until then it's as silly as gunning for 4k/30.

>>58815349
lol

>>58815365
I imagined a homosexual team in Microsoft put together because their all being gay somehow was strategically prudent.

>>58807270
Happens everyday. Many such cases!

>>58805829
I always knew Alex Jones has a side project aside from being a music producer.

>>58805525
>There's just no way this won't be yet another mild disappointment from AMD.
How much does Intel pay you to make these threads?

>>58807087
K10 was OK in terms of IPC, but had terrible clock speeds.
K10.5 was pretty good, had competitive IPC and overclocked really well. But it was just late to the party.

Bulldozer is where the massive problem lies. The architecture just fundamentally cripples IPC too much. It was a radical experiment that didn't live up to what they hoped.

>>58818689
Maybe it's like that movie with Arnold Schwarzeneger and Danny DeVito, where in this case Alex Jones gets the shit genetics.

>>58830257
It wasn't Rory Read that fucked AMD. It was Hector Ruiz.

>>58832965
They both did their part.

>>58833060
I'm pretty sure Rory wasn't even supposed to be the permanent CEO. He was only brought in to clean out some of the bloat within the company, and then left when he was done with that.

But Hector Ruiz? He became CEO when AMD were at their prime, with the Athlon XP doing well and the Athlon 64 about to release.
Then, he caused some of their best engineers to leave (Jim Keller, etc) by trying to make them take a pay cut, then basically sat on his arse not caring about a successor to the Athlon 64.
And when he inevitably got caught with his pants down (Core 2), decided to sell off a bunch of assets, such as selling their ARM division off to Qualcomm not long before the smartphone market took off, and selling off manufacturing to some arabs.

>>58833156
>I'm pretty sure Rory wasn't even supposed to be the permanent CEO. He was only brought in to clean out some of the bloat within the company, and then left when he was done with that.
The "bloat" he cleaned out was good engineers that warned against releasing the garbage that was Bulldozer, then further reduced the company's diversity and flexibility.

Lisa has spent the last 3 years restoring that. She is the only reason that AMD is still standing, and to be honest probably the only reason Zen is seeing the light of day. Rory was all about cutting costs and cutting corners, and that meant cheap, small cores (bulldozer) and low cost chipsets brought forwards from first generation bulldozer boards. It meant throwing away millions on bad deals with GloFo, and using their garbage nodes even when they leaked so badly that 125W-designed chips sucked down 150+W. Had he still been in control, Zen would've just been an even-more radically upscaled Bulldozer. Half the dev time, half the cost, half the performance.

>>58833259
> The "bloat" he cleaned out was good engineers that warned against releasing the garbage that was Bulldozer, then further reduced the company's diversity and flexibility.
It takes about four years to develop a new architecture from scratch. Bulldozer was already so far into the pipeline they couldn't just not release it by the time Rory became CEO.

>>58833281
Just looking at the timeline, Dirk Meyer was CEO from 2008 until August 2011. Bulldozer released in about October 2011, just two months after Dirk Meyer was kicked out.

>>58833281
>>58833302
From Wikipedia: AMD began planning the Zen microarchitecture shortly after re-hiring Jim Keller in August 2012.
Rory Read was CEO from August 2011 to October 2014.
So basically, he was CEO for the first stages of Zen development. And the high level goals and design would have been set before he left; You can't just drastically change how the design will go from there.

>>58805525
AMD fucking destroyed Intel in the early 2000's there is no reason they can't do the same again.

>>58833355
However, I do think Lisa Su is definitely a good CEO for AMD. The difference between Lisa Su and Hector Ruiz is massive.

>>58833366
>Germany destroyed Russia in 1914, there's no reason they can't do the same again in the 1940s

>>58805525
Retards forget Intel is under a leadership that makes AMD's worst management look godlike

>>58805643
The reason Pentiums got HT is because AMD is scaring Intel hard
>>58806027
Ford won LeMans after a 40+ years hiatus senpai
Also, literally the most important people go designed K8 were back for Zen
>>58807980
Internally no CPU runs x86, everything gets decoded to micro ops since a long while ago

>>58818192
Since the current AMD head brought Keller in in pretty sure she won't fuck up anytime soon

>Intel Reacts to Ryzen with new Kaby Lake Core i5 7640k and i7 7740K Processors

who got kaby here? how pissed are you?
i5 with HT, this is hilarious

>>58833398
>Germany destroyed Russia in 1914

germans were three countries away from russian borders during ww1, the stupid reparations was rush and stupid Lenin move

>>58833615
fucking the days of Pentium all over again

>>58833615
>i5 with HT, this is hilarious

What's wrong with that?

>>58834147
>>What's wrong with that?
Nothing, it's what we should have gotten years ago but are only getting it now since Intel's ass in on fire from both AMD and their retarded CEO

>>58834153
ah yes. God damn I'm not sure if I should upgrade my 3570K if Intel is just now going to bring the real stuff on the table

>>58805525
>The Zen core does have some limitations that make it less suitable for scientific computing, which accounts for 15–20% of the server market. It sacrifices floating-point and SIMD throughput to reduce area and power

>128-bit wide SSE/AVX unit
what now you flaming faggot?

>>58834239
Good thing they didn't skip on the integer units or they'd drop out of 80% of the market!
Also it's a good thing that 2x256bit fpus have a real chance to be in Zen+

>>58815365
Bruh Apple might do that for the Macbook 2017, but not the Macbook Pro. Apple's research has a long way to go before they can be at Intel's standarts.

>>58805797
If they would self destruct, they would have increased performance so as to keep amd off the competition (at least in the low tiers) and they would have kept selling a fuck ton of cpus (which would have been better for them).
The fact that now, after so many years, they push for the pentium with hyper-threading makes me believe it actually stalled.

>>58834239
Which real-world workloads require a wider than 128-bit AVX units?

>>58833416
My sister works for Intel, she has to sit through speeches from the CEO about how the latest company they bought and wasted money on its going to be the future. She is not impressed with management.

>>58834239
>>58834301
>>58835622
> 256b AVX

I'm not convinced that they will bother, since
> most processing is integer based and doesn't need it whatsoever
> they already do GPGPUs for dumb-as-bricks FP calculations
> MCMs with Zen will likely come out later this year
> Intel is starting to push AVX-512 with KL for complicated simulation loads anyway

It probably makes more sense for AMD to cede the tiny fraction of the market that needs beefy non-SIMT fp32/fp64 to Intel while expanding the heterogeneous MCM ecosystem.

>>58837172
the FPGA thing?

>>58838104
FPGAs on dies address a small but VERY profitable part of the market

>>58823888
Yeah we'll have to see. Because AMD will be a year late at getting any cards out with 1070/1080 performance, I'm hoping they'll be decently cheaper.

>>58838531
If Vega 10 is 1:2 FP64 then it's not all that late at all, the Nvidia GP100 PCIe competitor is only releasing in March.

As for gaming cards, if Vega 10 is based on the HPC chip then it's no wonder it's late for the gaming market since AMD aimed at lucrative server contracts before gaymers.

File: AMD-VEGA-10-VEGA20-VEGA-11-NAVI-roadmap.jpg (222KB, 1562x855px) Image search: [Google]

222KB, 1562x855px

>>58838575
Vega 10 won't have half-rate fp64.
That'll be Vega 20 in 2018.

>>58839774
Then AMD is just slow, hopefully the NVRAM thing pulls them through for a year.

>Intel reducing prices for their CPU drastically i5-7600k $214
>Nvidia GTX 1070 now in the low $300($309 approx.) range
Should I buy now or wait for Ryzen?

>>58840093
it's not even a full month anon, just wait those three weeks

>>58840223
I don't think a VEGA GPU will be faster than an equal priced GTX1070, not in a million years. Should I just buy the 1070 and wait for Ryzen?

>>58840300
>Vega not faster than a 1070.

Nigga what.

>>58840300
There's no way it won't beat a 1070. 1080 is more of a question but they'll have to price them appropriately anyway, especially given that AMD's cards will be a year late and may have to compete with a 1080 Ti.

>>58806027
Corvette did really well in 2015. Passed cars well above its tier.

>>58840300
>>58840684
Have you dumb niggers still not realized Vega is not the high end counterpart to Polaris but a whole new generation with a completely new architecture?

Even the early sample chip AMD demoed was faster than a 1080.

>>58807844
not to mention latency issues

>>58840300
Vega is literally at least twice the size of Polaris 10 and a new architecture with massive improvements on top of it.
It should have zero problems competing with a 1080, even overclocked.
The issue is how close will it get to a 1080Ti and how much will it overclock.
You can screenshot this and post it when it comes out.

Anyone know what Mhz RAM I should buy for Ryzen? Newegg is having a pretty good sale on RAM right now and I don't want to miss it.

>>58840300
Nigga Vega 10 is a 4096 shader HBM2 monster running in excess of 1500mhz. The question isn't whether it competes with the 1070, but whether it competes with the 1080 Ti. If anything Vega 11 is the expected 1070 competitor.

>>58840923
Wait for reviews to find out, high speed memory should have a notable effect on performance if AMD's memory controller is competitive.
And it probably is because they couldn't come close to the server market without good memory bandwidth.

>>58840933
>Vega11
>entry level shit competing with one of Nvidia's best GPUs
lol the delusion is hilarious with AMDfags.

>>58841178
Vega is an architecture built around HBM and fp16 performance, there's no 'entry level' Vega chip.

Vega 11 will be around 3072-3457 shaders.
That is enough to strike at the 1070.

>>58841178
you are just clueless about technology. Please lurk more

>>58840933
>10% faster than stock 1080 in OpenGL in best case Vulkan benchmark of doom (went equal with an oced 1080)

That puts Vega 5 to 10% behind 1080 in your typical dx11 games. Now even if the performance would increase with drivers we talk about "slightly faster than 1080" which is still 30% behind Titan X you retarded autist.

>>58841306
>demo 7 months from launch running Fiji drivers with the exhaust vents taped so the GPU was melting in its own heat

Oh yeah a legit comparison.
Do you think AMD will give you concrete performance figures HALF A YEAR before the release?

>>58841242
>muh bigger numbers
lol retard, if it was about that AMD would have been ahead for the past 15 years because their "architecture" is just "pile more shit on top of this 8 year old chip"

>>58841370
>Muh magic meme drivers will increase its performance by 100%

You can't make this shit up, /g/ is most illiterate or most hopeful tech board on planet Earth.

>>58841404
Your shitposting might pass on anandtech forums, try there.