Zen is a flop

6700k - 188 on cinebench
Zen - 179

AMD's supposed trump card has been exposed. You can all submit to the Intel and Nvidia jew. Welcome to reality.

Add in the cute anime girl bonus multiplier, and AMD is way ahead.

No source. Your thread is pointless

His source is anime, anything is possible.

>comparing a low clocked prosumer chip to a high clocked mainstream desktop chip

Kids these days.
If 3ghz~ Summit Ridge actually compared that well to 4ghz Skylake then intel would be shitting their pants.

Can't wait for Zen. Hopefully they'll price them to undercut Intel.

I'll then pair it with a GTX 1060, total zen.

So 5% slower IPC with more cores and I/O and lower price?

Anyone remember the bulldozer hype? 8-cores for the price of an i5, great performance, bla bla bla. Biggest flop I have ever witnessed in recent years. Phenom II was faster most of the time.

Well it was more of an issue of the software not being properly multithreaded than the hardware. In proper compute benches that utilized all 8 cores properly Bulldozer did pretty well, but most games back then relied pretty heavily on a 1-2 threads with some small stuff crammed onto other threads. AMD's own drivers were very heavily single threaded and Nvidia's weren't much better. Now with DX12 and Vulkan out people are noticing that Bulldozer chips are suddenly giving i7's a run for their money.

CMT was an all round shit design, don't try and defend it

AMD is so fucking incompetent these days, they are seriously dead.

Creating hardware that software can't use is stupid. Buying hardware that software can't use is worse. I don't know why AMD is going with it.
Look at their GPUs. If it's not DX12 or Vulkan they are behind because GCN was made with async in mind. But async is only now becoming somewhat relevant.

Except the plan is packaging Zen with the new GCN cores and HBM as a multi-chip module to make a run for the mainstream market, selling what's just shy of a full SoC that can play MOBA's comfortably. Neither Nvidia or Intel are going to have an answer for it by next year.

Bulldozer pulled ahead vs 4 core Sandy Bridge in select workloads because a physical core has better scaling than a logical thread. AMD didn't originally design Bulldozer to only be useful in these scenarios, the original incarnation of the arch was nothing like what launched in 2011. They simply failed to meet design deadlines, management rushed everything, and it was pushed out the door entirely unfinished.

CMT as a concept wasn't the issue, just AMD's execution of it.
The Bulldozer module was barely smaller than two K10.5 cores in Llano, and had considerably worse perf/clock especially in FPU throughput. They wanted higher compute power per mm2, and ended up going backwards.

Async has literally nothing to do with GCN's gaming performance in DX11 and other older titles.
AMD's arch for all their larger design is primarily limited in pixel throughput.

Why not a 490 when it releases?

This is the 480 all over again
No-one with a braincell expects it to be better than Intel's high end.

We HOPE that it'll be like 90% as good for half the price.
I expect that it'll be 60% as good for half the price.

> 490

WAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA

RX 490 is rumoured to be a dual card. Pretty much 2 RX 480 slapped together.

So screw up like the 295x2? For two 290 hawaii chips it should be far more powerful, but this is MAD we're talking about

>a new architecture is just two Polaris cards slapped together

stop

I want Amadaposters to get off this board.

Baseless WCCFtech nonsense.
That isn't going to happen.

>new architecture

I dont think the 490 will be vega my dude. That will nore than likely be the new fiji in the "fury II" and "furyX II" or rage or whatever it will be

It's not like it never happened before. Trust me, I don't want it either. AMD needs to get it's shit together and give nVidia a run for it's money.

> 6700k
It's well known that Zen will provide Haswell IPC for a competitive price.

Amada Kokoro? More like AMDmad Kokoroach!

AMD is AMDead

what ur forgetting to factor in is that zen has 2048 cores

>CMT as a concept wasn't the issue, just AMD's execution of it.
Actually it was an issue. Why would you do that when you could do the shit that Intel calls HyperThreading for cheaper and achieve like 70% of the benefit?

SMT doesn't provide a consistent uplift in throughput, CMT does.

Don't try to discuss architecture when you've never taken a class on it.

Nobody cares about your shitty preferences, dumbass. Real world benchmarks show that CMT is a dud. AMD spends like 2x the transistors on the same performance that Intel gets in practice.

Are you retarded? The 295x2 is a timeless card that's still at the very top tier of performance in most settings. A dual RX480 would be extremely close or better than a 1080 for less money.

Nothing I stated has to do with any "preferences" you tech illiterate child. I made a strictly factual statement.
CMT provides a consistent uplift in throughput because its physical hardware. Both CMT and SMT require additional die area, and SMT doesn't consistently increase performance in all workloads.

There is nothing wrong with CMT, just AMD's implementation of it in the Bulldozer family.
Stay in Sup Forums where you belong, uneducated child.

Now tip your fedora you autistic faggot

>tech illiterate retard gets called out for talking out of his ass
>hurr durrrf muh memes

Again, stay in Sup Forums where you belong.

Your factual statement is a statement of your preference, because it has been proven time and time again that this factual benefit does not materialize in real life.

SMT requires almost no die area when compared to CMT which practically builds a second processor core, you might want to learn something about this before you run your mouth.
AMD reduced the number of integer units to 2 per core from 3 to keep die sizes from growing huge. Intel still has 3 integer units and way better per core performance. With hyper threading active, Intel still has 1.5 integer units per thread.

Using SMT is a no brainer, because it is additional performance for minimal investment.

SMT core design is powered by demons, thats why Intel gives their processors satanic names

>tech illiterate retard desperately grasping at straws

An ALU inside of the integer core is not an "integer unit." Failing to understand what is essentially baby's first core 101 is more than enough reason to ignore all of your tech illiterate shitposting.

Also, ALUs are not divided evenly between threads, child.
Intel's latest Core i arch has 4 ALUs
The logical SMT thread is run on the last unoccupied ALU while the others are busy

You literally do not understand anything related to the topics you're trying to discuss.
Stay in Sup Forums.

>no source
>not even pretending to have a fucking model
>nothing about binning or price range

OP you should work harder to earn those rupees

Oh so now you're resorting to semantics to salvage your weak argument? Just admit that you don't know anything and don't understand CPU architecture and everyone will be happy.

Most games today still use very few threads

comparing frequencies across different architectures is completely fucking stupid, even more so when one of them is months away from release

Seriously fuck amd. I considered them until they went full botnet. If I'm getting botnet, I'd get the performance of intel with it.

>1060
why? the 480 is better

That isn't semantics, you tech illiterate retard. You flat out made horrendously incorrect statements demonstrating that you're clueless. Basic technical facts that you're 100% wrong about. Thats like saying V8s have 5 cylinders.
Trying to deflect and shitpost after making an ass of yourself won't fix the fact that you have shit for brains.

>i7 4790k - 178
So its all good. Its all about pricing at this point

Chances are AMD will release $100-200 4core-8thread CPUs. At which point, it would be better than i7s.

This is semantics. It's tech illiterate cretins like you why AMD is getting destroyed and becoming increasingly irrelevant. People who for some unknown reason want to achieve the same performance as the competitor as inefficiently as possible because some random workload is marginally faster in some rare situations.

>hurrrrrr guys it has 1.5ALUs per thread
>I totally know how SMT works!

Yeah, "semantics."
Tech illiterate retard.
Stay in Sup Forums.

It does have 1.5 ALUs per thread, retard. Do you know basic math? If a processor has 3 ALUs and 2 threads, it has the performance of 3 ALUs used by 2 threads. That's 1.5 per thread. This is an illustrative figure, not that one ALU is shared between the threads and 1 is dedicated. All 3 ALUs can be used by either thread, but if both threads are active you obviously aren't going to get the performance of 3 ALUs from each thread. You'll still get 3 total.

Just because you lack the ability to critically think about what is being said doesn't mean I'm retarded. It actually means you're retarded.

You're a tech illiterate retard desperately trying to save face. Under no circumstances are execution resources even distributed between threads, and if you understood this then you wouldn't have made such a stupid statement.
SMT has weak throughput specifically because the first physical thread gets the lions's share of resources, the logical thread uses only whats available.
Trying to discuss things you know nothing about then shitposting when called out for your ignorance is the behavior of a child, and you most certainly are just that. A small weak minded child.


Keep on digging yourself a deeper hole, Sup Forumsirgin.

but most of these discussions are centered around gayming but cpu isn't that important for gayming?

so Zen has a score of a little lower than the 6700k....so? you can have some old ass cpu and still run games well. zen will be priced cheaply. why get the expensive thing that offers power you'll never need when you can get the cheaper thing that does exactly what you need it to do.

Yeah, but AMD has a cute anime girl as their mascot while NVIDIA has a fat neckbeard and Intel has some asshole

It was good when sparc and ibm did it

>intel rapes amd in CPU market
>nvidia rapes amd in GPU market
Poor, poor Amada

Literally nothing is wrong with Amada, calm yourself

>AMD's mascot literally gets double penetrated every day

Fuck off, you know just enough to be dangerous when it comes to processor architecture

Are you legitimately retarded? I mean diagnosed with mental deficiency and legally disabled. You certainly can't seem to understand the what we are talking about. "Evenly distributed between threads" is something you came up with on your own through an inability to grasp abstract concepts.

You're not even arguing about the original topic, anymore which is,my insistent that CMT is shit because you need to dedicate huge amount of resources for it as opposed to for SMT where you invest peanuts and get tons of performance in return.

Do you get royalties from CMT patents or something? Why are you even defending this failed technology? Hurr durr we'll make 2 cores which share resources instead of assigning 2 threads to 1 core.

What will Amada do when AMD goes bankrupt?
>Tfw within a year Amada will be a poor anime girl left out on the streets with nothing to do but sell herself to Nvidia employees

>the tech illiterate retard is still grasping at straws

Again, child, both CMT and SMT require increased die area, the specifics of which depend entirely on implementation. For CMT's increased die area it always provides a consistent uplift in throughput where SMT does not.
You demonstrably have no idea what SMT actually is or how it works.

Yet again I refer back to my specific example of the Bulldozer module barely being smaller than two K10.5 cores in Llano. The issue is not with CMT itself, it is AMD's implementation that was the issue.

Stay in Sup Forums. You're not capable of discussing the basics of high level arch. Something even first year CS drop outs can manage. Thats how laughably pathetic you are.

Are we witnessing the true dead of amd?

I was waiting for so long. Still rocking 8350, 7970 but if amd dont deliver soon I think lot of peole, like me are gonna switch to intel/invidia.

For example I have athlon x2 and old core2.
Core 2 still rocking moms pc, athlon is unusable for modern browsing,

Zen better deliver.

>mom's pc
nuff said.

I switched from a c2d (E4400) to a 6700k. The c2d was absolute GARBAGE that couldn't stay below 60C browsing. On prolonged youtube vids (10mins+) it went up all the way to the high 60s/low 70s.

CMT requires more die area. Even if you make big CMT cores per module which are thread-to-thread comparable with a single thread single core with no SMT or shared resources like a Xeon with HT disabled in ALU performance, you're still making it much more complicated, rigid and pointless than just adding more cores with SMT into your processor .

SMT requires nowhere near the die space that CMT does, which is practically a second core... or literally a second core with some shared resources if AMD wasn't trying to cut corner and not making their 8 "core" desktop processors 8 core expensive.

Also stop being obsessed with Sup Forums you dumb redditor. I get it that you don't like it, you don't need to keep telling me that in every post you fucking autist.

ur a penis

Is it an early Athlon X2, as in the Athlon 64 X2? Because that's way older than a Core 2 Duo.

My dad is still using an Athlon II X3 unlocked to X4 and it's pretty good. I think I even have a SSD in that PC.

Yet again, you tech illiterate Sup Forumsirgin retard, how much die area they require depends on implementation. You keep talking about topics you know nothing about, making it clear that you're a 15 year old dipshit crossposing from Sup Forums.
I can implement CMT on the most rudimentary microcontrollers for less die are than it would take to implement another thread via SMT.
You are strictly wrong on this point as you have been on all others.


Tech illiterate retards belong in Sup Forums.

You have no idea what you're talking about and I think at this point I've given you ample opportunity to prove it.

>cinebench

>tech illiterate shitposter still desperate to save face after demonstrating he actually has no idea what SMT is or how it works

>Projecting

>being this retarded
And here I thought Sup Forums was smarter than Sup Forums.

>Zen
>Needs 8 cores to match a 4 core i7

Bulldozer 2.0. Enough said.

Navi is scheduled just in time for what would be the next Fury

>OP picks random numbers out of his ass, type'm and associate with another set of random numbers and letters that could represent the vulgar name or SKU for current or future consumer products

why is this thread being taken seriously?

>literally not telling what Athlon and C2D is it
This is some pretty bad shilling, specially since you could say the first Phenom is unusable for you compared to a older C2D due to the TLB bug
Also, by the time C2D's where out Phenoms were their counterparts