Zen is going to be so great. Cheaper and more powerful than Skylake plus with crazy OC potential.
I'm guessing very little OC potential if they are actually competitive with Intel. AMD will bin these chips to the limit.
>>55254788
>Cheaper and more powerful than Skylake plus with crazy OC potential.
Nigga we don't have confirmed information about fucking anything yet.
what makes you think there will be crazy oc potential?
>>55254788
>2016
>CPU's
Who cares. It's all about GPU's these days. (AMD WILL WIN, NOVIDIA IS FINISHED)
>>55254815
>>55254825
All of AMD's chips in the past have been very overclockable The FX-8150 has been overclocked to 8GHz before
>>55254788
>More powerful than skylake
Current optimistic estimates based on the +40% over Excavator figure AMD gave us would put Zen between Haswell and Broadwell IPC. Haswell level IPC is expected at minimum though for everything that isnt FPU limited.
>>55255011
Yes, but Summit Ridge will have twice the cores, plus it is twice as fast as the 8350
>>55255011
>muh IPC
There is more to CPU design than IPC.
>>55255086
Yeah, I am aware of that, I'm just saying to expect >Skylake level performance for Zen is folly when clocks and overclocking are unknown. Now for Zen+, it's expected, because +15% above Haswell would put it square within Skylake's territory.
>>55255006
I might have said that the bulldozer chips in particular are easily clocked up, but I wouldn't have suggested that ALL amd chips are like that. In my experience vishera chips tend to crap out at much lower temps than a comparable intel chip would have
My body is ready
>tfw fx 6300
>>55255006
Zen is geared towards efficiency and core performance parity with Intel over overclockability. I bet you anything that we'll see much less voltage tolerance and clock multipliers over Piledriver, similar to how Steamroller could not reach the same core frequencies as Piledriver.
>>55254788
Who made the pic op? The background isn't Zen. Wtf senpai
>>55254788
Fud. Zen will likely be haswell performance but with 8 real smt cores.
>>55254878
It's all about CPUs because there's virtually no concurrence to intel these days. Nobody can even question that intel's chips are better, even when measuring price/performance. Zen might change that, that's what we hope at least.
Will AM4 motherboards be cheaper than X99 boards?
>>55255898
>tfw fx 6100
>>55256805
I can't complain about the fx 6100 though. I'm not huge into gaming and use a tv @ 1366x768 60hz and the fx 6100 has yet to let me down in almost 5 years. Have it paired with a 650ti cuz it was dirt cheap after rebate when it came out. I upgraded from a 9500gt.
>>55256786
usually AMD motherboards are relatively cheap
>>55254788
Zen is a core architecture, not a line of CPUs.
Summit Ridge is competing against Haswell-E and Broadwell-E. It is not a mainstream desktop line. It will not compare to highly clocked Skylake chips for serial performance. Matter of factly it will not reach high clocks at all.
The mainstream competitor to be positioned against intel's i3, i5, and i7 lineup is the Raven Ridge APU.
>>55257568
>Summit Ridge is competing against Haswell-E and Broadwell-E. It is not a mainstream desktop line
AMD shills really believe this, huh?
Zen will NOT compete well with Haswell-E, or even Ivy Bridge-E. Their only saving grace is offering more cores for the money.
>>55257794
>more cores don't count
Intel bots actually believe this?
>>55257818
It would, if those cores were comparable to Ivy Bridge's. But they're not. A 6-core Zen can't keep pace with a 6-core i7-3960X, let alone a 6-core i7-6850K. AMD can only hope to steal buyers away from the i7-4790K/6700K, but you just know that it's going to be $100 more expensive than both those chips and not even 10% better at multi-threaded tasks.
>>55257794
Its hilarious that you're too stupid to even understand what you're saying.
IPC is around Haswell levels, Summit Ridge is an 8 core part, the top end Haswell-E is 8 cores, Summit Ridge has far more aggressive turbo, and can clock each core cluster independently.
Summit Ridge is not a mainstream CPU, it is a competitor to intel's i7E line, and the top binned part will likely retail for $500 under cutting the 6 core Broadwell-E parts.
Polaris disappointed and so will Zen
>>55257897
Disappointed so hard that Apple purchased millions of Polaris10 and 11 GPUs for new product lines. 4th gen GCN has stellar perf/watt.
Shitpost more, shill.
>>55257867
>IPC is around Haswell levels
PROTIP: it's not. Even the best estimates of Zen's per-core performance puts it just under Ivy Bridge.
>Summit Ridge is an 8 core part, the top end Haswell-E is 8 cores
The FX-8320 was an 8 core part too, so it must be a competitor with the i7-5960X.
>the top binned part will likely retail for $500 under cutting the 6 core Broadwell-E parts
And the i7-6850K will STILL beat an eight-core Zen. Top kek.
>>55257946
>trying to compare a minimal die area CMT arch to a big core SMT arch
>talking about bullshit "estimates" like you're not a clueless tech illiterate retard
Hilarious.
The only area that Zen will trail in is FPU throughput. No sane person would expect AMD to match intel here, and Su even directly stated that they'd address 80% of the enterprise market with Zen based Opterons. They're competitive where it counts.
Though I'm sure you're going to keep shitposting like a mad little baby. Seems thats all you limp dick shills can do.
>>55257897
What planet are you from? They're about to move a shit ton if the board partners don't gouge too hard.
>>55257897
>polaris disappointed
literally what
>>55257988
Except Zen IS an SMT-centric architecture, dumbass. And it STILL can't beat an old eight-core Haswell-E.
Face it, AMD is fucked ten ways to Sunday no matter how low they price those CPUs. They have to make up for all that R&D cost they wasted on making Zen, after all.
>>55258014
The RX 480 is slower than an ancient R9 290x in games.
>>55257818
If they believed that, why would they buy 8 core i7 extremes for 1000$?
>>55258015
>Except Zen IS an SMT-centric architecture, dumbass.
Way to prove you really are too retarded to read. Seriously, thats just pathetic that you're this stupid.
>>55258037
>comparing a shitty architecture like Zen to an outdated architecture that is Haswell-E
Wow, you really are digging into new depths, aren't you AMDrone?
>>55258072
You realize that you're just making yourself look like a worthless retard, right? No one thinks shitposting like an immature child is clever or funny.
You couldn't pretend to be an adult if your life depended on it.
>>55258102
Whatever makes you smile like a doofus, doofus
>>55257897
You probably one of those retards that thought the RX480 would beat the 1070 and touch the 1080
Once Zen arrives Intel is finally going to give us a real upgrade from Skylake. Intel has been cucking us for some time with basically the same performance CPU rehashed every two years because AMD Bulldozer was the worst architecture ever created. I hope ZEN succeeds and is alot more powerful then Skylake so we could have proper competition again.
>>55258586
I don't think anyone thought the RX 480 would beat the 1070... Possibly come close after an overclock with non-reference cooler and something heftier than the single 6 pin connector on the reference design, but not beat a card that costs almost twice as much.
>>55258798
>AMD Bulldozer was the worst architecture ever created
Nope that title is very much still held by Pentium 4 and Netburst in general. On a per-clock basis Bulldozer is at least 50% faster than Netburst, and on top of that it could overclock better.
Bulldozer may be a pile of shit compared to other chips released at the same time, but it does a hell of alot better than netburst.
>>55257866
Show me your Zen CPU and i'll believe you actually know the IPC and clocks.
>>55259286
Zen won't clock any better than Piledriver. This is practically confirmed after seeing the "progress" Steamroller and Excavator has made in core clock multipliers for the sake of efficiency. We're looking at a similar clock range envelope that Intel's Broadwell has (aka the worst overclocking Intel CPUs since the first Core is). Good luck getting a Zen past 4.4 GHz.
>>55259343
Summit Ridge is an 8 core chip, and those cores are comparatively much larger than anything in the BDver family. The base clock will likely be around 3ghz, with a ton of aggressive turbo states. Maybe one CCX will clock higher for more serial workloads, but you're not going to get anywhere near 4.4ghz on Summit Ridge without paying for it in heat output.
It is a counter part to the i7E line, not a mainstream desktop quad core.
Engineering samples that were spotted only had a 2.8ghz base clock.
>>55255006
8GHz with most cores disabled and with Liquid Nitrogen cooling, and 2GB of RAM I think.
Same has been done with Intel chips, 5GHz is the maximum that can be actually usable day to day.
>>55259598
There have been plenty of attempts at getting 8 with all cores enabled (#2 on the CPU-Z hall of fame is an FX-8370 at 8722mhz w/ all cores enabled and 8GB DDR3 in dual channel; http://valid.x86.fr/vytt61 ). Its the ones going for 9ghz that disable shit and go bare minimum. To date 9ghz has not been publically broken.
>>55259706
Yes but still, 8GHz is not stable for normal usage. That is in very very controlled settings.
>>55259796
Yeah I know that, Ive done my own experiments when it comes to suicidal overclocking.
While modern chips get more powerful, they get more complex, and that complexity (combined with transistor density) more than anything is what is impacting overclocking capability. I have no doubts that should intel, for whatever fucking reason, decides to make a couple dozen wafers of P4-2M chips on the latest node, the resulting ~150 million transistors on a really small chip would be able to hit the 10ghz Intel originally intended the Netburst series to eventually hit.
>>55259883
>10GHz
Holy shit that is fucking crazy.
I cant even imagine those speeds
>>55259924
Thats not the thing. Intel at one point before AMD and their own Pentium M slapped them silly with low clocks and high IPC, was seriously considering extending the unholy hell out of the netburst pipeline to 40-50 stages in their pursuit of higher clocks (7ghz *stock* was the intended target)
>>55260127
Really? Why didn't they do it? Would have destroyed AMD.
>>55260468
Because the housefire TDP that would have resulted from it and the fact performance would not have scaled much due to the vastly longer pipeline. Think FX-9000 TDP (220W) in an era where most chips were at 95-115W. There was also the fact intel had Penitum M as their pet mobile project that was beating the shit out of their HEDT and server chips in all fields, and was at parity with Athlon64 give or take a few percent.
If intel HAD been stupid enough to make Tejas and Jayhawk, not even the bribes they were flinging left and right would have saved them from losing the #1 spot to AMD.
>>55260468
Because the Prescott Pentium 4s required to be overclocked to 5GHz to be able to outperform a 2.6GHz Athlon 64 in every area.
if they did release that next Netburst arch, it would've been slower than a core 2 duo.
>>55258029
>The RX 480 is slower than an ancient R9 290x in games.
um, except that it isn't. It's between the 390x and Fury, depending on how much you overclock it.
It's only slower in something stupid like bitcoin mining.
>>55254788
Zen release date? 2050?
If netburst was remade in 14nm it would be the fastest CPU chip ever assembled by man.
>>55260793
by fastest you mean highest clock rate?
>>55260813
I mean fastest peroid. 20ghz with bulldozer IPC.
>>55260590
Oh okay I see mane.
>>55260600
Even at those insane clock speeds?
>>55260828
I doubt it could clock that freaking high.
>>55260828
yeah sure, there's no reason at all CPUs never passed the 5GHz cap
>>55260828
I don't think so anon. Everyone that tried speed demons abandoned it. And for a reason.
>>55260828
Branch misprediction on a deep pipeline sees your optimism and laughs at you. Its what killed P4, and its one of the things that punished Bulldozer.
Every modern chip that has a deep pipeline has a huge amount of hardware and die-space dedicated to minimizing the penalty of branch misprediction.
Clock speed doesnt mean shit if you're flushing your pipe too often.
it's going to be shit you better believe it
>>55260912
let's hope not otherwise the x86 market is pretty much fucked
Does anybody remember Tejas and Jayhawk that got scrapped shortly before their release?
These should've been the successor to Netburst, feature a 50 stage pipeline and clock to 7-10GHz.
>>55259883
>>55259924
>>55260127
>>55261143
Because back in the day, most computer users thought that gigahurrz was the measure of compute power. A CPU with insane clock speed would have sold better than one with lower clocks but better actual performance.
>>55254788
Zen is going to be so great. Cheaper than Skylake 6950x as powerful as a pentium 4 plus you'll have to underclock it to be stable.
Source: my ass.
>>55254788
More powerful than skylake? Thats wishful thinking. Ill probably buy it because im guessing its going to be haswell-broadwell performance at an 8 core and cheaper. I think its a bit extreme to assume it will put the hammer down on skylake though.
>OC potential
AMD already pre overclocks their "high end" to lightspeed just so they dont look shit. Good luck adding any more on top of that
>>55254788
id settle for one last am1 chip. those little fuckers oc like mad, for the price anyway.
will prolly just have to suck it up and open my wallet for zen...
Zen will be the new bulldozer
Expect high temperatures, huge tdp, shit performance compared to intel
>>55254788
I may be excited for zen, but no way in hell will it be better than skylake
