>Zen will have Broadwell IPC, not Skylake
>Inferring that skylake is barely better, if at all when it comes to IPC over even Haswell
but no one said it would have broadwell. It's gonna be well below that, bro - sorry.
I don't understand that pic
Zen a best microarchitecture
>>5398743
>soldered cpu
>>53984823
Wouldn't the wires increase latency?
>>53987975
not significantly.
>>53987975
Yes they would, >>53988164 = idiot.
>>53984823
If it will have Haswell or even Ivy IPC, that will still make it a good alternative depending on AMD pricing policy. I really hope they will not try to charge like 800$ for it. If it will cost more than i7 or even close to it, then fuck it.
>>53988217
tell me more? I'm guessing that because the topology of the pinouts on a CPU must correspond to a two dimensional array as it is, some pins will have more/less distance for their data to travel to and from the board, and thus there's already a need to account for the different distances/latencies traveled. Presumably the same system might be able to account for the small change in latency introduced by adding a few mm of copper to the path?
skylake does have better ipc over haswell.
a 4ghz skylake is roughly the equivalent of a 4.3ghz haswell overall. certain tasks skylake is greater than that of course.
not that huge of an increase but an increase nonetheless.
either way zen is a failure in the sense of dethroning intel as the performance king. zen will not cause a mass exodus of intel users to switch over. most haswell and above users won't move over. more so haswell-e users. they already have zen now.
the battleground will be for those on ivy bridge and lower. both will be fighting tooth and nail over those users. intel wanting to keep them intel users and amd wanting them to switch.
if amd is willing to sell their processors at a near lose they can battle with MOAR CORES. haswell to broadwell level of single threaded performance with 8 cores for $350 would really hurt intel.
but most likely amd won't do that. it be shareholder suicide if they did. if zen is really haswell to broadwell tier performance then their shareholders expect amd to market it as a premium product.
zen will most likely debut with a similar price structure to intel.
dual and quad core apu's for mainstream to compete with intel's mainstream. locked with unlocked versions.
mainstream platform with similar pci-express gimping as intel's with similar core count and "ht" style variations.
quads, hexa, and octo cores to compete with intel's enthusiast platform.
$350 quads
$550 hexa
$750 ocoto's
with no pci-express gimping and a modern platform.
>>53988361
Nah, AMD isn't that retarded when it comes to pricing. They know exactly what their CPUs compete with and will price them accordingly.
The real test will be whether AMD can deliver Broadwell or Skylake performance with the same power usage and heat generation.
Also, unlike Intel, AMD tends to decrease prices of old stuff with time.
>>53988412
Zen was never about dethroning Intel, it was about becoming competitive again at a lower price point
It's about appealing to the masses and leaving the power users to Intel
A 40% uplift in IPC over Excavator is nowhere near Broadwell.
Stop regurgitating stupid reddit bullshit based on passmark scores.
>>53988457
>power users
Meaning the people who buy the thousand dollar Xeons? A Skylake i5 is more than what the vast majority of people have in their homes.
>>53988392
Not him.
But as far as I understand the distance between the CPU and the RAM is critical.
Maybe the CPU pins closest to the RAM slots are connected to the outermost RAM chips and those on the opposite side to the inner RAM chips? - I think that would pretty much equal out any distance.
>>53988412
I don't see Zen selling at msrp with all the demand. Mediocre skylake is just now getting back down to msrp levels.
>>53988217
yeah by fucking 100 picoseconds
hes surely an idiot...
>>53988392
if your clock and data signals are the same length its fine, looking at that pic it looks like they are equal lengths
on the other hand this setup is more prone to picking up noise, and radiating noise...
the different distances to pins are fixed by making wiggly traces, so to match a short trace to a long trace they put squiggles in the short wire so the lengths are equal
>>53988498
>A Skylake i5 is more than what the vast majority of people have in their homes
Those are the people AMD wants
And I'm talking about the people on X99 as well as higher end Xeon users. Those people have no reason to use AMD, it would be a downgrade
>>53988494
AMD said they have exceeded those expectations.
lmao @ pic
imagine all the interference btw the touching wires
>>53988619
Where? Last I heard they were expecting near 1st gen Haswell as a best case scenario
>>53988412
amd's apu's with haswell level of performance would be great alternatives to intel's. similar single threaded performance along with stronger integrated graphics.
$150 quad core with no HT apu with ~370x or greater level of graphics performance would make for an amazing budget gaming build and not only hurt intel but nvidia as well.
i don't see amd releasing a $350 enthusiast quad. zen will fail hard considering current haswell-e ships hexa's cores for $380 and broadwell-e will continue the $400 entry hexa core.
i see it more as:
$400 hexa
$700 octo
release them as you said with with no pci-express gimping and a modern platform and it be a decent success for amd considering intel's $400 broadwell-e hexa core will continue the pci-express gimping.
a interesting enthusiast line up if they really want to tackle intel would be:
$300 quad
$400 hexa
$700 octo
$300 quad + ht + no pci-express gimping would really hurt intel's enthusiast line. but it might also hurt amd's mainstream apu line as well since a $300 enthusiast quad would compete with their own $300 quad apu.
>>53988689
>http://www.overclock3d.net/articles/cpu_mainboard/amd_s_zen_will_have_a_greater_than_40_ipc_improvement/1
>http://wccftech.com/amd-am4-motherboards-launch-q2-2016-for-bristol-ridge-apus/
take it as you will considering they said AM4 will launch "march 2016."
well its now april and no launch happened.
>>53988730
$300 quad is not realistic because it is very close to current i7 skylake with definitely better performance. 6700 is around $310-320 and 6700K is around $360-370, and then goes 5820K for $390. Getting worse performance for saving extra 20-70 bucks does not make any sense. Ofc AMD much likely will not follow intel's locked-ulocked CPU model, but a decent motherboard to overclock zen will cost you extra. I think zen quad can be around current i5 pricing range around $200, followed by hexa around $300 and octa maybe around $400.
>>53984823
>this picture
IS THIS REAL LIFE?
>>53984823
Why
>>53988802
>well its now april and no launch happened
I'll be surprised if it's out by the end of the year
Intel will probably launch Kaby Lake and Broadwell-E before AMD does anything
>>53988539
TIL about mobo epididymes, thanks /g/
>>53989025
AM4 platform was supposed to come before zen. It's a pity because I was looking forward to anything carrizo and oems seem to have screwed them completely.
>>53989099
Been using the Athlon X4 845 on a motherboard for about a week now; carrizo's cash, yo
(it's a pretty mediocre CPU desu)
Zen will probably be AMD's last CPU anyways even if they succeed just because x86 is fucking dead.
>>53989301
I keep seeing you people post that x86 is dead. Why is it dead?
>>53989327
>327â–¶
>>>53989301
>I keep seeing you people post that x86 is dead. Why is it dead?
They're wrong, but I think their argument would be that "most people" use devices for purposes that could be solved just fine by ARM, which is better-suited to a competitive, saturated, mobile, and ignorant market than x86 for a variety of reasons.
But they're fucking wrong. x86 and hell, even POWER will be around for a long time, doing important things, and perhaps even growing.
>>53988619
No they didn't.
In a slide specifically about DATACENTER chips
AMD stated that their next generation would have over a 40% uplift from the current generation. AMD's "current" generation of Opterons are the same Piledriver based chips that have been available for years.
AMD never, ever, ever even remotely implied a greater than 40% uplift in IPC over Excavator.
You retards need to be culled from the population.
>>53989536
Well it seems to me that PC gaming and content creation seems to be growing every year, I think in part due to Millennials getting decent jobs now that they're out of college.
So yeah I agree that heavy processing and computing requirements aren't going anywhere
>>53988412
>a 4ghz skylake is roughly the equivalent of a 4.3ghz haswell overall. certain tasks skylake is greater than that of course.
not even remotely true.
Skylake is a complete dog, bringing <5% IPC gains for everything but a few AES and specialty integer math operations.
>mfw waiting for Zen
>>53989710
4 to 4.3GHz is a 7.5% increase so he wasn't that far off if the 5% claim is true.
>>53989740
something like 80% of use cases will see improvement in the 2% to 4% range, which is realistically only about a third of the claimed difference.
>>53989710
>>53989740
>>53989838
Not sure exactly how CPU Z benchmarks, but this is my OC'd 4770k next to a 6700k
No reason to upgrade
>>53989710
read the claim more carefully.
An 7-10% IPC jump from *haswell* to skylake is entirely feasible, whereas broadwell to skylake would be BS.
But yeah, Haswell was the last OK architecture improvement by Intel, while Sandy Bridge was the last actually GOOD one.
Coming anywhere between Broadwell and Haswell IPC will be good enough for Zen to be very competitive, frankly.
>>53989715
Only 6 more months. I'm saving up my NEETbux.
>>53987430
>doesn't know how to solder his own cpu
>>53984823
>zen
>broadwell ipc
lolno
90% of haswell IPC as PROMISED, but
>amd promises
fucking kek
>>53990427
>90% of haswell IPC as PROMISED
source?
>>53989918
Pretty much "silicon wall". But 6700K wins in power efficiency, having enough room to store more cores for future 8c/16t consumer grade models.
>>53989598
I can't wait for Zen to release with 10% IPC increase over Kabylake and btfd both Intel and their shills like you.
>>53984823
>that one wire unsoldered and hanging there
TRIGGERED MY AUTISM
>>53988539
Are you implying that he's not an idiot?
At 4 GHz the cycle time is 250 picoseconds.
>>53989710
/g/ says Skylake didn't bring any improvements, but in Videogames benchmarks comparing Haswell and Skylake I can often see the Skylake get ahead by almost 20FPS in games that aren't GPU bound.
They might not be getting better IPC but they are definitely more efficient when it comes to do task.
>>53992362
>in Videogames benchmarks comparing Haswell and Skylake I can often see the Skylake get ahead by almost 20FPS
sauce, please.
If anything, one of the unexpected mid-range vidya champs are the broadwell-based i7-5775C/i5-5675C, due to their 128MB L4 being able to fit most of a game's working set.
>>53992362
Some games are improved by higher speed RAM, so it is the DDR4 at work.
>>53990184
He doesn't know the difference between CPU and Chipset
>>53990184
That's not a CPU, idiot.
Your mother would be ashamed of what you're posting in the basement.
>>53992362
A good number of those improvements were from fast DDR4, which gets negated if you already have fast DDR3.
>>53992525
>>53992628
I thought /g/ always said ram speed didn't matter?
Fuck the comparative analysis from your inflated ego fed by your fucking OCD of specs. Build the best pc you can if you're not happy fuck off and build another
>>53992662
It's not that RAM speed doesn't matter, it certainly does matter, but it shouldn't directly result in framerate increases because your GPU should be doing pretty much all the heavy lifting there.
Basically shitty engines.
>>53987975
huge fucking increase.
AMD will have double the coarz for same price
>>53988457
>these image
Seems legit.
>>53992662
It does matter. It doesn't matter if you compare 1333 MHz to 1600 MHz, but when it comes to over a 1000 MHz difference, assuming you are using applications which are intensively using your RAM, you will definitely see a difference.
>>53991514
Also, lower temps. I just built a knew PC with an i5-6600k, though I fully intend to upgrade when there is a 6/8 core Skylake alternative. If that never happens, I'm quite happy with its performance. My system is cool and quiet as fuck and I'm not even water cooling.
>>53988802
There are people outside of AMD using zen CPUs right now so there was a "shipping"
>>53988730
The extra graphics don't matter. Its limited by memory bandwidth.
>>53989536
POWER8 is bad desu senpai. 15% better performance on *some workloads* for 2x the power. POWER9 *maybe* better. I hope so. Its OpenSource which is awesome. ARM is gonna eat a lot of the world. Its too power efficient. Apple A9X has caught up too 5th gen mobile i7's
>>53989598
Yes they did
>AMD never, ever, ever even remotely implied a greater than 40% uplift in IPC over Excavator.
they did numb skull.
https://en.wikipedia.org/wiki/Zen_%28microarchitecture%29
Pic related is the current laptop version of Excavator in benchmark vs skylake laptops
remind me again how this has anything to do with haswell
intel shills can't even get info right lol
>>53994990
Your pic is not haswell though. It looks to be broadwell parts in that comparison.
>>53995127
>broadwell Laptops
>>53992263
Hey, you do realize that the bus clocks are unbelievably lower than 4 GHz? Like the clock entering the CPU is about 100 MHz, and the RAM is communicating at around 800 to 1200 MHz and not even necessarily on one line. The extra length really doesn't matter. The length also varies severly motherboard to motherboard.
Idiot.
>>53987975
Electric potential waves travel at nearly the speed of light, so no.
>>53995270
1/3rd speed of light
>>53992662
doesn't matter how much you upgrade your rig, game is shit very poorly optimized
>>53995305
>>53995270
its slower than that in a small chip with a lot of potential impedance
>>53992609
> That's not a CPU, idiot.
> Your mother would be ashamed of what you're posting in the basement.
>Not your motherboard
Missed stupid pun opportunities
>>53992662
'Speed' (or MT/s, which is what you are referring to as 'speed') matters less, latency is the killer.
Still, that doesn't mean don't get high'speed' ram, just make sure you're getting good shit that is going to decrease (or at least equal) your old latency.
see pic related, most ram since DDR1 has been SHIT SHIT SHIT.
Latency (in REAL TIME) has been worse than the AthlonXP/Pentium4 days until very recently.
>>53996330
While I understand that latency increases as bandwidth increases, doesn't the increase in MT/s make up for the increased latency, especially in DDR4?
I would consider an AMD CPU if they can deliver me something with 8 cores for 400 € that can compete with Haswell-E. More lanes and less money.
I remember when I bought a Phenom over Core, good old times. Make AMD great again. Currently you have no other choice than Intel unless you're retarded or aiming for a poorfag/lowend rig.
>>53996444
AMD is going to have to pull some fucking miracles if they hope to catch up to normal Haswell, let alone Haswell-E
Remember, Intel will most likely lauch Broadwell-E before Zen is even close to launching
>>53995270
I'm pretty sure some of the newer models of the Xbox 360 revisions had troubles because they were too fast and microsoft had to do some roundabouts to increase the latency to match the older models.
>>53996394
See the pic related, very good DDR3 and DDR4 can make up the difference, but most of the time no.
total latency for a row access (tRAS) is measured in nanoseconds at the bottom.
Common DDR1-400 modules tRAS was 30ns (6 clocks cycles at 5ns per cycle), common DDR2/DDR3 was actually worse than that.
Some really nice DDR3 (like my tridentx 2400 with stock 10-12-10-31 timings is 31 clocks for tRAS at 0.83ns per clock cycle - 23.33ns for tRAS) and DDR4 is better, but most shit isn't.
When looking to buy new ram, do the calculations yourself and just make sure what you're buying isn't shit (like most RAM sold)
>>53997119
most stock timings for decent ddr3 1600 modules is 9-9-9-24
good 1600 kits can hit 7-7-7-21 with a volt bump to 1.65
really good higher rated (1866, 2133 etc) kits can do 1600 7-7-7-18
my RAM currently set to 1423@7-8-7-19, which translates to a 26.7 tRAS response time
1800@9-9-9-24 would be 26.6
Also CPUs have always and will forever be horribly bandwidth constrained - modern L1$ have throughputs on the level of 1TB/s, L2$ being about half as fast, and L3$ generally being ~200GB/s
RAM latency is not nearly as important as you think, and shaving off 10ns total response time for a 20% drop in bandwidth is not advisable.
>>53992662
it generally doesn't matter at all for gaming. fallout 4 is a weird outlier.
>>53997608
Crysis 3 gets almost 15FPS more with faster DDR3 ram and this is with a 2500K.
>>53984823
Broadwell and Skylake are both 5% IPC jumps though
>>53992662
It doesn't matter, pretty much every test out there shows it doesn't matter, Digital Foundry seems like they're trying to find situations where it matters, but you can run single channel RAM and you won't see any real difference in games.
>>53988412
AMD doest sell processors that cost more than $200. Wth 8350s going for $175 there's no way Zen will be anything more than $250
>>53994990
That's a fucking GPU benchmark you idiot
>>53988457
>he thinks zen will be competitively priced
AMD sold piledriver chips dirt cheap because of the massive public backlash over it, originally they wanted to sell them for just as much as intel sells CPUs.
>>53997696
The Athlon 64 blew the fucking pants off of Intel and they sold for upwards of $600
>>53997727
And once Zen is the focus of said public backlash for not delivering on its promises, it will be dirt cheap as well
>>53992662
Fallout 4 is just a bad game (optimization wise) and the rule obviously only applies to games that are "made" well.
>hurr durr this program I made that benefits from faster RAM benefits from faster RAM
Don't apply that logic. Ever.
>>53997696
the 8150 was $250 at launch
lets also not forget the comical 9590 which literally melts motherboards and launched at the price of $900
>>53997745
i don't think AMD will fail to deliver on zen, but i guess there is still a chance AMD could be forced into selling them cheaply if it doesn't live up to what the fanboys want. i really just want 8 core + ht at a price that isn't >$1000
>>53997774
Expect AMD to slot in ~$50 behind Intel in almost every offering, unless they have some magic up their sleeves.
We know it's a slightly inferior manufacturing process on a slightly inferior performing part.
>>53997774
Different anon here, but I'm planning on doing a new build if zen is any good; expecting to pay up to $650 for the CPU.
>>53997774
What I want is ECC support and other shit so that I don't have to pay out the dick for another fucking Xeon-based system.
>>53984823
skylake is a fucking joke
>>53997784
More like $800 for top tier. No way can AMD sell a processor for any more than that
It's almost comical, how AMD are almost in the same position in 2016 as they were in 1996
Can they pull a rabbit out of the hat for 1̶9̶9̶9̶ 2019?
>>53994990
You are literally too retarded for words, my god.
Can you even read at all?
will Zen be able to battle GTAIV's horrible optimization? my FX-8350 surely isnt
>>53995270
http://www.wolframalpha.com/input/?i=speed+of+light%2F3GHz
also it will destroy all high frequency signal due to increased capacitance/inductance
>>53988412
>Zen will only be at Haswell performance
>Gotta wait for Zen+ in order to match Intel
>inb4 just gotta wait for Zen+ bro :^)
Fuck this bullshit, I am upgrading right now.
>>53992662
/g/ is a bunch of retards that would have you buy older, inferior tech (DDR3) when DDR4 is the same in price nowadays. Because new tech is bad and useless according to /g/.
Never listen to the retards in here.
>>53988412
Intel is a price setter.
AMD cannot compete at those prices.
The octo's will be 500 max, probably with an insane clock
>>53989301
They are in a much better position than Intel is then. They don't manufacture anymore, GloFlo does their manufacturing at a discount. They just design/R&D and are flexible enough to move beyond x86.
Intel is heavily invested in x86 and handles it's own manufacture
>>53999691
I have DDR3-2133 and don't see a reason to upgrade, my board does only run DDR3. Won't be upgrading until 2018 or when a 8 core 5820K level CPU arrives, maybe I'll take a look on AMD if they really deliver something that can compete. I was always on AMD systems but nowadays you are forced to use Intel if you are not retarded.
Also faster RAM like 2133 is not more expensive, its 5 € sometimes, the only expensive thing are extreme modules like 2800.
>>54000606
>I have DDR3-2133 and don't see a reason to upgrade, my board does only run DDR3.
Good for you. However, I keep seeing retards here on /g/ suggesting DDR3 for brand new builds and chimping out if someone wants to use DDR4 instead.
Everyone who does that should consider suicide.
>>54000606
I forgot to mention, its not only for gaming, it will speed up conversation and zipping speeds. One of the reasons why I upgraded from 1600 (Except that I wanted 2 x 8 instead of 2 x 4), I use Winrar a lot and also convert Blurays.
>>54000631
I see a lot of builds with i7-6000, this is retarded when you can get a 6 core 5820K (Or the 6XXX soon) for almost the same money. DDR4 more lanes and most people won't need the GPU.
The new processors have 10 % performance increase to my OC'd 3770, so the best thing is buying used or -E. I also see a lot of people with 2600Ks, you can get these for 100 € nowadays! Was already thinking of making a mediacenter PC with a 2000K, but I suppose that I'll use my 3770K someday when it gets too old but can still compete a future i3 for media streaming.
>>54000699
>almost the same money
The boards are much more expensive though.
>buying used
I keep my hardware for a long time so buying used is not an option.
Honestly, I am still on a Q9550 and it's pure shit nowadays. But I am waiting in case AMD makes a comeback with Zen.
Otherwise I'll just get a 6500 or 6600 with a DDR4 board.
>>54000740
You shouldn't buy cheap boards, the DDR support isn't the only issue with them.
If you are a long time user then consider a 5820K, more cores, more lanes and almost the same price as a 6700K. i5 is fine if you don't need the threads.
I bought my 3770K for the price of the i5 when Haswell was around, no regrets.
>>54000822
>cheap boards
I can get really good 1151 boards from 120 to 150€. 2011-3 boards cost 200€ minimum.
Whatever the case, I am not getting an i7 anyway. Intel jacked up their prices too much since they have no competition so fuck them. An i5 is good enough.
ZEN
E
N
JIM MOTHERFUCKING KELLER
I
M
THESE ARE WORDS TO LOOK FOR IN 2016, 2017 AND BEYOND
>>54002602
The Jim Keller who left AMD a couple of months ago?
Guess I'll stick to getting a 5820k in a few months.
>>54002681
He left because he was DONE.
Jim leaving isn't a bad thing for Zen, it's a bad thing for AMD's plans for the architecture after Zen. Now is about when you should start planning for and designing 2020's architecture.
What this says is that AMD does not have an 'after Zen' architecture.
Zen is it.
>>53984823
If AMD can release an 8 core CPU with per core performance being the same as a 4790k then they'll win. The 6700k and the 4790k are performancewise identical anyway.
https://www.pugetsystems.com/labs/articles/Haswell-vs-Skylake-S-i7-4790K-vs-i7-6700K-641/
>>54000844
Then the i5 is the better option. But most people buy i7s. You can't go i7 when you don't have the money for a proper mainboard.
>>54002733
This or the 6820K, it will come soon, and newly bought you would get more value from the 6xxx. Maybe more lanes or cache, but at least 14nm.
>>54002817
So does the 3770K and 2600. They were using 1600 DDR3 which should explain the differences.
The current AMD FX model can almost compete in multithreading, but it sucks in single core and needs 40 GHz to archieve that!!! The power draw is heavier than Haswell-E.
>>53996557
At $2000 sure
>>53992727
The engine is fine. Bethesda performance issues are related to large cells with thousands of unique actors running at once.
It's not a GRAFFIX thing, it's a simulation thing.
>>54004173
>Then the i5 is the better option. But most people buy i7s. You can't go i7 when you don't have the money for a proper mainboard.
I did. The good motherboards that provide all the functionality of a relatively standard Z170 chipset start at $200 minimum. The motherboards themselves are in the spotlight this generation. The CPUs aren't that better than the older gen. You can get an Asus Z170A for $150 which has basically anything you can ask for within reason. Full speed m2 ssd/pcie3, ddr4/oc upto ~3200mhz, sli/cf, tons of sata ports, usb c, new heatsink designs, much better bios, lan etc
I would need to shell out about $800 for a good E/mobo combo. If you factor in everything else (psu/ram/case/cooling) you'd reach 1200 easy imo. Rather have all the goodies of the motherboard and an i7 instead of needlessly shelling out 200-300 for a motherboard. It's just way over the budget of some people, even those that buy i7's.
>>54002787
AMD is doing with Zen what intel did with core. They will make small revisions to it for many years while building up a war chest to support the R&D of their next arch, which will likely be debuted in 10-15 years.
>>54004346
>The engine is fine.
Surely you jest?
>>54004346
>The engine is fine, Bethesda performance issues are related to the engine being shitty
>>54004346
you're kidding right?
>>53988457
I hope you know that that image is a fake
>>54004553
I want to upgrade somewhen, but only with a performance increase. The Maximus V Z77 board I have was top-end in 2012 and is still decent except fast m2 and NVMe support. Don't forget you'll get money back for your old stuff. Don't have it laying around unless you plan to do something useful with it.
>>54000631
I am one of those retards. Why should I move to DDR4 every time?
>>53984823
Ivy bridge is like 10% slower than Haswell
Haswell is 2% slower than Broadwell
Browadwell~Skylake.
Even an Ivy bridge is good enough only 15% slower than current gen intel ipc.
>>54004689
Zen is probably their last architecture. Maybe another die shrink <10 nm and they will hit the wall. Graphene will not use conventional architectures used on Si.
>>54005549
Work on Zen+ was already underway when AMD gave their financial analyst presentation in early 2015.
>>54005490
Not really, its more 5% Haswell<>Ivy Bridge. Maximum 10 % Ivy Bridge<>Broadwell.
I get 30 points less than a 4790K in some benchmarks, at the exact same clock.
>>54005549
>Graphene will not use conventional architectures used on Si.
Are you implying that it wouldn't be gates made of transistors?
>Raven Ridge Excavator will be the sub-$200 APUs
>4 core Zen will be $250 APUs and $200 CPUs
>6 core Zen will be $300+ CPUs
>8 core Zen will be $500+ CPUs
Zen will be a disaster. No one wants to pay $500 for an 8-core 125W TDP power hog that has the same multi-threaded pull of a 6-core Intel Xeon-L running at a sub-90W TDP.
>>54002787
If Zen is done, why isn't it in production yet?
>>54008147
Raven Ridge is not Excavator.
Raven Ridge is Zen.
Way to not have a single idea what you're talking about.
>>54008177
Bristol Ridge, then
Who gives a fuck aout AMD codenames anyway?
Kaby Lake, Cannon Lake, and Skylake are the only codenames you have to care about
>>53997608
doesn't it do the same for gta 5?
>>54008112
I'm implying that cramming more cores higher clocks and little bit higher efficiency will hit the was ~5-10 years after graphene is mainstream. CPU architecture will be different because all other parts will be diffrent.
>>54008295
Graphene is not going to be the major component of CPU design for a long, long time.
It's very fucking difficult to build graphene transistors that don't short each other out.
We will see III-IV, III-V, Ga-Ar type materials and Gate-All-Around topologies before graphene is a viable production choice.
My guess is that Zen will be something like Phenom II again. It'll be overshadowed by it's somewhat inferior performance to Skylake, even if it's a way better bang for buck. Regardless, neither of these 2 are making compelling options for someone with at least a Sandy Bridge, unless you need moar cores. I know lots of people who are still happy with their 2500k.
>>54008747
AMD was still at least in a veritable position against Intel with PhenomII - even returning to those desktop market share levels would be a big win.
>>54002681
>hurrdurr it's summer how does jim keller work I donno what 4chin is
>>54008295
I think it 10 years RAM willl probably be gone. There will be like 16/32/64 gb L5 or L6 cache in CPU itself, so no memory controller. also everything will be pci-e or something that will change pci-e.
The way I see things it is - 16-32-64 cores, more tiers of cache, ssd or memristors or whatever will be the jazz at the time on the PCIe slot directly and fuckton of PCIe lanes. PCs will be more modular because mobos will be just pci hubs and everthing will be on slots.
>>54008767
Phenom II were great, also for upgraders, when Cores only accepted DDR3 Phenoms could use both (DDR3 was more expensive back then).
I really hope they get anything that can somewhat compete, the current FX are a complete disaster. But even if they get something on i5 level for half the price without being a 220W housefire, people would buy it. Better would be more powerful CPUs with at least 80% single core performance of current Intels.
