https://wccftech.com/intel-moster-1 2-core-hedt-skylake-x-co

>Revamping Extreme Editions

BURN IT DOWN GOY

File: C5qPYNbUoAAVYlc.jpg (111KB, 1000x1000px) Image search: [Google]

111KB, 1000x1000px

>>60012278
dont fire me pls

File: s.aolcdn.com.jpg (194KB, 600x392px) Image search: [Google]

194KB, 600x392px

>>60012278

I wonder if you can upgrade via software voucher

>>60012278
>12 cores
is this because AMD has higher ipc now? lol pathetic, hope intel collapses under its own weight soon, maybe buying a few more shitty startups for 50 billion each will help

>>60012334
>All Skylake-X chips will feature a rated TDP of 140W (with the possible exception of the 12-Core die)

>All chips will be marketed as the Core i7-7000 series processors and utilize the LGA 2066 socket.

>2066
>4 core kabylake
>same 16 pcie lanes, 4 CORES

who will buy it? I'd like to see that person.

File: .jpg (26KB, 239x268px) Image search: [Google]

26KB, 239x268px

>>60012365
>startups for 50 billion each

File: 1492046443601.jpg (634KB, 2096x1344px) Image search: [Google]

634KB, 2096x1344px

>1500$$$ (projected) price tag

File: mudshitintelshill.jpg (10KB, 225x225px) Image search: [Google]

10KB, 225x225px

>>60012383
>who will buy it? I'd like to see that person.

>>60012383
>The Kaby Lake PCH will have 24 PCIe 3.0 lanes, 8 SATA Gen3, 10 USB 3.0 ports and connected via 4 DMI 3.0 lanes to the processor. Intel LAN (Jacksonville PHY) will also be present on the chipset.

my bad, still no USB on CPU
AMD fucked southbridge hard

>>60012393
>"As you’ve heard me say, others predict the future. At Intel, we build it," Intel CEO Brian Krzanich wrote in a letter to employees.

after buying mobileye, you can't make this shit up

>>60012399
the goyim will unironically think this is a good deal because the 10c 6950X was $1700

>>60012393
https://www.crunchbase.com/organization/intel/acquisitions

spending all that money they have, would not be surprised if intel declared bankruptcy do to these shit decisions and this is just the surface of intels shit management over the last 5 years

>>60012428
>McAfee $7.68B in Cash
haha
I guess in 2010 it seemed like a good idea.

>>60012365
AMD doesn't have higher IPC though. Their IPC is almost at Intel's level, but Intel has higher clock speeds. Right now, Intel is best for single threaded performance, and merely OK in multi threaded. AMD is ok in single threaded performance, but crushes in multi threaded.

File: 1489636946449.jpg (101KB, 614x387px) Image search: [Google]

101KB, 614x387px

>>60012450
things that are less than 10% equal margin of error
IPC is not constant number you can apply
it's better in some cases, and worse in others but not by a HUGE margin

>>60012480
DELID DIS

>>60012450

not him, but of course they have higher clock speed when they have been using the same architecture for 4 years.

the jump from 6700K to 7700K is literally 3% performance increase because they use the same architecture. This is worse way than AMD's Rebrandeon 580

File: 1492127495787.png (871KB, 801x1500px) Image search: [Google]

871KB, 801x1500px

>>60012278

>>60012513
>when they have been using the same architecture for 11 years.
ftfy

>>60012278
>tfw people will expect this to clock to 5Ghz like Kaby can
ayy

>>60012513
>literally 3% performance
>3%
no its closer to 1%

>>60012513
>This is worse way than AMD's Rebrandeon 580
you do know even current pascal cards are still based on the G80 gpu that was used in the 8800 right? they'll never stop rebranding that card, it literally made nvidia what they are today.

>>60012528
the IPC is 0% or even negative depending on the workload, it's just a clockspeed increase

>>60012278
MOAR CORES

MOAR TDP

Intel have finally fulfilled the prophecy and become AMD.

File: 1491835275725.png (234KB, 882x758px) Image search: [Google]

234KB, 882x758px

>>60012278
>/Housefirelake-X/
PLEASE STOP

>>60012540
>current pascal cards are still based on the G80 gpu
I don't think so Tim.

Nice, Intel.

But you still got far to go

Zen HEDT CPU's are called Threadripper(good job marketing..)
Each CPU will include 64 PCI-E Lanes!
It includes 4 CCX's.
Lower SKU(Probably 12/24) 140W TDP, Higher SKU (Probably 16/32) 180W TDP.
Socket will be an SP3 LGA
Platform's name will probably be X399
Chips will be B2 revisions.
32MB L3 Cache
ES's are 3,3 or 3,4 Ghz base and 3,7 Ghz Boost
It is aimed for Retail SKU to have 3,6 Base/4 Ghz Boost
ES's that are in the wild have 2500 CB R15.
Infinity Fabric can have a bandwidth up to 100GB/s <-(double of what we have now)
Announcement; COMPUTEX at Taiwan, sales will start after 2-3 weeks following COMPUTEX.

>>60012650
you can be as delusional as you'd like, it doesn't change the fact.

next you're going to tell me pascal isn't literally a nodeshrink of a rebrand of a gimped fermi architecture

>>60012278
Intel will only be able to see these 6+ core CPUs if the price is reasonable.
There aren't many people who will drop $500+ for a 6 core

>>60012685
Pascal is still a 32 wide superscalar layout that was used in Fermi, but most of the frontend stuff from Fermi changed with Kepler.

To keep it short Kepler was Nvidia's last big architecture change.
But it wasn't a complete overhaul like Fermi was

File: 1486247368242.jpg (39KB, 650x650px) Image search: [Google]

39KB, 650x650px

>>60012685
>literally a nodeshrink of a rebrand of a gimped fermi
I honestly hope you are just pretending to retarded

>>60012685
If Nvidia doesn't change their architecture, then how do they manage to gimp their older architectures by encouraging game devs to design their games to run poorly on their older architectures?

>>60012670
Fuck. Should have waited.

>>60012728
kepler literally has all the disabled double precision cores still on the chip

>>60012804
gimping parallel compute

>>60012831
Your point? Kepler doesn't have hotclocks, which alone makes it a enormous change over Fermi

>>60012841
minor tweaks on an existing architecture don't make it a new architecture.

the canvas is still g80, meanwhile, for example GCN, was a completely new architecture over terascale

>>60012590
Best part is extreme edition cant be delided, permahousefire

>>60012728
>kepler
>big architectual change
no it wasn't, kepler was literally fermi with DP gutted

the last big arch changes for nvidia were the original Tesla arch and Maxwell

I'm no chip engineer, but is revising architectures every 18 months or so really that bad if we're still getting performance and power efficiency improvements? I don't care if Pascal is basically die shrunk Maxwell if it means the 1070 is as powerful as the 980ti for ~$400 and only needs one 8 pin power connector.

>>60012278
Why bother releasing a 12 core when the Ryzen 16 core will BTFO it so badly?

File: 1488420066835.jpg (42KB, 898x886px) Image search: [Google]

42KB, 898x886px

>>60012412
underappreciated

How many more years is Intel stuck with this decaying arch?

Should make for great entertainment

>>60012670
They will also probably have an 1850X(tentative) that's also 4 CCX, though 8c/16t like the 1700-1800X.
Same 3.6/4.1 clock speeds, but with the more PCIe lanes and L3 cache.

>>60012278
Soooooo they are not fine with reusing 2011 socket? 6950x is not compatible out of a sudden??

File: fire supression system.png (48KB, 1280x720px) Image search: [Google]

48KB, 1280x720px

>>60012278
>>60012590
>>60012870
Geez, just mount a diy fire supression system inside your case.

File: 1325277361193.jpg (46KB, 634x600px) Image search: [Google]

46KB, 634x600px

Nah, I'm waiting for the next generation of Intel cpu's to hit and then I will jump on a used 6700k set with 3200 mhz ram.

Should be a nice upgrade from my 2600k with 1866 mz ram.

>>60012278
AMD POO'D THE FUCK OUT

>>60012670
>HEDT CPU faster than mainstream stuff in games
while demolishing in productivity

Finally, ABOUT FUCKING TIME
Faster IF speeds is all we've been waiting for

File: 1489122339460.png (495KB, 1070x601px) Image search: [Google]

495KB, 1070x601px

>>60012296
>>60012523
>>60012527
>>60012558
>>60013081
SHUT UP

THANK YOU BASED INTEL HOW CAN AMD 8 CORES EVEN COMPETE

>>60014066
By going up to 32 cores.


t. Soon to be a former corelet with a 3570k

>>60014066
higher ipc and not being housefires

>>60014107
Lmao

>Author: Agner Date: 2017-04-12 00:52
>The single-thread instructions per clock rate of Ryzen is higher than for any Intel processor, except for 256-bit vector code. I am testing the Ryzen right now and the test results are coming soon. Please be patient.

http://www.agner.org/optimize/blog/read.php?i=49#829

>>60014164
If Agner Fog is saying so there must be some truth to it.

>>60014185
Difference in decoders, Ryzen has 4 complex ones, Intel has 1 complex + 3 simple.

>>60012523
This pic is a masterpiece.

>>60012950
Revising architectures is a good thing, since the first generation(s) can often be improved significantly with relatively minor R&D expenditure. It becomes questionable when the only improvements are more cores and higher clocks due to better manufacturing process rather than better architecture.

Intel's memelake is WAY past its best-by date, though.

>>60012412
I hate this smug Intel worshipper

>>60014844
le mudslime intel merchant

>>60012412
>One person try to de-hype that ryzen 5 video
Im sorry, wat ? Pay me bitch

>>60012416

>Mobileye is an Israeli technology company that develops vision-based advanced driver assistance systems providing warnings for collision prevention and mitigation.
>Israeli

>>60015141
1 5 b i l l i o n
1
5
b
i
l
l
i
o
n

>>60012670
Oyyy-vey.

File: Maisie_Smith.jpg (93KB, 634x725px) Image search: [Google]

93KB, 634x725px

>>60012278

Fuck that, what I want is a 7700k with a soldered ihs, 6 cores and no igpu.

I can't have it, because, jews.

>>60015285
That's called "ryzen 5 1600".

>>60015285
that will be $1200 goy

File: 1492771991243.png (162KB, 633x900px) Image search: [Google]

162KB, 633x900px

>>60012278
>/Housefirelake-X/
DELID THIS

>>60015285
It won't clock as high.

File: maisie_smith-2a-a.jpg (70KB, 485x700px) Image search: [Google]

70KB, 485x700px

>>60015326

The brain-dead clock limited CCX interlink on Ryzens is a deal killer for me.

Maybe they will fix it with Ryzen 2...

>>60015412
How so?

>>60015417
Someone on youtube whined about it

>>60015417

Because limiting the CCX link freq to the memory controller freq is a design fail.

If they fix it, I'll buy.

>>60015463
Why do you even care about the internals of a CPU? Isn't it more important that it just works and gets shit done?

File: Screenshot from 02-04-2017 21:00:52.png (39KB, 826x775px) Image search: [Google]

39KB, 826x775px

>>60015463
It's irrelevant, even for largely single threaded applications.
Stop shitposting, you ignorant child.

>>60015412
>still performs great
>wont buy because he doesnt like the internal design

>>60015463
>omnidirectional crossbar instead of ring bus is a design flaw
Back to Israel with you.

>>60015463
Oh man I wonder what you'd say if you looked at Intel's interconnect fabric

>>60015463
Stop pretending you have any idea how the average CPU works, you dumb /v/tard. What you mean is "somebody said it was bad so I now parrot that opinion".

>>60012527
Server chips are soldered so the 6 core may hit higher clocks due to thermals

+200mhz and -8w at equal voltage
from 14nm to 14nm+

http://www.anandtech.com/show/11232/intel-launches-xeon-e3-1200-v6-family

>>60015541
You still need a waterloop to cool it down at 5 niggahurtz.

>>60015541
Only a small part of server chips are BGA, everything including Intel's own MCC and LCC dies are LGA
That also goes the same for AMD

>>60015541
Those tdp are pretty fucky. The ones with iGpu only consume 1 more watt? Bullshit. 15 watts minimum is what iGpu will consume.

>>60015568
>I have no idea what "TDP" means

>>60015463
b8

>>60015568
You're clueless what TDP means.

>>60015579
It means whatever the manufacturer wants you to believe

File: 04b02ee2ef6b313d9dd849d644ee5ad2.png (174KB, 696x678px) Image search: [Google]

174KB, 696x678px

>>60012278
>Mega-Tasking

File: 1457716466110.png (10KB, 710x423px) Image search: [Google]

10KB, 710x423px

>>60012450
It's only about 25% advantage in single core. The next version of Zen will increase the IPC by 10% and that single threaded advantage will almost vanish ..

>>60013113
2020 at the earliest. Probably 2021. And their new arch breaks backwards compatibility so it might just end up being Itanic 2.0 http://wccftech.com/intel-developing-new-x86-uarch-succeed-core-generation/

File: 1392104226142.png (359KB, 792x540px) Image search: [Google]

359KB, 792x540px

>>60013186
You're going to wait until 2021 to upgrade your 2600K?

File: single socket holocaust.jpg (28KB, 550x550px) Image search: [Google]

28KB, 550x550px

>>60014491
Fucking this, Intel has milked their current arch to death. There's nowhere to go but near useless die shrinks and higher TDP. Housefires are back, baby.

>>60015412
It's a bit offputting you know, but the results speak for themselves.

I don't like nvidias business practices either with Pascal literally being Maxwell on Steroids, but the results on their cards - again - speak for themselves. Who cares if it has some bullshit interlink. It is apparent that it still BTFOs Intel who use a monolithic design.

>>60015762
penitium on laptops were absolulty unsuable, ven with fans or external cooling 30 min of gaming was max.
I fuckikng rembere 2 of my rich freinds laptops

>>60015412
Everyone thought it was a dealbreaker and Ryzen's achilles heel at first, but then it turned out not to be such an issue. It was mostly shit motherboard BIOS and unpatched games causing the majority of issues.

>>60015877
>shit motherboard BIOS
this is OK

>unpatched games
see this is where I see an issue. Why do devs need to optimize their shit so that it runs on AMD's special snowflake architecture.
Why can't it just be an x86 CPU with x cores/threads and the Ryzen microcontroller or at worst the OS will take care of the rest of the scheduling?
Why do individual applications now need to worry on WHICH core they are putting stuff so they don't get limited by the CCX interlink?
This should be handled more transparently

>>60013129
>1850X with more PCIe lanes and L3 cache.

guaranteeably not habbening.

the entire purpose of Zen is a top-to-bottom platform based off 1 die, and Zeppelin is already limited to 32 IO lanes and 16MB L3:
>>60004483

Everyone go fucking home, Skylake-X has no IPC improvement
BWE-E @ 3.6
https://browser.primatelabs.com/v4/cpu/2593799

Sky-X @3.7
https://browser.primatelabs.com/v4/cpu/2596026

>>60015945
Lol

>>60015412
The crossbar will be clocked at full DDR4 rates on HEDT and Naples instead of half (so ~70-100 GB/s instead of 35-50 GB/s).

It's only at half rate for the low end parts to save power, but it needs to be cranked all the way up for parts with 4+ memory channels.

>>60015945
The only thing that improved are crypto scores due to fixed function hardware (like Ryzen has) while integer and FP scores remained the same

Oh wow

>>60012383
I honestly have no idea what a 4 core chip is supposed to do on a HEDT platform. Hell, aren't the Coffee Lake mainstream chips supposed to have 6C too?

>>60015945
>Skylake-X has no IPC improvement

Why were you expecting one? Intel's HEDT line has always just been an old architecture with more cores. It just happens to have caught up to the mainstream platform due to the fact that Intel are now spinning their wheels.

>>60015945
The monster L2s (256kB -> 1MB) will shine in some tasks though despite the L3 cut (2.5MB -> 1.375MB per slice).

Tiny benchmarks aren't where it will do best, so expect Intel to reemphasize more rendering shit, etc.

>>60015906
>AMD's special snowflake architecture
Intel is the one with a proprietary implementation of SMT because they wouldn't be arsed to pay IBM for royalties of the proper implementation of the feature, and the one distributing compilers that generate binaries that shit the bed whenever they detect they're not run on "Genuine Intel(TM)", but whatever.

>>60016023
>Why were you expecting one?
Massive cache overhaul from 256kb core to 1MB, but also a huge loss in shared L3.

>>60016003
retards who think 4c needs 4ch DDR4, or people who want maximum performance in 2013 DX11 games.

>>60015906
>special snowflake architecture
It's just new and different, Nehalem had many of the same issues when it launched. Stop being a fucking autist.

>>60016045
Not saying Intel is the good guy, I'm simply wondering.

Do individual devs now actually have to care that their threads run on a single CCX? Because to me that seems like something that theoretically should be taken care of on much lower level.

Because I'm sure not every dev will do that.

>>60012278
so now both company are going muh core meme?

>>60016076
Windows should've been patched by now (Wasn't there some patch already?)

>>60016092
With the difference that, while AMD decided to redesign their architecture from the ground up to support multiple cores without punishing the power consumption or IPC, Intel is almost literally just throwing more Pentium 3's at the problem and expecting them to stick

>>60016033
Difference is, while L3 can be used by all cores, L2 can not.

Unless the L3 is a inclusive, which is fucking retarded at that small size

>>60015906
>shit motherboard BIOS
>this is OK

shall i remind you about burning $1000 chips on x99 launch week?

>>60016239
I was merely saying by this that this means it's not the CPU's/AMD's fault

>>60016046
>but also a huge loss in shared L3.
wait, intel went shared L3? I thought it was 2MB per core on browadwell-e.

that's actually good thing for AMD, people will start to program for it.

>>60016188
Shifting area from L3 -> L2 just means more potential redundant data between L2s. If L2s end up having mostly shared contents, that just means that programs benefit by having more of what they need at lower latency anyway.

The real tradeoffs are area, power, and remote cache fill latency. It's faster and energetically cheaper for a core to pull cache lines from a L3 victim pool than from a 2nd core's L2.

>>60016352
L3 is still shared. Potentially no longer on a fat ring bus though. Definitely cut by nearly half though (2-2.5 MB -> 1 3/8ths MB per slice).

>>60016399
sounds counter intuitive
is it done for power reasons?

>>60016352
Intel's L3 has always been shared with 64-byte wide cache lines.

>>60016281
>this means it's not the CPU's/AMD's fault
>what is AGESA

File: 1486284675460.png (6KB, 390x470px) Image search: [Google]

6KB, 390x470px

>>60012523
kek'd heartily

Smaller L3 doesn't seem a smart idea for multi socket systems.

>>60015945
Please tell me you're joking, I expected IPC to go up due to higher L2 cache

>>60016507
It will _IN_SOME_APPLICATIONS_(like CPU-Z) but you're also looking at higher latencies, and IPC is 95% latency.
Most current applications have issues filling up Ryzen's 512kb L2, it won't be much better for the 1MB of Intel's L2

>>60016420
lol no, this will crank power consumption up notably just for fewer memory stalls.

should help SMT performance a bit too I'd expect.

victim L3 is actually only very useful if the eviction logic is lucky/smart enough to discern between once-and-done data vs. moderately recurrent data.

shit can get easily fucked if working set sizes exceed L2+L3 size, but if you can keep your L2 hot, you're golden.

>>60012523
be a good goy and DELETE THIS

>t.
>|
>|>
>|3
>|

do you think they rushed it because there is no validations staff left?
"eh will work" kind of approach?

>>60016583
Ye, prepare for another errata storm.

>>60012670
>Platform's name will probably be X399
>Intel can't use X399 two years from now
>Intel can't use B350 for their Cannon/Coffee Lake business chipset anymore
What is this dickery?

>>60016604
AMD is having a good chuckle as usual. Member P-rating?

>>60016569
Better than anything out of india

>>60016604
It's effective, Intel's next line loses consistency.

Unless they name it B450 then AMD has to make something up later

>>60015285
never going to happen goyim

File: grizzlyconductonaut.jpg (436KB, 1612x829px) Image search: [Google]

436KB, 1612x829px

>>60015285
Liquid is better than solid when its the same material

File: 8aff7d6ebd15e8d5c1cfb1ec01b9d9db6c730fb36d8184a6981298462a87038d.jpg (38KB, 435x183px) Image search: [Google]

38KB, 435x183px

Can anybody confirm pic realted?

Anyway, this smells like full damage control from the Ryzen showdown.

>>60016727
>140W

If it is this is hilarious.

But it's likely not.

>>60012278
>Launch price ~$1500
OK

>>60015945
DELET DIS

File: 1482515713843.jpg (184KB, 1125x1119px) Image search: [Google]

184KB, 1125x1119px

>>60015945
>mfw

>>60016727
>i5-8600K 3.5GHz 4.0GHz
>i7-7700K 4.2GHz 4.4GHz
>i7-6700K 4.0GHz 4.2GHz
>i7-4790K 4.0GHz 4.4GHz
What did they mean by this?

>>60016727
>140W
Kekus maximus

>>60016727
Yeah, nah, cache sizes nor TDP fit.

>>60016727
most likely fake, chances are they can't bring coffee lake down to under 100w for 6 cores at reasonable clock on mainstream and 10nm is late

expect 4.6Ghz i7 8700K 4 core. again.

I can expect them intentionally lower i5 clocks so it would make 6 core look better though, we are in uncharted waters
hilarity will be whole year round starting with skylake x, intel gets retarded when not ahead

>>60016820
L3 might fit, if amd can give you 16mb for $190
why can't intel?

>>60016887
Because lmao monolithic design in 2016+1.

File: worried '''''merchant''''''.png (129KB, 354x504px) Image search: [Google]

129KB, 354x504px

>>60015945
DELETE THIS

File: housefire inside.png (126KB, 357x330px) Image search: [Google]

126KB, 357x330px

>>60012523
Over 9000 hours in MS Paint

>>60016887
Because moving from 2 to 2.5MB per core on the same architecture makes no sense.

File: amdbtfo.png (310KB, 503x868px) Image search: [Google]

310KB, 503x868px

when will they learn

>>60016988
you only need 2 cores goyim

>>60016851
Numbers don't seem to match up

1280 v5 80w 3.7ghz 4 core
1280 v6 72w 3.9ghz 4 core
A v7 Quad should be something like 4.1ghz 64w or just 4.1ghz 72w at the least for 14nm++

1280 v7 108w 4.1ghz 6 core
or
1280 v7 96w 4.1ghz 6 core

6x2.5MB from 6x2.0MB?

>>60016988
What exactly are you trying to argue here?

File: shocked-intel-shill.png (246KB, 882x758px) Image search: [Google]

246KB, 882x758px

>>60012278
MOAR COREZ! MOAR GIGAHURDZ! MOAR BIBELINES!

>>60012670
>Infinity Fabric can have a bandwidth up to 100GB/s
It's over Intel is finished.

>Chinese
It's a fucking rumor. Do you even need to ask?

>>60017051
and your a fucking faggot intelcuck

>>60016727
>140W mainstream part

Yeah, nah, not happening.
Intel isn't that insane yet

File: 1478352363789.jpg (37KB, 281x321px) Image search: [Google]

37KB, 281x321px

>>60017019
here is this thing about new xeons

>>60017067
I meant to quote >>60016727

>>60017099
>28c@2500mhz is 205W rated TDP
Hot. Gotta throw moar ring buses.

File: 12345241332.png (438KB, 1064x1619px) Image search: [Google]

438KB, 1064x1619px

>>60017099

According to the 7700T specs if you scaled down to 2.9ghz, you get 35w TDP for 4 cores ( TDP doesn't include turbo )

256+2048 vs 1024+1408 cache per core

18C 200W at 3.0ghz base seems around right
6136 must be made up?
6154 seems more realistic

>>60017263
>6136 must be made up?
It's not made up when you find out some Skylake Xeons don't have 6 channel memory nor AVX512

>>60012735
Well Fermi was an flatout failure on Nvidia side.

They launched Kepler without the hardware schedulers and async computing and moved more into energy efficiency. AMD on the other hand, invested on improving the schedulers and async computing.

Now Nvidia is trying to make time to patch the hardware back again, and try to beat AMD and not launch a Fermi 2.0 with DX12 @14nm

Thing is, I hope Nvidia can be as efficient as AMD on that regard, if this happens, Nividia will have a hard time with their gimpworks approach since the way the schedulers work would also benefit AMD. They would have to implement a "GENUINE_Nvidia" kinda of thing, and we all know what would happen.

>>60017496
>GENUINE_Nvidia
As if gimpworks weren't bad enough.

How much will the Skylake hexacores be?

I'm strongly considering one after the failure that is 1800X

>>60017613
>after the failure that is 1800X
oiiiiiiiiiiii veeeeeeeeeeeeeey

File: maxresdefault.jpg (27KB, 1440x1080px) Image search: [Google]

27KB, 1440x1080px

>>60017613

>>60012893
>>60012728
>>60012685
>>60012650
>>60012540

From the 7000 series to the 200 series it was basically ADD MOAR CUDA CORES.

Yes, the 400 series was a massive arch change that didn't play out as expected, then we got Kepler wich was another big arch change, not massive since it was mostly a rolldown, yet we can all say they are "the same arch" just as we do with intel and the Core series.

Then Mantle/Vulkan/DX12/Async/Ps4Xbone/AMDGNC happened and Nvidia was caught with their pants down.

Now Nvidia took Maxwell and added some frisky neat tricks to emulate the Async hardware @ driver level but that ins't paying out any better, Volta must be an extremely efficent Async ready card.

I really expect a 230-250w TDP "1180" and a 250-270w TDP Titan/80Ti...as best case scenario.

They would have to invest SO much money to have great yields, a good amount of DP power and the Async hardware.

As i said on >>60017496

AMD invested 3-4 gen's of GCN in making that hardware efficient enough, it would be a miracle for Nvidia to make it in 1.5 gens...its possible, but i can't imagine the money involved.

t. RX470 owner

>>60012278
Isn't AMD still rumored to be releasing a 16-core HEDT platform?

>>60017613
Enjoy your Broadwell IPC

>>60017662
W-why would they d-do that?
16 cores? What good is 8 cores? Nobody really needs more than 4 cores and 2 cores is optimal

>>60017662

All i want is a 6 core 6 thread APU with a gpu between the 450-460/1050

>>60018316
Raven Ridge tops at 4/8.

>>60018316
Maybe next year.
The current 12CU Raven Ridge APU seems to have somewhere around the 460's performance though.

>>60017660
>Apparently Nvidia is bad at making arches and just makes moar coars
>Still better than AMD
really gets my noggin joggin

>>60019027
Well...you got far too little reading comprehension and just stuck with my first sentence.

.0005₹ have been added to your account.

File: 32521.png (9KB, 213x237px) Image search: [Google]

9KB, 213x237px

>>60019059
>Y-You're just a shill

>>60019078

Well if you read, i wrote a pretty comprehensive explanaition about the nvidia arch and never mentioned AMD besides the gamble they took with improving the Async hardware.

If all i should be an Nvidia fanboy since im expecting Volta and have an 470 due to my GTX 670 dying last year.

Then again you are a shill for having 0 reading comprehension.

>>60012523

>140W
>10c/20t chip

Intel is out of its god damned mind. All AMD has to do is release a 12c/24t chip at 3.4GHz and it'll match Skylake in single-threaded performance and utterly BTFO it in multi-threaded performance, while still being well under 120W.

>>60017099

>28C/56t
>2.5GHz
>38.5MB Cache
>205W

Uh... AMD can release a 32c/64t Naples chip at 3GHz for 180W with an even higher cache size. Xeons gonna get BTFO in their own game across all specs.

So where's the x299 boards?

>>60020407
July

>>60018028
To run 4k or even 8k massively threaded coding is required. Yes, 16C/32T would be necessary for 8k.

>>60020454
Higher resolutions are less CPU intensive though

>>60019835
>>60019389
What excuse will you guys come up with when naples inevitable doesn't beat xeons
Is it gonna be microsofts fault? Nvidia? Intel? You're using it wrong?

>>60020746
My god, someone's this retarded.

>>60020773
It's true though
Or are we talking about encoding?

>>60020765
It already beats it in 5 out of 6 metrics, I/O, memory capacity, bandwidth, and power efficiency.
Only thing left is pure performance and there's no way Intel will have any decent clocks at 28+ cores.

>>60020797
4 out of 5

>>60020797
So what is your excuse going to be when Naples inevitably loses to xeons

File: AMD-Ryzen-7-1800X-UnixBench-whetstone-Benchmarks.jpg (69KB, 856x548px) Image search: [Google]

69KB, 856x548px

>>60020811
It's okay anon, you can still take preemptive painkillers, it'll hurt less.

>>60020853
So what is your excuse going to be when Naples inevitable loses to xeons

>>60020876
anon, are you okay? short memory stutter there?

>>60020916
Will you have an excuse or are you going to have to make something up on the spot?

>>60020930
it's already wining vs current xeons
new xeons are not coming until mid autumn

>>60020853
Why not the E5-1680 v3 or v4 ( 8 cores and frequency closer to the 1800X )

E5-1660 v3 / v4 should be vs the 1700

>>60020930
You seem unstable, please disconnect from the internet for a few hours.

>>60020964
>>60020951
We should blame Microsoft
Seems like the way to go, or say Intel is sabotaging the server line

>>60020963
Why don't you ask Patrick Kennedy? He runs with the server hardware he has.
I mean did you really miss the 16 core xeons and dual 10 core Xeons there? There's even XeonPhis and AWS servers.

>>60020853
>what are real world tests
>gimme them synthetics boi

>>60020985
https://www.servethehome.com/amd-ryzen-7-1800x-linux-benchmarks-paying-for-speed/

File: Screen Shot 2017-04-16 at 23.51.07-fullpage.png (2MB, 1480x8356px) Image search: [Google]

2MB, 1480x8356px

>>60021001
Doesn't count amritie

>>60021048
>1800x crashed
>no xeons
Lmao

File: Screenshot from 17-04-2017 00:18:16.png (212KB, 977x688px) Image search: [Google]

212KB, 977x688px

>>60021048
>>60021048
Lol 6900k

>>60020985
nvm
1800X at 73000 ( 3.6 - 4.0 ) + XFR?
1700X at 61000 ( 3.4 - 3.8 ) + XFR?
E3 1660 V3 at 67000 ( 3.0 - 3.5 Haswell )
E3 1680 V3 is 3.2 - 3.8
E3 1660 V4 is 3.4 - 4.0

>>60021110
E5-1680 V4

>>60021103
>Bulldozer vs $1000 BDW-E
Lmao

File: 1471111385635.png (98KB, 612x491px) Image search: [Google]

98KB, 612x491px

>>60012383
140 watt TDP, you serious ?
What the actual fuck do they need that much power for ?
That is worse then the oc'ed fx-8350 i had

Finally.

My next pc was waiting for Naples vs skylake-x . Good to see progress. In the PCIE dept I want actual benches before I make up my mind.

>>60021156
This
Tired of AMD selling me promises
I want results

File: Screenshot from 15-04-2017 01:35:14.png (72KB, 939x891px) Image search: [Google]

72KB, 939x891px

>>60021088
>post Loonix benches
>YOU CAN'T COMPARE 8 CORES AGAINST 16+ CORES IN MT BENCHES BECAAAAAUSE NOT AT THE SAME CLOCKS INTELS SCALING WITH TWO RING BUSES ISN'T SHIT I SWEAR

>post Wangblows day 1 benches on Tom's shitware
>YOU CAN'T COMPARE 6900K BECAUSE ITS NOT XEON BAAAAAAW


Any more assimplosions from you?

>>60021103
I wonder if there is anyone on earth who actually bought a 6900k?
I seriously don't think anyone bought it

>>60021103
>>60021184
It's hilarious but it's probably some bug, which didn't get fixed for whatever reason.
Really weird since it's part of SPEC

>>60020853
>>60021110
~70000 for it was a Broadwell at V3's frequencies
E5 1660 V4 is 3.2 - 3.8 should overtake the 1800X
E5 1680 V4 at ~78000-80000

>>60021183
What are you talking about aspie?

>>60020765
>>60020876
>>60020930
>>60020951
>>60021001
>>60021088
>>60021209
This is goy of EXCELLENT quality.

>>60021215
What?

>>60021183
The xeon version of the 6900K would beat the 6950X by 4fps due to higher clocks
3.4 - 4.0 vs 3.2 - 3.7

>>60012278
INTEL DID IT! THEY RELEASED THE 7700K AGAIN WITH NO IGPU, HIGHER TDP AND AT 350 DOLLARS

PS: YOU HAVE TO BUY A NEW EXTREMELY EXPENSIVE MOTHERBOARD


AHAHAHAHAHAHAAHAHAHAHAHAHA

File: Intel-Xeon-E3-1240-V6-NAMD-Benchmark.jpg (83KB, 858x555px) Image search: [Google]

83KB, 858x555px

>>60021206
>should
>if
>but only if I overcloked my housefire Xeon it would win but use twice the power!

If you want your shitty corelet at high frequencies he did tests for that too, surprise it does crap in a bench that uses a lot of cores( 50k cores in fact)

File: 1485199459745.jpg (83KB, 600x617px) Image search: [Google]

83KB, 600x617px

>>60021137
To be fair, FX-9590's initial price was like $900.
So it is a more-or-less fair comparison.

>>60020811
>>60020765
Just kill yourself already.

>>60021299
It's also a 2013 CPU that now costs like $200

>>60021299
it wasn't, MSRP was ~300
there was some retail fuckery so real price got to $900 for some magical reason

File: Intel-Xeon-E3-1240-V6-Sysbench-Single-Thread-Benchmark.jpg (88KB, 852x587px) Image search: [Google]

88KB, 852x587px

It's really weird how a Intel optimized SINGLE THREAD bench Intel gets murdered, the application can probably make use of AMD's 4 complex decoders.

File: Intel-Xeon-E3-1200-V6-Value-Comparison-Summary.jpg (90KB, 1079x234px) Image search: [Google]

90KB, 1079x234px

>>60021279
https://www.servethehome.com/intel-xeon-e3-1240-v6-linux-benchmarks-review-first-kaby-lake-xeon-e3-cpu/
Fresh outta the foundry, literally released days ago.

>>60015463
>ryzen comes out and suddenly every intel shill on /g/ is an engineer

>>60021341
>the application can probably make use of AMD's 4 complex decoders.
Thank you based GCC

>>60021404
GCC is fucking garbage ICC beats it any day of the week

>>60021436
I'm gonna cripple your face like ICC crippled AMD.

>>60021436
>mouth frothing Intel in delirium

http://www.agner.org/optimize/blog/read.php?i=49#829

>>60021374
>4c/8c
oh intel when will you learn

File: Xeon-E7-v4-Broadwell-EX-CPU-Block-Diagram.png (75KB, 1743x940px) Image search: [Google]

75KB, 1743x940px

J-just wait, a-a-a-anoter ring b-bus!

>>60021552
Ring buses are circular. CCXs are square. Circles > Squares. It's over AMD is ded.

>>60021571
Flawless logic AMD is dead

File: AMD.png (91KB, 653x726px) Image search: [Google]

91KB, 653x726px

I should wait for zen+

File: pic2.jpg (26KB, 434x203px) Image search: [Google]

26KB, 434x203px

>>60021552
>the city of Ring Bus
28 cores look good on paper but when they're barely faster than 22 cores because the ring bus can't traverse enough data without using retarded amounts of power it's just a waste of silicon.

>>60021341
0.75 vs 0.90 isn't too far away with the clocks and ipc deficit and even 2400 vs 2133 memory

File: Screenshot from 22-04-2017 16:06:05.png (158KB, 2033x909px) Image search: [Google]

158KB, 2033x909px

Oh guys have you seen how awesome those shiny new Skylake Xeons are? Based Intel

>>60021571
They're both rectangular.
I guess Intel is two rectangles.

INTEL WINS AGAIN

>>60021571
there's a reason why they made the LHC a ring and not a square, electrons can't turn sharp corners easily

AMD is bankrupt

>>60021676
Nice try fag, Intel 28 cores only has six rectangles, AMD 32 cores is eight rectangles.

AMD wins

>>60021624
>naples is bretty guud :----D
>what do we do
>make a larger bingbus :--------D
>ebin :----------------D

>>60021655
so they're achieving better multi-core perf with 60 threads than they did with 66 threads?

Vega Reviews when?

File: Screenshot from 05-03-2017 23:38:48.png (62KB, 957x532px) Image search: [Google]

62KB, 957x532px

>>60021655
MASSIVE SHOCK!

Wait a minute while I regain control of myself, I was expecting like 20% IPC!

>>60021713
At higher clocks, on a 'new' architecture 2 years latter.

10% MT performance per 2 sockets isn't gonna help them shit, no Haswell rollout would even care about upgrading.

>>60012334
ROASTED

>>60020853
Dayum, this bench took 50 minutes to finish on my i7-4720HQ

>>60021650
What are you talking about? Look at the regular 1700 there, it's single core turbo is 800MHz lower than the 7700k and it still roasts it.

File: 1489873082986.png (357KB, 640x480px) Image search: [Google]

357KB, 640x480px

>>60015671
>Making categorical statements about IPC
>When IPC is not a constant measurement
What is it that this creature could have intended to convey with such actions?

>>60021279
>>60021849
Meant the other graph

>>60012278
> extreme performance
... in highly threaded workloads and not much better than the 8c16c Ryzen on Linux.

> Mega-Tasking
Mega cringe

> Ultimate Platform
Ultimate as in, this is last one of this socket. Next processor will need a new socket, so prepare your shekels goy!

>people in this thread keep trash talking Intel
>this is after years of AMD being trash (since Pentium 4, so 17 years in total so far)
>Ryzen comes out
>"wow it's actually good!"
>AMD actually came up with something good after 17 years
>people are mad at Intel for not having a direct response to Ryzen because they assumed Ryzen would be a fluke (and I don't blame them)
>Intel suddenly on full damage control
>they actually have to come up with something good
>they can't lower prices because that would just be admitting that Ryzen is better
>they have until November to make sure that Coffee Lake beats Ryzen
Mark my words, when Intel comes out with the first 10nm desktop process, you guys are going to be pissed as fuck when Intel:
>comes up with something many times better than Ryzen
>makes it more expensive than Ryzen
>blows AMD the fuck out such that AMD won't have a response to Intel's new architecture, not even with Ryzen 2
>Intel owns the CPU market for many years to come while jewing the fuck out of everybody

>>60022848
>comes up with something many times better than Ryzen
>blows AMD the fuck out such that AMD won't have a response to Intel's new architecture, not even with Ryzen 2
intel have the money but they don't have the management or engineering talent to do something like that.

the only reason they've dominated the market is because bulldozer was a flop, not because they were better. they've been recycling the core architecture for over a decade now

>>60022848
I'm terrified of a canceled desktop Skylake dieshrink
Like literally shaking and shit

>>60022877
>don't have the engineering talent
Neither did AMD, and they were so desperate they literally had to use a good chunk of their money to hire Jim Keller to come up with a CPU that many are wrongly calling the Intel killer.

You also have all of these blind fanboys saying that Skylake-X is something Intel has had waiting in warehouses in anticipation of Ryzen, which is totally wrong. If that were the case, Intel would have started releasing them not too long after Ryzen's release and for the same price, if not cheaper, as well.

Intel actually has to come up with a response for once, which is shocking not only to the company, but to the rest of the world.

>>60022945
>Neither did AMD
oh they do, they hired jim keller to lead the teams not to design the architecture.

people give him way too much credit because he wasn't -that- involved in the design, he lended his expertise to the pre-existing talent at the company

>>60022992
Good management is crucial if you're dealing with some 100+ people designing a high performance SoC
There's no better guy than Keller for this role.

>>60012278
>Intel is for reals going the MOAR CORES route

>>60023011
which was exactly what mark papermaster thought. he's an incredibly smart guy and his move to bring keller back was significant in boosting the spirits of the talented engineering teams. there are amd employees who have openly said they probably would have left the company were it not for jim keller joining again and restructuring shit

>>60021254
Hey there, newfriend :)))))
I'll help you out, little guy!
In Hebrew, "goy" means sheep :^)
So go have fun out there on the 4chans!

>>60022848
>Mark my words, when Intel comes out with the first 10nm desktop process, you guys are going to be pissed as fuck when Intel
SLOWER THAN 14NM

>>60022848
They already announced 10nm will be slower than 14nm at first you fucking kike cocksucker. Intel has no tricks up their sleeve. You're the worst kind of fanboy that just assumes Intel will magically improve their dead-end architecture despite all evidence to the contrary.

>>60023093
>In Hebrew, "goy" means sheep :^)
And you're calling me a newfag? goy means someone who isn't jewish you fucking tard

>>60023258
>Newfag doesn't know the real meaning of "goy"
W E W L A D
E
W
L
A
D

>>60023258
goy is short for goyim which refers to a non-jew, hence why jews say "good goy"
There is no "real meaning", you just got BTFO and couldn't think of anything else to say, fucking moron

>>60023344
>>60023293

BORN JUST IN TIME TO SEE THE BIRTH OF THE INFINITE CORES ERA

20 core Ryzen when?

>>60023364
Never?
High core AMD line is Zen
wonder if that means Ryzen will be considered corelet tier

>>60023364
No such thing.

Ryzen is only gonna be 12, 16, 22(big doubts of this happening), 24, 28 and 32 cores.

Anything in between is better re purposed as not an Opteron.

>>60023360
The sheep definition is a meme, get with the times.

>>60021143
>>60012383
Finally I can make use of my 650watt PSU

>>60015877
The main issue is, believe it or not, these retarded game Devs check the CPUID, assume since it's AMD it's bulldozer, and optimize for bulldozer which fucks Ryzen performance.

It, yes, wasn't the CCX.

Most game devs should be lined up and shot.

>>60016076
>Do individual devs now actually have to care that their threads run on a single CCX
No. Applications make requests to the OS to do the scheduling, even with User Mode Scheduling.

Windows should have already patched by now to prefer to keep applications threads on the same CCX when possible.
But some games run like shit by assuming Ryzen is Bulldozer because of shit spaghetti code.

>>60016727
>Can anybody confirm pic realted?
They just stole it from my posts on 4chan.

It seems like the likely strategy if Intel wants to remain competitive, but who says they want to remain competitive?
It's just as likely that they simply increase clocks by 100MHz, make TIM worse, and make bigger housefires.

File: Reply_or_your_mother_dies.jpg (48KB, 676x426px) Image search: [Google]

48KB, 676x426px

Where is the ring bus meme coming from?

>>60025031
Ryzen's CCX units are designed to scale. Intel's ring bus really isn't. The more cores they add, the worse performance and the higher the power draw.

>>60021374

>integrated GPU Xeon

For what purpose

>>60025266
Just worse performance, ring bus is actually not that power hungry, but due to lacking throughput, it scales terribly.
Crankings its clocks up is beyond counterproductive as it does literally what it was never designed to do, sip more power than the cores its managing.

>>60012383
I have a 800 wat pcu. What do I care? Head room for days, amdumbs just mad they lost the core race.

>>60025601
Can't tell if trolling or just retarded.

>>60021552
>>60021624
Ringbusses are flawed and ATi learned about their downsides years ago, which is why they've never used it since.

>>60012278
How many thousands will this cost? Wait it won't because someone seems to be pressuring Intel. All the more reason to support amd.

>>60012365
And make sure to hire based solely on race and sex. Business is definitely a place for feel good decisions on key staff members.

>>60012443
That bought a lot of coke, guns, and hookers.

File: 14nm.jpg (186KB, 1529x836px) Image search: [Google]

186KB, 1529x836px

>>60023158

Cannonlake-U clocks better than Kabylake-U

>>60026107
http://www.infoworld.com/article/3186781/processors/intels-cannonlake-pc-chip-shipments-may-slip-into-next-year.html

In theory, the 10-nm chips should be faster than the 14-nm chips. But the first low-power Cannonlake chips will have fewer transistors, and won't be comparable to the mature 14-nm chips with more transistors.

File: 131825923493.jpg (19KB, 347x464px) Image search: [Google]

19KB, 347x464px

Intel's 8 core currently cost $1100 and 10 core $1900
At that point going xeon might be a better idea

>>60026359
meanwhile amd will probably release their 16c/32t HEDT at $1300 tops

>>60026390

Doubt it. If Intel releases a 10c/20t Skylake-X to the consumer market, AMD's answer would be a 12/24t chip to counter that via Zen2 next year for $600-700. A 4c/8t jump to 16/32t @ 1600 would be way too big a jump.

I imagine that a 16c/32t HEDT part would come in closer at $1000, $1200 max w/ a 18c/36t part.

Zen has basically caught Intel with its pants down and a dildo up its butt. Its also gotten really fat over the years with no competition in the market. As a result, while they have gobs and gobs of money to be able to counter Zen. They unfortunately lack the time to design a new microarch that can do price/perf AND core/thread better than Zen can--because Intel's been playing the same tune that fucked them during the Pentium 4 days; GHz wars.

Skylake-X and Kabylake-X both scream of "we need to buy TIME to design a countering microarch to Zen that can match core/thread at perf/watt; so here's a stop gap part until that happens." A 10c/20t Skylake-X is devouring 140W TDP. Meanwhile, Zen's 8c/16t 1800X is chilling at 95W TDP. 2c/4t between 1800X and Skylake-X but with a 45W divide.

If AMD downclocks Zen to 3GHz, they can throw a 16/32t part at Skylake-X in the consumer market at 140W TDP. What's Intel gonna do? Launch a 16c/32t part too at 200W TDP? The market will eat them alive, and mobo manufacturers will tell them to fuck off.

>>60026532
https://www.extremetech.com/computing/246232-rumor-amd-may-planning-16-core-32-thread-ryzen-titan-challenge-intels-broadwell-e-skylake-x

https://www.pcgamesn.com/amd/amd-ryzen-16-core-cpu

https://www.techpowerup.com/231911/amd-16-core-ryzen-a-multi-chip-module-of-two-summit-ridge-dies

also, the HEDT version is at 180W, the naples one will use less power, at 120-140W

>>60026609

>$1000

Right on the fucking money. All I want is for AMD to release a Zen2 R7 2700 with 12 cores and 24 threads at 3.4GHz. I will cruise on that platform all the way into 2022 if that happens. That's enough IPC per core and enough cores with SMT threads to maximize Vulkan/DX12 scaling for the entirety of the Scorpio/PS4 Pro life cycle.

And then I'll upgrade again with Zen3/4 to a 18c/36t or 24c/48t chip again and the cycle will repeat. Not to mention, its damn near guaranteed that if Xbox Scorpio and PS4 Pro are Jaguar/Polaris, then Xbox Scorpio II and PS5 will be Zen2 with high end Vega.

>>60026690
>I'll upgrade again with Zen3/4 to a 18c/36t or 24c/48t
anon

>>60026718

What? I said 2022. Its fair to expect a 18c/36t consumer chip by 2022. Five years. So I imagine we're looking at either a mature 10nm node or an introduction of a 7nm node. If AMD managed to do 8c/16t at 95W on a new 14nm node, a mature 14nm node should be able to do 16c/32t at 140W TDP easy.

Drop from 14nm to 10nm, and you'd be able to scale to 18c/36t at 120-140W TDP at equal to slightly lower clocks.

>>60026718
well, there will be a new socket after AM4, it will probably come in DDR5 too

>>60026609
Snowy Owl will be <100w

>>60026690
you're retarded.
2700 will be 8 cores still.
12 and 16 core are still first generation Zen parts

>>60026359
>Intel's 8 core currently cost $1100 and 10 core $1900

The 10-core 6950x is $1,650. Still overpriced but lets not start making shit up.

File: prices.png (11KB, 1328x103px) Image search: [Google]

11KB, 1328x103px

>>60027015
apologies, i expected the price in dollar to be close to european prices

>>60027059
Remember you guys have ~20% VAT included in the price, while in the US sales taxes are customarily not included in the list price, since they vary so much from state to state. Some states don't even have one.

>>60012278
Go back to school. Maybe then you will learn something. For example: placement of commas.

>>60015141
you'd have to be a fucking moron to do any sort of business with an israeli tech company. probably one of the managers is personally benefiting from it and he doesn't care about intel more than to suck it dry

>>60022848
>10nm
SLOWER THAN 14NM AHAHAHAHAHAHHAAHAHHAHAHHAAHHAHAHAH

>>60023037
>his name is actually mark papermaster
lol

I'm still on a [email protected] and just waiting for better than 7700K HEVC hardware decoders and iGPU which means I would probably have to get an Intel CPU since AMD just simply lacks iGPU.

>>60027841
Raven Ridge, stupid.

>>60027929
"what if" amd cpu.

>>60027943
what

>>60027943
>if
HUEHUEHUEHUE

File: Pepe_Reeeeee.gif (1MB, 680x680px) Image search: [Google]

1MB, 680x680px

>>60012278
Tired of this shit cores and more cores what I want is a processor that goes to 10Ghz with only two cores 4 at most.

Since 2005 there is no improvements in the technology they are just tricking people adding cores.

We must accept that silica id dying, there is no way of breaking through 5Ghz without melting your house.

>>60012443
>>I guess in 2010 it seemed like a good idea.
McAfee's name was in the gutter since long before 2010.

>>60021700
AMD moves electrons in crosses
DEUS VULT

god on AMD side. can't lose.

>>60022992
>>60022945
>tfw you realize papermaster is the real mastermind strategist and zen arch was lead by a woman with actual lead architect - michael clark

keller played a big role, but it wasn't as MAJOR as everyone makes it out to be

>>60028556

AMD's CPU division under the bulldozer arch and subsequent revisions was pretty much cancer. Keller was bought in to restructure the division into a form that could design the thing that became Zen. It's almost guaranteed that Zen would have been impossible without Keller, Certified Shit Wrecker, walking in through AMD's front doors and basically telling every mid and senior level manager that they're a buncha fucking retards with no clear direction. Then redoing everything so that communication was tip top from top to bottom and bottom to top. So that talent wasn't wasted, and so that every engineer could excel within the limited scope of resources available.

Also, to have Jim Keller on board leading the Zen project no doubt did wonders to the morale to AMD's engineering teams that did work on Zen. Every engineer went to work in the 3 years that Keller worked at AMD, wanting to put in 200%.

Zen was a team effort, but Keller is like a hero unit. Using him can turn the tides of war if used correctly.

>>60027184
Maybe you will learn how not to fall for jewish tricks goyim

>>60021215
Intel can use JUST WAIT, its a AMD trademark