AUDIBLE FUCKING KEK INTEL The Platinum 8180 costs $12000 a

>>61103494
ok

File: JUST krzanich.png (1MB, 1002x1480px) Image search: [Google]

1MB, 1002x1480px

>>61103494
YAMEROOOO

File: Screenshot from 27-06-2017 12:53:49.png (386KB, 1794x901px) Image search: [Google]

386KB, 1794x901px

>>61103494
Also it was already obvious but memory bandwidth AMD demolishes.

File: buy now.png (227KB, 328x330px) Image search: [Google]

227KB, 328x330px

>>61103494
>>61103511

>>61103494
>>61103511
DELET

Intel is stable

>>61103972
El ohe l

>>61103972
Are they? According to who? Because Intel has been fucking up a lot lately

>>61103494
E7 Xeons scale up to 8 sockets while EPYC is only 2 socket max.

File: stateofintel.png (458KB, 615x450px) Image search: [Google]

458KB, 615x450px

>>61103972
yes, their Temps are stable @100°C

NOOOOOOOOOOOOOOOOOOO

>>61103998
And their scaling per socket is shitty for anything that needs to cross the QPI links.

>>61103998
>>61104451
>MOAR SOCKETS!
lmao. This needs a spurdo edit.

>>61103998
oh no. more useless 8p servers.

>>61104024
https://chiru.no/u/average_intel_owner.webm

>>61103494
>>61103511

ryzen already punted for the server market which is where the cake's at, explains the rushed release of x299 and dmg control

workstation epyc fucking when

>>61104759
http://www.thinkmate.com/inside/articles/amd-epyc-7000

>>61103972
LEL LEL LEL

https://hothardware.com/news/critical-flaw-in-intel-skylake-and-kaby-lake-hyperthreading-discovered-requiring-bios-microcode-fix

>>61104805
Those are racks, OEMs haven't released EPYC towers yet, supermicro has them upcoming.

https://www.supermicro.nl/products/nfo/AMD_SP3.cfm?pg=SS

>>61104853
It's absolutely nothing, retard

>>61104873
KEK
Being stable is easy with eyes closed

>>61104910
Do you compile OCaml on model 78 or 94, stepping 3? No?

>>61104759
Tweet to AMD here https://twitter.com/AMDServer I already sent them one asking them to make a workstation motherboard with decent power VRM and the ability to set the CPU multiplier. They're kind of clueless about the workstation market, they don't even understand people would buy 32 core epyc workstation if they provide a solution.

>>61103494
hmm, to buy a single intel cpu or 2 complete epyc systems?

File: (((intel))) customer.jpg (21KB, 300x290px) Image search: [Google]

21KB, 300x290px

>>61104873
hahahahahahahahahahahahahaha *inhales* hahahahahahahahahahahahahahahahahahahahahahahahahaha

>>61104873
Like the bricked atoms? freezing skylakes? disabled TSX? sata corruptions? usb sleep bugs? 2MB bugs?

Intel inside, errata outside

>>61105137
also FDIV bug

>>61103494
http://vocaroo.com/i/s1p98Qh2if6B

>>61105152
https://en.wikipedia.org/wiki/Pentium_F00F_bug

People seem to forget the F00F one

File: 1496968759987.jpg (33KB, 597x430px) Image search: [Google]

33KB, 597x430px

>>61103998

>>61104984
Harry, can you do an inventory on whether our company ran any business critical processes where the compiled code ran short loops of less than 64 instructions and used the AH, BH, CH or DH registers as well or their corresponding wider register (e.g. RAX, EAX or AX for AH) on a Skylake or Kaby Lake platform?

If we did then we might have silently gotten faulty data without ever knowing. Please report back on monday, thanks!

>>61105349
No problem! The patch was out in April, and it would have crashed the system when compiling, not months later, boss!
>>61105098
There is no difference between you and the fags calling out ryzen for having to have patches to get the memory working properly.

>>61105384
>The patch was out in April
Not on all steppings. No fixes for most the kaby lake motherboards.

>>61105400
It will be resolved. I'm old enough to remember windows 95 was a no-go on k6.

>>61105384
>No problem! The patch was out in April
Sorry, that's not satisfactory. The management and the whole analytics department has been using HP Skylakes since last summer. I need you to go back and verify ALL the data since then.
>it would have crashed the system when compiling, not months later, boss!
I have short patience with my employees lying to me Harry. I'm this close to firing you now.

Sorry to put all of this in your lap Harry, maybe if Intel would have notified us of the potential magnitude of this problem instead of trying to sweep it under the rug and hope no one noticed... Trust is a fragile thing Harry, remember that.

>>61105484
It's a good thing we picked Intel over opterons, that hypervisor bug would have bankrupted us, or that vesa local bus bug on piledriver, or the agp issues with athalons. Quit kidding yourself, boss.

File: 1433276589653.gif (2MB, 388x356px) Image search: [Google]

2MB, 388x356px

>>61105570
>athalons

>>61105570
Harry, I'm tired of your excuses for not delivering. You've become a liability so I'm moving you to the basement and we're reclaiming your red stapler too. In addition I'm moving everything to Power8 + Hurd.

>>61105609
Hey boss, can I have Harry's red stapler? Also, I think we should be making up migration plans to Temple OS.

>>61105609
I like the cut of your jib. Let's order 50 Talos workstations.

>>61105625
Sure, make sure the TPS reports evaluting the move to Power8 + Hurd + TempleOS are on my desk by monday, and the stapler is yours. You will find that the migration will happen and it will cost no more than $1 million. Thanks.

>>61105609
I'm leaking the pics of you and Mike at the company retreat "playing around" in your fursuits on the pool table.

>>61103494
so the flagship Epyc literally beats the best silicon Intel has, bar none. that's kinda fucked up.

can't wait until GloFo moves to 7nm and Zen+ clocks reach KBL levels. shit's going to get ridiculous.

>>61103494
OY VEY
THIS CANT BE HAPPENING
DELET THIS GOY

>>61103494
excuse me i dont know much about servers but why do you need to get four intel cpus to beat two amd cpus

>>61106034
Intel's CPUs scale like shit. Their multi-CPU setups scale even worse.

>>61106081
As if there was any other option

>>61106034
Intel has just been caught with their pants down- they have an old design that they've been iteratively improving and then AMD came around with a brand new design.

>>61106102
There is now.

>>61106034
The interconnect between the processors is inferior so performance is negatively affected whenever the application has to share resources between processors. Zen has some amazing interconnect that has almost 1:1 scaling.

>>61103494
>>61103511

OY to the VEY

File: 1490730560541.jpg (318KB, 882x758px) Image search: [Google]

318KB, 882x758px

>>61103494
NOOOOOOOOOOOOOOOOOOOOO

>>61103494
B T F O

>>61103494
>2 32 cores barely beats 2 28 cores
wow it's nothing

>>61106711
please read op you fucking intelfag
>The Platinum 8180 costs $12000 a pop and it's getting stomped by a $4000 EPYC
>$12000 is getting stomped by $4000

>>61106750
>$12000 because it's 4S+ scalable
>the $4000 EBIN isn't
Fucking retard

>>61106782
Intel 4S is dogshit though.

File: big_intel-xeon-platinum-8176-cinebench-r15-benchmark.jpg.jpg (281KB, 980x942px) Image search: [Google]

281KB, 980x942px

>>61106782
>muh scalable
>scales like shit on dual socket let alone on quad socket

>>61106711

check the clock speed

>>61106818
>>61106804
At least it can scale to 4/8S
Some datacenters simply need as many cores as they can stuff into a single blade server, memory and IOs be damned.

>>61106711
why would you give a shit about >muh cores at this point
AMD is cheaper and faster

>>61106841
80% of the server market is 2S so I think AMD is making the right call here purely from a marketplace perspective.

I hope the stock drops on Q2 earnings so I can buy a fuckload of it. I'm pretty sure Ryzen 5 won't have changed anything, not even miners should have changed it that much.

>>61106924
t. cuck that didn't buy in at $2

>>61106942
Yes, it still haunts me. I wanted to buy too, but I was a poor student at the time. Now I'm a full blown wagecuck with some money to spare.

>>61106924
If it drops it won't drop much.
If ER is good it will go over $18 and you'll be in a even worse situation

File: 1413913323099.png (162KB, 1000x1000px) Image search: [Google]

162KB, 1000x1000px

>>61103494
delid dis

>>61107040
t. rajesh

File: india inside.jpg (353KB, 1500x1164px) Image search: [Google]

353KB, 1500x1164px

>>61107341

>>61107440
they merely adopted the designated shitting streets
AMD was born in it, molded by it

File: f.gif (564KB, 800x430px) Image search: [Google]

564KB, 800x430px

>>61104873

File: 1497891461086.jpg (54KB, 679x758px) Image search: [Google]

54KB, 679x758px

>>61103494
PLZ BUY OUR CPU, GUYS!!!!

>>61103494
>Average Score: 283.13 GB/s
>Average Capacity: 162.00 MB
What are these metrics?

>>61107642
Made up numbers to make Intel look worse than it is

>>61107655
Good goy! Remember to buy our Xeon Scalable Family processors. They are proven and battle-tested!

>>61106818
If you're going to post about how badly it scales, you should include some information to deduce that. What's the corresponding 1S score?

>>61103494
While it's impressive that they're managing the beat Intel at those clocks, I have to say I was expecting higher clocks from EPYC given the supposed efficiency of Zen.

>>61107800
Uncore is not free and they want to fir into 180w TDP. Intel's 28c is 205w.

>>61103494
DELID DIS
>DELID DIS
DELID DIS
>DELID DIS

File: 1498560837050.png (18KB, 246x120px) Image search: [Google]

18KB, 246x120px

>>61103511
wait a minute...

>Literally no competition for 10 years
>Implying intel doesn't have a lineup of massive performance improvements to dish out as soon as they get some real competition
I'm going to enjoy buying up cheap intel stock and selling it for big bux.

>>61109072
Yes-yes, goy, we TOTALLY have a completely new super-incredible x86 uarch. Believe us.

>>61103506
>White CEOs
>ever
LMAO

>>61109041
is this... a...

>>61109072
Source: My uncle works at Nintendo.

>>61109072
So that's why Intel decided to create Skylake-X. They wanted to heat things up before releasing this new uber architecture that they totally have

>>61109310
>heat things up
Pun intended.

>>61109072
This idea only shows a complete lack of understanding of how the semiconductor business works. It's never a good idea to sit on performance improvements. If you can make a smaller chip that performs better, you always do so because it means you can use each wafer more efficiently and thus make more money. No competition means you can price those parts at a premium, so it's not as if you're losing any money by increasing performance year over year. In fact, if you can increase performance each year, you incentivize consumers to upgrade every year, or two years, etc.

There isn't a reason to sit on semiconductor tech, ever.