...
INTEL IS BANKRUPT AND FINISH
Caleb Hernandez
Other urls found in this thread:
servethehome.com
anandtech.com
tomshardware.com
tomshardware.com
techreport.com
agner.org
ark.intel.com
twitter.com
Juan Hughes
...
Logan King
...
Isaiah Sanchez
...
Jose Johnson
>Up to 3.2 GHz
Aiden Evans
...
Jaxon Butler
being compared to an intel chip with 3.6ghz turbo
;^)
Aiden Walker
...
Connor Powell
...
Brody Murphy
kek
Ayden Wilson
...
Angel Thomas
IT'S OVER!
ATTENTION ANONS, IT'S OVER!
Ayden Collins
...
Henry Torres
Stop, you piece of shit!
Jacob Green
>single core turbo speeds relevant on servers
Zachary Gonzalez
b-b-but what if.. one query is big for 1 of the 100000 hits per second you get?
Julian Hernandez
>21% more integer performance from one EPYC than 2 (TWO) Xeons
Isaiah Powell
Can it run crysis?
Daniel Morales
this will be huge for thermal performance.
Intel deserves this for sitting on their heels when AMD was working on APUs post-bulldozer
Carter Thomas
Yes, since it's literally four R7 1800X on one socket
Carter Perez
HOLY SHIT
Kevin Green
>2TB
>per socket
>of encrypted-on-the-fly
>DDR4 ECC
Grayson Campbell
Kikes are dead.
Parker Clark
Where are you getting this OP? Are you in the audience? If so godspeed you beautiful bastard, thanks for making Sup Forums great again
Noah Wright
tfw u will never have an ebyn chip
Joshua Robinson
Anand's live blog.
anandtech.com
William Torres
HOW MUCH FOR AN 8 CORE EPYC
I NEED PRICES
I NEED THAT CPU FOR A HOMESERVER
Jayden Johnson
...
Hunter Foster
holy shit intel is dead and finished
Kayden Rivera
$400.
James Gutierrez
Jesus christ that's like a 1700x at launch
THEY CAN'T KEEP GETTING AWAY WITH THIS
Bentley Lee
>Unironically cheaper than consumer i7s with less cores
Michael Gonzalez
You need to buy one before OEMs do, user.
William Baker
PRICE BRACKETS
Brody Bailey
>one epyc draws 17% less power than two xeons
lol fuck
Ayden Hill
...
Isaiah Bell
FORREST LOOKS TOO FUCKING SMUG RIGHT NOW HOLY SHIT
Parker Moore
Oh fuck.
Brandon Sullivan
INFINITY FABRIC
Angel Nguyen
POWER CONSUMPTION
Leo Gonzalez
post yfw Intel got irreversibly btfo'd
Logan Hernandez
...
James Sanchez
It bears mentioning that Intel publicly posts its SPEC benchmark data, and AMD's endnotes indicates that it reduced the scores used for these calculations by 46%. AMD justified this adjustment because they feel the Intel C++ compiler provides an unfair advantage in the benchmark. There is a notable advantage to the compiler, but most predict it is in the 20% range, so AMD's adjustments appear aggressive. - See more at: tomshardware.com
Gabriel Taylor
Bad graph aside, this is honestly very impressive.
Bentley Richardson
wait intel fucking cheats with their compiler and they STILL got btfo??
Ethan Wilson
It'll still be ahead of Xeons even at 20%.
Aiden Young
From what I gather AMD presented comparisons to Intel specs that THEY had modified to how they considered fair. So no
Wyatt Scott
$15 per stock by tomorrow.
BUY BUY BUY BUY
Angel Thompson
servethehome.com
>With the level of power/ performance of the new systems, you can essentially replace four Intel Xeon E5-2600 (V1) servers with a single dual socket EPYC node and get more performance (in most cases) in a single node that uses half the power. That is absolutely stellar.
>The AMD EPYC platform is still seeing major updates to BIOS for power and performance which is why we are calling these preliminary results. At the same time, we are already seeing some impressive figures.
Daniel Rogers
And the best part is?
For AMD, they are getting rid of 4 bottom of the barrel, throwaway dies for $400.
Julian Myers
Actually the graph makes sense if you DON'T expect 1 × 2 to be 2 (ie, inefficiencies at scaling). They're showing off that they have very close to perfect scaling. It would be better if they contrasted an Intel scaling graph that obly went to 1.7 or 1.8x
Colton Morris
Imagine if a 32-core Xeon die was so badly damaged only 8 cores worked. Intel would have to throw it out. The multi-chip shenanigans are what let AMD bin this low.
Jackson Allen
>50% more perf/watt
ded
DED
DEEED
Nicholas Evans
I really do want an official statement on what their yields are. I imagine with all of this binning it's a major exception to have to throw out a die.
Lincoln Johnson
post yfw AMD will have entire wafers where every chip can be sold
Leo Perry
Ryzen is 80% so I would expect this as a minimum.
Justin Perry
...
Jonathan Gonzalez
In 2010 Intel had its first perfect wafer, it was a Nehalem wafer.
Juan Sullivan
>Nehalem
god tier
Isaiah Carter
It would have to have less than 4 useable cores across the entire die. Or maybe 2+ dead cores per CCX.
Carter Hill
...
Eli Clark
...
Justin Roberts
>17% less power consumption than two CPUs
That's a lot of power.
Jace Sullivan
...
Adam Long
YAMEEEEEEEEEEEEEEEEEEEEEEEEEROOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO
Chase Russell
...
Liam Carter
AYmeD has completely shat on intel
how long until bankruptcy
James Garcia
WEEKS IF THEY'RE LUCKY
Lucas Williams
...
Colton Cox
>The Power Of Simplicity
>that direct shot at Intel
lol holy fuck AMD is on fire
Aiden Rogers
next cpu will be named AMD DYCK
David Bailey
128 cores
16 dies
6 gorillion dead
Nathan Long
...
Eli Moore
Again, AMD lying: techreport.com
Adrian Ward
Intel will be back and genocide all poojets and chinks.
Evan Turner
DAMAGE CONTROL
DAMAGE
CONTROL
Angel Martin
>With Intel's compiler results adjusted this way, two-socket Broadwell systems generally come out slightly behind to moderately behind versus Epyc parts when running SPECint_base_rate2006, especially at the lower end of the market. AMD's systems still win out, but the differences in performance seem closer to our real-world examinations of Zen and Broadwell performance.
AMD still wins, faggot.
Benjamin James
>chinks and poojets lie
That's a shocker.
Joshua Mitchell
GCC for AMD and ICC for Intel
What's the fucking problem, you're not gonna run a 3 billion grid array using a compiler that's worse performing, is this guy a fucking idiot?
Landon Gonzalez
Jesus Fucking Christ, Should've waited for EBIC
Brayden Anderson
>3.2 GHz
Daniel Edwards
It's not the ICC is more efficient on Intel, it's that ICC INTENTIONALLY disables certain instruction sets on "Non genuine" Intel CPUs.
This has been known for years now.
Grayson Morgan
can't wait till 256 dimm mobo lole....
Ayden King
ICC cheats in benchmarks, user.
Jackson Ross
are you retarded?
Logan Williams
still gets rekt by my i7 7700k@ 5hgz in games
Luke Reyes
Repostan:
It doesn't matter if Intel delivers 32 core Xeons now. If they don't have full AES-128 across the entire 2-4TB memory pool, AMD is better. If they can't deliver 128 PCI-E lanes across ALL CPUs regardless of price, AMD is better.
If they can't deliver 8 channel memory across ALL CPU ranges regardless of price, AMD is better.
Intel basically has to sell $6-8,000 chips now at $2-4,000 pricing if they want to be able to compete with AMD. Further more, they'll need to drop their original $2-4,000 chips into the $400-1,000 price range in order to be able to compete.
AND AND AND, on top of that, ALL THESE chips, need to offer the same feature set across the stack on a unified motherboard, socket, and chipset.
IF INTEL CANNOT DO ALL OF THE ABOVE, ACROSS THE ENTIRE STACK, THEY'RE FUCKED.
Because AMD can, for less, while delivering more, with less power, heat; better performance, availability. Its god damn black magic up in this bitch.
Intel has to pull black magic out of its ass, OR;
IT
CANNOT
WIN
Colton Cox
he's just grasping at the last straw he has
Andrew Howard
Jacob Taylor
Itt Muslims from reddit
Owen Harris
>it's INTEL IS FINISHED thread #59250250235
>it's lel Intel are JEWS thread #50443503534 when AMD also has extensive facilities here in Israel
*yawn*
Samuel Perez
Lets not pretend i do not read Agner.
David Mitchell
Lexis nexis is now on board with epyc
Brandon Murphy
lol @ those base clocks
AYMMD fanbois BTFO with their bs about scaling
underwhelming desu
Oliver Anderson
>With the level of power/ performance of the new systems, you can essentially replace four Intel Xeon E5-2600 (V1) servers with a single dual socket EPYC node and get more performance (in most cases) in a single node that uses half the power.
Holy shit, the absolute state of Intel
Dylan Gray
Can't tell if trolling or just retarded.
Henry Wood
looking forward to a writeup on this one and it's benefits and how it do dis