This kills Intel.
What in God's name is that
>>60924328
Jewripper.
>x kills intel for the last 20 years
ok
i9s are DOA
>>60924328
Threadripper, get it.
>>60924711
>no 8c/16t, 128 PCIe Lanes, 8-channel Memory desktop monster
JUST
>>60924731
but there is a 8c/16t version of ripper.
>>60924743
how fast can i shitpost on /g/ with it?
>>60924757
/g/ is single threaded. get a g4560.
>>60924757
That picture is too spooky and I am offended
>>60924711
What the fuck have Intel been doing all this time. They really gonna be taken down by some AMD poo in the loo guy? kek
>>60924782
>robert hallock
>pajeet
Wut.
>>60924782
CONTRACT KELLER
>>60924782
Intel is too busy with (((diversity))) programs.
K E L L E R
E
L
L
E
R
>>60924787
>>60924711
>finger on the contacts
Holy shit I know its likely just a prop but god damn that really fried my bacon.
>>60924931
And? The dude designed R300 (which kicked so much ass it resulted in Dustbuster), R700 and R800.
https://www.youtube.com/watch?v=qspdnAYiiug&t=44s
I've yet to see something more loud than that.
>>60924963
Holy fuck look at the size of those fins.
Who thought that shit was a good idea?
>>60924971
>Who thought that shit was a good idea?
Nvidia's damage control department. Yes, 9700pro was THAT good.
>>60924979
Thank got AMD exists.
I cant wait for them to put out something for me to replace my 3570k with.
And ThreadRipper looks like the perfect meme to do it with!
>>60924963
fx series were an even bigger blunder than fermi
>>60924992
>https://www.youtube.com/watch?v=WOVjZqC1AE4
At least nVidia tried to be ironic or something instead of damage controlling like crazy and doing autistic screeching at Larrabee and R700/R800.
>>60924931
Poo powered GPUs are terrible
>>60925009
I'll skullfuck you with my 5850.
>>60924992
The FX-6300 was one of the best values out there. Not profitable to AMD but certainly great for me at $100.
>>60925016
He means GeForce FX, anon. The ones that couldn't into DX9.
>>60924761
>/g/ is single threaded
But there are lots of threads on /g/
>>60924353
kek
>>60925023
Fuck me. I see, thanks.
>>60924782
https://i.4cdn.org/g/1497578142502.png
Making more ringbuses. Instead of having 2-3 ringbuses, the next gen Skylake chips are going to have 11 in a grid.
>>60925074
Literally
>make moar bingbus :--------------DDDDDDDDDDDDDD
Fucking hell Intel you need a cure from bingbus fever. Ghost of ATi should help them.
>>60925074
>>60925086
It looks like fucking prescott's pipeline meme is coming back from the grave.
>>60925129
What's next, 3D mesh of bingbuses?
>>60925157
Dont give them ideas, you know they'll do it too.
>>60924353
>jewripper
>>60925179
He he, Haifa team got you covered.
>>60924939
Are you blind? Nowhere in this picture is anyone touching the contacts.
>>60924731
thats called a low-end server you fucking putz
i really hate you op
we get that amd is pro now
who cares
>>60925720
>who cares
>stop picking on me goyim!
nah
>>60925720
>Stop talking about technology on the technology board!
Fuck off.
>>60925230
>Pinches CPU between contacts and heat spreader
When is the embargo lifted? We need to see them benchmarks and prices
>>60926157
On Kikeripper? Who knows.
EPYC NDA lifts on 20th of June, along with OEM racks reveal afaik.
>>60923795
>literally two Poozens glued together
>latency between the two CCXes in each Poozen is horrid
>now there's even more latency between the two dies before even hitting either CCXes
Meanwhile, Intel has a Mesh that just werks
No interdie latency penalties or crossdie penalties, just pure bandwidth
>>60927264
Latency? What are you talking about?
I've heard of this before but nobody ever supplied more information.
>>60927294
He's shitposting. Latency is irrelevant and IF on Threadripper runs in 2:1 mode so there's enough bandwith for 4 memory channels.
>>60927294
>>60927314
>more latency to pass through metal layer than to access DRAM
Hoo boy whoever did the measurements should gas himself.
>>60927332
Why do you think the Ryzen 5 1500X is so bad?
>>60927342
1500x is very good you fucking retard. Its 1400 with 4mb of victim L3 that's meh. Are you really that tech illiterate?
>>60927372
If the 1500X was so good, then why does a 1700X with one CCX disabled and the same clock speeds beat it?
>>60927372
>4mb of victim L3
It's 8mb and it doesn't make much of a difference.
>>60923795
This kills AMD
>4.7GHz on 10 cores at 1.25V
>faster single-threaded performance than the 7700K
>scales like crazy on all 10 cores
>no more bingbus
>>60928275
It's LCC. It still has bingbus and it pulls 370w.
>>60928295
lol, mad AMDrone detected
>>60928308
Are you gonna post sources or just shill like the retard you are?
>>60928308
370W BOYO. ITS HOTTER THAN THE SUN.
>>60928318
Google, nigga
>>60928324
SMALL PRICE TO PAY FOR A CPU THAT ISN'T LATENCY-GIMPED SHIT
>>60924328
solar panel
>>60928343
WHAT FUCKING PRICE ITLL BURN DOWN YOUR HOUSE FAGGOT. ITS WORSE THAN THERMI. WORSE THAN 5800 ULTRA. WORSE THAN 9590. ITS THE ULTIMATE HOUSEFIRE.
>>60928324
ridiculous power consumption
idle consumes half the power of a ryzen on full load
>>60925230
In the video past the thumbnail he does
>>60928324
AHAHAHAHAHAAHAAAHAHAHAHAAHH
>>60928324
fake benchmarks
>>60929067
Go to bed Trump. You're drunk
>>60924353
top kek
>>60924992
I would argue that Intel deciding to stop innovating and stick with what works was an even bigger blunder in the long run.
CERTIFIED
U
N
TERMINATOR
>>60925016
i was bamboozled with the 8350 but i've been riding that bitch for the last 5 years. paired with a Fury it's actually pretty decent still.
i have everything for a Ryzen build except the chip itself. looking for a little bit of a drop for the 1700X
>>60928275
>faster single-threaded performance than the 7700K
>scales like crazy on all 10 cores
with the amount of power that chip pulls you could power my block for two days
>>60927303
Is this verified?
WIDER
>>60923795
T H I C C
>>60929719
>"Latency is irrelevant"
The only thing being verified is the damage control AMDrones undertake when Ryzen weaknesses are highlighted.
>>60930907
>The only thing being verified is the damage control AMDrones undertake when Ryzen weaknesses are highlighted.
I don't. What? How does that answer the question?
>>60929719
On a serious note I'd love to see a citation
I too have heard the twice-the-speed rumor about Threadripper's Infinity Fabric, but I have never read a credible source.
>>60931005
That's the only way you can have enough IF bandwith to support 4 memory channels.