AMD Threadripper bench leak - 7.16.17

>>61250115
We already know it's performance just double ryzen and you're done.

Same arch, same process, same speeds. The only minor difference will the the chip config, ir which parts were disabled from epyc

>wccftech
gr8 b8

>>61250170
meh, check the sources of their posting and stop being a child.
wccftech = b8 is vocalized by brainlets who don't know how to evaluate information on a per article basis.They're a great aggregator for info that's on the wire and they tend to get it out in a much more timely fashion than others.

> inb4 100 other sites carry the same info and then is deemed legit

>>61250163
Sorry, things never scale in such a fashion. The benches reflect that.

>>61250210
The Zen Infinityfabric architecture scales way better on multithreaded tasks than Intel's ringbus.

>>61250210
compute scales exactly like that.

benches like the above are not linear anon, they are logarithmic.

>>61250115
Not this shit again. There are 1800X's that can score higher than these numbers. They made the exact same bullshit the last time some dope uploaded fake results to Garbagebench. Jesus buttfucking christ.

>>61250334
Link related, it's the last time these morons got fooled. And the results last time were even higher. Get fucking real. WCCFturd is rumor mongering zero research garbage. http://wccftech.com/amd-ryzen-threadripper-1950x-specs-performance-leak/

>>61250334
>>61250170
On second look, you're right. fucking wccftech.
me = BTFO

File: 1498863373594.jpg (17KB, 274x540px) Image search: [Google]

17KB, 274x540px

>>61250115
>That pin count

>>61250452
>pin
those are pads, summer

>>61250115
1. wccf

2. geekbench

3.TR better be $850.....

>>61250115
Who is paying """"'"reviewers""''"""" to use Geekbench?

Literally no one respects that shillware but it keeps showing up in the news.

>>61250115
>Geekbench
is it the benchmark that uses ICC?

>>61250115
Leave it to /g/ to care about processors over a few hundred dollars.

>>61250115
I just wanna know if we'll see any high-clocked unbuffered ECC DDR4 any time soon. Teh thought of Threadraper on 3600 MHz ECC makes me hard.

Workstations don't need Chipkill (i.e., interleaving 4 channels with all x4 chips so you can lose any one without being automatically fucked), so simple 9x8b DIMMs using fast unbuffered DRAMs should only cost ~12% more than normie non-ECC DIMMs.

File: 1496544992352.jpg (33KB, 597x430px) Image search: [Google]

33KB, 597x430px

>>61250210
just shut up

Wait, AMD is moving to pads?

>>61252582
They've used LGA sockets for enterprise parts for years now. This is basically just the same thing but a consumer/prosumer part instead of enterprise.

Only mainstream consumer parts are PGA.

>>61250115
It's literally two dies glued together?

>>61250115
>Almost the same scores as 1800X
Wccftech is trolling right?

>>61250318
intel doesn't use ringbus on skylake-x

>>61252668
Dubble da Bower :DDDDD

They should have given AM4 32 lanes, I hope the AM4+ has all of them enabled.
I don't need 16c.

>>61252720
No, they're just that stupid and/or desperate for clicks.

>>61250115
Your entire argument is flawed
The IO connections on Threadripper by itself is worth $200 more than the 'equivalent' intel part.