Threadripper in servers?

>>61103756
Only boards for it so far are e-atx

why buy threadripper when you can buy epyc, it sounds much cooler

>>61103791
for small server threadripper would be better because clocks on one socket matter?

>>61103756
Honestly no.
Be a good goy and buy the Intel Shit because is more proof tested. Separate some resources to buy a threadripper server when it launches and do some testing.

Because just like Ryzen its microcode will need some update, and update == downtime which is not ideal in a server.

File: chika.jpg (57KB, 640x480px) Image search: [Google]

57KB, 640x480px

Yes, wait for Threadripper. You'll get the slightly newer Ryzen stepping

>>61103801
Intel sure is proof tested
http://www.tomshardware.com/news/hyperthreading-kaby-lake-skylake-skylake-x,34876.html

>>61103815
Pls, imma stupid. What does this mean?

>>61103756
>mid-tier tower server
ryzen 1700 not enough?

>>61103834
Do you compile things with OCaml?

>>61103862
more I/O's maybe

>>61103875
No, but a Hyperthreading bug should cause more than that

File: H11DSi-NT-16-9-28d9677be4b2a38d.jpg (547KB, 1600x901px) Image search: [Google]

547KB, 1600x901px

>>61103862
Waiting for Supermicro I guess.

I wouldn't wait for threadripper to stir up pricing, I'd wait for threadripper to buy a threadripper.

For two reasons:
1. Intel won't drop prices
2. Even if they did, threadripper would still be more cost effective.

File: 2017060813504454_src.png (2MB, 1000x1000px) Image search: [Google]

2MB, 1000x1000px

>>61103903
Of course, one could settle for a Gigabyte.

>>61103898
Your wrong. And a patch is already out, and has been.

>>61103815
He means Xeons.

>>61104006
Some Xeon is affected.

>>61103971
So there was/is a bug?

>>61104012
Actually, nothing is affected if you just update your microcode.

>>61103931
Shouldn't Epyc placed for entry level and mid tier servers too? If they provide a good enough alternative, prices should drop.

>>61103971
>no fix for a few skylake steppings
>kaby lake fix will be provided by vendors JUST WAIT

>>61104028
There are plenty of errata on every cpu. Skylake has half haswell does, for instance.

>>61104040
The patch is already out. You won't encounter the bug anyway.

>>61103756
threadripper is not for servers

>>61104058
>JUST WAIT
hahahahahaha.

>>61104070
It was issued in April.

>>61104082
>issued in april
>no fix for old skylake steppings
>kaby lake fix provided by mobo vendors - never in a few cases
You tried.

>>61104061
Epyc is, but than it is the same argument, just with AMD's server CPU line

>>61103756
People using servers are usually hooked to Xeon. Only a trendy startup with no infrastructure that somehow isn't using the cloud would try AMD

>>61104094
It doesn't affect older models. You can update your microcode anytime you want. And you won't run into this error if you don't compile OCaml.

>>61104050
So both AMD and Intel is trash?

>>61104116
If they provide a good alternative to Intel, people will start to buy them after a while. Somewhere they need to start.

>>61103756
Pricing will likely be similar between TR and Epyc, some of the Epyc pricing is insanely cheap already. Also, Epyc socket is known to support Zen2 since they stated it already. We don't know the future of the TR only socket. Buy Epyc board, get decent Epyc proc, then leave yourself available to upgrading everything later instead of dropping in a TR and having half the RAM slots, half the IO, and unknown future support.

>>61104142
Every cpu ever made contains errata.

File: Screenshot_20170627-062932.png (462KB, 1440x2560px) Image search: [Google]

462KB, 1440x2560px

>>61104157
What's this Intel being bug free thing about?

>>61104187
No one has ever claimed that except you.

>>61103801
>Because just like Ryzen its microcode will need some update
it's exactly the same core

>>61104282
Threadripper's the B2 stepping. The question is if they just updated the on-chip microcode or tweaked the actual silicon.

>>61104116
Did you not see the EPYC launch conference?
They have tons of buyers lined up already.

>>61104116
Kek I would say the reverse is true but not even trendy startups are stupid enough to go with intel

>>61103756
Maybe. It does support ECC and supermicro is supposedly working on server motherboards for Threadripper.

lmao ryzen doesn't even run esxi anyone wanting to build a server with these is retarded

>>61104116
>>61104511
>>61103801
The shills are put in full force today. Brian must have gotten mad after Intel's stock dropped just from AMD's EPYC announcement.

>>61104511
>esxi
Fixed on the lastest agesa.

>>61104511
>esxi
Yep, because every server needs access to furry porn.

>>61104550
>furry porn
>esxi
what?

>>61103756
Why not a used dual Xeon built?

>>61104655
I never said I would like to buy an AMD CPU for the server. All I am aiming for is a price drop when AMD releases new CPUs.

>>61104655
Why not predigested McDonald's?

>>61104511
>esxi
That thing still exists?

>>61104727
What do you use? Proxmox?

>>61103862
Was for me desu

>>61104388
When has a stepping not been at least a metal layer change?

>>61105051
It's always a metal layer change, different clocks compared to B1 guarantee it.

>>61103756
No ultra low power, no CUDA. A useless shite just for gaymers.

>>61105114
>no CUDA
Threadripper probably would score higher in CUDA operations than Titan X

>>61104784
Plain Libvirt+QEMU on any half-decent Linux distro, even comes with a GUI.

>>61105135
Oh lord.

>>61105135
>>61105114
I don't know who's more retarded.

>>61105135
http://docs.nvidia.com/cuda/floating-point/index.html

>>61104029
You need EFI flash as well

>>61104116
That's because no one else had good server CPUs.
Now amd does.