(((Intel))) Xeon and AMD Poozen Epyc Eternally BTFO!

5.0 GHz 32 CORES NO CHILLER!
oracle.com/us/products/servers-storage/sparc-m8-8-ds-3864183.pdf
x86 take this L

Attached: OwOwhatsthis.png (570x333, 68K)

>$583,500 list price before the $70,000 annual support contract

>One random number generator (One per processor)
>One per processor
Fake cores

What the fuck is an Oracle Number?
If you want a risc server, just buy POWER, or those new cavium ARM servers that are finally competitive.

shitty ipc

>Six 3,000W hot-swappable AC power supplies with N+N redundancy

Attached: 1515896901860.jpg (583x439, 35K)

>md5 is an industry-standard cryptigraphic algorithm

Attached: 1499838292490.jpg (640x628, 55K)

It is for a lot of things, juts like base64.

I guess whoever is still using it must really like hash collisions.

If you're simply hashing SALTED passwords is OK, anything else is a waste of cycles.

Using md5 for hashing passwords (even salted) is a bad idea because if it leaks then you're using a fast hashing algorithm which cuts down an attackers bruteforce time.

Yeah but even if they found a collision, they can't just plug that shit on a login form because the salt will mess it up.

What happens if my passwords are frosted instead?

that's true but I'm talking about a leak. you should always plan for the worst-case scenario. in this case, it's not md5's collision resistance that hurts it but it's speed. Also, there are a fuckton of rainbow tables for md5.

>baby's first ASIC

Wow! It's only 1100 USD a core!

Is Oracle behind this?

You don't need a RNG to have a functional CPU, therefore it's not necessary per core.

>fake cores
notice that it's listed under architecture, the generator itself is not a core

>implying the average Jow Forumsnigger has had to deal with oracle as a company
Let them have their fantasies as they babysit their lamp stacks.

MD5 is no longer useful for anything except checksum in a trusted system where the extra space needed to store an SHA256 is prohibitive but CRC-32 isn't a strong enough integrity guarantee. So, basically nowhere.

You should be using PBKDF2/scrypt/argon2 for password hashing. These algorithms are intentionally slow key derivation functions which make marginal passwords secure passwords.

You can still use MD5 in things like HMAC, etc. without it being an issue

In combination with other techniques, md5 can be made secure. But md5 by itself is not an industry standard cryptographic algorithm. And the hmac is for authentication, the hardware reference is an encryption accelerator

Its actually used heavily in authentication of network devices, both RADIUS and TACAS+ use it as their data in motion encryption algorithm.

>TACAS
TACACS+

Can it run Crysis?

[spoiler]no
[/spoiler]

If you look at the wiki page on radius, it says that even though md5 is baked into the rfc, you should be using it with other techniques like ipsec since md5 is considered insecure. Additionally, tacacs+ is known to not be secure. Additionally both pure RADIUS and pure TACACS+ aren't fips compliant because of their use of md5. if you turn on fips compliant mode in your security policy, i think windows will throw security exceptions with RADIUS and md5 without something like ipsec. And once again, md5 is only secure in combination with other techniques, it is not by itself an industry standard cryptographic algorithm.
The only purpose where I could see md5 being okay is for something that must be purely pre-image resistant and you better hope your adversary isn't someone with even a modest amount of computational power.

how many zen+ TR is that worth?

you obvioisly dont know what hmac is nor how it works. hmac is by def only as secure as the underlying hash algo, i.e md5 and therefore doesnt improve shit