SPARC64 vs x86
Images are sometimes not shown due to bandwidth/network limitations. Refreshing the page usually helps.
You are currently reading a thread in /g/ - Technology
You are currently reading a thread in /g/ - Technology
Is silicon security better than OS security?
>>52154562
you mean security by obscurity?
>SPARC
>Secure
>Made by an American company (Oracle)
>>52154562
>SPARC
>Oracle
Get out newfag
SPARC hardware has some very nice properties like being 100% manageable from a serial port and not needing to pretend to be a 1981 16-bit microcomputer for a few cycles every time it boots up. Unfortunately it gets raped by x86_64 on single threaded performance, power usage, and MIPS/$.
>>52154683
SPARC and POWER haven't been relevant since the dot com bust in the early 00s
>>52154653
>>52154666
> Made by Fujitsu, Texas Instraments, Atmel as well,
> completely open source unlike Intel chips
You seriously are one ignorant faggot
>>52154919
>completely open source
ahahah, only the old as fuck stuff is open
In March 2006, the complete design of Sun's UltraSPARC T1 microprocessor was released in open-source form at OpenSPARC.net and named the OpenSPARC T1. In 2007 the design of Sun's UltraSPARC T2 microprocessor was also released in open-source form as OpenSPARC T2.[1]
>>52154700
The fourth most powerful computer IN THE WORLD is SPARC
https://en.m.wikipedia.org/wiki/K_computer
>>52154995
And yet it's a diamond in the rough as x86_64 has 4/5ths share of the TOP500/
>>52154919
There hasn't been a new OpenSPARC chip in like 8 years, Oracle put a stop to that pretty quick.
>>52155009
Because the ape developers can only handle Intelshit.
>>52155089
How does it feel when Xeons are cheaper, faster and more power efficient than SPARC
>>52155104
>more power efficient than SPARC
The thing about SPARC is all the threads. You can spin up like 2-4x more VMs with a SPARC chip vs x86 because of the threads. In that regard SPARC would be the more efficient one because you would need more processors just to reach the equivalent amount of threads.
Though admittedly SMT really tends to suck at scaling past 2 threads anyway so SPARC would likely lose to x86 when they have roughly equivalent hardware.
>>52155190
I really want to love Sparc but Oracle ruins everything
>>52155190
My overall complaint about SPARC is not necessarily with the hardware, but with the overall health of the ecosystem.
2 vendors, declining sales etc => only sparc shops will continue to buy the h/w. It's slowly going the way of Itanium.
>>52155395
>Oracle ruins everything
they still managed to execute pretty well on their sparc launches compared to sun
SPARC64 is total trash. Daily remember that if your company purchased a SPARC box you should quit and find a new job because your company now works for Oracle NOT for its CEO.
>>52155431
>My overall complaint about SPARC is not necessarily with the hardware, but with the overall health of the ecosystem.
>2 vendors, declining sales etc => only sparc shops will continue to buy the h/w. It's slowly going the way of Itanium.
I think Sun foresaw this happening too which is why they released all the specs to the public. Maybe they were hoping to bring new life to the architecture. Then of course Oracle happened and that fucked that plan up because Oracle hasn't released jackshit since.
>>52155650
>Oracle happened and that fucked that plan up because Oracle hasn't released jackshit since.
>managed to push out several new sparc chips with better performance
yeah
>>52155904
I was referring to OpenSPARC. Decade old designs aren't enough to get other companies interested. If there's no interest in your platform then the software will dry up.
It could be.
It is easier to obscure hardware, so predicting the output of a hardware random generator is harder.
Authenticating might be faster in hardware so you can do more complicated things and thus make security better but all this is harder to modify, so when you find a flaw, you have to replace all hardware instead of updating the software.
But if the thing is secure, it can't be altered to an insecure version or tricked to do something else than it was designed for.
It is not an easy choice
>>52155952
It's nice to see that IBM finally woke up with OpenPOWER, but it's probably too late.
>>52156125
I think they might have a chance. Intel has hit the wall hard on x86_64 single threaded performance, with Haswell still within 10% of its current and upcoming microarchitectures in IPC. OpenPOWER lets people keep their Linux, KVM, and Java stuff (IBM has their own JRE for POWER8) from x86 land on much faster silicon.
