Intel® Xeon Phi™ Coprocessor Infographic

sheroo

sheroo

New member
Intel® Xeon Phi™ Coprocessor Infographic.

I wonder, if there was a suitable folding client, how these bad boys would perform?

xeon-phi-for-researchers-infographic.jpg
 
Last edited:
I don't think a merge is handy for any board,
especially Folding At Home(F@H)...

Don't think anyone on here is going to use it :p;)
al tho, i really would try it

Ray......
 
SnW said:
I don't think a merge is handy for any board,
especially Folding At Home(F@H)...

Don't think anyone on here is going to use it :p;)
al tho, i really would try it

Ray......
Click to expand...

If I win the lottery dude, I'll buy you one.
 
" The passively cooled Xeon Phi 5110P (60 cores) is shipping for revenue at Intel now, and will be generally available on January 28 to the rest of us for $2,649. The actively cooled Xeon Phi 3120A (57 cores) card, which is hotter and yet has less memory and bandwidth, will be available sometime in the first half of 2013 with a price that is expected to be around $2,000."

4 of those should give Tom some nice F@H numbers. :)
 
Writer's Block said:
" The passively cooled Xeon Phi 5110P (60 cores) is shipping for revenue at Intel now, and will be generally available on January 28 to the rest of us for $2,649. The actively cooled Xeon Phi 3120A (57 cores) card, which is hotter and yet has less memory and bandwidth, will be available sometime in the first half of 2013 with a price that is expected to be around $2,000."

4 of those should give Tom some nice F@H numbers. :)
Click to expand...

I havn't looked at how these work and how they may intergrate into programs, but 1 Teraflop is quite behind what graphics cards are crunching and with nvidia being the graphics card of choise when it comes to folding, would these actually perform any better.

I don't know how 'many' of the flops a 680 GTX is capable of get 'used' buy fah. If it's a case of 30% where as the Phi could use 99% then the advantage is obvious. On the otherhand the Phi could be totally redundant for anything other than Intel specific tasks.
 
I think the advantage to these may be in how they are programmed; they run - at least my understanding is that - standard code. NVidia for instance use CUDA and openCL.

OF course these aren't competing with the consumer range - gfx cards such as a 680 & 7970 aren't it's target; it's the likes of Firepro and Quadro.

Xeon Phi is not a graphics card, but a coprocessor with totally different functions; their function is boosting cpu intensive tasks - rendering is one example; they are expected to be used in the scientific community.

http://www.theregister.co.uk/2012/11/12/intel_xeon_phi_coprocessor_launch/ is an interesting read.
 
Writer's Block said:
I think the advantage to these may be in how they are programmed; they run - at least my understanding is that - standard code. NVidia for instance use CUDA and openCL.

OF course these aren't competing with the consumer range - gfx cards such as a 680 & 7970 aren't it's target; it's the likes of Firepro and Quadro.

Xeon Phi is not a graphics card, but a coprocessor with totally different functions; their function is boosting cpu intensive tasks - rendering is one example; they are expected to be used in the scientific community.

http://www.theregister.co.uk/2012/11/12/intel_xeon_phi_coprocessor_launch/ is an interesting read.
Click to expand...

Don´t forget Tesla!! ;) Firepro and quadro are more fore workstations! Nvidia Tesla are dedicated to intensive calculation!! :)
 
Writer's Block said:
I think the advantage to these may be in how they are programmed; they run - at least my understanding is that - standard code. NVidia for instance use CUDA and openCL.

OF course these aren't competing with the consumer range - gfx cards such as a 680 & 7970 aren't it's target; it's the likes of Firepro and Quadro.

Xeon Phi is not a graphics card, but a coprocessor with totally different functions; their function is boosting cpu intensive tasks - rendering is one example; they are expected to be used in the scientific community.

http://www.theregister.co.uk/2012/11/12/intel_xeon_phi_coprocessor_launch/ is an interesting read.
Click to expand...

The advanted Intel speak is that the Phi can run x86 code. However, that code still needs to be recompiled. Since CUDA is C and C++ the only real adventage Intel may have would be programmers using machine code. Something they would have to learn for CUDA but may already have experience in for x86.

Like I said, I don't know great specifiics on the hardware and I don't like to talk in detail about things I don't really know about, which I why I was talking with an overview at just the flops.

Lets not beat around the bush, 1 teraflop is still alot of power and I'm very certain that the Phi is more flexible than todays GPUs.

Thats why I was only talking about folding and crunching folding work units.

As an aside, have youseen the amount of memory bandwidth the Phi has? WOW
 
Pickster said:
The advanted Intel speak is that the Phi can run x86 code. However, that code still needs to be recompiled. Since CUDA is C and C++ the only real adventage Intel may have would be programmers using machine code. Something they would have to learn for CUDA but may already have experience in for x86.

Like I said, I don't know great specifiics on the hardware and I don't like to talk in detail about things I don't really know about, which I why I was talking with an overview at just the flops.

Lets not beat around the bush, 1 teraflop is still alot of power and I'm very certain that the Phi is more flexible than todays GPUs.

Thats why I was only talking about folding and crunching folding work units.

As an aside, have youseen the amount of memory bandwidth the Phi has? WOW
Click to expand...

I can't say I had; I'm still trying to find out information and put into context what I find out.

Numbers, for me, are meaningless if I can't get a frame of reference; sometimes, even knowing a stat is awesome doesn't truly demonstrate it without context.
 
You must log in or register to reply here.

Similar threads

tinytomlogan
Intel Xeon w9 3495X and w9 3475X Review
Replies
2
Views
2K
FTLN
F
K
9070XT auto overclock
Replies
3
Views
659
KingNosser
K
J
Is LTT forum is choke full of spooks ?
Replies
5
Views
387
hmmblah
hmmblah
P
Do some people never learn from the past?
Replies
2
Views
766
Pugsport
P
P
Memory error
Replies
4
Views
4K
Warchild
W
Back
Top