Launch of the AMD FX 8 Core processor - Page 4

Christenson · 2011-10-24, 21:51

I think we said that if we had 8 integer cores and 4 FPUs on Bulldozer,
then optimum would be to run 4 copies of integer-only mfaktc (a siever) to feed the external GPU (or, if you like, P-1) and 4 more cores of heavy duty FP, say, P95 LL tests.

moebius · 2011-10-24, 23:30

Quote:

Originally Posted by Christenson

I think we said that if we had 8 integer cores and 4 FPUs on Bulldozer,
then optimum would be to run 4 copies of integer-only mfaktc (a siever) to feed the external GPU (or, if you like, P-1) and 4 more cores of heavy duty FP, say, P95 LL tests.

Sure, but I currently run 4 Threads LL on a AMD Phenom X4 955 BE (%CPU: 390-394%) and 1 Thread CudaLucas (%CPU: 3-8%) which leads to 8 LL Results in the 44-52M Range per Month.

Christenson · 2011-10-25, 03:36

Quote:

Originally Posted by moebius

Sure, but I currently run 4 Threads LL on a AMD Phenom X4 955 BE (%CPU: 390-394%) and 1 Thread CudaLucas (%CPU: 3-8%) which leads to 8 LL Results in the 44-52M Range per Month.

So the value of bulldozer for you is either some threads of mfaktc, or possibly 4 threads of P-1 on the four additional integer cores. Worth it? Depends on the price point, power consumption, and external costs (surrounding Mobo, mem, power supply, etc) that you may incur.

Jeff Gilchrist · 2011-10-31, 17:06

Looks like someone over clocked Bulldozer to 8.46GHz. I wonder if at that speed it can keep up with the latest Intel...

http://www.engadget.com/2011/10/31/a...ed-to-8-46ghz/

LaurV · 2011-11-01, 02:29

Awesome! But can it run Android?

(stolen from the comments on your link)

henryzz · 2011-11-07, 15:49

Quote:

Originally Posted by LaurV

Awesome! But can it run Android?

(stolen from the comments on your link)

Once Android has been converted back to x86 yes. From memory android 4 should run on it.

R.D. Silverman · 2011-11-07, 19:15

Quote:

Originally Posted by Jeff Gilchrist

Looks like someone over clocked Bulldozer to 8.46GHz. I wonder if at that speed it can keep up with the latest Intel...

http://www.engadget.com/2011/10/31/a...ed-to-8-46ghz/

There is mention of some liquid helium........

Just how did they manage to cool the device????

fivemack · 2011-11-08, 13:37

These runs aren't being done for very long; I think you build a dam around the processor, fill it with liquid nitrogen (or in the AMD pure-advertising demo, liquid helium) and run until the liquid evaporates. It implies that some interesting attention has been paid to matching coefficients of thermal expansion ...

xilman · 2011-11-08, 19:14

Quote:

Originally Posted by fivemack

These runs aren't being done for very long; I think you build a dam around the processor, fill it with liquid nitrogen (or in the AMD pure-advertising demo, liquid helium) and run until the liquid evaporates. It implies that some interesting attention has been paid to matching coefficients of thermal expansion ...

OTOH, if anyone were particularly interested, and were prepared to pay the power bill, the chips could be run for as long as required. The technology and engineering are all well known. In particular, liquid N₂ is very easy to work with and very easy to produce. The major things to look out for are to avoid blockages from solid H₂O and CO₂ and to avoid creating liquid O₂ without appropriate precautions. Solid Ar can also be a problem --- been there, done that --- but usually isn't because of its low atmospheric concentration.

Paul

Christenson · 2011-11-08, 19:47

What about all that frost (H20, CO2) forming on the mobo? How do you fight that? I would assume that the solids wouldn't be the same density as the N2, so would sink or float, not be too hard to separate....

xilman · 2011-11-08, 21:38

Quote:

Originally Posted by Christenson

What about all that frost (H20, CO2) forming on the mobo? How do you fight that? I would assume that the solids wouldn't be the same density as the N2, so would sink or float, not be too hard to separate....

Why would you want to fight them? At 100K or below both are non-conductive rocks.

As long as the mobo is dry before immersion in the liquid nitrogen you don't need to worry about ices forming in creivces between components.

Paul

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
(windows) How to launch mfaktO/C with processor affinity	swl551	Software	1	2012-09-24 23:29
Intel unveils 50-core maths co-processor card	stars10250	Hardware	16	2011-06-25 17:24
Prime fatal error on 3D launch	kaeptn_kork	Hardware	11	2009-03-17 01:58
Product launch	xilman	Factoring	12	2006-07-11 16:23
Intel Core Duo processor	drew	Hardware	5	2006-05-29 07:00

2011-10-24, 21:51	#34
Christenson Dec 2010 Monticello 5×359 Posts	I think we said that if we had 8 integer cores and 4 FPUs on Bulldozer, then optimum would be to run 4 copies of integer-only mfaktc (a siever) to feed the external GPU (or, if you like, P-1) and 4 more cores of heavy duty FP, say, P95 LL tests.

2011-10-31, 17:06	#37
Jeff Gilchrist Jun 2003 Ottawa, Canada 3×17×23 Posts	Looks like someone over clocked Bulldozer to 8.46GHz. I wonder if at that speed it can keep up with the latest Intel... http://www.engadget.com/2011/10/31/a...ed-to-8-46ghz/

2011-11-01, 02:29	#38
LaurV Romulan Interpreter "name field" Jun 2011 Thailand 41·251 Posts	Awesome! But can it run Android? (stolen from the comments on your link)

2011-11-08, 13:37	#41
fivemack (loop (#_fork)) Feb 2006 Cambridge, England 1936₁₆ Posts	These runs aren't being done for very long; I think you build a dam around the processor, fill it with liquid nitrogen (or in the AMD pure-advertising demo, liquid helium) and run until the liquid evaporates. It implies that some interesting attention has been paid to matching coefficients of thermal expansion ...

2011-11-08, 19:47	#43
Christenson Dec 2010 Monticello 5×359 Posts	What about all that frost (H20, CO2) forming on the mobo? How do you fight that? I would assume that the solids wouldn't be the same density as the N2, so would sink or float, not be too hard to separate....