Graphics card fans

Do you have a question? Post it now! No Registration Necessary

Translate This Thread From English to

Threaded View
Why do graphics cards only monitor the speed of one fan? If the other one fails, it won't know!

Re: Graphics card fans
On Tue, 14 Jul 2020 16:10:17 +0100, "Commander Kinsey"

Quoted text here. Click to load it

Are graphic cards used mostly for games? And maybe bitcoin mining?
It's weird that one PC can contain more compute power than existed on
Earth in 1970, and be used for games.

I suggested to Mike E that LT Spice should use a graphic card for
computation, but I guess that's not going to happen now.

A modest Windows PC can spin Solidworks 3D images around just fine.



--  

John Larkin         Highland Technology, Inc

Science teaches us to doubt.

We've slightly trimmed the long signature. Click to see the full one.
Re: Graphics card fans
On 7/14/2020 11:36 AM, snipped-for-privacy@highlandsniptechnology.com wrote:
Quoted text here. Click to load it

Is SPICE trivially parallelize-able in that way?

Graphics cards have thousands of compute cores. Most operations on 3D  
mesh vertices and pixel "shading" are trivially parallelize-able; the  
operation pipeline is programmed for a given task and then each core  
runs its algorithm on a given vertex or pixel of the scene without  
needing any information from the others.

Early graphics cards didn't have re-programmable pipelines there was a  
somewhat fixed set of operations with some configurable options that  
could be applied in series to vertices and pixels.

Modern GPU code is written in a dialect of C with some features not  
relevant to single instruction multiple data operations removed, like  
pointers

Re: Graphics card fans
On 7/14/2020 11:50 AM, bitrex wrote:
Quoted text here. Click to load it

Every frame if the scene is in motion the lighting reflections and  
shadows have to be re-computed, as one example.

Re: Graphics card fans
Am 14.07.20 um 17:50 schrieb bitrex:

Quoted text here. Click to load it

No. Inverting the conductivity matrix is hard because you
cannot do the pivoting in advance. The necessity shows
up during work.

For transient analysis, every time step builds on the previous one(s)
and you cannot parallelize a lot of them because you don't know
the starting condition of the future ones.

It has been tried often, a working solution would have been worth gold.
I remember the Weitek array coprocessor back in 80386 times and
a try with the NS16032. They never got a factor of more than 2 or 3.

Everything really interesting is np-complete.  :-(

Cheers, Gerhard


Re: Graphics card fans
On 7/14/2020 12:14 PM, Gerhard Hoffmann wrote:
Quoted text here. Click to load it

Quoted text here. Click to load it

That's what I figured.

There are probably ways to leverage GPUs in the process somehow but I  
expect it's going to be a 3 or 4 times speedup over using a general  
purpose CPU not like a ten thousand times speedup in the way rendering  
scenes is.

Re: Graphics card fans
On 7/14/2020 12:14 PM, Gerhard Hoffmann wrote:
Quoted text here. Click to load it

Quoted text here. Click to load it

Another problem of practical value that's NP-complete is the pen-plotter  
problem or the "postal-route inspection" problem; how do you connect  
vertices of a vector image with lines such that the total Manhattan  
distance the plotter head covers in the process is minimal.

as opposed to the shortest path problem on directed and directed graphs,  
exact solution to that one is np-complete. There are heuristics that do  
pretty good

Re: Graphics card fans
On 7/14/2020 12:25 PM, bitrex wrote:
Quoted text here. Click to load it

Quoted text here. Click to load it

it's similar but distinct from the traveling salesman problem

Re: Graphics card fans
wrote:

Quoted text here. Click to load it

LT Spice can already use multiple cores, so something is
parallel-izable. The petaflop computers, used for weather and physics
simulation, have thousands of CPUs.

Spice is usually fine, but once in a while I want 1000x or so more
speed.



--  

John Larkin         Highland Technology, Inc

Science teaches us to doubt.

We've slightly trimmed the long signature. Click to see the full one.
Re: Graphics card fans
On 7/14/2020 12:27 PM, snipped-for-privacy@highlandsniptechnology.com wrote:
Quoted text here. Click to load it

GPU cores aren't general-purpose CPUs, they're serial-pipelined and  
optimized for SIMD-instructions.

A mutlicore general-purpose CPU has good cache coherency there's a fast  
on-die cache for all 4 or 8 cores or w/e that any of the processors can  
look at data the others are working on quickly.

It's hard to achieve that kind of cache coherency with thousands of  
cores. If core #127 needs to "see" what core #562 is working on it  
usually has to go out to video RAM. which is not nearly as fast as  
on-die cache.

Re: Graphics card fans
On 7/14/2020 12:27 PM, snipped-for-privacy@highlandsniptechnology.com wrote:
Quoted text here. Click to load it

People with advanced credentials in e.g. meteorology or computational  
biology or physics plus the computer science of optimizing  
multiprocessing systems get paid the biggo buckos

Re: Graphics card fans
On 2020-07-14 12:14, Gerhard Hoffmann wrote:
Quoted text here. Click to load it

Quoted text here. Click to load it

It probably could be, if you changed the scheme so as to impose a  
speed-of-light propagation limit.  That way you could divide the  
schematic up into chunks, do time steps locally, and then propagate the  
changes to adjacent chunks.

That gets rid of every node having to know about every other node on  
every time step, and makes FDTD codes such as my POEMS facility  
parallelize well.  (It works that way.)

Linear algebra also can be made to vectorize well on the right hardware.

Cheers

Phil Hobbs

--  
Dr Philip C D Hobbs
Principal Consultant
We've slightly trimmed the long signature. Click to see the full one.
Re: Graphics card fans
Am 14.07.20 um 20:54 schrieb Phil Hobbs:
Quoted text here. Click to load it

Quoted text here. Click to load it

Quoted text here. Click to load it

This was my proposal at Z80 / AM9511/AM9512 times, just one node
per square mm of DUT chip. I also tried to port Spice 2G6
to Interactive Unix on my 80286/287 Bullet board. What a fiasco.
64K segments conspiring with f2c as a Fortran compiler. Never could
have worked.

But this computer had 2 MB and a 70 MB Fujitsu disk. That was pure
hubris in the hands of a EE & CS student. Our VAX11 at the semiconductor
institute had 2 300 MB Fujitsu Eagles for all people together. :-)
And we made real chip designs on it.

Later I had a T800 transputer cluster, that would have mapped nicely
to this problem. But I never could find a customer for any T800
solution I proposed. All went X86.

The only exception was smuggling a Parsytec cluster to east Berlin.
But little Gerhard did not dare to. Few did I know. Some weeks
later, all the sudden, was the German re-unification and nobody would
have cared anymore about smuggling technology to an east-German
railway company that went belly-up anyway. Sigh.

Quoted text here. Click to load it

Quoted text here. Click to load it

Gerhard

Re: Graphics card fans

Quoted text here. Click to load it

How about Monte Carlo simulation ?


Re: Graphics card fans
On 2020-07-14 11:36, snipped-for-privacy@highlandsniptechnology.com wrote:
Quoted text here. Click to load it

GPUs are limited to single-precision floating point IIRC.

Cheers

Phil Hobbs

--  
Dr Philip C D Hobbs
Principal Consultant
We've slightly trimmed the long signature. Click to see the full one.
Re: Graphics card fans
On 7/14/2020 11:53 AM, Phil Hobbs wrote:
Quoted text here. Click to load it

Was true circa 2005.

Many modern GPUs can do double precision floating point, how well any  
particular one does it depends on the particular architecture, though.

Re: Graphics card fans
On 7/14/2020 11:53 AM, Phil Hobbs wrote:
Quoted text here. Click to load it

That is to say they can do double but they're in general not optimized  
for it.

Re: Graphics card fans

Quoted text here. Click to load it

Some are.  I always buy the ones that are, since Milkyway at home loves them.

Re: Graphics card fans
On 7/14/2020 3:31 PM, Commander Kinsey wrote:
Quoted text here. Click to load it

Quoted text here. Click to load it

I suppose one has to buy the "Pro" variant rather than the  
gamer/consumer variant.

Moychendizing, moychendizing

Re: Graphics card fans

Quoted text here. Click to load it

I get them second hand, so I don't know what they were aimed at originally.  Four of them are R9 280X.  I thought those were high end games cards in their day.  It could have been before they stopped putting double in everything.  But it could be that the double precision ones are made in smaller quantities, so you don't get the mass production discount.  Or it could be that it's more expensive to make them.  But the reason they make them without is that games don't make use of double precision.  Less double processing means space for more single processing on the die.

Site Timeline