How much memory is Nvidia's X *really* using?
on
Nvidia's NV20
·
· Score: 2
I too have an Nvidia TNT2 card and I'm using
Nvidia's driver with XFree 4.
Here's what top says:
Size RSS Share
252M 252M 2024 S 0 1.7 100.4 5:18 X
Seems excessive, doesn't it? Well, I've only
got 256M on my machine, and guess what?
NO SWAP SPACE IS USED.
PS tells a different story:
VSZ RSS
276408 12704 ? S 10:00 5:21 X:0 -bpp 32
VSZ is the VIRTUAL size of the process, 276M
12.7M is what it actually uses.
Another poster in another forum explained that
this apparently huge virtual size was due to
virtual-memory-mapping of various bit planes and color depths in the VRAM into virtual memory.
12.7M is still pretty high, but hardly burdensome on my 256M machine.
PeterM
What about the ecological impact? Worse than OIL?
on
Wave Driven Generators
·
· Score: 1
What is the ecological impact of putting wave generators all over the coastline to generate power? How many beaches and coastline ecosystems will be destroyed? Will fisheries be reduced in productivity?
How much does it cost to build a wave plant? How survivable are the plants to storm damage?
Unexpected power-off is NOT the only thing which can happen to a filesystem. What about these disasters?
1) Bad block takes out part of your disk unexpectedly.
2) Your OS screws up and spews a mess onto your filesystem before it crashes. (there ARE bugs in the kernel!)
3) You have a minor headcrash which takes out one of your tracks, but the disk is still functional.
What're you gonna do? Tux2 isn't gonna help you.
You could restore your latest dump. You could
also attempt to repair the filesystem.
You need fsck or some other means of filesystem repair.
Someone else helpfully supplied this quote from the US constitution:
Article 6, Clause 2: This Constitution, and the Laws of the United States which shall be made in Pursuance thereof; and all Treaties made, or which
shall be made, under the Authority of the United States, shall be the supreme Law of the Land; and the Judges in every State shall be bound thereby,
any Thing in the Constitution or Laws of any State
^^^^^^^^^^^^^^^^^^^^
to the Contrary notwithstanding.
The obvious interpretation of this is that a
treaty supercedes a STATE law or STATE constitution. Any other inerpretation simply makes no sense.
My plan is to keep my own data on "live" media: I'll keep everything on my computer HD (and a copy on the mirror HD), and whenever I upgrade, I'll copy all the old stuff of significance onto the new huge hard disk.
And BTW, in reply to the guy claiming $30 for 5.2G is a good price: nope, it's not. 20G disks are $84 now. That's $4.20/G. Your slow 5.2G DVD-RAM costs you $5.76/G, not counting the cost of the drive.
I'd rather have the hard disk for backup purposes.
I can easily see more latency coming out of RAM compression/decompression: you have to actually do all the computation. However, if the data coming across the memory bus is compressed 50%, then you can get 2x throughput, so that you can double your bandwidth at the cost of some extra latency.
The big question is how bad is the latency? If it is too bad, then performance will suffer. If it is fast enough, then performance may actually *increase*.
Our plasma simulation group has several simulation codes which would be pretty good as part of an open-source floating-point benchmark suite--*provided* this benchmark suite is distributed under the GPL or Berkeley license.
We considered giving our codes to SPEC, but SPEC wants to be able to *sell* their benchmark suite for $500 a copy. This caused us legal headaches so rather than deal we didn't try to participate in SPECfp2000.
We can offer C and C++ codes which exercise the FPU and memory subsystem heavily: they tend to be cache friendly though.
The industry's been working on low-k dielectrics for a long time.
The real issue with low-k dielectrics is that they reduce the capacitance of the onchip wiring.
This has several benefits:
1) Wiring capacitances in general will be signficantly reduced.
2) Wires running parallel to each other will have less crosstalk. I.e., if the voltage on one wire switches, it'll have less tendency to drag the voltage of wires close to it with it. This is a result of neighboring wires having mutual capacitance. --> EASIER DESIGN
3) Since capacitances are reduced, less power will be needed to cause metal wires to switch voltage. --> COOLER CPUs, -->LOWER POWER
4) Because of the reduced capacitances, it requires less current to switch wires. --> FASTER CPUs.
This is a really significant advance by IBM, maybe more signficant than copper metallization.
One possible problem is poor heat conduction from die to package, but the tradeoff is probably worth it.
Millions of lines? I can't see it as worth the effort for the Open Source community to bother with what is probably an intractable morass of buggy garbage.
It'd be MUCH more useful to force them to accurately publish their API, and split them into an Apps division and an OS division, with the Apps division using the same documentation as everyone else.
As with Mozilla, the world doesn't really want to share responsibility for a project which wasn't theirs to begin with and was arguably ruined before the world ever got access.
I, for one, will be VERY displeased the only remedy forced on MS is to open their source. This measure would be woefully inadequate on its own.
Because the CS department admins had tired of repeated breakins to unsecured Linux machines, they created a modified RedHat version which turned off most of the daemons, and included new rpms for common security tools.
As far as I know, though, they don't have a snappy name for the "new" distribution.
DRAM access is 50ns, or so, and that is just to get stuff off of the chip once it's asked for at the chip.
In reality, it takes more than 120ns to do this: 1) Miss the cache 2) Ask for the data on the bus 3) Get the data off of the chip 4) Ship the data back to the CPU 5) stick it in a register
6ns is just the clock period of the memory bus on the very fastest SDRAM interfaces. 133MHz memory does NOT imply a 6ns access time: it means that you can put data across it every 6ns.
I looked at VAResearch. They have quad P-III Xeon and quad Xeon machines, but no quad P-II machines. I think the other posters are correct: there ARE no quad-P-II boards.
Well, we certainly use Linux for our plasma simulations. We use Linux/Alpha too, though that we've had trouble with. The major thing missing from Linux from our point of view are some very good optimizing compilers. I've tried some of the commercial offerings and they didn't work well on our C++ code, which vendor compilers on other Unixen compile without trouble. egcs/g++ also work well.
Many of my colleagues here at UCB also use linux heavily for research, and SLAC has just christened a cluster of 16 dual P-IIs for distributed computing.
Slashdot Mirror
I too have an Nvidia TNT2 card and I'm using
:0 -bpp 32
Nvidia's driver with XFree 4.
Here's what top says:
Size RSS Share
252M 252M 2024 S 0 1.7 100.4 5:18 X
Seems excessive, doesn't it? Well, I've only
got 256M on my machine, and guess what?
NO SWAP SPACE IS USED.
PS tells a different story:
VSZ RSS
276408 12704 ? S 10:00 5:21 X
VSZ is the VIRTUAL size of the process, 276M
12.7M is what it actually uses.
Another poster in another forum explained that
this apparently huge virtual size was due to
virtual-memory-mapping of various bit planes and color depths in the VRAM into virtual memory.
12.7M is still pretty high, but hardly burdensome on my 256M machine.
PeterM
What is the ecological impact of putting wave generators all over the coastline to generate power? How many beaches and coastline ecosystems will be destroyed? Will fisheries be reduced in productivity?
How much does it cost to build a wave plant? How survivable are the plants to storm damage?
When is THIS one scheduled to smash into the earth?
Right. It is not the fault of the person asking, but a fault in the editing.
I would like to see this tried. It seems like the moderation and meta-moderation infrastructure is already largely here.
At the very least, a much-rejected story should be referred to the editors for review.
http://oss.software.ibm.com/developerworks/opensou rce/afs/
This story was on slashdot three days ago!
60G hard disk ($300): .5 cents/M
10G extern. USB ($155): 1.5 cents/M
1.44M floppy: ($.10): 7 cents/M
64M flash: ($150): 234 cents/M
The USB external hard disks seems like the big
win to me, for moving stuff around!
Unexpected power-off is NOT the only thing which can happen to a filesystem. What about these disasters?
1) Bad block takes out part of your disk unexpectedly.
2) Your OS screws up and spews a mess onto your filesystem before it crashes. (there ARE bugs in the kernel!)
3) You have a minor headcrash which takes out one of your tracks, but the disk is still functional.
What're you gonna do? Tux2 isn't gonna help you.
You could restore your latest dump. You could
also attempt to repair the filesystem.
You need fsck or some other means of filesystem repair.
Someone else helpfully supplied this quote from the US constitution:
Article 6, Clause 2: This Constitution, and the Laws of the United States which shall be made in Pursuance thereof; and all Treaties made, or which
shall be made, under the Authority of the United States, shall be the supreme Law of the Land; and the Judges in every State shall be bound thereby,
any Thing in the Constitution or Laws of any State
^^^^^^^^^^^^^^^^^^^^
to the Contrary notwithstanding.
The obvious interpretation of this is that a
treaty supercedes a STATE law or STATE constitution. Any other inerpretation simply makes no sense.
PeterM
My plan is to keep my own data on "live" media:
I'll keep everything on my computer HD (and a copy on the mirror HD), and whenever I upgrade, I'll copy all the old stuff of significance onto the new huge hard disk.
And BTW, in reply to the guy claiming $30 for 5.2G is a good price: nope, it's not. 20G disks are
$84 now. That's $4.20/G. Your slow 5.2G DVD-RAM
costs you $5.76/G, not counting the cost of the drive.
I'd rather have the hard disk for backup purposes.
PeterM
I can easily see more latency coming out of
RAM compression/decompression: you have to
actually do all the computation. However,
if the data coming across the memory bus is
compressed 50%, then you can get 2x throughput,
so that you can double your bandwidth at the
cost of some extra latency.
The big question is how bad is the latency? If
it is too bad, then performance will suffer. If
it is fast enough, then performance may
actually *increase*.
PeterM
The fix for 32GB disks was done earlier than
/scratch1bak
2.2.15 pre 6 or 7.
Here is my 2.2.14 kernel machine with a 40G
disk on it:
alfven [2] uname -a
Linux alfven 2.2.14 #3 SMP Fri Apr 7 16:48:33 PDT 2000 i686 unknown
alfven [3] df
Filesystem 1k-blocksUsedAvailableMounted on
/dev/hdb1 38005262 5960733 30043635
Maybe some people just do open source software to make the world better? Greed and self-aggrandizement are not the only human motivators.
Giving software costs the giver only his time, and for his time he may improve the lives of millions of people.
There's also the pure satisfaction of creating something of worth.
Open software people are simply a class of people for which there is more to life than looking out for number 1.
http://www.tuxedo.org/~esr/gif2png/
On the other hand, the cost of sending email is negligible.
Spam would be a lot more manageable if it cost them $.33 for each spam they sent out.
Now all we have to do is find a way to charge spammers.
PeterM
Their web site says their satellites are in geosynchronous orbits.
Geosynchronous orbit is 37500 km. This translates into a minimum roundtrip speed-of-light latency of 250ms.
Any netrek player will tell you that that amount of latency is too high for effective gaming!
Our plasma simulation group has several simulation codes which would be pretty good as part of an open-source floating-point benchmark suite--*provided* this benchmark suite is distributed under the GPL or Berkeley license.
We considered giving our codes to SPEC, but SPEC wants to be able to *sell* their benchmark suite for $500 a copy. This caused us legal headaches so rather than deal we didn't try to participate in SPECfp2000.
We can offer C and C++ codes which exercise the FPU and memory subsystem heavily: they tend to be cache friendly though.
PeterM
The industry's been working on low-k dielectrics
for a long time.
The real issue with low-k dielectrics is that they
reduce the capacitance of the onchip wiring.
This has several benefits:
1) Wiring capacitances in general will be
signficantly reduced.
2) Wires running parallel to each other will
have less crosstalk. I.e., if the voltage on
one wire switches, it'll have less tendency to
drag the voltage of wires close to it with it.
This is a result of neighboring wires having
mutual capacitance. --> EASIER DESIGN
3) Since capacitances are reduced, less power
will be needed to cause metal wires to switch
voltage. --> COOLER CPUs, -->LOWER POWER
4) Because of the reduced capacitances, it requires less current to switch wires.
--> FASTER CPUs.
This is a really significant advance by IBM,
maybe more signficant than copper metallization.
One possible problem is poor heat conduction
from die to package, but the tradeoff is probably
worth it.
How large is the source code for Win2k?
Millions of lines? I can't see it as worth
the effort for the Open Source community to
bother with what is probably an intractable
morass of buggy garbage.
It'd be MUCH more useful to force them to
accurately publish their API, and split them
into an Apps division and an OS division, with
the Apps division using the same documentation
as everyone else.
As with Mozilla, the world doesn't really want
to share responsibility for a project which
wasn't theirs to begin with and was arguably
ruined before the world ever got access.
I, for one, will be VERY displeased the only
remedy forced on MS is to open their source.
This measure would be woefully inadequate on
its own.
PeterM
Because the CS department admins had tired of
repeated breakins to unsecured Linux machines,
they created a modified RedHat version which
turned off most of the daemons, and included
new rpms for common security tools.
As far as I know, though, they don't have a
snappy name for the "new" distribution.
DRAM access is 50ns, or so, and that is just
to get stuff off of the chip once it's asked for
at the chip.
In reality, it takes more than 120ns to do this:
1) Miss the cache
2) Ask for the data on the bus
3) Get the data off of the chip
4) Ship the data back to the CPU
5) stick it in a register
6ns is just the clock period of the memory bus
on the very fastest SDRAM interfaces. 133MHz
memory does NOT imply a 6ns access time: it
means that you can put data across it every
6ns.
IBM a while ago issued a press release claiming
they've developed the "holy grail" of memory:
SRAM speed (10ns access)
DRAM density
nonvolatility
If they realize the promise of this type of memory, many computers may not need hard disks at all in 10 years, when 4Gb chips are being manufactured.
I looked at VAResearch. They have quad
P-III Xeon and quad Xeon machines, but no
quad P-II machines. I think the other posters
are correct: there ARE no quad-P-II boards.
Well, we certainly use Linux for our plasma
simulations. We use Linux/Alpha too, though that
we've had trouble with. The major thing missing from Linux from our point of view are some very good optimizing compilers. I've tried some of the
commercial offerings and they didn't work well
on our C++ code, which vendor compilers on other
Unixen compile without trouble. egcs/g++ also
work well.
Many of my colleagues here at UCB also use linux heavily for research, and SLAC has just christened a cluster of 16 dual P-IIs for distributed computing.