Save cash on eletricity , by using the proper VGA card

[Kiriakos-GR]

One nice link about computer graphic cards  , and power consumption . 

http://translate.googleusercontent.com/translate_c?hl=el&sl=de&tl=en&u=http://ht4u.net/reviews/2009/asus_radeon_eah_5770/index11.php&rurl=translate.google.gr&usg=ALkJrhie8o9SlwJ572wNGJz7vC0BU9klAQ

[DJPhil]

Neat stuff!
I like how you can mouse over any of the cards and the display changes to percent. I'll have to remember that trick.
I wonder if there's an easy way to test on board video too. I'd guess that onboard video would use less power just by the absence of the fan and extra circuitry, but I imagine it's hard to say.
Thanks for the link, I was just thinking about picking up a newer graphics card.

[Mechatrommer]

my GPU is Radeon 4650, listed closest is only 4670? how do i know how much 4650 consumes?

[DJPhil]

my GPU is Radeon 4650, listed closest is only 4670? how do i know how much 4650 consumes?

There's an article about these on tomshardware. It explains some of the basic differences, and near the end of the first page they have some power information that suggests that your 4650 uses less power overall in most of it's incarnations. It's probably not conclusive, but it might be a start.

[Mechatrommer]

There's an article about these on tomshardware. It explains some of the basic differences, and near the end of the first page they have some power information that suggests that your 4650 uses less power overall in most of it's incarnations. It's probably not conclusive, but it might be a start.

thanx. just re-open my old box, it turned out mine is gigabyte version, in the link there is AMD, Sapphire, PowerColor, etc. hmm, makes me more spinning. should i care? i dont know, i just use it for some purpose to make me happy, thats all i care right now

[saturation]

Video cards are heavy duty minicomputers these days, its a reason CPUs can do more, the video processing is off loaded to a dedicated processor, with its own OS and memory.  It can draw a lot of power, depending on what needs be done: tip, look at their heatsinks!

[Mechatrommer]

Video cards are heavy duty minicomputers these days, its a reason CPUs can do more, the video processing is off loaded to a dedicated processor, with its own OS and memory.  It can draw a lot of power, depending on what needs be done: tip, look at their heatsinks!

well... mine got the heatsink.... and the fan too!
i'm not really into gaming and 3d stuffs (not anymore), but thats the cheapest decent GPU i can get locally.
maybe the solution for power saving, is using the MB GPU, but that kinda too slow for me either.
save cash for electricity, save cash for the graphic card itself as well

[Kiriakos-GR]

Well, currently I have an HD3850 AGP , that is a nice card for games and video , but i am forced to switch to PCIE , and I have to start looking for an replacement VGA  , under those Watt comparisons , it looks that the HD5770 its the card for me.

In direct comparison of the  HD3850 VS HD5770 , both needs the same wattage at idle and load ,
with the only deference , that the second its 5 times faster ... 

 

[orbiter]

I'll be looking for a reduction in GPU power consumption next time around too. At present with 2x GTX295 (Quad SLi) and a 4.2GHz CPU overclock.. my system is pulling over 750W from the wall  I'm just hoping that as technology moves on and dies get smaller, the power requirements for such parts will be reduced whilst still being able to perform spectacularly.  

[Kiriakos-GR]

My plan are to exchange  the single core with an modern quad , that has the same power consumption..
The "old" VGA, with the five times faster one , with the same power consumption ..

And the true saving will come , as soon I replace my SONY 19" CRT ( 160W ) ,
with an new modern 23" screen ( 45W) .

And this is the best that I can do, so to save the planet, and stay happy .       

[Mechatrommer]

I'm just hoping that as technology moves on and dies get smaller, the power requirements for such parts will be reduced whilst still being able to perform spectacularly.  

dont worry pal, i can assure you that!

[Mechatrommer]

My plan are to exchange  the single core with an modern quad , that has the same power consumption..
The "old" VGA, with the five times faster one , with the same power consumption ..

And the true saving will come , as soon I replace my SONY 19" CRT ( 160W ) ,
with an new modern 23" screen ( 45W) .

And this is the best that I can do, so to save the planet, and stay happy .       

quad = $$$
5x vga = $$$
23" lcd = $$$

sure you can save electricity!
ps: i tried saving my time harder

[TopherTheME]

 I'm just hoping that as technology moves on and dies get smaller, the power requirements for such parts will be reduced whilst still being able to perform spectacularly.  

Don't count on it. Future processors will probably be a little more efficient without much of an increase in computing power. CPU design is getting very close to hitting a technological wall meaning that in order to make significant increases in processing power, dies have to go 3D instead of the conventional 2D.

[NiHaoMike]

Don't count on it. Future processors will probably be a little more efficient without much of an increase in computing power. CPU design is getting very close to hitting a technological wall meaning that in order to make significant increases in processing power, dies have to go 3D instead of the conventional 2D.

There is a lot of room for improvement. The biggest part is to abandon x86 in favor of better designs. Depending on what you want, ARM, MIPS, and Cell are probably the best bets.
http://www.silentpcreview.com/forums/viewtopic.php?t=49884&highlight=&sid=aa2474b860cb4f78b90725897a93e9b2

Part of my job is developing embedded software for a custom board with a low power ARM7 microcontroller (clocked at 48 MHz), but we don't run an operating system at all. My understanding is that MHz for MHz, an ARM will generally beat an x86 because of all the legacy cruft in the x86 instruction set and architecture which use up gates in the CPU.

http://forums.reghardware.com/forum/1/2008/10/14/arm_vs_intel_umd_biz/

Intel CPU's have a fast risc core and huge number of transistors to convert X86 instructrutions into risc instructions. ARM (and MIPS) are risc cores, so the do not need to waste transistors and power converting a legacy instruction set into something that can be decoded efficiently.

Intel and Via CPU's use power to get data to and from the north bridge. The north bridge uses power to get data to and from the memory. ARM (and MIPS) CPU's have a memory controller built in like AMD CPU's, so it only takes half the power to get data from memory to the CPU compared to an Intel or Via chipset.

These two advantages more than make up for the better manufacturing processes that Intel uses for Atoms. ARM (and MIPS) give better performance per watt than X86. Also you do not need to pay for the massively over-priced Inhell Inside sticker, so you get better performance per £ too.

I have even "designed" a PC based on the BeagleBoard.

Of course, that image is a bit outdated now that there's PandaBoard.
http://pandaboard.org/

[Mechatrommer]

I have even "designed" a PC based on the BeagleBoard.

what the speed like? qualitatively
what OS?

[NiHaoMike]

what the speed like? qualitatively
what OS?

The original BeagleBoard is a 600MHz Cortex A8. The new BeagleBoard xM upgrades that to a 1GHz Cortex A8, and the upcoming PandaBoard upgrades it even further to a 1GHz dual core Cortex A9. Not enough for serious development work but more than enough for many common PC uses like web browsing, document editing, and media playback. For something that needs more than one of those boards, a board based on a Cell would be great, but I'm not aware of any.

And I think Linux is the best general purpose OS. Everything can be tweaked if you know how to. For network-heavy applications, FreeBSD can often be an even better choice.

[Mechatrommer]

thanx for the idea. i'm eyeing the cortex right now

[alm]

There is a lot of room for improvement. The biggest part is to abandon x86 in favor of better designs. Depending on what you want, ARM, MIPS, and Cell are probably the best bets.

If you want to trade power for performance. I wouldn't choose an x86 CPU for something like a cellphone either, they're clearly optimized for performance. But I wouldn't buy a PC with any current ARM or MIPS either, not enough bang (as in performance) per buck. I've nothing against alternative architectures, but they're clearly optimized for a different market. Optimizing something for power or performance is very different, you can't expect one CPU to be good at both. One of the reasons most architectures other than x86 now focus on niche applications is that they failed to compete on performance. Same with Via.

http://www.silentpcreview.com/forums/viewtopic.php?t=49884&highlight=&sid=aa2474b860cb4f78b90725897a93e9b2

Part of my job is developing embedded software for a custom board with a low power ARM7 microcontroller (clocked at 48 MHz), but we don't run an operating system at all. My understanding is that MHz for MHz, an ARM will generally beat an x86 because of all the legacy cruft in the x86 instruction set and architecture which use up gates in the CPU.

The problem for competitors like ARM is that the cost of this x86-to-RISC translator is basically constant, so with the increasing number of gates in modern CPU's and decreasing cost and power per gate, the extra costs are becoming less significant. They've been doing this since the P6 (Pentium Pro) I believe. The number of transistors used in the whole Pentium Pro is small (5.5M according to Wikipedia) compared to a modern CPU (382M for Clarkdale according to Wikipedia), and the translation logic was only a small part of the Pentium Pro. The implementation might be more advanced in modern CPU's, but I can't imagine it being significant in terms of transistors or power.

http://forums.reghardware.com/forum/1/2008/10/14/arm_vs_intel_umd_biz/

Intel CPU's have a fast risc core and huge number of transistors to convert X86 instructrutions into risc instructions. ARM (and MIPS) are risc cores, so the do not need to waste transistors and power converting a legacy instruction set into something that can be decoded efficiently.

Let's drop the RISC vs. CISC debate, it's not the 1990's anymore. Modern processors are somewhere in between.

Intel and Via CPU's use power to get data to and from the north bridge. The north bridge uses power to get data to and from the memory. ARM (and MIPS) CPU's have a memory controller built in like AMD CPU's, so it only takes half the power to get data from memory to the CPU compared to an Intel or Via chipset.

This is not true anymore since the Core i3/i5/i7/Pine Trail.

[Kiriakos-GR]

My plan are to exchange  the single core with an modern quad , that has the same power consumption..
The "old" VGA, with the five times faster one , with the same power consumption ..

And the true saving will come , as soon I replace my SONY 19" CRT ( 160W ) ,
with an new modern 23" screen ( 45W) .

And this is the best that I can do, so to save the planet, and stay happy .        

quad = $$$
5x vga = $$$
23" lcd = $$$

sure you can save electricity!
ps: i tried saving my time harder

I got the Quad Q6600 8M cache  ( used )  100EUR
The VGA its another 130EUR
And the monitor  around 240 EUR

The old SONY CRT G420 , had cost me at 2001  1070EUR .
Its manufactured in UK , and only the last year had start to show some signs , that something its not very healthy in the video controller in it (RGB ) .
Unfortunately if it gets permanently damaged , no one are able to repair it in Greece.  

[NiHaoMike]

In the portable segment, ARM easily beats Atom. In the high performance segment, I don't think there's any x86 that can compete with Cell. X86 does well in the middle, but the gap is closing from both sides.

I think there should not be a hardware decode stage (or make it bypassable) and let the compiler do the hard work. A software-based compiler will not be restricted by complexity limits.

[alm]

In the high performance segment, I don't think there's any x86 that can compete with Cell.

Last time I checked, x86 CPU's (eg. Opteron) were doing pretty well in the HPC market. In the supercomputer top 10 of June 2010, numbers 1, 2, 4, 6, 7, 10 are Intel/AMD powered. Number 3 is hybrid Opteron / Cell, and numbers 5, 8, 9 are something else (Blue Gene). Cell is hardly dominating there. I wouldn't call supercomputing a real threat for Intel/AMD, since it's a tiny niche market compared to PC's. I'm not aware of anything PC-like using the Cell, but maybe I missed it?

One issue with complex CPU's like the Cell is that it's hard to get software developers to use them efficiently. The PS3 is a closed platform, but getting desktop software developers to optimize their software for the Cell is much less likely.

X86 does well in the middle, but the gap is closing from both sides.

It's likely that the Atom netbooks will lose marketshare to ARM, but remember that the Atom is really slow compared to an entry-level Core i3/Phenom, probably about the performance of mainstream CPU's ten years ago or so (at least the low-end Atoms). ARM is still a long way of from competing with the AMD/Intel mainstream products. I don't see the Cell competing in the high-end-PC/workstation market, are there any workstation level applications yet? A major issue in the PC market is backwards compatibility, people still want to run the crappy VB6 app from many years ago. Getting them to switch will require a major advantage in some other important area for the user.

I think there should not be a hardware decode stage (or make it bypassable) and let the compiler do the hard work. A software-based compiler will not be restricted by complexity limits.

They tried that with the Itanium. Ask HP/Intel how that worked out. Turns out compilers didn't improve as fast as Intel had hoped, and didn't handle the complexity very well.

[Kiriakos-GR]

As long  the user pays for his software  ( any true working application ) , it becomes his property.
And so its totally logical that no one will scrap his money , because some fellows they like to use an new platform, so to play with.

I have lots of legal software that works on XP-Pro , I do not move forward even an inch, so to not loose my money that I have invest in software.

By my believes , any company that it does sell software , it should offer the new version at the current customers for free.
Operating system or what ever.

As long they do not do that, and they charge for its little improvement that they make ,
they are the ones , who influence the market , by an negative statical way.

I am loosing my money by getting the latest hardware, to loose also more money about getting the latest compatible software , I call this stupidity , because I loose double times , and they win double profits .    

 

[Mechatrommer]

if you buy new OS? then be ready to buy new softwares, and new hardwares too. i have bad experience when upgrading from win98 to winXp. i have to tossed most of my hardwares (no compatible drivers). so now i have tens of thousand worth of equipment that work in xp, think i'm going to toss them? no way! and think i'm going to learn the new vb.net? no way! i suck at it! let it belong to the new kiddies on the street. i dont want to spend the rest of my life to keep learning and playing.

there is current paradigm that most people still dont get behind it... with newer super latest pentium chip (or whatever processor brand that you like) we will have super fast processing power! well ok. and then the microsoft, the game, graphics, 3D developers, even the spreadsheet and word processing softwares come out with the latest software that utilize every bit of the latest features, and put alot of very very nice useless graphics, colors, animations and doggies. in the end, we will get what we get used to be, in term of performance and achievement... just "perceptually" happier.

having sex with a slave and having sex with a princess is just the same if you think about it (with condition the slave looks the same to the princess on their face , and... not thedigitalprincess ), its just the paradigm that makes it different. and we with our computer with thousands of millions processing power per second, we still cannot do any good compared to mr thomas alva edison. let alone to do anything basic nowadays... without a computer. we are getting pampered, more than ever before, and most kiddies nowadays have lost their true direction due to this technological advancement.

just my mumbling of jumbling rubbish during "boring time at work"

[NiHaoMike]

Why pay for software when for many of the common ones out there, there's a free alternative that is as good or better? And emulation is the way to go to make sure the applications will always work.

They tried that with the Itanium. Ask HP/Intel how that worked out. Turns out compilers didn't improve as fast as Intel had hoped, and didn't handle the complexity very well.

Do a pre-release so the GCC developers can get something done.

[Kiriakos-GR]

Long time back I was an member at betanews.com ..
I was spending daily time over there,  so to trace  the update of any small application, that I got for free..

All those trainee developers , they do an small change , and next thing are to download over and over and over,  the software again , so to stay updated .. 

Nop, wasting time about hunting updates of trainee developers , its not my style any more.

I bet that is a matter of age ,  after 30s or more , you are looking to find more productive ways , so to save time and effort.
 

[Kiriakos-GR]

Well , other than video cards that they targeting on saving energy , and the modern CPU's , and the motherboard manufacturers like GIGABYTE they started to act alike,
GIGABYTE  are using an new technology called  DES ( Dynamic Energy Saver) .

Under some tests , it looks to save just 7W per hour , that actually its not sound much,
but it is something as gain in a year base.
    
I got the EP35C-DS3R used, and its on the specs ..

Personally the last two years I am looking more and more about getting devices that they do save energy ,by the one or the other way ...  Even when choosing my  8ports 1GB  switch hub, I got an smart model by ASUS , that it does power off the unused ports , or use an low speed mode at the ones that they stay inactive for long periods.

The only power hungry device next to me its the  SONY CRT , that belongs in the past decade.






.
Go Green with GIGABYTE P45 Motherboards / ?? ?



Launch your Dynamic Energy Saver today! / !

[Mechatrommer]

my gigabyte M912 netbook is a power hog! cant leave a fully charged battery in it during switched off, tomorrow the batt will be depleted!

[saturation]

On the side, shafri, that sounds like the battery just need changing.  As for new PCs, the new generation netbooks get 6-8hr of continuous working power, over 15 hours of off on use.  I'm using it now and never shut it off, just let it 'sleep.'

my gigabyte M912 netbook is a power hog! cant leave a fully charged battery in it during switched off, tomorrow the batt will be depleted!

[Mechatrommer]

On the side, shafri, that sounds like the battery just need changing.  As for new PCs, the new generation netbooks get 6-8hr of continuous working power, over 15 hours of off on use.  I'm using it now and never shut it off, just let it 'sleep.'

battery is OK, i've made a test... full charged the battery during netbook ON, and then leaved it on the netbook ... and full charged the batt and take out of the netbook. the netbook consumes power when turned OFF, maybe i got a lemon netbook not suppose to be like that right? the M912 is quite an old model already (few years back). a total waste of electricity! i've seen many laptop/netbooks that dont behave like that.

[Kiriakos-GR]

My ACER laptop ( all subsystems by Intel ) Travelmate 2414 , came with an Celeron as CPU, that was had an fixed speed 1.5G , I replaced the CPU with an 1.6 Pentium that can do speed step  , and also did and an " simple " modification that works on this motherboard , and now the CPU works at 2,1G ( top speed )  - 800MHz ( idle )     , or I can set the speed to 800 on demand by software.

There is no need to say that the battery last more , and that is so fast , equal to an single core Pentium 4 3.2G  

[NiHaoMike]

Out of curiosity, what is the typical transistor count for a modern ARM or MIPS core? I'm under the impression that it is possible to fit a lot of ARM and/or MIPS cores in the silicon footprint of a single x86 core.

[joelby]

One estimate puts the dual core ARM Cortex A9 at 26 M transistors, compared to about 47 M for a single core Intel Atom.

[NiHaoMike]

So roughly 15 Cortex A9 cores would fit onto the same silicon footprint as a single i7 core. (Forget the Atom, that's a really lousy CPU.) And I'm sure the 15 A9s would outperform the single i7. The A9s will definitely do much better for power, particularly at low load where all but one of the cores can be powered down.

So if one ARM core couldn't outperform one x86, the fact that you could fit many more ARM cores in the same amount of silicon would mean ARM would deliver more performance per chip production cost, assuming the same process.

[joelby]

So if one ARM core couldn't outperform one x86, the fact that you could fit many more ARM cores in the same amount of silicon would mean ARM would deliver more performance per chip production cost, assuming the same process.

Yes, with a big but - either your problem needs to be highly parallelisable or you need to be highly skilled at multi-core programming to use all of the cores efficiently. In general-purpose computing there are currently diminishing returns from additional cores, partly because compiler technology hasn't caught up and partly because it's more difficult for programmers to think about concurrency, but in specialised or embedded situations it's easier to conceive of a situation where all of the resources can be used maximally.

[Mechatrommer]

Forget the Atom, that's a really lousy CPU.

who's you calling lousy? if it is, then ARM is too!
http://netbooked.net/blog/arm-vs-atom-size-vs-power-vs-performance/
http://www.geek.com/articles/chips/arm-posts-cortex-a9-vs-atom-performance-video-intel-should-be-worried-2010016/
anyone care to compete Atom and ARM in FP intensive such as 3D and game? i did surf internet with my Pentium1 long years ago. no worries!

And I'm sure the 15 A9s would outperform the single i7. The A9s will definitely do much better for power, particularly at low load where all but one of the cores can be powered down.

and now u wanna compete with i7. who's going to handle the multitasker and resources manager... u? if its proven, then why they (the manufacturer) dont do it... today! if they are so brilliant. i'll surely 100% buy one.. the fanless multicored GHz'er ARMs... for my WinXP.