Page 2 of 3

WMV HD can not play?

Posted: Fri Jun 02, 2006 1:56 pm
by chinna_n
I was going thru article and I wondered how come systems mentioned could not play 1080P HD material or can hardly play 720P. Is it same material as WMV HD clips provided by Microsoft? or something else.

Why I am asking is, my desktop with P4 [email protected] can easily play 1080P with a whimpy MX440-SE video card. ( ofcourse my LCD monitor is 1280x1024). Even my Laptop with Turion ML-37 can play 720P with less than 50% CPU load and with 1080P about 60-70% CPU usage.
None of the test systems was capable of playing either of the HD samples correctly, although the EPIA probably came the closest to playing the 720P clip smoothly. Only an intermittent stutter in the audio track caused us to notice that the clip was not played back properly. Neither of the other two systems — or even my A64 3800+ main system — was capable of playing the 720P clip without frequent dropped frames. Oddly enough, using Window Media Player instead of Media Player Classic degraded performance significantly, and the EPIA was not capable of playing the 720P video smoothly when Window Media Player was used. None of the test systems were capable of playing the 1080P clip properly
So how come even sempron 3400+ struggled and it is also mentioned Athlon64 3800+ struggles with 720P!!? :roll:

Posted: Fri Jun 02, 2006 2:59 pm
by Devonavar
Yes, I used the Microsoft samples.

I've been looking for a coherent answer for why even my main system can't play HD files for some time, and nobody has been able to give me even close to a reasonable answer. For a while, I thought it was my VGA card, but your report shows that that is probably not the case. If anyone else can shed some light on this, I'd be most greatful. The only system I've ever seen that can play 1080P smoothly was a Pentium D system that we built in the lab to test the effectiveness of one of the cases that we reviewed.

Posted: Fri Jun 02, 2006 9:51 pm
by Slaugh
From Dailytech:
VIA Launches EPIA-CN
Ultra low power and embedded systems friendly

VIA has been on a roll as of late with the launch of new EPIA mini ITX motherboards. The latest addition to the EPIA family is the ultra low power VIA EPIA CN. Two EPIA CNs will be available—a 1.3 GHz CN13000 and a 1.0 GHz CN1000E. Average power consumption is around 16 watts which make the new EPIA CNs perfect for the embedded PC market.

Other notable features include the VIA CN700 northbridge which features VIA UniChrome Pro AGP graphics. The UniChrome graphics support hardware MPEG-2 acceleration via the Chromation CE engine and dual display support via DuoView+. Strangely enough, VIA claims SATA II support with the EPIA CN, though it seems impossible as the VIA VT8237R lacks SATA II support.

Additionally the EPIA CN features a single PCI slot, S-Video, composite video, S/PDIF audio outputs, serial, parallel, 10/100 Mbps Ethernet, and USB 2.0 ports. There’s also support for two parallel IDE and two SATA connectors with support for RAID 0 and 1. Memory support is limited to 1GB max as the board only has a single 240-pin DIMM slot to accommodate DDR2 400/533 MHz memory. VIA’s PadLock Security Engine is also integrated for hardware accelerated encryption processing.

Pricing is expected to be $193 while availability is expected in June.

Posted: Sat Jun 03, 2006 5:15 pm
by defaultluser
Good review.

The product itself is not bad, but it is not all that great for $270, unless you absolutely need THE LOWEST power consumption on the market, aside from using a notebook.

You still have to pony up for ram, a hard drive, powersupply and a case, plus an optical drive if you want to use it as a full-fledged computer. And when you try using it as a full-fledeged computer, you realize that the performance just isn't there. The performance, instead, is 6 years in the past.

The C7 a die-shrunk C3 with increased cache, and a new bus interface. The C3 is based on the Winchip, and hasn't seen any serious architectural changes since Via bought Centuar. They added branch prediction with Nehemiah, and a full-speed FPU, but these are minor changes. The processor still only has a SINGLE integer pipeline and SINGLE floating-point pipeline, which is the reason for the poor performance.

The improved bus and the larger cache helps performance, but you can only squeeze so much out of an underpowered processor. If you need something more powerful than a P3 600 (with a Savage 4 [aka UniChrome] graphics card), look elsewhere.

Posted: Sun Jun 04, 2006 2:50 am
by Bluefront
The fact that the sys temp usually registers higher than the CPU temp, is nothing new. Usually the CPU has some active cooling, along with a big heatsink. The sys temp has no standard spot of measurement. Sometimes it can be measured at a place of little airflow, with no heatsink. The ASUS board in my Aria project is like that....the sys temp always measured almost 10C more than the CPU. Only under CPU benchmarking did the CPU temp exceed the sys temp.

Well I was curious and somewhat concerned.....so I checked every board component while the computer was running, using a non-contact temp reader, which uses a laser beam. This makes it easy to get into tight places for a reading. The South-bridge immediately became suspect because of a high reading....nope. The north-bridge had active cooling and measured about the same as the CPU. In the end I determined the Windbond chip on the far corner of the board, was where the sys temp sensor was located. I put a small heatsink on the chip, and modified the airflow so the winbond chip got some airflow.

That did the trick. The sys temp is now usually a max of 3C over the CPU temp. Did it do any good, was it worth the trouble? Heh....... :lol:

Posted: Sun Jun 04, 2006 3:08 am
by Firetech
Bluefront wrote:T was it worth the trouble? Heh....... :lol:
Of course! It proved that there was nothing "seriously wrong" with your board either... :D

Posted: Sun Jun 04, 2006 9:28 am
by joolz
Hello,
My first post.

I have a question for Devonavar, how is the picture quality from the VGA port? I have seen som people complaing on other boards about picture quality. Mostly on older models. Picture quality is verry important for med because i want to use this board in a HTPC. thanks

pics:
http://gbire.free.fr/epia/epia_sp_vga.php

Posted: Sun Jun 04, 2006 12:08 pm
by Devonavar
joolz, I'm glad you asked. I meant to comment on that in the review.

VGA quality was quite poor. Although it is *capable* of outputting a 1600x1200 @75 resolution (my preference), text was unreadable at this level, and I had to back things down to 1024x768 before I got an acceptable picture. Even then, the display was noticably blurry.

I would highly recommend getting the DVI daughter card if VGA quality is important to you.

Posted: Sun Jun 04, 2006 12:56 pm
by joolz
alright, thanks for the info. i can't get a DVI daughter card because i must use the one pci-slot for the pvr tv-card. i think its really strange VIA doesn't fix this, because VGA quality should be important to most of the users.

anyway, thanks for the info. Maybe one day it will be possible to build a silent HTPC :)

/joolz

Posted: Sun Jun 04, 2006 2:13 pm
by Devonavar
The DVI daughter card doesn't occupy the PCI slot; it attaches directly to a port on the board via a ribbon cable. The integrated graphics are still being used — only the means of output changes. However, because DVI is a digital signal, image quality should no longer be dependent on the VGA port.

Posted: Mon Jun 05, 2006 1:50 am
by joolz
yes but i still have to install the dvi port on the back of the case right? I havent found a mini-itx case that have two slots for extension cards on the back.

Posted: Mon Jun 05, 2006 2:08 am
by Bluefront
Heh...that's what a dremel or a tin snip is for. Or you could get a low-profile tuner card, and use the top of the pci port for the DVI output.

Posted: Mon Jun 05, 2006 8:31 pm
by CityK
Devon,

Couldn't agree more with some of the things you wrote. Specifically:
No matter what benchmark is used, VIA's processors take a beating compared to even low-end offerings from Intel and AMD.

On the other hand, even the lowest-performance offerings from Intel and AMD are overkill for many — perhaps most — computing tasks, which is why VIA's offerings are viable at all. Performance alone is no longer king, and performance-per-watt is increasingly important. In this arena, VIA is a pace or two ahead of both of the big boys. The original C3 processor was designed with power efficiency as a primary design goal.

...[snip]...

testing by other sites such as SFFTech has shown conclusively that the 1.2 GHz Eden processor falls far behind even the lowly Sempron when conventional benchmarks are used. On the other hand, such benchmarks do not yield much insight into whether the EN12000E is powerful enough; they merely show where it stands in relation to other products on the market. If all of the available products are powerful enough for the task at hand, the differences uncovered by the benchmarking are somewhat artificial.
And from your conclusion:
for a general purpose PC where the performance requirements are unpredictable, the EPIA is almost certain to run into the occasional task where a little more oomph would be appreciated.
Which highlights the very nature of these platforms:
Overall, the EN12000E seems best for a purpose-built system.

So long as the performance is up to that purpose, the EPIA can probably do it cheaper, cooler, more efficiently, and more quietly than an AMD or Intel-based system.
When it comes to mini-itx platforms, I'm rather fond of a quote from an epiacenter review, but which can be universially applied:
We'd advise buyers to be very clear on the way the board will be used, and plan software accordingly
Incidently, much of what you have implied is quite complimentary to what the founder of Centuar has previously stated:
The idea I had, which actually was hard for people to accept could be successful, was, "Let's not try to make the world's fastest. Let's look at all the characteristics. Speed is one, cost is another, power consumption is another, footprint is another. And let's make something that wins on footprint size, cost, and power, and is fast enough to get the job done for 90 percent of the people but is not the fastest thing in the world."

The last 10 percent of performance is a huge cost. And not just the hardware side, but also in design complexity. So, in fairness, our parts are slower than Intel's. On the other hand, our parts are fast enough to get the job done for most people. Other than the marketing disadvantage of having slower parts, our parts perform quite well. And they have much lower power, and a much smaller footprint, and they cost much less. Those characteristics appeal to a number of applications.

So, that's our theme. The marketing guys don't like me to say "fast enough" performance, but you know, we're not as fast, head-to-head, as Intel is. But, we are fast enough to do 90 percent of the applications that are done, using a processor. Maybe even more than that.
Now some constructive criticism:
It worked perfectly for DVD playback
It would have been operating in software mode, as MPC's internal MPEG-2 decoder does not support offloading to the CN700 for hardware acceleration. In fact, currently, no app under a Windows OS is able to leverage this feature of this Via chipset.

With the former CN400 and CLE266 chipsets, to the best of my knowledge, only WinDVD and PowerDVD were able to exploit this feature on the Windows platform (although it is my understanding that PowerDVD 7 is a bit buggy in this respect) ... Perhaps configuring MPC to use either the Cyberlink or Intervideo filters would work for these older chipsets too, but I'm uncertain. Anyways, regardless, the hardware acceleration feature is currently a no show for the CN700 on Windows.

The flipside, of course, is that the seamless DVD playback operation (via software decoding) you observed helps affirm your question – “whether the EN12000E is powerful enoughâ€

So close...

Posted: Wed Jun 07, 2006 4:17 pm
by dougz
First-rate review from Devon and CityK's comments were very useful.

To-date the only Mini-ITX solutions that are suitable for multimedia seem to be Intel platforms. I really admire the engineering in the Centaur/Via CPUs, but the products never seem to be quite good enough for general computing.

Just a bit too slow for reliable multimedia. Driver support is always a bit dodgy, doubly so for Linux. And finally, the products are always late and expensive. Mini-ITX seems to be best suited for industrial embedded computing and case-modding hobbyists.

More and more I have to admire the brilliance of the Apple Mac Mini. Even the AOpen clone can't quite touch it.

Posted: Thu Jun 08, 2006 11:15 pm
by aidanjm2004
defaultluser wrote:The product itself is not bad, but it is not all that great for $270
Although, if you are running it as an htpc or server or even desktop machine that is left on 24/7, over a few years wouldn't you get that money back and more in saved electricity costs?

Posted: Fri Jun 09, 2006 2:18 am
by jaganath
Although, if you are running it as an htpc or server or even desktop machine that is left on 24/7, over a few years wouldn't you get that money back and more in saved electricity costs?
Depends on the cost of electricity in your country. It's like the choice of diesel vs petrol engines, diesels are more efficient, but cost more initially, petrol-engined cars generally get less miles to the gallon, but are cheaper to buy.

Posted: Fri Jun 09, 2006 7:03 am
by Tibors
I'm going to mix $ and € in the next calculation, but for a rough approximation that shouldn't matter.

A Sempron + mobo with integrated graphics and a cooler can be had for as much as $180. The VIA board for $270. Price difference $90. If you calculate the fact Devon is using an expensive PSU add another $50.

The sempron system will run about 50W idle and 76W load. The VIA system will run 28W idle and 33W load. Things like fileservers, gateways and desktops left running run idle most of the time. In the recent power consumption article on SPCR a ratio of 9/1 was used for idle/load.

With the fairly high cost of electricity here in the low lands by the sea of €0.17/kWh this gives you a price saving of:
50W vs 28W = 22W = 193kWh/y -> €33/y (idle / no CnQ)
76W vs 33W = 43W = 377kWh/y -> €64/y (load)
10% load + 90% idle -> €36/y

So for me it would take 2.5 year to earn it back or almost 4 years if you factor in the expensive PSU. I believe most people in North America pay about half what I pay for electricity. Now factor in the fact that I lost two (out of four) VIA mini-ITX boards in the last year to "aging". They certainly had run less hours than 2.5 years 24/7. So choosing this board for financial reasons doesn't sound like a good bet.

If you start undervolting and underclocking the sempron system, then you can get an even lower wattage from that machine.

Posted: Fri Jun 09, 2006 10:59 am
by jaganath
If you start undervolting and underclocking the sempron system, then you can get an even lower wattage from that machine.
This is very true. I run my Sempron 3000+ at 1.6Ghz, 1.1V all the time with an 80mm Yate Loon @ 5V on the stock heatsink, it never goes above 35C.

Posted: Fri Jun 09, 2006 11:12 am
by QuietOC
jaganath wrote:
If you start undervolting and underclocking the sempron system, then you can get an even lower wattage from that machine.
This is very true. I run my Sempron 3000+ at 1.6Ghz, 1.1V all the time with an 80mm Yate Loon @ 5V on the stock heatsink, it never goes above 35C.
Yes, and that's not even a real limit. A 90nm K8 can run at less than half of that power even at a reasonable clockspeed (say, 1.2GHz)!

Underclocking w/ Yate Loon on stock cooler

Posted: Mon Jun 12, 2006 3:36 pm
by dougz
This is very true. I run my Sempron 3000+ at 1.6Ghz, 1.1V all the time with an 80mm Yate Loon @ 5V on the stock heatsink, it never goes above 35C.
Now you really have my interest. I've looked at Via's C7, but I like streaming audio & video too much. Via has inferior price/performance. Underclocking an AMD looks like a much better strategy.

I'm looking to build an Nvidia 6100/6150 mobo box in late summer. I had been planning to go Sempron, but I understand that only Athlons support virtualization, which would be worth the difference in price to me.

Am I correct in assuming that the cheaper mobos (like Asrock) would give me fewer options for underclocking? Should I be looking to Biostar and MSI? Other brands?

I'd be willing to go for closeout 754 or 939, as it looks like AMD is moving rapidly to AM2. My needs are kind of "trailing edge" but I do want cool-n-quiet.

Re: Underclocking w/ Yate Loon on stock cooler

Posted: Mon Jun 12, 2006 7:01 pm
by Felger Carbon
dougz wrote:I'm looking to build an Nvidia 6100/6150 mobo box in late summer. I had been planning to go Sempron, but I understand that only Athlons support virtualization, which would be worth the difference in price to me.

Am I correct in assuming that the cheaper mobos (like Asrock) would give me fewer options for underclocking? Should I be looking to Biostar and MSI? Other brands?
I too was very interested in undervolting. I heard that Asrock boards in general supported it. I was also interested in the 6100, so I downloaded the manual for the Asrock K8NF4G-SATA2 mobo. The manual clearly shows the CPU voltage as adjustable down or up. So stupid me, I bought the board. Guess what?

The board has absolutely no way to adjust the voltage. The problem is in the bios settings program, not the bios itself. The old settings program is not available for download (nor is the new one). So I've got a mobo stored away that I feel I was tricked into buying. :(

Posted: Tue Jun 13, 2006 2:01 am
by dougz
The manual clearly shows the CPU voltage as adjustable down or up. So stupid me, I bought the board. Guess what?

The board has absolutely no way to adjust the voltage. The problem is in the bios settings program
What CPU are you running? Important point --
Changing the Vcore from BIOS is only available for CPUs with Cool 'n' Quiet Technology, i.e. Athlon64 or Semprons with PR over 3000+, and even for these you can only go as high as 1.45V. If you want more or have a lower Sempron you must use the hardware mod. Older CPUs on 130 nm are another story, since they have a default Vcore of 1.5V. http://www.ocworkbench.com/ocwbcgi/ulti ... 04925;p=23
FWIW, the Asrock I run does what they said it would, which wasn't that much. Reliable, yes. Tunable, not very. Got what I paid for.

It appeared to me that the tuning options of budget (or second tier) mobos was distinctly inferior to what the premium boards offered. I just want to verify which boards deliver good underclocking capabilities.

Posted: Tue Jun 13, 2006 3:07 am
by jaganath
It appeared to me that the tuning options of budget (or second tier) mobos was distinctly inferior to what the premium boards offered. I just want to verify which boards deliver good underclocking capabilities.
Check out the list of undervoltable motherboards; normally if a BIOS offers undervolting it will also have underclocking options.

Undervoltable Motherboards

I'm using a Winfast Foxconn 760GXK8MC mobo that I picked up for £20 on Ebay; it allows undervolting and multiplier change (from 5x to 10x I think), so it's not necessarily only the expensive mobos that have good BIOS options.

Posted: Tue Jun 13, 2006 5:05 am
by QuietOC
dougz wrote: It appeared to me that the tuning options of budget (or second tier) mobos was distinctly inferior to what the premium boards offered. I just want to verify which boards deliver good underclocking capabilities.
Biostar TForce 6100.
Much better than some premium boards.

Yes, it is not perfect. It simply will not set the 0.800 or 0.850V it shows as options in the BIOS. The 0.900V setting works out to 0.88V (it undervolts slightly by default.) That setting will run my E6 Sempron 2500+ stable at its default clockspeed of 1.4GHz. No need to underclock (and no means to underclock either--200MHz is the lowest HT speed selectable in the BIOS.)

Also, Selecting any CPU voltage in the BIOS except "default" locks out any voltage control by the C'n'Q mechanism. So, I can't configure my 2.7GHz overclocked Sempron to idle at 1.2GHz @ 0.8V. I can still make it adjust down to 1.2GHz, but the vast majority of the potential power savings is from lowering the voltage.

The Tforce BIOS does give you the ability to adjust to the lower whole multipiers on C'n'Q processors, but this is only slightly useful. The board has to actually POST at the higher default multiplier before it will adjust to the lower selected multiplier. This prevents effective undervolting and underclocking since whatever voltage you set has to be able to run the CPU at its default multiplier for a second or two.

Currently, I have my E6 Sempron 3100+ set to 8X in the BIOS and default voltage (to gain voltage control using C'n'Q). I used to be able to get this to work at 300MHz HT speed (8x300=2.4GHz), now 299MHz is flaky, but 298MHz works reliably, and its not becasue the CPU is flakey at that speed. It certainly is not. It because it is flaky at 2.7GHz at 1.4V, but just stable enough to POST slightly slower than that.

If you have a C'n'Q processor just do the undervolting and any underclocking in Windows using CrustalCPUID or RMClock. Any BIOS options for voltage and multiplier are really irrelevant for you. If you have a new C'n'Q eanbled K8 you will probably find you can't undervolt under 1.1V setting which is a real shame.

The coolest running option is a K8 CPU and motherboard that will get you under that 1.1V limit. Examples: my E6 Sempron 2500+ will run at default speed of 1.4GHz at an actual 0.88V on my TForce 6100 and a D0 Sempron 3100+ undervolted/underclocked to 0.8V using C'n'Q via CrystalCPUID in a ASUS T2-AE1.

Posted: Tue Jun 13, 2006 6:59 am
by jaganath
If you have a C'n'Q processor just do the undervolting and any underclocking in Windows using CrustalCPUID or RMClock.
I found that my Semp 3400 wasn't stable using RMClock to "mimic" C'nQ with personalised settings. Not saying it won't work, but BIOS undervolting is working much better for me.

Posted: Tue Jun 13, 2006 7:41 am
by QuietOC
jaganath wrote:I found that my Semp 3400 wasn't stable using RMClock to "mimic" C'nQ with personalised settings. Not saying it won't work, but BIOS undervolting is working much better for me.
It really is the same C'n'Q mechanism as the AMD driver. It just allows you to set your own settings. You have to test those settings of course! I haven't had any problems using CrystalCPUID on several different systems. (other than that 1.1V limit using the E6 processors).

Even motherboards are using the C'n'Q mechanism to change CPU voltage and multipier. Nothing wrong with that--it is all software after all.

Biostar TForce 6100

Posted: Tue Jun 13, 2006 4:29 pm
by dougz
Selecting any CPU voltage in the BIOS except "default" locks out any voltage control by the C'n'Q mechanism
Didn't know that. Key point. Thanks, QuietOC.
If you have a C'n'Q processor just do the undervolting and any underclocking in Windows using CrustalCPUID or RMClock.
Unfortunately, I'm one of those darn Linux users... :?

AnandTech likes the Biostar 6100 boards pretty well -- http://www.anandtech.com/mb/showdoc.aspx?i=2539&p=1

The socket 754 and 939 boards are very similar, except for the number of DIMM sockets (2 & 4, respectively). I downloaded the 939's manual (have to boot windows to unpack) and will read up on BIOS options. Cool'n'Quiet might turn out to suit me pretty well.

If I keep procrastinating on this, I might end up with the Socket AM2 model... :) http://www.biostar.com.tw/products/main ... 6100%20AM2

Update to this post -- "Biostar TForce 6100 AM2 Motherboard Review" --
http://www.pcstats.com/articleview.cfm?articleID=1966

FWIW, the SPCR list of undervoltable mobos looks like it needs an update.

Re: Biostar TForce 6100

Posted: Tue Jun 13, 2006 5:16 pm
by QuietOC
dougz wrote:Unfortunately, I'm one of those darn Linux users... :?
There must be a utility for Linux that gives you the same control...

Re: Biostar TForce 6100

Posted: Wed Jun 14, 2006 4:53 am
by Tibors
QuietOC wrote:
dougz wrote:Unfortunately, I'm one of those darn Linux users... :?
There must be a utility for Linux that gives you the same control...
No, there isn't. You have to hack the code to do that.

Re: Biostar TForce 6100

Posted: Wed Jun 14, 2006 5:23 am
by QuietOC
No, there isn't. You have to hack the code to do that.
So, user configurable C'n'Q is possible in Linux. It's more difficult, but it is linux after all. There are even kernel patches available to ease the process. Well, you get what you choose.