Toshiba SpursEngine SE1000/Leadtek’s WinFast PxVC1100 [Archive]

View Full Version : Toshiba SpursEngine SE1000/Leadtek’s WinFast PxVC1100

lucassp

12th September 2008, 13:42

http://www.xbitlabs.com/news/multimedia/display/20080909151207_Leadtek_Readies_Image_and_Video_Processing_Card_Based_on_Toshiba_s_SpursEngine.html

It seems like CELL's SPEs are on their way to the PC (looks like a small Larrabee). Is there a way to make x264 encoding faster using these SPEs? How does its API compare to CUDA?

Sagekilla

13th September 2008, 21:22

I've already asked before if it'd be possible to use the Cell on a PS3 for encoding, which is more or less the same concept you speak of. It is possible to encode using a Cell processor, but it wouldn't be very fast because there's not a lot of optimizations present for Cell. Unless they gave these things out like candy to everyone I doubt anyone would be really interested.

Blue_MiSfit

13th September 2008, 23:25

Well there's a couple problems...

1) The Linux OS for the PS3 doesn't expose all the SPEs, so all you get is the "control" CPU, which isn't incredibly fast.

Even if we were to overcome this by using one of these Toshiba machines with unlocked SPEs..

2) x264 is heavily optimized for x86 (and x86_64). Porting over this assembly to the Cell's architecture would be... non trivial to say the least ;)

~MiSfit

14th September 2008, 01:55

1) The Linux OS for the PS3 doesn't expose all the SPEs, so all you get is the "control" CPU, which isn't incredibly fast.
Six of the seven (or eight) SPEs are available in Linux and IBM has made the SDK and documentation (like this introductory article (http://www.ibm.com/developerworks/power/library/pa-linuxps3-2/?ca=dgr-lnxw03PS3-Linux-P2)) freely available. Projects like spu-medialib (http://wiki.ps2dev.org/ps3:spu-medialib) already use the SPEs for video processing.

Only the GPU (RSX) remains locked except for basic framebuffer access and mode switching (although there have been some successful hacks to gain more control).

tre31

16th January 2009, 08:45

G_M_C

16th January 2009, 09:30

I sent an email too Leadtek and even though this card is currently only selling in Japan (at this time as far as I know), the customer service representative replied saying they could arrange one for me if I so desired.

Just a little fyi for those that are interested in realtime h264...

That, at least, looks like a sing that they might be more open for suggestions. If they will be as cooperative as nVidia has shown to be to Neuron2, it might come somewhere. But at this time, as the article says, there is hardly any software thats makes use it.

Golgot13

16th January 2009, 15:40

FireCoder Blu is the version of SpursEngine from Thomson.
And it work with Edius software from Canopus (Thomson's company).

poisondeathray

16th January 2009, 17:00

There are some benchmarks, links to reviews here

http://forum.doom9.org/showthread.php?t=142918

tre31

18th January 2009, 08:25

That, at least, looks like a sing that they might be more open for suggestions. If they will be as cooperative as nVidia has shown to be to Neuron2, it might come somewhere. But at this time, as the article says, there is hardly any software thats makes use it.

http://www.avsforum.com/avs-vb/showthread.php?p=15252430#post15252430

Looks like they have pretty good software backing too me, quite a few things on that page...

popper

18th January 2009, 19:03

it seems they are releaseing the SDK soon too

http://forum.beyond3d.com/showpost.php?p=1258308&postcount=89
"17-Jan-2009,One said:Leadtek Japan says they're going to release the SDK for WinFast PxVC1100 around the end of this month. It contains Toshiba libraries except for hand gesture recognition and upconversion.
http://www.leadtek.co.jp/news_release/sdk_pxvc1100.html

Also, CRI Middleware (Japanese game middleware developer) is offering CRI SpursCoder which is a free H.264 encoder for SpursEngine.
http://criware.jp/spurscoder/ "

http://translate.google.com/translate?prev=hp&hl=en&u=http://www.leadtek.co.jp/news_release/sdk_pxvc1100.html&sl=ja&tl=en

http://translate.google.com/translate?prev=hp&hl=en&u=http://criware.jp/spurscoder/&sl=ja&tl=en

http://translate.google.com/translate?prev=&hl=en&ie=UTF-8&u=http://pc.watch.impress.co.jp/docs/2008/1120/tawada157.htm&sl=ja&tl=en&history_state0=

theres a V100_04\CriSpursCoder.txt in the spurscoder zip that seems to indicate the cli options you can use but i cant seem to find a good translator that can make it clear....google didnt do a good job of that.....

tre31

19th January 2009, 02:03

Using systran's textbox translator after text was opened in office using correct symbols (must have Asian text packs installed), then coped from office too the systran textbox, select Japanese source - English destination.

/****************************************************************************** * SpursEngine console encoder “CRI SpursCoder” - gratuitous edition -
* Copyright (c) 2008 CRI Middleware Co., Ltd * Version: Ver.1.00.04 * Date : 2008-12-26
*
----------------------------------------------------------------------------* < Notice > * (a) being gratuitous, only private utilization it can utilize this product.
* Commercial business utilization other way please inquire desired one.
* (b) As for copyright of this product “corporation there is a CRI middleware”.
* (c) The work writer does not owe responsibility altogether in regard to whatever trouble due to the use of this product.
* Please use in the responsibility of the user.
* (d) Analysis, alteration and second distribution of this product are prohibition.
* (e) As for general basic use permission condition other than describing please refer to description below.
* http://www.criware.jp/spurscoder/NDA_spurscoder.htm
* (f) as for reprint of this document please be modest.
* *****************************************************************************/

=============================================================================== ■1.　CRI SpursCoder (later, this product) SpursEngine (TM) (*) you used concerning CRI SpursCoder, it is the encoding tool. It operates on the console.

　[Gratuitous edition] it can do only H.264 video encoding.
　[Compensatory edition] we would like to put out to the next year last third, is.

(*) SpursEngine is corporation Toshiba trade mark.

=============================================================================== ■2.　it does not support this product concerning support as a general rule. Please use with the user making use of the bulletin board coping as a prerequisite.
　However, because the developer has passed through the eye to the bulletin board, when question and demand etc you write, perhaps there is a some reaction.

=============================================================================== ■3.　Operational verification environment it did the operation verification of this product with environment below.

　CPU Intel (R) Core (TM) 2Duo T9400 (2.53GHz)
　Memory 4GB
　OS Microsoft Windows XP Professional Version 2002 Service Pack 3
　SpursEngine (a) Leadtek WinFast PxVC1100
PxVC1100.sys : 1.2.2.11
spdu.dll : 1.1.0.2
candi.dll : 1.2.1.0
(b) THOMSON Canopus FIRECODER Blu (MHX-C1)
mhx_c1_kml.sys: 1.2.2.11
spdu.dll : 1.1.0.2
candi.dll : 1.2.2.1
(c) Qosmio
?

　It is not something which guarantees the operation with this environment. Please acknowledge.

=============================================================================== ■4.　Installation installation method of this tool is shown below.

　(step1) It copies in the folder where pass passed by this tool. (*1) (*2)

　(step2) The necessary file is copied.
　　　　　Description above (step1) in the folder, please copy the file below.

　　(a) In case of Leadtek WinFast PxVC1100
　　　C:\Program Files\Common Files \ Ulead Systems \ Mpeg \ candi.dll
　　　C:\Program Files\Common Files \ Ulead Systems \ Mpeg \ specdc_trans2.ssb

　　(b) When THOMSON Canopus FIRECODER Blu (MHX-C1) is,
　　　C:\Program Files\Canopus\FIRECODER WRITER \ candi.dll
　　　C:\Program Files\Canopus\FIRECODER WRITER \ specdc_trans2.ssb

　(*1) how to pass of pass
　　[Property of system] - [details setting] - [environment variable] - [system environment variable]
　　Variable “PATH” lastly “; C:\MyTools; ” The way it adds.
　　You adjusting to your own environment, you will describe the part “of C:\MyTools”.

　Unless (the *2) it passes through pass?
　　After moving to the folder of this product, unless it is, it cannot execute.
　　If shortcut is made, in order also for the [te] to move through pass, it is possible.

=============================================================================== ■5.　The basic using it shows the basic using below.

　> CriSpursCoder.exe - in=filename.avi [- out=filename.264] [subcommands]

　[Example]
　　> CriSpursCoder.exe - in=sample.avi - bitrate=500000
　　> CriSpursCoder.exe - in=sample.avi - mbps=0.5 - pr=24000 - yuv=ccir709

　　(*) subcommand is our company individual name of generally known “option”.

▼5.1 input/output format (1) input format
(a) Microsoft AVI (yes above 2GB) * recommendation *
Video : Non compressed RGB type (BMP suitable)
Audio: Non compressed PCM type (WAVE suitable) < - having entered, you ignore,
(b) Sequence number still picture
24/32bit BMP of color (non compression)
24/32bit TGA of color (non compression/run length)

(2) Output format
H.264 element stream (H.264-ES)

▼5.2 subcommand

<Subcommand>
- in : Input file name
- out : Output file name (default: Replacement extension of input file name)
- help : Indication of help message
- verbose : Setting of log output {ON | OFF} (default: ON)
- proc_prio : Process priority class setting (default: NORMAL)
{IDLE | BELOW | NORMAL | ABOVE | HIGH | REALTIME}

<Subcommand for video encoding>
- bitrate : Bit rate /avr [, max] [, min] (default: 12000000,1440000)
- kbps : Kilobit rate/avr [, max] [, min] (1k=1000)
- mbps : Megabit rate/avr [, max] [, min] (1M=1000000)
- PR : The number of pictures per 1000 seconds (default: 29970)
- START : Encoding start frame number (default: The 0= first)
- end : Encoding end frame number (default: - 1= terminal)
- crop : Crop/w and h [, x, y]/
(w and h: After the processing picture size/x, y: Offset)
- scale : Resize/w and h/(w and h: After the processing picture size)
- scope : Margin addition/w and h [, x, y] [, r, g and b]/
(w and h: After the processing picture size/x, y: Offset/r, g and b: A little more than white)
- yuv : RGB->YUV conversion system/{CCIR601 | CCIR709 | etc} (Default: CCIR601)
- idr_rate : Period of IDR picture/1…32767 (default: 60)
- ipic_rate : Period of I picture/1… 1023 (default: 60)
- bpic_rate : Period of B picture/0… 2 (default: 2)
- sps_rate : The period of SPS /0… 1023 (default: 15)
- pps_rate : The period of PPS /0… 1023 (default: 1)
- h264_profile: H.264 profile/{Baseline | Main | High}/(Default: High)
- h264_level : H.264 level/{1.1… 4.1} (default: 4.1)
- h264_entropy: H.264 entropy coded mode/(default: CABAC)
{CABAC | CAVLC}
- h264_ratectrl: H.264 rate control mode/{VBR | CBR} (Default: VBR)
- h264_aspect: H.264 aspect ratio/{0… 3} (default: 0)
(0=AUTO and 2=4: 3, 3=16: 9)
- h264_framescan: H.264 frame scan mode/(default: progressive)
{progressive | bottom | top}

=============================================================================== ■6.　The question and answer to be good a certain question and answer may be good, is shown below.

　-----
　(Q.01.) error appears in the execution time, but it is
　Perhaps (A.01.) installation is not done just.
　　candi.dll “load_code (CANDI_MODE_STAT) failed.”When it is indicated,
　　Setup of the necessary file does not do.
　　Please refer to the installation method of this product.
　　There is information that in the folder where Japanese enters it does not move well.

　-----
　(Q.02.) non compression is input what with something. The [ro] where is not useful and is!
　(A.02.) encoding quality how you prepare the clean material, is important.
　　The customer of our company is that the one which quality quite is attached importance is many.
　　Mistaking, in order not to prepare the outrageous material, it designates non compression as prerequisite.
　　While paying attention to merit value, please challenge the one which is not made the air, to there.

　-----
　(Q.03.) - yuv=raw420 [tsu] [te] what?
　(A.03.) it is subcommand of the �� [me] which inputs the YUV material.
　　Because input has designated the RGB material as prerequisite, usually the conversion of RGB->YUV is done.
　　Because this processing is the processing whose load is very high, inevitably operation becomes slow.
　　From first when you can prepare the YUV material, when - yuv=raw420 it is appointed, conversion processing dynamic kana
　　Because to be, you can encode fast. FourCC YV12 or IYUV can be input.

　-----
　(Q.04.) - scale it is dirty?
　(A.04.) it is. Like the is.
　　Being the simple pixel even, it has gone.
　　It is to be a tool which becomes the base of this product, but you say that [kanpake] is input,
　　It developed with policy. Because of that, as for filter processing when, [iza], help
　　It is the positioning, [ke] function.
　　At this point in time improvement is planned, but you cannot work directly. Even fortunately AviSynth
　　Because it is possible, to receive output, if you can exceed with there, you think.

　-----
　(Q.05.) “it cannot process with the filter which presently is used to end,”??
　(A.05.) perhaps, when using DirectShow, it is fate.
　　The number of frames of the material the taking �� everything to do this product beforehand, be able to process number of frames
　　It is the specification which continues encoding to the [ru].
　　But whether or not you can acquire the number of correct frames, it depends on the efficiency of the filter.
　　For example “the number of frames is 100”, that however it was said, really only 97 it is
　　When it is to be. In this case, it waits for the fact that remaining 3 is read, but it is
　　When time elapsed, “truth it was [oshimai] with this, don't you think? it is”, that with to judge
　　It is.
　　Actually it has finished encoding all frames supposing, that, similar to the normal time
　　Have tried to advance processing. Perhaps there is no problem in the data which it is possible.
　　As for error as for the reason where it is indicatory, “without being the operation which this product expects” the user
　　Is in order to inform.

　-----
　(Q.06.) “Q.01 the error which is different comes out”, perhaps (A.06.) encoding parameter outside the range.
　　The fact that I am caught well is, when resolution it is smaller than 320x240.
　　In regard to the parameter perhaps it reaches the point where in the future you can use, intentionally error check
　　You exclude there is a ��.

=============================================================================== ■7.　Renewal past record 2008-12-26 Ver.1.00.04
(a) You reset on the basis of the default value of encoding parameter.
(b) - h264_framescan was released.
When bottom and top are used, unless - yuv=ccir601int, - yuv=ccir709int is used
Field blends.
(c) At the time of RGB->YUV conversion, the problem where color slips depending upon resolution was corrected.
(d) - In case of h264_aspect=1, 1: Because 1 was not, it concealed. Just 0,2,3 release.
(e) Upgrading SRK. The expectation which reaches the point where it moves even with Qosmio.

2008-12-12 Ver.1.00.03 [human pillar edition]
(a) - Misprint of help indication was corrected. (Misprint in document)
- vpr - > - PR
(c) You reset on the basis of RGB->YUV conversion. Because (there is a bug in the processing with Ver.1.00.02,)
(d) - When out is not appointed, output file name becomes same as input file name
Bug was corrected.
(e) Default value of encoding parameter continuing resolution. (- As for the indication of help while they are former times)
- idr_rate: 60 - > 150
- ipic_rate: 60 - > 15
- bpic_rate: 2 - > 1
- bitrate: 120000,1440000 - > 6000000,7200000
(f) In the midst of SRK interchangeable test.
(g) In the midst of plural board corresponding tests. Unless (there is MHX-C1, perhaps useless)
The plural boards 挿 when the [tsu] [te] it is, equal to board quantity amount can start this product simultaneously.

2008-12-05 Ver.1.00.02
(a) - h264_aspect was added.
Referring to resolution, analyzing in that appearance, it sets default.
VGA = 4:3 and TV-SD = 4:3, TV-HD = 16:9
(b) - h264_ratectrl was added.
(c) - help indication was rearranged.
(d) Indication of bit rate was designated as the two of avr and max. Because (min is 0 fixing, you turned off)
(e) It accelerated RGB->YUV conversion. The expectation which becomes fast according to the number of cores. Precision concern.
(f) The fact that of the stream the rubbish has adhered lastly was corrected.
(g) Reading timeout occurring, without making error ending it tried to continue processing.
Timeout occurring as a condition, are acquired from DirectShow the number of frames which fact
You suppose when it is more. Therefore, when it occurs, you call error
From the margin normality which is rubbed (you could acquire all frames which exist) with the judgment which is said.
Once because it is problem, making similar message to the user until recently, it notifies.

2008-11-26 Ver.1.00.00
New compilation

=============================================================================== ■8.　Development delusion (2008-12-26 presently) the tool which specializes in console edition probably will be made. First TS correspondence goal.

2008-12-下 ~ 2008-01-中
(a) AAC encoding correspondence
As for encoding library compilation being completed.
(c) MP4 and M2TS correspondence
As for encoding library compilation being completed.
(d) MPEG-2 Video correspondence
Perhaps, there is no problem, probably will be.
(e) MPEG-2 TS correspondence
Unless holiday is used, how time does not take. Holiday going to work prohibition is painful.
You start from the place where you inspect whether the sound gap is what kind of feeling.
(f) Try to move even with Qosmio
You think that it could correspond with Ver.1.00.04.

=============================================================================== ■9.　While investigation examination (2008-12-26 presently) looking at the bulletin board, you are troubled, in. Memorandum.

(a) Plug-in of AVIUTL
-> Image not to have grasped completely.
(b) MPEG-1 input
Experimental �� Isao who uses →Sofdec in the decoder.
(c) MPEG-2 input (TS it includes)
-> With only playback experiment there is no development. There is no time.
(d) It corresponds to the MPEG-2 encoding function of SpursEngine
-> Encoding was possible. Not to increase subcommand, as for the [te]….
(d) The frame of AAC is cut down, (re- withoutencoding same period we would like to do to adjust)
-> Not yet commencement.
(e) Picture quality improvement (as for technique while variety thinking)
-> As for experiment the 1 success.
(f) Japanese pass?
(g) Interlace retention
The air which is produced with →-h264_framescan does.
As for personally owned color conversion routine interlace correspondence being completed. Intention.
(h) AC3 correspondence
->…

Worked out alright :D

plonk420

19th January 2009, 04:37

Only the GPU (RSX) remains locked except for basic framebuffer access and mode switching (although there have been some successful hacks to gain more control).

it's only locked if you installed the upgrade that gives WMV/DivX5 (XviD) support. 1.91 or 1.92?

CruNcher

19th January 2009, 04:53

Using systran's textbox translator after text was opened in office using correct symbols (must have Asian text packs installed), then coped from office too the systran textbox, select Japanese source - English destination.

[/CODE]

Worked out alright :D

Jep most know about CRIs Encoder though nobody can test it :P as the Hardware is not available and the Japanese that actually have the Card available don't release any Quality Comparisons or any other Tests so all this is still very sketchy. The most interesting thing i saw so far from the SpursEngine so far is the Power Consumption http://translate.google.com.br/translate?u=http%3A%2F%2Fwww.watch.impress.co.jp%2Fakiba%2Fhotline%2F20081206%2Fetc_toshibaev2.html&sl=ja&tl=en&hl=pt-BR&ie=UTF-8 ,though also that is hard to judge without testing it as the demo here also was very low described you can't just say i run the Encode and i had this Power Consumption without telling the exact complexity level of the Encode. And they for sure didn't compared vs x264 here more likely Itunes ;)

consulting

28th March 2009, 22:19

May I reanimate this thread? Sorry if not interesting.

I got two PxVC1100 cards from Leadtek in Taiwan for investigating and testing (press tv). The included software is only for end user (interVIDEO player, COREL/Ulead DVD Movie Factory with transcoder feature, VS 11.5).

I searched for engine-close tools on the CD/DVDs but didn't find anything.
Later I could download (some tricky) something from CRI that I lost by losing an USB stick (was stolen).
No idea what it was but I guess it was the freeware for adressing the SpursEngine.

We downloaded the SDK from Leadtek but there are (in advance) two softwares needed we don't have (and aren't going to acquire, so long).

Now we got mail with attached instructions how to address the card for converting files and to measure needed times ('SpursEngine Codec Reviewers’ Guide (Rev1.1)').

Requirements
Leadtek WinFast PxVC1100 - okay / installed
Codec sample application program - ???
Movies to convert（MPEG-2 format）- okay / masses

The guide tells a lot on how to execute certain commands, e.g. "C:\Program Files\TOSHIBA\SpursEngine\tools>shtool.exe".
Sounds well. But where is the shtool.exe?
Also for a "sample_app.exe" is asked.
We could not find anything like this in our machine or HDDs that we reserved for investigating this card.

We remain cool. Coz the development will go its way, and 3 or 6 months later the world is another. The main thing: The card will remain up-to-date.

So we can wait. But we also can ask.
You.
What to do.

Regards,
Manfred

tre31

29th March 2009, 01:05

CruNcher

29th March 2009, 03:31

Dear Manfred,
As tre31 already mentioned try to get it either from Leadtek or Toshiba, you could also use the CRI Coder like mentioned above :) the newest version is http://www.criware.jp/spurscoder/CriSpursCoder201_00.zip the free version has only Video support and Bitstream output but should be enough for your purpose :) of course you testing more or less CRIs Encoder for Spurs then and im not sure what Toshibas own Encoder is capable of compared to it, that most likely reviewer like you are using :)

PS: What's the System Software Encoder Configuration or Hardware you want to compare it against ?

edison

29th March 2009, 16:42

I have PxVC1100 on hand too. according my test, the video's quality encoded by it is very bad compare to x264 at this time.

poisondeathray

29th March 2009, 16:47

I have PxVC1100 on hand too. according my test, the video's quality encoded by it is very bad compare to x264 at this time.

If you have the chance, could you please post more information: ie settings, comparisons, encoding time, examples, etc. ?

Thanks for sharing your results.

edison

29th March 2009, 17:55

http://www.viddler.com/pcinlife/videos/

x264_dx_sd_fast_1p_1mbps_vbr_hp_4.1_1_audio_track (http://www.viddler.com/explore/pcinlife/videos/7/) ( megui preset DX-SD-Fast, i mod it as 1 pass ABR, and set the level as 4.1)

x264_dx_sd_balanced_1p_1mbps_vbr_hp_4.1_1_audio_track (http://www.viddler.com/explore/pcinlife/videos/6/) ( megui preset DX-SD-balanced, i mod it as 1 pass ABR, and set the level as 4.1)

tmpegenc_spursengine_vbr_1000_3000_18_18_128kbps (http://www.viddler.com/explore/pcinlife/videos/5/) (the number "18" is gop setting and this is default value in spursengine plug-in)

tmpegenc_mainconcept_vbr_1000_3000_default_128kbps (http://www.viddler.com/explore/pcinlife/videos/4/)

you can download the "original encoded" file by click the "downlaod this" button that below the player .

the bitrate setting and VBR setting of spursengine plug-in seems does not work correctly.

consulting

30th March 2009, 02:07

At first: Thanks to all.

FYI: I asked Leadtek for two transcoder test/evaluation devices and got them directly from Taiwan.
The scope should be to see what the hardware is able to do with the included software,
what we may be able to create for transcoding any applicable video file from one format to another (a little indepent tool with - maybe - a suitable GUI).Leadtek should not have any costs and the right to give the result away as freeware.
Also: We would like to get out whether or not it would be possible at least to create proxy files for editors like MAGIX - in a short time. MAGIX doesn't support proxy editing. But if you are able to create interchangeable (frame accurate) video files it works (you can get it running...).

Leadtek is knowing all this but did only offer their SDK to download. But you can use that SDK only if either purchasing a certain basic software from M$ (why should we?) or using a freeware that only works with command line inputs (what we don't like).

So what? - We pay and others have the benefits?

What we tried so long: To get VHS-, S-VHS-, Video8-, Hi8- (all converted to DV AVI) and miniDV/D8-files upscaled from their original SD format up to AVCHD with 1920x1080 in a high quality. That worked.
DVD MovieFactory uses the SpursEngine to transcode DV AVI in SD format to 7mbps MPEG-2 VBR in SD format. These MPEG-2 files you can upconvert to AVCHD using the SpursEngine. And the result is amazing! Really. (Maybe I create samples for uploading so one can download them for viewing.)

But the other issues: There should be more support than only copying and attaching a sheet from a reviewers guide doc.

Next week we'll try the CRI software.

Our goal, again: To create an independent tool that can address the transcoder Card directly. It may enable the user to convert/transcode any format to another format, possibly with using parameters for certain quality standards.

If we don't succeed we'll forget it.

I think in the summer the LoiLo Inc will present a SuperLoLoScopePRO that is able also to use the SpursEngine from Toshiba on a Leadtek transcoder card. On the other hand: NVIDIA is working on integrating the SpursEngine on graphic cards. With or without CUDA any appliaction that is able to use hardware acceleration then will be able to be supported by the engine. NVIDIA will try to convert any PC into a number cruncher by only replacing the graphic card.

Means: Our efforts are not really needed. Wait a while and especially being a video editor you'll get a new world. A fast world. NVIDIA and ATI (AMD) will make war against Intel. Coz Intel is trying to put CPU and GPU together with their own graphic. So one doesn't need any graphic card further. But you need a new CPU and a new mainboard.

Especially NVIDIA will try to turn the table. Could become their best business ever.

So, it's a long story now, but I owe you this - I think,

Regards,
Manfred

edison

6th April 2009, 15:38

my Winfast PxVC 1100 review post, here is the link:
http://www.pcinlife.com/article/graphics/2009-04-06/1239027193d770_5.html

Chinese->English:
http://translate.google.com/translate?u=http%3A%2F%2Fwww.pcinlife.com%2Farticle%2Fgraphics%2F2009-04-06%2F1239027193d770_5.html&sl=zh-CN&tl=en&hl=en&ie=UTF-8

poisondeathray

6th April 2009, 16:42

Thanks for posting your review and sharing the results, edison.

CruNcher

6th April 2009, 20:50

Nice compare edison it also shows that badaboom isn't that far from x264 (what i always said) :) also you see quite clear that X264 has one of the best RCs even ABR :)

There is some stuff i think that could be added though :)

1. Power Efficiency Compare (Comparing power output @ Encoding of all Platforms SupursEngine,CPU (x264) and GPU)
2. Tuneing x264 based on the previous results for higher speed :)
3. Adding CRI Spurscoder CLI Encoder to the compare
4. Useing Nvidias Cuda 2.2 Driver for the Badaboom Compare
5. Adding Neros GPU Encoder results when it becomes available (any day now)

edison

9th April 2009, 14:29

update:
add CRI SpursCoder 2.01 IQ results:
http://www.pcinlife.com/article/graphics/2009-04-06/1239027193d770_8.html

deekey777

18th April 2009, 15:32

Pegasys To Offer TMPGEnc 4.0 XPress Plug-In For SpursEngine™ Technology Yielding Up To 588% Faster Processing Speed For High Definition Video Encoding (http://tmpgenc.pegasys-inc.com/en/press/09_0415.html)

Opinions?

consulting

18th April 2009, 16:48

At present I'm some irritated.
Yesterday I transcoded a HDV video to an AVCHD video using the PxVC1100 card.
Time: 16:50.

Then I loaded the same HDV video with SuperLoiLoScope MARS and
"rendered" it to an AVCHD video with very high resolution (25mbps).
Time: 15:30.

I used CUDA (9500GT card).

Now I'm guessing:
For this Graphic card is to pay EUR 60 (Germany).
For the Leadtek card is to pay EUR ~170.
Have a nice weekend,
Manfred

CruNcher

18th April 2009, 18:47

@deekey777
somehow i doubt Pegasys is developing their own Encoder :P here i guess it's based on either FixStars or CRI R&D waving around with a lot of numbers but without showing quality differences isn't very interesting @ all.

@Manfred
it's more complex then this every Encoder can be tuned in different ways every option has a impact on speed, but this stuff is generally something you as a consumer doesn't see :)
in this case it really depends on the Core Encoder not on the Hardware and the research for the Spurs Encoder is still ongoing and seeing the results of CRIs Encoder it does make indeed progress there is also another company called FixStars that develops their own Spurs Encoder (CodecSys), @ the bitrates you primarily work @ the difference is marginal anyways between all of them, because the main Compression limits are going to catch here and HVS wise it's harder to see a difference especially for untrained viewers (but you can very well see heavy problems in the rate control) (@ those regions Power Efficiency becomes interesting) :).
Though i saw the price for the Leadtek card it costs 170 € as you might already know (a 4 core SPE Device) and IMHO that is still to expensive for the quality that you can reach with other solutions and without a lot more power draw @ those bitrates AVCHD you talking, imho toshiba just tries to lower their loses they driven with HD-DVD going the Spurs way the war against Sony cost them a lot especially now in the crisis they have been heavily hurt, interesting it gets if you combine all of these technologies that is really powerful in a Video Editing and Post Production environment (of course the powerdraw will be very high, but you could balance that out depending on what you need or want) :)

also a nice weekend for you as well :)

PS: Currently neither CodecSys nor CRIs Encoder reach x264 Visual Quality :) but the development is still ongoing and soon it could be :)

consulting

18th April 2009, 21:39

Thank you very much.
-------------------------------------
For getting more informations on CUDA, for initiating a useful discussion on our side I'm going to install a special CUDA forum.
It isn't official yet because I'm preparing it. And at first I must feed it with some facts.

But the is already one topic that may be of interest for you:
HERE (http://cuda.foren-city.de/topic,8,-ein-beispiel.html).
I think you'll have access without registering.

Nice weekend,
Manfred

Dark Shikari

18th April 2009, 23:02

Nice compare edison it also shows that badaboom isn't that far from x264Is this a joke?

The test basically shows that Badaboom is worse than SpursEngine, and that x264 is 70% better than SpursEngine quality-wise when using DXVA-SD-Fast. At such a vast quality gap, speed is completely meaningless.

By SSIM:

(1 - 0.7755) / (1 - 0.8682) = 1.70333839 = 70.3% better.

SpursEngine is competitive with Xvid, not x264.

dansus

28th April 2009, 20:08

Then I loaded the same HDV video with SuperLoiLoScope MARS and
"rendered" it to an AVCHD video with very high resolution (25mbps).
Time: 15:30.

I used CUDA (9500GT card).

What is SuperLoiLoScope ? I cant find any reference through google searches.

poisondeathray

28th April 2009, 20:39

What is SuperLoiLoScope ? I cant find any reference through google searches.

http://forum.doom9.org/showthread.php?t=146381
http://loilo.tv/