VideoHelp.com Video Benchmarking

Faustus · insane. Joined: 11 Apr 2002 Location: Dallas, TX

I've never had a problem with any info that Tom's has posted.

BJ_M · Patron Joined: 30 Jul 2002 Location: Canada

Faustus · insane. Joined: 11 Apr 2002 Location: Dallas, TX

BJ_M · Patron Joined: 30 Jul 2002 Location: Canada

and also toms ... anyway , there are lot of news of this -- was even discussed here ..
_________________
"Each problem that I solved became a rule which served afterwards to solve other problems." - Rene Descartes (1596-1650)

rallynavvie · Posted: Jan 16, 2005 10:04

I'm already looking for a new way to implement this test. I had hoped to get at least a couple dozen responses to it so that I could use results and drop 2 off either end (high/low) to get a better median, but that would mean at least 6 results per processor clock. The main fact of the matter is we shouldn't have such a wide variety of results being returned. Chipset, memory, hard drives, other processes: these things may only vary the result by a few seconds either way, sometimes not even that much.

I'm going to look into the checksum method and then we'll do a private submission until we get enough results to make a decent spread. Can one of the mods help me out and host a simple form on the site that submits results to me (or one of us) in an email so that we can cull from there? I'll come up with the fields we'll need on it.
_________________
NE CEDE MALIS SED CONTRA AUDENTIOR ITO

BJ_M · Patron Joined: 30 Jul 2002 Location: Canada

checksum will not work for 2 reasons -- one, divx (or other) decoder (version) used and its settings (because of preprocessing) and version of tmpgenc used -- will yield different results ...
_________________
"Each problem that I solved became a rule which served afterwards to solve other problems." - Rene Descartes (1596-1650)

thecoalman · Member Joined: 25 Feb 2004 Location: Pennsylvania

And just to throw another monkey wrench into your results.... I didn't writ them down but I OC'd my 3.0 to 3.45 with a Bus speed of 933? and still only got 1:10. That was with 22 proccesses running in the background, the very latest chipset drivers and everthing else I could find to update. I'll post full results if you want but I have since backed off the OC because my achine crashed a few times.
_________________

Nepadigital Video Articles

rallynavvie · Posted: Jan 16, 2005 10:41

Now I'm really interested in getting at some computers hands-on in a couple weeks. I just tested my old workstation and got 48 seconds out of it. The only thing in it that's faster than my main workstation is that all my SCSI hardware is in the old machine.

Anyway, the old machine is very similar to mrtristan's except that I'm running two 1900 MPs on a Tyan mainboard. CPU-Z detects the same AMD 762 chipset, though the Tyan 2460 I have is the only board that runs the 760 chipset so I think CPU-Z never programmed this one board in. I'm running Windows MCE 2005 on that machine, 1GB PC2100 ECC/registered memory (enabled). I'm reading from one 15k U320 drive and writing to another, though that should hardly explain the fast times.

So now we're in a pretty mess, aren't we?

New test will use DV AVI. I assume there is only one codec for that built into Windows? Maybe I should include a dump from AVI Codec to see which codecs are being used.
_________________
NE CEDE MALIS SED CONTRA AUDENTIOR ITO

BJ_M · Patron Joined: 30 Jul 2002 Location: Canada

not a good idea to use dv either -- for a couple of reasons - but could be used..
_________________
"Each problem that I solved became a rule which served afterwards to solve other problems." - Rene Descartes (1596-1650)

vhelp · ..Semi Neutral.. Joined: 31 Mar 2001 Location: New York

@ rallynavvie

On the subject of PM 'ing accounts for numbers ...

I would do this w/ the MODs, because its probably eaiser and less chances of
anyone tainting their own results. I'm assuming that all MODs are honest

This way, you have a working Gauge to sart with. But, calling on certain
users might make things biased or unfair or something personal

Anyways.

I really don't think that you should limit results to *only* those high
numbers you posted at top of page 1 (IMO) You are just throwing everyone
else out of the game. Its probably a no wonder why there are very few
reports back.

I think you should include ALL numbers High and LOW. It makes me want
to leave. You see, I want to see all the numbers so that I have a
better perspective of things. If you only post highest, what makes you
think that I want to go out and buy a whole new system, just to get
those numbers. Thats not how things work all the time in this area.
.
At the moment, I'm pretty strapped w/ what I got.. an XP 1800+ is my
highest CPU processor. I could probably go for something a little
higher, *but* if you post only those hestest number, those that can't
afford the CPU for *those* number will not be able to participate or
bother.. as it's on the boarderline of "bias". No effense. Really

.
After reading the above posts and seeing results, if there were LOW
number (in addition to HIGH ones) I would probably look into minor CPU
upgrade

Like maybe a 2400 clock vs. my 1800+ clock.
.
I think that if you want to get *more* numbers, then you gotta stop
scaring people off. The above scares even me off. Sorry, but that is
how I see things, at the moment. But, I like this tipics test and things.
It's all very interesting reading.

On the "score" of things ...

I re-ran the test on my given system. I won't post the results, because
at the moment, all I do see is "high" numbers.. (those that you only want
highest in the teens) ..but I'll say that my timeing was 1:39 FWIW.
(In my first test runs, I didn't realize that you were doing 2pass encodes)
I would throw in more numbers, but I don't know how persuaded you are, in the
specifics of BEAFy CPUs you want

Sorry, no affense.

Also, on your first page, its not clear enough about HOW and WHERE to obtain
those items for your count. I would post some sample pics of CPUz's window
and paint some big numbers of the areas you are interested in for the numbers
to use in our posting here.
Bare in mind, that there are THREE forms of obtaining numbers from CPUz's
window.. PICs; HTML; and text dump. Not everybody is using the *same*
method of obtaining the numbers.., and in consistant order.
.
I would also request them in specific order so that you have a
basic consistancy through out. And it would make things easier for us all
to see, in the things of numbers

If you want to make things easier, and more users to post their numbers,
I would ask that they post certain "data ellement" as a trial for a basic
table to display results with. The other more interesting numbers you
could use in another table or finalized graph or whatever. Asking for
too much numbers makes things too aukward and the thread pages get a bit
busy w/ a lot of junk (not that they are junk, but just too much data
dumped all over) Just ask for a few *basic* data element to lay out in
a column or rows. This would make it easier for you (and others that
have ideas in mind of their own) to extract those basic elements and
come up with a Table or semi-Graph for starters

Cheers,
-vhelp 3014

rallynavvie · Posted: Jan 16, 2005 11:58

Simply: I don't have enough results yet to make any sort of graph/chart. And now with some odd results we're getting it's difficult to decide which to keep on the chart. It certainly won't help anyone to see that one rig gets really good results if the benchmark wasn't done properly.

I'm playing with my four PCs to come up with a better benchmark. As for using the mods to keep track of things it would certainly be up to them. I'm trustworthy enough since I don't really have anything to gain or lose from it, and I would hope that I haven't imparted anything that would speak otehrwise. Besides, I was the one who volunteered to put this on so it's only fair I take care of the workload

_________________
NE CEDE MALIS SED CONTRA AUDENTIOR ITO

Cobra · Moderator Joined: 06 May 2002 Location: UK

How to create a nice, neat CPU-Z report

Firstly, download CPU-Z from here:

http://www.cpuid.com/download/cpu-z-126.zip

Launch CPU-Z. A window will open up - notice the tabs across the top:

Now, click on the tab labelled "1". Click the button marked "2" and save the document as something.htm - I did "cobra.htm". You can then upload it to VideoHelp like this:

Click on the Browse button (labelled "1") and select your .htm file. Then just type the file name in with the same URL markers in square brackets that you see in the image.

When you click "Submit" hey presto! You get something that looks like this:

cobra.htm

How's that, vhelp?

Cobra
_________________

Forum Rules :: PM me with any questions/problems/suggestions!

thecoalman · Member Joined: 25 Feb 2004 Location: Pennsylvania

You can also just select all the text on the html report and paste.

BTW I think you can trust my results since they suck so bad compared to lesser machines.
_________________

Nepadigital Video Articles

vhelp · ..Semi Neutral.. Joined: 31 Mar 2001 Location: New York

@ rallynavvie

Then I think its important to take everything down - all the text dump that
the user includes in their resonse

..weather pics or text. Just so long
as they put them up. I think that everything needs to be kept then (since
you are saying (ISMW) that there seems to be a missing "key element" here
that is just not yet found.
.
I have an idea to narrow that down, but would need more data posted here too.
.
I think that posting only minor elements is ok if only for demo purposes,
but for real BEAF of inner workings and unlocking potential, *all* data
elements need to be inlucded in everyones posts (feedback)
It's better to have too much, then too little.

I'm all for DV. Seems like a lot of us are using it now-a-days. And TMPG
handles Type 2 w/out any issues. It's the codec that we need to worry about.
And the key to proper encodign w/ DV is to *NOT* select any options in TMPG's
Environment tab. The three key fields should be left as is, in this way:

[x] Enable reading referrd format..
[ ] Interpolate YUV data..
[ ] set equation for color space [ ... ]

The above will pass the DV AVI file for encoding to the actual DV CODEC
to handle the decompression.

If you use either fourCC strings "dvsd" or "cdvc" they will both be
handled properly by the codec responsible (pending which one(s) is installed
and turned on) (note, there is an app (name escapes me) that can change
by turning on/off multiples of the same codec (ie, DV's) and you can then
be certain that a given DV codec is being used throughout the encoding
phase - paying attention to the above param settings

So, I would keep posting results. But, (IMO) not limit to slower numbers,
rather keeping all, because even those users who are stuck w/ a certain
CPU (ie, myself and my XP 1800+ cpu) could still benefit from this Table
because we could see an area or two that could be tweaked.. take for
instance, in Heywould3's feedback. His system number are almost
exactly as mine, but his numbers are different from mine. By viewing his
elements (computed numbers) I would make fine-tunes to my setup and quite
possibly gain the same number (or higher) as a result. But remember, I
said if some of us (and I"m sure there are many of us) can't afford to
upgrade at this time, then this Table could become invaluable in our
current stage of setup. But we could perform a little tweak here and there
and gain a little bit for the time being. That's my direction.

-vhelp 3015

vhelp · ..Semi Neutral.. Joined: 31 Mar 2001 Location: New York

@ Cobra

Regarding your wonderful report howto ...

If that is based on the *complete* set of data elements that each user
includes, then that's a great output to use (submission'wise)

But, there seems to be missing something here. That's one report. How's
it gonne look when, say 5 people post reports ??

Can you go further in this ??
Thanks. (oh, but the report looks great) Cheers

-vhelp 3016

Cobra · Moderator Joined: 06 May 2002 Location: UK

That reports everything to do with the current machine. It will be a standard layout for every user since it is generated by the one application.

I'm not sure how all the reports could be taken together but I'm sure rallynavvie can take the critical details from it and integrate it with the time the machine achieved. I don't know.

I just hope people use that HTML report option so we can get standardised machine specifications in!
_________________

Forum Rules :: PM me with any questions/problems/suggestions!

vhelp · ..Semi Neutral.. Joined: 31 Mar 2001 Location: New York

So, how would we users see others' reports when they respond by using
this method ??
.
I mean, is there a webpage or subpage on vcdhelp, where all these reports
get placed, that we might be able to view them and time ??

Thanks,
-vhelp 3017

Cobra · Moderator Joined: 06 May 2002 Location: UK

As I say, ralllynavvie would have to continually edit his post to provide:

- User
- Machine specification
- Time achieved

...or alternatively he could put together a document in Excel or something where results could be placed (again, he'd need to continually update this).

If anyone knows how to get a website to "read" the CPU-Z report and bring data submissions together then that would be the best thing - users could add their scores straight away.
_________________

Forum Rules :: PM me with any questions/problems/suggestions!

vhelp · ..Semi Neutral.. Joined: 31 Mar 2001 Location: New York

I was doing a little testing and just looking at the HTML, and did a
coupld of copy and pasts. When one copies an area of the report, the
final output that gets pasted

Its just a matter of the app that "imports" this data to filter and
line up the data elements to respective fields. And from there, can
make an Access database and import the Excel data, and build from their,
the final conclusions. Or, can just cook up an offline parser w/ a
stand-alone app, and then expert/import to database, etc. well, you all
get the picture.

This would be nice for everybody cause they could see everything and
use whatever values they find useful in their setup. And, if more
users contribute, then the more range of motion (options) everyone might
have in system tune-ups

.
I thought this idea about cataloging Mobo/CPU/Chipset from the forum
users were suppose to take off some years ago, (2001 maybe) I think.
I had this same idea then, only still now..hehe. I thought it was
finally gonna happen. Oh well, anyways.

Thanks again,
-vhelp 3018

rallynavvie · Posted: Jan 16, 2005 14:47

vhelp · ..Semi Neutral.. Joined: 31 Mar 2001 Location: New York

@ rallynavvie

Sorry for not responding sooner. I took a break, cause I was exhausted from a
project I was working on (in the background) Anyways.

Below is all relevant info.