Here, here! So many problems are directly or indirectly related to hardware capability. I think this continues to be an extraordinary resource, and am very happy it is up to CS5.
I am on an underpowered system, and not part of the CS5/hot system landscape. But I have vicariously appreciated the input of many of you (Bill, Harm, Scott et al, adobe folk) on hardware issues.
I would declare this "Bill and Harm Day" in the lounge, but the last time I did, it got all carried away on favorite food and wine and snakes.
I have run the benchmark three times on two different systems.
On Steps 3&5, I was never prompted by the Statistics.vbs script to
provide the absolute path to the project file... the first question posed
was "Personal or Computer ID".
I presume the absolute path is required to calculate the values for:
;MPE off result
In my "Output-MPE" and "Output" text files these values are missing.
Have I done something wrong?
There is no requirement to enter the path, since all files are assumed to be in the same ..\\PPBM5\ directory as Statistics.vbs
MPE result and MPE gain are left blank, because those values are only calculated automatically when entered into the spreadsheet. They are just placeholders to make our work on data entry a bit easier. You will see the same with Performance Index.
We have automated data entry to a large extent, including a lot or error checking and this way makes it a lot easier to enter the correct data in the proper fields.
You have not done anything wrong. We merge both files into one, then with a couple of simple keypresses we:
Fetch the data from the Output file, add the New results for inspection, add Test results to the spreadsheet, reSort the results and Update the pivot table.
CTRL + F, N, T, S, and U makes our life a bit easier. Then publish and we're done.
I was confused by the parenthetic hint in Steps 3&5:
"Use Copy address in Explorer before Double clicking the
.VBS file and on the first two questions use Paste"
The first two questions are Personal or Computer ID
and Computer Manufacturer.
Joe thanks for this tip. We missed it and I have corrected it now. It will be included shortly in the ZIP file.
Let me give you two examples of things that need to be checked or corrected:
Someone enters "5.0.1" for the CS5 version, another one enters "5.01". In Excel the first is a string, the second a number, so these have to be corrected.
Someone enters "gtx 480" for the video card, another one enters "GTX-480" and a third one enters "GTX 480". Three different spellings, where the same card was intended. This has to be corrected, or the MPE Performance chart is all messed up.
These kind of entry errors have to be caught at data entry, or the results go down the drain. We wanted to have that phase automated to the largest degree possible, and no doubt we will still run into slight problems now and then, but overall, we think we have made a major step forward, thanks to your remarks as well.
Hi Harm, just wondering if you have any idea why more RAM in the machine is affecting the MPEG-DVD (CPU test) results in PPBM 5. I'm the guy who sent Bill my results with a score in that section of 39 sec.
I sent you a PM. It will explain our confusion and I'm waiting for the answer to my questions. As soon as I know, I'll let you know.
Discretion is the better part of valor.
Just have to thank you guys(Bill Gehrke,Harm Millaard and a good number of beta testers) for this resourse , it is so invauable and a help to a person that wants to biuld a system PPBM4 and 5 it could be refered to" As the Bible of HW for CS4/5 ", If you had to go with Adobes info you would be at a loss.
Interesting the slow down with H.264 and dual processors.
I wonder if the same apllies to Windows Media Encoding.
Could anyone test
lets be clear with the dual Xeons.
GHZ vs GHZ a dual Xeon will always outperform a single.
its a tad hard to compare a 2.8GHz Xeon system
with a 3.33GHz or 4GHz OCed 6 core.
we have 2 x 3.33GHz dual Xeons being built this week so we should find time to benchmarks one of them.
if not Intel just sent us a pair for testing anyway.(we were whining about not having them for benchmarks)
if i could just get Eric to actually do some work :-)
I have some work laid out for him if he has a couple of 4 G sticks around...
We should have some Harm. If not I have a Dual Xeon 5680 system with 48GB or 96GB of ram coming down the pipe soon I can run what testing you want on.