Martin Mocko (vid, Slovakia)
Intel Virtualization (VT-x) Tutorial
2 GB part0:
http://tokk.biz/fasmcon2009/M2U00503.MPG
550 MB part1:
http://tokk.biz/fasmcon2009/M2U00504.MPG

Madis Kalme (Madis731, Estonia)
SMP initialization and Interprocessor interrupts
1,5 GB
http://tokk.biz/fasmcon2009/M2U00506.MPG

František Gábriš (Feryno, Czechoslovakia)
Turning off Hypervisor and Resuming OS in 100 Instructions

2,0 GB part0
http://tokk.biz/fasmcon2009/M2U00507.MPG

120 MB part1
http://tokk.biz/fasmcon2009/M2U00508.MPG

I strongly suggest you to read the updated and corrected presentation available here:
http://board.flatassembler.net/topic.php?p=100097#100097

Erm, 8GB of video files (and maybe more to come?)! Ouch.

Tomasz Grysztar (Poland)
Ideas for fasm 2, a new fasm fork
0,9 GB
http://tokk.biz/fasmcon2009/M2U00510.MPG


The final 'The world biggest battle for beer' has only 300 MB.
Will be available in 1 hour.
Must be seen!!!

Ender (Poland)
Fast Fourier Transform & what really fast can we do it

1,4 GB part0 (some small part of begin is irreversible missing...)
http://tokk.biz/fasmcon2009/M2U00511.MPG

100 MB part1
http://tokk.biz/fasmcon2009/M2U00512.MPG

The world biggest battle for beer
260 MB
http://tokk.biz/fasmcon2009/M2U00513.MPG

Must be seen even in case you missed all presentations!!!

Thank you, Feryno.

As usual, I am working backwards from the rest of the world, starting with episode 13, and working my way to the front....

The beer party was good. I recognized most folks. However, I wasn't sure if f0dder was hiding behind a beard, or that was someone else....?

Looked like a lot of fun. Beer seemed cold. Water, wet. What more could one ask for on a hot summer day?

I watched part 1, i.e. part 2 for those of us inexperienced with writing FFT's in assembly, and confess, I could comprehend neither the question, nor the answer. I will try again, after watching part zero, i.e. part 1. It is now downloading, and will take about 4 hours on my slow DSL connection....Oops, time for basketball. Back in a few hours....

hope everyone is having fun touring the organs in the churches of Myjava.
Cheers,
tom

Wow a lot of files. Thanks Feryno!

One on FFT I'm gonna get that right now.

Hmm, are other people downloading the files or watching them online? They don't seem to be downloading and when I try to watch them they don't load... Edit: got it working with wget.

How come there are two parts to each; with one being big and one being small?

pal wrote:


2 parts = limitation of camera
camera uses FAT filesystem, file size limit = 2 GB
if the video is longer then the camera saves the file in 2 GB fragments

now I'll try to merge files and convert them to smaller files (but then FFT will be unreadable because of small text, it is on the edge of visibility even under current video format, other presentations use larger fonts and should be readable after impairing their quality by recompression) - my brother helps me I hope (he is better than me for that)

I just wanted for you to be able to watch videos immediatelly after the conference. If I succeede with merging files and recompression, then presentation will be available for guys with slowlier connections also.

To tom tobias - f0dder wasn't there.

Sounds good as long as you can read the text

Any chance of getting the powerpoint presentations? (Or whatever the presentations were done with if they were done via a computer.) I think that would be a good thing to run next to the video.

I just finished watching the fasm2 presentation. A few questions:

Didn't anyone ask about inline macros?
Why 128bit?
Will it have floating point arithmetic?
Can constants be tagged as floats/integer/pointers?

OK here my presentation, less than 1 MB, ppt format
http://fdbg.x86asm.net/Turning_off_hypervisor_and_resuming_OS_in_100_instructions.ppt
I suggest you to read the presentation, I corrected some mistakes there and put some additional explanations also and answered the question of vid at the end of the presentation recorded in video

use this if you don't have open office either ms office:
(only 1,9 MB size)
http://www.microsoft.com/downloads/details.aspx?FamilyId=428D5727-43AB-4F24-90B7-A94784AF71A4&displaylang=en

Just finished downloading and watching part zero of Feryno's FFT lecture. Now getting ready to start downloading Tomasz' talk, i.e. the main event.... hurrah.

Umm, a couple questions about the FFT discussion:

Thank you very much Feryno for this presentation on FFT. Two questions, both related to the title of your talk, i.e. speed:

a. With regard to the fewest arithmetic operations, i.e. multiplications and additions, I was under the (perhaps mistaken) impression that the original Duhammel and Hollmann split radix, (derived from the Cooley-Tukey algorithm) as modified by Burrus and his colleagues, including Sorensen, was the fastest transposition available, computing the DFT (O(N^2))in O(N*logN) operations. Are we then to understand your presentation yesterday as summarizing the improvements by Guo and Burrus, or have you improved upon the results described in Johnson and Frigo's paper in IEEE Transactions Signal Processing (2007)? I missed the beginning part of your talk, so I am not sure whether you were summarizing the process, or explaining a novel innovation.

b. While Burrus and Sorensen and their colleagues were working with a different cpu, made by TI, one can use their same logic with the Intel cpu. However, the floating point coprocessor is not the best resource, as I think you pointed out at FASMCON two years ago, in beautiful Brno, so, I was hoping that you may have devoted some portion of your talk at this year's conference to novel ways to implement the Burrus split radix with the newest architectural modifications on a quad core cpu.

A quad core cpu, opens up some interesting challenges with the split radix!!! So, I was hoping you were going to share with us the relative advantage of multicore integer performance on FFT, versus SS3/4 computations....

Next year....

Thanks again.

tom: FFT lecture was given by Ender, another guy from Crackow (besides Tomasz). Feryno was talking about turning off the hypervisor.

Thanks Martin,Thanks Ender, Sorry Feryno. I wish I could blame my confusion on mere senility, but, unfortunately, it has been a lifelong burden!!!!



Looks like it was a great meeting, vid, outstanding job, organizing it. Thank you very much....


cheers,
tom

William Whistler (UK), Karel Lejska (MazeGen, Czechia)
A crash course on Manual x86 Instruction Disassembly
286 MB part0:
http://tokk.biz/fasmcon2009/M2U00501.avi
16 MB part1:
http://tokk.biz/fasmcon2009/M2U00502.avi

Martin Mocko (vid, Slovakia)
Intel Virtualization (VT-x) Tutorial
286 MB part0:
http://tokk.biz/fasmcon2009/M2U00503.avi
76 MB part1:
http://tokk.biz/fasmcon2009/M2U00504.avi

Madis Kalme (Madis731, Estonia)
SMP initialization and Interprocessor interrupts
211 MB
http://tokk.biz/fasmcon2009/M2U00506.avi

František Gábriš (Feryno, Czechoslovakia)
Turning off Hypervisor and Resuming OS in 100 Instructions
286 MB part0
http://tokk.biz/fasmcon2009/M2U00507.avi
16 MB part1
http://tokk.biz/fasmcon2009/M2U00508.avi
updated and corrected presentation in ppt format:
http://fdbg.x86asm.net/Turning_off_hypervisor_and_resuming_OS_in_100_instructions.ppt

Tomasz Grysztar (Poland)
Ideas for fasm 2, a new fasm fork
122 MB
http://tokk.biz/fasmcon2009/M2U00510.avi

Łukasz Szańca (Ender, Poland)
Fast Fourier Transform & what really fast can we do it
190 MB part0 (some small part of begin is irreversible missing...)
http://tokk.biz/fasmcon2009/M2U00511.avi
15 MB part1
http://tokk.biz/fasmcon2009/M2U00512.avi

The world biggest battle for beer
36 MB
http://tokk.biz/fasmcon2009/M2U00513.avi
Must be seen even in case you missed all presentations!!!

vid did, the day before conference.

Just as 64-bit was chosen initially for fasm supporting 32-bit code generation, I think 128-bit should be right thing for the 64-bit capable one.

Undecided.
Nope, they'd all be just binary data, as with fasm 1.

Aw, now you're teasing us! What was the answer? My guess is "no", right? Indeed. I wonder how many people will think that it is for SSE and then ask why not 256 bit for AVX? Without tagging I think that the float arithmetic would be difficult to implement. With tagging then things like type conversion can be automatically done during arithmetic operations.