Why is Intel VTune so shamefully expensive?
c0d1f1ed
VTune 7.1 currently costs $699. If it was one tenth I would actually
buy it. It would even be for their own good to make it free, because
it can be used to optimize every application for Intel specifically.
I'm seriously considering, for this reason only, to buy an AMD chip
next time.
Bryan Parkoff
Intel's different type of CPUs? Then, AMD is probably just the same to
optimize AMD's Instructions each AMD's different type of CPUs. It does not
do any good.
If you can't afford to buy Intel's VTune Performance 7.1 unless you have
trial version, you still can be able to change the date to 7 days earlier
that it fools VTune Performance thinking that trial version is already
expired.
--
Bryan Parkoff
"c0d1f1ed" <nicolas...@hotmail.com> wrote in message
news:f76a7c17.04061...@posting.google.com...
Jim
If you have a good business case for your work, you can contact
Intel's developer relations and get on their Early Access Programme.
That includes VTune and a lot of other tools and hardware which are
very useful.
Jim
Tim Roberts
Why? Because it is a complicated tool that required a large amount of
development effort to create, and continues to need additional development
effort and tuning as each Intel procssor is released. Combine that with
the fact that the potential market is very, very, small, and you have the
recipe for an expensive product.
Also, Intel is not in the software business. They have demonstrated that
innumerable times over the years. They do not NEED to make this a best
seller.
--
- Tim Roberts, ti...@probo.com
Providenza & Boekelheide, Inc.
Betov
news:ib6lc016epihgk1r1...@4ax.com:
> nicolas...@hotmail.com (c0d1f1ed) wrote:
>>
>>I don't understand...
>>
>>VTune 7.1 currently costs $699.
>
> Why? Because it is a complicated tool that required a large amount of
> development effort to create, and continues to need additional
> development effort and tuning as each Intel procssor is released.
> Combine that with the fact that the potential market is very, very,
> small, and you have the recipe for an expensive product.
>
> Also, Intel is not in the software business. They have demonstrated
> that innumerable times over the years. They do not NEED to make this
> a best seller.
What i have read about 'VTune' is saying to me that
this is nothing but what i call a 'Symbolic Profiler'.
A "Symbolic Profiler" is nothing but a simplistic Debugger
flaging the Calls and measuring the times spent in between
each Call and Ret. Then, if i understood it correctly it
displays the various times in relative colored Bars, or
something like this (sorry i did not looked more closely,
but, i am afraid this is simply the main important
functionality of such Tools). The optional fnctionality
of following the Sources, depending on the Language
seems to me to be nothing but a selling argument and
a selling strategy, designed to make as much money as
possible form programmers unable to understand what a
Code Address is (a great feature nevertheless... that
you can get in the most advanced Debuggers...).
So, definitively,... No. It is not "a complicated tool
that required a large amount of development effort to
create". I plan to implement this feature in RosAsm,
and i would be much disapointed if this implementation
would take me more than, say, one month of regular work.
As RosAsm already integrates a state of Art Debugger
and a TreeView of the user'Source, re-using those
two, to build a "Timings-TreeView" with colored Bars
aside each Label, to show the botlenecks should be
a breath to write.
Usualy the price of programming Tools depend first
on the Market state. When there is no competitor,
the price is high. Period. ;)
This last point is also a massive argument in favour
of Assembly as a main Language: The offer is wide,
and the price is zero. :)) :)) :))
Betov.
John Miller
when optimizing hand-crafted IA-32 code, especially MMX, SSE and SSE-2. It
can identify "bottlenecks" on a per-instruction basis. It can identify if
the bottlenecks are due to such things as partial register stalls, cache
delays etc. It can simulate other processors in the IA-32 family.
And it *saves a lot of time*.
Time = money.
QED.
If it were simply a "symbolic profiler", I'd just use Visual Studio's
built-in profiler. Good luck with your month-long project. Don't expect me
to use it.
"Betov" <be...@free.fr> wrote in message
news:XnF9506749113...@212.27.42.74...
Robert Wessel
> this is nothing but what i call a 'Symbolic Profiler'.
>
It's a bit more than that, it understands performance counters and
can, in addition to the basic profiling function, generate information
like "this line of code is generating most of your cache misses",
"that line keeps generating branch mispredictions".
Betov
news:bea2590e.04061...@posting.google.com:
Thanks. I did not saw this in the pages i took a look
at. Nevertheless, even with this other feature it does
not justify a so high price. The interrest of such a
feature is for Code Level Optimizations. This is to say
significatively bound to a given Processor, and of
interrest only in exceptional cases.
Now, looking at the "selling Strategy", the other way
round, from the seller point of view, they might explain
the price by same argument:
Very few potential users >>> Very high price. Well... ;)
Betov.
< http://betov.free.fr/RosAsm.html >
Tim Roberts
>
>Now, looking at the "selling Strategy", the other way
>round, from the seller point of view, they might explain
>the price by same argument:
>
>Very few potential users >>> Very high price. Well... ;)
Which, coincidentally, is exactly what I said.