[evla-sw-discuss] [Fwd: RE: Green Hills tool set]
Bill Sahr
bsahr at cv3.cv.nrao.edu
Thu Apr 18 14:17:15 EDT 2002
-------- Original Message --------
Subject: RE: Green Hills tool set
Date: Wed, 17 Apr 2002 18:27:29 -0600
From: "Jim Clure" <jclure at ghs.com>
Reply-To: <jclure at ghs.com>
To: "Bill Sahr" <bsahr at cv3.cv.nrao.edu>
Bill,
We do have TC1v1.3 support and support for the
PCP2 coprocessor.
> - Processor & chip support: Is the TriCore V1.3 architecture
> specifically supported ? Is the TC11IB specifically supported ?
Yes for TriCore v1.3.
This means the compiler will support the v1.3 as a generic core,
but the user will be responsible for writing his own link maps,
talking to peripherals, etc.
> I've also been reading the literature published by one of your
> competitors. In their literature they mention tool chain support
> for silicon defects identified by Infineon. This support takes
> the form of what they call "code bypasses" and checks on instruction
> sequences in their assembler & compilers for all identified TriCore
> CPU functional problems. Does the Green Hills support for the
> TriCore architecture include this feature, i.e., tool chain support
> for Infineon identified silicon defects ?
Yes.
We can provide a list of our workarounds if you want.
We have compiler options for turning particular workarounds
on and off, which, when used carefully, can save code size and
execution time. I don't know if our competitors can do this.
> Your web page literature seems to do a good job of addressing the
> issues of far function calls, and optimization of instruction and
> data fetches. Those same pages also indicate that your tool chain
> does include support for the PCP & PCP2 coprocessors. I do have
> one question in this area - is the macro-assembler which supports
> the PCP2 co-processor separate from or combined with the assembler
> for the TriCore processor ?
It's a separate assembler; TriCore and PCP code must be in separate
modules. They are combined at link time into a single binary.
NOTE: THE PCP AND PCP2 ARE NOT SUPPORTED BY THE DEBUGGER.
> The pages on TriCore Family support mention an instruction set
> simulator, named SimTri. Can you supply more information on
> SimTri ? It sounds as if SimTri operates independently of any
> RTOS kernel. Does this limit SimTri to simulation of single
> tasks that do not make any OS service calls ? Is SimTri limited
> to programs written in assembler or are C, C++, and EC++ programs
> also included ? Does the simulation environment also include the
> PCP2 co-processor ?
SimTri is an architectural simulator provided by Infineon and
integrated with MULTI. It simulates the TriCore architecture in
enough detail to run applications. It supports the usual GHS system
call interface. Since simtri operates on compiled binaries,
it is language-independent...so C, C++, and EC++ are all supported.
The simulator does not simulate peripherals...so the PCP2 is
not supported.
Additional Note (WJS): SimTri is meant to work with a "bareboard",
i.e., no OS, and no OS calls in the code. If we choose ThreadX
as the rtos, a simulation of ThreadX is available which can interact
with SimTri, allowing code with OS calls to be simulated. I have
no idea how well all of this works or how useful it might be to
us, especially in the absence of the ability to simulate the
network software.
> - Hosts: Our possible hosts are 1) Windows NT Workstation 4.0 SP6a,
> Windows 2000 Professional, SP2 & SP3 when it is released, Sun
> Solaris 5.8, Linux (Red Hat 7.X). Will the TriCore version of
> your tool chain run on all of these hosts ?
Yes, all these hosts are supported...though the current release does
not support Linux.
Additional Note (WJS): There is a problem with the license manager
for floating licenses if the license manager is running on a
Windows 2000 system.
> Performance Profiler - the web pages indicate that this feature is
> configuration dependent and may require custom integration. Would
> you please elaborate on this statement. Dependent upon what aspects
> of the configuration ? What sort of custom integration, and done
> by whom ?
The current TriCore release should already support our usual
profiling with simtri. I'm not sure if hardware profiling works
with jioserv, but I suspect it does not.
> Version Control System - I've briefly discussed a point or two on
> this subject with Stewart Reid of Green Hills. He indicated that
> the version control system is proprietary, but that it can integrate
> with CVS. Integrate to what extent and in what manner ? What is
> required to achieve the integration ?
CVS integration is fairly automatic in newer versions of the GHS
tools.
> Remote Target Connection - The Infineon chip includes a JTAG
> interface. I am not familiar with JTAG, but plan to become so.
> Does your tool chain include JTAG support. Is JTAG support
> specific to a board, or to a chip ? If board specific, would
> support be provided for the Infineon starter kit for the TC11IB,
> part number SKTC11IB ? If board specific, how involved would it
> be to develop support, by your tool chain, for the TC11IB daughter
> board we plan to develop ?
We support TriCore boards that use Infineon's funky JTAG interface
in which the wiggler is basically part of the board.
Additional note(WJS): For the TriCore, the only way to download
software from host to target is via parallel port to JTAG cable.
Since Sun workstations do not have parallel ports & Linux is
not supported by the current release of the Green Hills tools,
we may be limited to using Windows.
> We plan to use either Nucleus PLUS or ThreadX as the RTOS.
As we discussed, you will need to obtain ThreadX directly from
Express Logic. Once they have finished the port we can then build
the KAP (Kernel Awareness Package). Since this is not off the shelf
available for your architecture their may be a charge
for this from our Advanced Products group. I'll have to check.
> - Compiler Issues:
> - The TC11IB chip does not appear to include a floating point
> processor. Is a floating point library for the TriCore V1.3
> architecture included in your runtime libraries for assembler,
> C, C++, EC++ ?
Yes. We have full software floating point support (single and
double) for TriCore for all of the above.
Best Regards,
Jim
-----Original Message-----
From: bsahr at aoc.nrao.edu [mailto:bsahr at aoc.nrao.edu]On Behalf Of Bill
Sahr
Sent: Friday, April 12, 2002 6:47 PM
To: clurej at ghs.com
Cc: stewart.reid at ghs.com
Subject: Green Hills tool set
[snip]
- Pricing & Support: What are your licensing models/policies ? Do
you
offer both node-locked and floating licenses ? What is the cost for
each type of license ? What are your policies re support ? Is some
term of support included in the original purchase price ? If yes,
for how long ? What is the cost for additional periods of support ?
Are software upgrades included in the support ? What form does
support take, i.e. phone, www, fax, email ? Do you have any policy
concerning the maximum turnaround time for response to a support
issue ?
Additional Note (WJS):
For Windows, three types of licenses are available: dongle, node-locked,
and floating. Dongle & node-locked licenses are $5.9K per seat,
floating licenses are $7.9K/seat. For Solaris, only floating licenses
are available, at $9.9K/seat.
If ThreadX is chosen as the rtos, there is an event analyzer available
that understands ThreadX events. The event analyzer is not a part
of the toolchain. It's pricing is $2.4K/seat for a floating license,
and $1.9K/seat for a node-locked license.
- Evaluation copy: Stewart Reid mentioned that an evaluation copy of
the software is available. We might be interested in receiving an
evaluation copy. What needs to be done to obtain one ?
Additional Note(WJS): My question concerning an evaluation copy,
along with several other issues I raised were not addressed in the
response to my email. By phone, it was mentioned that Green
Hills can do an interactive web based demo of the tool set. All
that is needed at our end is one or more PCs with internet connections
and, preferably, speakerphones. We would be able to see, hear,
and interact with the people doing the demo as well as view the
demo.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: TriCore_Support_v0600.pdf
Type: application/pdf
Size: 102261 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evla-sw-discuss/attachments/20020418/b9ea974d/attachment.pdf>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ThreadX.pdf
Type: application/pdf
Size: 1777611 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evla-sw-discuss/attachments/20020418/b9ea974d/attachment-0001.pdf>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: compiler.pdf
Type: application/pdf
Size: 1318345 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evla-sw-discuss/attachments/20020418/b9ea974d/attachment-0002.pdf>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: EventAnalyzer_ThreadX.pdf
Type: application/pdf
Size: 188963 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evla-sw-discuss/attachments/20020418/b9ea974d/attachment-0003.pdf>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: tcp-ip.pdf
Type: application/pdf
Size: 93568 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evla-sw-discuss/attachments/20020418/b9ea974d/attachment-0004.pdf>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ftp.pdf
Type: application/pdf
Size: 123981 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evla-sw-discuss/attachments/20020418/b9ea974d/attachment-0005.pdf>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: options.pdf
Type: application/pdf
Size: 49634 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evla-sw-discuss/attachments/20020418/b9ea974d/attachment-0006.pdf>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: nichefile.pdf
Type: application/pdf
Size: 29196 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evla-sw-discuss/attachments/20020418/b9ea974d/attachment-0007.pdf>
More information about the evla-sw-discuss
mailing list