#8300 closed patch (wontfix)
Disable unaligned accesses on Alpha
Reported by: | SF/mellum | Owned by: | fingolfin |
---|---|---|---|
Priority: | normal | Component: | Ports |
Keywords: | Cc: | ||
Game: |
Description
On Alpha, unaligned accesses don't crash, but are
incredibly
expensive.
Ticket imported from: #847463. Ticket imported from: patches/405.
Attachments (1)
Change History (8)
by , 16 years ago
Attachment: | alpha-unaligned.patch added |
---|
comment:1 by , 16 years ago
comment:2 by , 16 years ago
Status: | new → pending |
---|
comment:3 by , 16 years ago
Status: | pending → new |
---|
comment:4 by , 16 years ago
The point is to avoid slowdowns of a factor of 100 and spamming
the syslog file with warnings. For this, the patch shortcuts
the test
which tries to check whether one can get away with unaligned
accesses.
The proper method would IMHO be to have a list of architectures
which are known not to have any problems with unaligned
accesses,
but that was turned down when I suggested it.
comment:5 by , 16 years ago
Owner: | set to |
---|---|
Resolution: | → wontfix |
Status: | new → closed |
comment:6 by , 16 years ago
A hardcoded list of architectures hardly ever is a "proper way". It
means that you get troubles whenever somebody tries to get stuff
working on an so far untested system.
The reason that autoconf is so successful and flexible is that it
checks for features, not for operating systems/processors/.... I
very much strongly disagree with your statement that "a list of
architectures" is the "proper method" to handle this. It most
definitely is *not* the proper way, isiolated.
However, *augmenting* an automatic detection by a few hard
coded cases is an option. However that shouldn't be implemented
by relying on some preprocessor flag in the C check, rather the
script should check the uname value to detect the Alpha
architecture.
comment:7 by , 13 months ago
Component: | → Ports |
---|
So what exactly is the point of this patch? What does it do, and
why do you think it should be added to ScummVM?