MozillaZine

SeaMonkey 2.46 requires CPU with SSE2 support

Discussion of general topics about Seamonkey
libertyernie
 
Posts: 114
Joined: January 20th, 2015, 12:29 pm

Post Posted February 20th, 2017, 7:14 am

Someone could probably build a release off of 2.42.x branch as a "final version" for people who don't have SSE2.
I don't have any computers that don't have SSE2 so I couldn't test if it worked or not.

frg
 
Posts: 265
Joined: December 15th, 2015, 1:20 pm

Post Posted February 20th, 2017, 11:22 am

There is a chance that an unzipped 2.48 package will work. 2.48 is done with VS2015 update 3 and not sure if the compiler flags for Windows where already changed. I have a VS2015 update 3 SeaMonkey 2.42.7 I use for occasional regression testing but it needed a bunch of backports to get rid of the most annoying errors in 2.40.

Personally I would just call it a day and get a new PC or retire the old. Unlike old cars old computers don't age well unless you are into vintage gaming. And for the reason: The compiler can just optimize the code better with newer instructions.

FRG

therube

User avatar
 
Posts: 16873
Joined: March 10th, 2004, 9:59 pm
Location: Maryland USA

Post Posted February 20th, 2017, 1:19 pm

Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US; rv:1.9.1.19) Gecko/20110420 SeaMonkey/2.0.14 Pinball CopyURL+ FetchTextURL FlashGot NoScript

frg
 
Posts: 265
Joined: December 15th, 2015, 1:20 pm

Post Posted February 20th, 2017, 1:36 pm

I know because there is a block in the installer but the unpacked SeaMonkey from the zip might work: http://ftp.mozilla.org/pub/seamonkey/ni ... .win32.zip

LordOfTheBored
 
Posts: 232
Joined: December 7th, 2005, 8:36 pm

Post Posted February 21st, 2017, 4:56 am

cuppettcj wrote:I should also point out that Windows Vista is still an officially supported OS, albeit for only a couple more months. I find it strange that many applications continued to support XP for years after support ended, but are quick to drop support from Vista even before it's support window closes.

Not that strange, really. Microsoft's OS support policies have no effect on anyone's development cycle.
Application development is targeted to goes the platforms people are using. XP had an unusually long real-world usage run, in part because people didn't want to use Vista.

therube

User avatar
 
Posts: 16873
Joined: March 10th, 2004, 9:59 pm
Location: Maryland USA

Post Posted February 21st, 2017, 9:04 am

there is a block in the installer but the unpacked SeaMonkey from the zip

Oops, should have mentioned, I don't "install", rather manually unpack the .exe installer.
And with that manually unpacked .exe installer, seamonkey.exe does not run without SSE2.

(There is no difference between the .exe installer & the .zip, except for the actual install routine. Really don't see why .zip are even built. .exe are smaller & can easily be unpacked.)
Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US; rv:1.9.1.19) Gecko/20110420 SeaMonkey/2.0.14 Pinball CopyURL+ FetchTextURL FlashGot NoScript

frg
 
Posts: 265
Joined: December 15th, 2015, 1:20 pm

Post Posted February 21st, 2017, 12:00 pm

Ok,

then SSE2 is a must.

>> Not that strange, really. Microsoft's OS support policies have no effect on anyone's development cycle.

There is a lenghty discussion about continuing Vista support in the Bug:

https://bugzilla.mozilla.org/show_bug.cgi?id=1130266

Used it a lot for testing SeaMonkey in VMs because the licenses were/are cheap. But with support even faster dropped than XP and almost non existent userbase not worth to still support it beyond ESR 52. SeaMonkey 2.49 will be ESR too and so continue to support it for another year.

cuppettcj
 
Posts: 29
Joined: June 19th, 2008, 10:41 am

Post Posted February 21st, 2017, 1:10 pm

BenoitRen wrote:Inexcusable. There's no good reason for requiring this. Instruction support should be detected, and similar instructions should be used that aren't as fast if SSE2 isn't supported.

frg wrote:And seriously its 2016 almost 2017 not 2001 and there are good reasons to do this.

BenoitRen wrote:No, there are not. Nothing will be lost by not requiring SSE2 and making it optional instead.

Also, 2001? My CPU is from 2004, as AMD was really late to the SSE2 party. And it's not like I can just pop a newer CPU into my motherboard that has SSE2 support.

frg wrote:And for the reason: The compiler can just optimize the code better with newer instructions.


Question 1:
So who's correct here? Is it not possible to compile the source in such a way as to support SSE2 instructions for those machines that can use it, but fallback to similar instructions on those machines that can't?

Question 2:
What is the last relatively stable version of SeaMonkey that still supports machines without SSE2? 4td8s mentioned "Adrian Kalla's unofficial SM 2.45 release build" as being compatible with non-SSE2 processors. Can anybody confirm that this is compatible? If so, can he/she provide a link to this build? libertyernie mentioned forking 2.42.x release branch. Why is that a better option than 2.45? If 2.45 won't work, then can somebody provide a link to last relatively stable build that will work?

Question 3:
Is there any large obstacle that would prevent a highly motivated person with a reasonable level of programming skills from learning how to compile SeaMonkey from source and compiling a version that doesn't require SSE2 instructions for all future releases? If not, then I may give this a shot in my spare time. I know of at least 3 people on this forum alone that would benefit.

libertyernie
 
Posts: 114
Joined: January 20th, 2015, 12:29 pm

Post Posted February 21st, 2017, 1:48 pm

I can't answer most of those questions very well, but the reason I suggested 2.42.x is because then it could have more current security updates - at the cost of an otherwise older code base (I imagine there are more bugs.)

cuppettcj
 
Posts: 29
Joined: June 19th, 2008, 10:41 am

Post Posted February 21st, 2017, 2:48 pm

libertyernie wrote:I can't answer most of those questions very well, but the reason I suggested 2.42.x is because then it could have more current security updates - at the cost of an otherwise older code base (I imagine there are more bugs.)


Thanks for your response, but I would think that 2.45 would have a newer code base than 2.42.x, unless there is something I'm missing. What I may be missing is that 2.45 is not actually compatible with non-SSE2 processors. I have no way to confirm because I don't have a link yet to the 2.45 build that 4td8s referred to.

therube

User avatar
 
Posts: 16873
Joined: March 10th, 2004, 9:59 pm
Location: Maryland USA

Post Posted February 21st, 2017, 2:57 pm

FF 49 is where Mozilla required SSE2.
FF 49 is akin to SeaMonkey 2.46.
So you would expect the prior version, FF 48 or SeaMonkey 2.45 to not require SSE2.

SeaMonkey jumped from 2.40 to 2.46, there were no intervening releases.
Though if you dig around the "latest" directories, I'd imagine you'd find something that says, 2.45a#, perhaps.

http://ftp.mozilla.org/pub/seamonkey/nightly/
http://ftp.mozilla.org/pub/seamonkey/ni ... mm-aurora/
http://ftp.mozilla.org/pub/seamonkey/ni ... comm-beta/
http://ftp.mozilla.org/pub/seamonkey/ni ... ral-trunk/

See what might be there.
(And then not sure which to tell you to get, but I guess the latest [most recent] that you seen in the 2.45 line.)


[IMO, when using a LOT of tabs/windows & a LOT of RAM, 1500 MB+, SeaMonkey 2.40 works MUCH better then 2.46 [or 2.48], FWIW.]
Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US; rv:1.9.1.19) Gecko/20110420 SeaMonkey/2.0.14 Pinball CopyURL+ FetchTextURL FlashGot NoScript

libertyernie
 
Posts: 114
Joined: January 20th, 2015, 12:29 pm

Post Posted February 21st, 2017, 3:19 pm

cuppettcj wrote:
libertyernie wrote:I can't answer most of those questions very well, but the reason I suggested 2.42.x is because then it could have more current security updates - at the cost of an otherwise older code base (I imagine there are more bugs.)


Thanks for your response, but I would think that 2.45 would have a newer code base than 2.42.x, unless there is something I'm missing. What I may be missing is that 2.45 is not actually compatible with non-SSE2 processors. I have no way to confirm because I don't have a link yet to the 2.45 build that 4td8s referred to.

I meant to say that 2.42 has an older code base than 2.45, but it also might have security fixes that 2.45 would not.

cuppettcj
 
Posts: 29
Joined: June 19th, 2008, 10:41 am

Post Posted February 21st, 2017, 4:11 pm

libertyernie wrote:
cuppettcj wrote:
libertyernie wrote:I can't answer most of those questions very well, but the reason I suggested 2.42.x is because then it could have more current security updates - at the cost of an otherwise older code base (I imagine there are more bugs.)


Thanks for your response, but I would think that 2.45 would have a newer code base than 2.42.x, unless there is something I'm missing. What I may be missing is that 2.45 is not actually compatible with non-SSE2 processors. I have no way to confirm because I don't have a link yet to the 2.45 build that 4td8s referred to.

I meant to say that 2.42 has an older code base than 2.45, but it also might have security fixes that 2.45 would not.


Ah, I read your previous reply again and realized that I misread what you were saying. I get it now, thanks.

LordOfTheBored
 
Posts: 232
Joined: December 7th, 2005, 8:36 pm

Post Posted February 22nd, 2017, 1:29 am

cuppettcj wrote:Question 1:
So who's correct here? Is it not possible to compile the source in such a way as to support SSE2 instructions for those machines that can use it, but fallback to similar instructions on those machines that can't?

As with many such discussion, who is right depends on what you think is most important.
It is possible to compile the program with an alternate "fallback" code path to increase compatibility, but it results in additional complexity, a larger executable, and I think it places limits on what kinds of optimizations the compiler can explore.
And in this case, the NEWEST affected processors are well over a decade old.

I am actually surprised at how much trouble the SSE2 requirement has caused. It warms my heart to see how many elderly computers are still chugging along.

Return to SeaMonkey General


Who is online

Users browsing this forum: No registered users and 1 guest