Re: Does SSE2 usage still need to be conditional?

On Mon, Feb 1, 2016 at 10:04 PM, Benjamin Smedberg wrote: > It might be worth reconsidering now: I'm not willing to throw away 0.5% of > our users without good cause, but perhaps there is a good cause to be made > here? What would the performance gain be for the remaining 99.5% of users, > realizi

Re: Does SSE2 usage still need to be conditional?

On Fri, Feb 5, 2016 at 10:31 AM, Ralph Giles wrote: > Would it result in drama if it's just a bug with an easy fix? Distros' > support policies are different from ours as their upstream. Perhaps we > should just try it without the `-C target-feature=-sse2` for linux32, > but pass it for win32? T

Re: Does SSE2 usage still need to be conditional?

On Thu, Feb 4, 2016 at 11:45 PM, Henri Sivonen wrote: >> We're actually building it for 32-bit MacOS X too, but all x86 macs have >> SSE2. > > Is there a plan for adding Android builds and 32-bit Windows and Linux builds? We plan to do so. Nathan Froyd has been working on both of those. Win32 i

Re: Does SSE2 usage still need to be conditional?

On Wed, Feb 3, 2016 at 4:58 PM, Chris H-C wrote: > So I reran the analysis, this time breaking down by OS the users who we > can't say for certain have SSE2: > https://gist.github.com/chutten/e4ccd0d5a46b782bae53 > > This was on a 25% sample of users reporting in from release Firefox on Jan > 21.

Re: Does SSE2 usage still need to be conditional?

On Wed, Feb 03, 2016 at 10:41:03AM -0800, Bobby Holley wrote: > On Wed, Feb 3, 2016 at 10:32 AM, Ehsan Akhgari > wrote: > > > On 2016-02-03 12:50 PM, Martin Thomson wrote: > > > >> On Thu, Feb 4, 2016 at 2:21 AM, Milan Sreckovic > >> wrote: > >> > >>> 99.77% of the users on all channels have SSE

Re: Does SSE2 usage still need to be conditional?

On 2016-02-03 1:41 PM, Bobby Holley wrote: On Wed, Feb 3, 2016 at 10:32 AM, Ehsan Akhgari mailto:ehsan.akhg...@gmail.com>> wrote: On 2016-02-03 12:50 PM, Martin Thomson wrote: On Thu, Feb 4, 2016 at 2:21 AM, Milan Sreckovic mailto:msrecko...@mozilla.com>> wrote:

Re: Does SSE2 usage still need to be conditional?

On Wed, Feb 3, 2016 at 10:32 AM, Ehsan Akhgari wrote: > On 2016-02-03 12:50 PM, Martin Thomson wrote: > >> On Thu, Feb 4, 2016 at 2:21 AM, Milan Sreckovic >> wrote: >> >>> 99.77% of the users on all channels have SSE2 support; >>> 51.7% of all users are on 32-bit Windows; >>> 0.44% of all users

Re: Does SSE2 usage still need to be conditional?

On 2016-02-03 12:50 PM, Martin Thomson wrote: On Thu, Feb 4, 2016 at 2:21 AM, Milan Sreckovic wrote: 99.77% of the users on all channels have SSE2 support; 51.7% of all users are on 32-bit Windows; 0.44% of all users on 32-bit Windows do not have SSE2 support. Those numbers wouldn't justify a

Re: Does SSE2 usage still need to be conditional?

On Thu, Feb 4, 2016 at 2:21 AM, Milan Sreckovic wrote: > 99.77% of the users on all channels have SSE2 support; > 51.7% of all users are on 32-bit Windows; > 0.44% of all users on 32-bit Windows do not have SSE2 support. Those numbers wouldn't justify a change to me. When we make decisions about

Re: Does SSE2 usage still need to be conditional?

On Wed, Feb 3, 2016 at 5:23 AM, Henri Sivonen wrote: > My understanding is that the new > MP4 demuxer that's written in Rust is currently only being built in > the x86_64 case, so, AFAICT (I'd love to be wrong!), Rust code in > 32-bit Firefox isn't a solved problem yet. We're actually building i

Re: Does SSE2 usage still need to be conditional?

Right - I mentioned the number earlier, but let me summarize: 99.77% of the users on all channels have SSE2 support; 51.7% of all users are on 32-bit Windows; 0.44% of all users on 32-bit Windows do not have SSE2 support. — - Milan > On Feb 2, 2016, at 23:06 , Martin Thomson wrote: > > On We

Re: Does SSE2 usage still need to be conditional?

So I reran the analysis, this time breaking down by OS the users who we can't say for certain have SSE2: https://gist.github.com/chutten/e4ccd0d5a46b782bae53 This was on a 25% sample of users reporting in from release Firefox on Jan 21. The tl;dr is that it's mostly WinXP. So much so that it's al

Re: Does SSE2 usage still need to be conditional?

On 2016-02-03 8:23 AM, Henri Sivonen wrote: For Rust code that doesn't explicitly try to use SSE2, are we going to use the default official rustc which emits SSE2-requiring code and, therefore, make Firefox require SSE2 on 32-bit x86? Or are we going to use rustc in a non-default configuration so

Re: Does SSE2 usage still need to be conditional?

On 2016-02-03 14:23, Henri Sivonen wrote: As for the consequences of requiring SSE2 unconditionally, I'm personally more worried about a conflict with Linux distros that don't already require SSE2 (even if near 100% of their users actually had SSE2-enabled hardware; this concern is not about the

Re: Does SSE2 usage still need to be conditional?

On Fri, Jan 29, 2016 at 10:29 PM, Chris H-C wrote: > tl;dr - Around 99.5% of Firefox Desktop clients on release channel > represented by (a 20% sample of) pings submitted by on January 21, 2016 had > "hasSSE2" detected. > > Here's the analysis and results on github. Please feel free to check my >

Re: Does SSE2 usage still need to be conditional?

On Wed, Feb 3, 2016 at 12:54 AM, Benjamin Smedberg wrote: >> Do we have any, say telemetry, that would move this from speculation >> into numbers? Sure, numbers aren't necessarily perfect, but I'm sure >> that they would help. > > Milan is quoting numbers from telemetry data. The last time I calc

Re: Does SSE2 usage still need to be conditional?

Right, what Benjamin said - the “if I’m reading the data correctly” meant I was looking at the telemetry results from the same link that I included in one of the earlier replies: http://people.mozilla.org/~danderson/moz-gfx-telemetry/www/#view=system 51.9% 32-bit Firefox on 32-bit Windows 44.7%

Re: Does SSE2 usage still need to be conditional?

On 2/1/2016 7:35 PM, Martin Thomson wrote: On Tue, Feb 2, 2016 at 10:42 AM, Milan Sreckovic wrote: Surprisingly, perhaps, there are a lot of people using Firefox on 32-bit Windows. If I’m reading the data correctly, more than half. A small percentage of those don’t have SSE2. Do we have

Re: Does SSE2 usage still need to be conditional?

On Tue, Feb 2, 2016 at 11:31 AM, Chris Peterson wrote: > On 2/1/16 3:56 PM, Mike Hommey wrote: >> 64-bits Firefox was only officially released recently, and AFAIK, we're not >> offering 32-bits Firefox users an upgrade to 64-bits Firefox if their >> system permits. How about we started doing that?

Re: Does SSE2 usage still need to be conditional?

On Tue, Feb 2, 2016 at 10:42 AM, Milan Sreckovic wrote: > Surprisingly, perhaps, there are a lot of people using Firefox on 32-bit > Windows. If I’m reading the data correctly, more than half. A small > percentage of those don’t have SSE2. Do we have any, say telemetry, that would move this f

Re: Does SSE2 usage still need to be conditional?

On 2/1/16 3:56 PM, Mike Hommey wrote: > 64-bits Firefox was only officially released recently, and AFAIK, we're not > offering 32-bits Firefox users an upgrade to 64-bits Firefox if their > system permits. How about we started doing that? There are two steps planned to bring 64-bit Firefox to norm

Re: Does SSE2 usage still need to be conditional?

On Tue, Feb 2, 2016 at 11:09 AM, Nicholas Nethercote wrote: > On Tue, Feb 2, 2016 at 10:55 AM, Xidorn Quan wrote: >>> >>> Surprisingly, perhaps, there are a lot of people using Firefox on 32-bit >>> Windows. If I’m reading the data correctly, more than half. A small >>> percentage of those do

Re: Does SSE2 usage still need to be conditional?

On Tue, Feb 2, 2016 at 10:55 AM, Xidorn Quan wrote: >> >> Surprisingly, perhaps, there are a lot of people using Firefox on 32-bit >> Windows. If I’m reading the data correctly, more than half. A small >> percentage of those don’t have SSE2. > > It's not surprising to me. We didn't have Firefo

Re: Does SSE2 usage still need to be conditional?

On Mon, Feb 01, 2016 at 06:42:30PM -0500, Milan Sreckovic wrote: > > > On Feb 1, 2016, at 18:00 , Xidorn Quan > > wrote: ... > > > > It seems to me if we do, whether enabling SSE2 on x86 doesn't really > > matter unless we have a good reason. Fewer and fewer people would > > stick on x86, especi

Re: Does SSE2 usage still need to be conditional?

On Tue, Feb 2, 2016 at 10:42 AM, Milan Sreckovic wrote: > >> On Feb 1, 2016, at 18:00 , Xidorn Quan wrote: >> ... >> >> It seems to me if we do, whether enabling SSE2 on x86 doesn't really >> matter unless we have a good reason. Fewer and fewer people would >> stick on x86, especially who cares a

Re: Does SSE2 usage still need to be conditional?

> On Feb 1, 2016, at 18:00 , Xidorn Quan wrote: > ... > > It seems to me if we do, whether enabling SSE2 on x86 doesn't really > matter unless we have a good reason. Fewer and fewer people would > stick on x86, especially who cares about performance. Surprisingly, perhaps, there are a lot of pe

Re: Does SSE2 usage still need to be conditional?

SSE2 is always available in 64bit CPUs, it's included in the AMD64 specification. So I'd be surprised if compilers didn't automatically use SSE2 for 64bit targets. Another interesting data point is that Windows 8.1 doesn't work without SSE2 [1], but I guess most people that use Windows 8.1 ha

Re: Does SSE2 usage still need to be conditional?

I don't think there are any compilers that support x64 without SSE2. SSE2 registers are required for passing float parameters in both MS and System V ABIs. -Jeff On Mon, Feb 1, 2016 at 6:00 PM, Xidorn Quan wrote: > On Tue, Feb 2, 2016 at 7:04 AM, Benjamin Smedberg > wrote: >> >> >> On 1/29/201

Re: Does SSE2 usage still need to be conditional?

On Tue, Feb 2, 2016 at 7:04 AM, Benjamin Smedberg wrote: > > > On 1/29/2016 2:05 PM, Cameron Kaiser wrote: >> >> On 1/29/16 9:43 AM, Ashley Gullen wrote: >>> >>> FWIW, the Steam Hardware Survey says 99.99% of users have SSE2 (under >>> "other settings"): http://store.steampowered.com/hwsurvey >> >

Re: Does SSE2 usage still need to be conditional?

On 1/29/2016 2:05 PM, Cameron Kaiser wrote: On 1/29/16 9:43 AM, Ashley Gullen wrote: FWIW, the Steam Hardware Survey says 99.99% of users have SSE2 (under "other settings"): http://store.steampowered.com/hwsurvey For that to be valid, one must assume that the population of Firefox users and

Re: Does SSE2 usage still need to be conditional?

Telemetry reports 99.77% with SSE2… http://people.mozilla.org/~danderson/moz-gfx-telemetry/www/#view=system — - Milan > On Jan 29, 2016, at 15:33 , Kartikaya Gupta wrote: > > I also want to highlight the thing at the

Re: Does SSE2 usage still need to be conditional?

On 1/28/2016 12:35 PM, Henri Sivonen wrote: It's been a while since the previous SSE2 thread. I have some questions: * Does Firefox (for Windows and Linux) still run on non-SSE2 hardware? * If it does, do the usage statistics justify it continuing to do so? * Do Linux distros that ship Fir

Re: Does SSE2 usage still need to be conditional?

I also want to highlight the thing at the end of the gist linked above - the majority of the non-SSE2 population are on 43.0.4. That is, they're keeping up-to-date, and would likely be affected by this more than somebody stranded on an old version. On Fri, Jan 29, 2016 at 3:29 PM, Chris H-C wrote

Re: Does SSE2 usage still need to be conditional?

tl;dr - Around 99.5% of Firefox Desktop clients on release channel represented by (a 20% sample of) pings submitted by on January 21, 2016 had "hasSSE2" detected. Here's the analysis and results on github. Please feel free to check my work: https://gist.github.com/chutten/4959c873d7fbbec0785a Kee

Re: Does SSE2 usage still need to be conditional?

On 2016-01-29 2:05 PM, Cameron Kaiser wrote: On 1/29/16 9:43 AM, Ashley Gullen wrote: FWIW, the Steam Hardware Survey says 99.99% of users have SSE2 (under "other settings"): http://store.steampowered.com/hwsurvey For that to be valid, one must assume that the population of Firefox users and

Re: Does SSE2 usage still need to be conditional?

On 1/29/16 9:43 AM, Ashley Gullen wrote: FWIW, the Steam Hardware Survey says 99.99% of users have SSE2 (under "other settings"): http://store.steampowered.com/hwsurvey For that to be valid, one must assume that the population of Firefox users and Steam users are sufficiently similar. I don't

Re: Does SSE2 usage still need to be conditional?

FWIW, the Steam Hardware Survey says 99.99% of users have SSE2 (under "other settings"): http://store.steampowered.com/hwsurvey On 28 January 2016 at 18:35, Henri Sivonen wrote: > It's been a while since the previous SSE2 thread. > > I have some questions: > * Does Firefox (for Windows and Lin

Does SSE2 usage still need to be conditional?

It's been a while since the previous SSE2 thread. I have some questions: * Does Firefox (for Windows and Linux) still run on non-SSE2 hardware? * If it does, do the usage statistics justify it continuing to do so? * Do Linux distros that ship Firefox by default run on non-SSE2 hardware? * Do w