Topic: Would you fund development of an x86_64 SME Server 7.0?

[gordonr]

I would like to gauge the interest in an x86_64 version of SME Server 7.0

This release would be functionally identical to the x86 SME Server 7.0 release, but would be built for the x86_64 hardware which is now becoming common.

Doing such a release would involve significant time and effort from the development team, both for the initial release and for ongoing support of two separate releases. This work will need to be funded.

So, if you are interested in x86_64, would you fund such a release?

Thanks,

Gordon

(unsticky, for now)

[chris burnat]

Gordon,
I believe that it may be useful having interested people putting their names down rather than just voting anonymously.
chris burnat $250.00
Regards
chris

[jameswilson]

Hi i wouldnt mind funding it but i dont see the advantage of moving to 64 bit apart from keeping pace with technology. I think ultimatly say 2 yrs time a 64 bit version might be essential but what can 64 bit do that 32 bit cant in 'most' situations. I know when we moved from 16 b

to 32 bit there were performance reasons, but i personnaly am yet to need over 4 gig of ram (or 8 gig whatever the limit is)
Unless im missing something huge why would you want to support and develop 2 completly different versions as from what ive read some of the libs cannot co-exist.

I think 32 bits has a got a few years left yet. On another point how many people run sme on brand new hardware?

[gordonr]

Unless im missing something huge why would you want to support and develop 2 completly different versions as from what ive read some of the libs cannot co-exist.

64 bit processors such as the Athlon 64 are becoming common and are price competitive with the 32 bit processor/motherboard combinations. The games market is pushing these into the mainstream and I suspect 64 bit will become commonplace within a year.

These machines can run the 32 bit version, but performance is likely to be significantly better on the 64 bit version, leaving more room for growth. However there are costs, such as the library compatibility issue you mentioned. That's not likely to affect most people, but will certainly affect the development team on an ongoing basis (hence the reason for the poll).

I think 32 bits has a got a few years left yet. On another point how many people run sme on brand new hardware?

I think you'll find quite a lot of people do, especially for new installs. One of the big benefits of 7.0 is support for new hardware. I too think 32 bit has quite a few years left, but once 64 bit becomes flavour of the month  I suspect we will see a very rapid shift.

[jameswilson]

I stand corrected. I agree with the shift to 64 bit, and yes in time 64 bit will rule and 32 will be classed as legacy. (does the new core duo have 64bit extensions in it) i know amd are fondly followed and i believe in the next few months we will start seeing dual socket am2 boards (for quad core machines...mmmmm) then this will probably be more relevant. If most people use new hardware then it makes sense, but i can imagine the comments (mistaken of course) of dropping support for 32 etc etc

[gordonr]

Gordon,
I believe that it may be useful having interested people putting their names down rather than just voting anonymously.
chris burnat $250.00
Regards
chris

Chris,

Many thanks for standing up to be counted publicly and for your generous pledge. My initial thought was that we could gauge the interest with an anonymous poll as I'm hoping that people will only vote for a figure if they are prepared to come up with the money if we decide to do an x86_64 release.

If there is sufficient interest and funding and we decide to go ahead, my plan is to acknowledge all people who fund the project (unless they ask to remain anonymous).

I've set this poll to last for 28 days, so let's see where we are in a month. Please note that I have intentionally not set a figure for "sufficient funding - if you're interested, please vote.

Thanks,

Gordon

P.S. I have added a "Yes - More than $500" option to the poll. The options are now out of order, but I don't want to change the existing items.

[sonoracomm]

As for the availability of 64 bit hardware, it's probably more common than you think!

We are a reseller and OEM of computers and servers.  Not a big one, but representative I suspect.  Virtually every machine, desktop and server alike, that we have sold in the last year is 64-bit.  We sell mostly AMD-based workstations, but even the Intel-based machines are 64-bit too.  We don't even really mention it to people since we aren't really brave enough to install 64-bit Windows anyway.

I also suspect there is a lot of SME installations on new hardware.  Frankly, if you are charging for your time (or paying for someone elses time), it is almost never economic to mess around with used hardware.

I've been running 64-bit Linux on my home and office desktops for going on three years now.  Yes, it was painful at first, but it's surprisingly functional and easy now.

I think a x86_64 SME Server would be a great idea.  It might even help convince the rest of the world that the move to 64-bit is really worth it.  

Though it might be dificult to quantify (benchmark) the advantages of x86_64, if 64-bit SME just came out in a stable form, that would be fantastic.

G

[GPete]

Add another voting option:
                 Currently running SME on x86_64
?

[gordonr]

Add another voting option:
                 Currently running SME on x86_64
?

I could certainly add this, but is it a "Yes" or a "No" to funding the development of an x86_64 release? I suspect there will be both - those happy with the existing release running on x86_64 hardware and those who want a native x86_64 release.

[CharlieBrady]

These machines can run the 32 bit version, but performance is likely to be significantly better on the 64 bit version, leaving more room for growth.

I doubt that you will see any significant performance difference. The big difference with the 64bit version is its ability to handle very large processes. We try to avoid those.

We have very scarce development resources, and I don't think a 64 bit version should be near the top of out priorities. We have very many feature requests already in the bug tracker, and plenty of other improvements we could think of. I'd like to see people step forward and fund development of some of those.

[kruhm]

I vote this can be shelved for some time. There are certainly other areas of more importance that need to be developed before this.

@CharlieBrady
Maybe you could give a rough top ten list that you see as most important.

[CharlieBrady]

Maybe you could give a rough top ten list that you see as most important.

My list doesn't matter. What matters is what various people are prepared to develop themselves or pay to have developed for them.

[kruhm]

People who develop have neither:
-an avenue to put forth or get ideas as to what's needed most (other than the forums and hence this thread).
-nor an idea of what their reward would be (a pat on the back or $1000).

People who pay to have improvements developed for them have:
-no easy avenue to collectively put funds toward a particular improvement.
-no idea who to contact if they need something developed or supported.

A combination of these facts probably explains why donations are low.

[gordonr]

A combination of these facts probably explains why donations are low.

I agree, but I think we should take that to another thread. I raised this poll as a specific example of a clearly identified work item which could be funded separately. It's certainly not high on my list, but it is refreshing to see that people are willing to ante-up reasonable sums of money.

I believe we need an active marketplace where funders and developers can get together. A bounties system has been proposed in the past, but hasn't progressed. I'm also not sure that donations are ever likely to fund serious development efforts. IMO this is best handled by a more direct relationship between the developer(s) and the requestor(s).

To be clear, we are at a serious crossroads with development funding. We are fortunate that e-smith/Mitel have provided a lot of funding/developer time over the years for infrastructure development.

My small consulting company funded a lof of the work for SME Server 7.0 by my decision to allocate time to this release instead of taking other paid work. My bank balance will not allow me to continue to do so.

In order for this distribution to continue and grow, we need a solid financial base so that developers can be rewarded for their efforts and those needing additonal features have an avenue to ask for them.

Maybe we should open a "Funded development and bounties" forum and take this there?

[kruhm]

To be clear, we are at a serious crossroads with development funding...

In order for this distribution to continue and grow, we need a solid financial base so that developers can be rewarded for their efforts...

Would you contact me via email and explain the development funding (unless you want to continue in this thread). I *may* be able to help and provide corporate sponsorship but I'll need honest numbers.

In any event, the next step would be to identify revenue streams.