DC++ 0.880 will introduce new build names, require SSSE3

DC++ has already introduced CPU opcode feature requirements in the past when those features were widespread enough in various PC hardware the program had been used on. We already require SSE3 since DC++ version 0.863 and SSE2 since 0.861. They have brought siginificant advantages and optimizations to the code as you see in the linked posts in detail. We added those optimizations carefully back then, knowing all the features in case had been already widely introduced in all CPUs manufactured in the previous 10 years or more so it shouldn’t have caused issues for the vast majority of users who care to keep their system and software up to date.

The obvious step forward is to require SSSE3 to gain more advantages but there’s a caveat: while Intel introduced this feature in its client CPUs in 2006, AMD has added it only surprisingly later, in 2011. Aging of PCs are knowlingly slowed down in the recent years so requiring this feature would make the latest versions of DC++ unusable on many old but still working PCs and we’d like to avoid that. At the same time, if we decide to require at most ~10 year old CPUs now then we can go much further with optimizations as there are plenty of other new CPU opcodes are supported by the processors that has been manufactured in this timeframe.

Therefore we decided to make a difference in optimizations between the two available builds of DC++: the 64-bit build, named “Optimized” from now, will require a CPU that is manufactured in the last 10 years going forward whereas we plan to make the 32-bit build (now called “Legacy”) remain usable on pretty old (currently 15+ year old) computers and 32-bit client editions of Windows. We plan to release this build as long as there’s some form of official support of 32-bit Windows editions exist.

As a first step foward the Optimized build of DC++ 0.880 will require an SSSE3-capable 64-bit CPU, which is basically anything newer than Core2 from Intel and FX series (Bulldozer architecture) from AMD. We plan to add more feasible CPU opcode optimizations in the subsequent releases of DC++.

This also means that from version 0.880 and on, users running 64-bit Windows versions on older, non-SSSE3-capable hardware will have to use the 32-bit Legacy build of DC++, even on 64-bit CPUs.

We added an automatic, completely seamless mechanism to the installer of DC++ that decides what build is best for the users’ system and that will be offered by default in the options at install time. Those, however, who like to use the portable releases have to be cautious. We’ll update the informative readme file in the download folders for help the decision of what build to download.

We release these two builds from now as predictably there would be not much demand for 64-bit releases targeting older CPUs, even less for 32-bit ones targeting newer processors. However, if you’re in this unlikey situation for some reason then do speak up in the dev hub and your request won’t be denied.

About emtee
I started to use DC using DC++ in 2003 when its version number was around 0.260. Since then I've been amazed by the DC network: a professional but still easy-to-use way of P2P file sharing. I was invited to the DC++ team in 2006 where - in the beginning - I had been doing user support and some testing only. A few years later I started to add small contributions to the DC++ code as well so for many years I'd been doing mostly bug fixes, testing, feature proposals and improvements. At the same time I worked on improving the documentation for both DC++ and ADCH++ as well. These days I'm trying to maintain the whole code and the infrastructure behind to keep these software secure and usable for a prolonged time. My ultimate goal is to help making the DC network as more user friendly as possible.

Leave a comment