SuperMalloc: A Super Fast Multithreaded Malloc for 64-bit Machines

Someone wrote: post #2

Code at https://github.com/kuszmaul/SuperMalloc . Paper seems to be there, too, at https://github.com/kuszmaul/SuperMalloc/blob/master/paper/cp...

"SuperMalloc is an implementation of malloc(3) designed for x86 Hardware Transactional Memory (HTM). It turns out that the same design also makes it fast even without HTM. "

Does anyone know if there's a cpuid feature flag for HTM? Do all haswell-generation SKUs support the feature?

Re: SuperMalloc: A Super Fast Multithreaded Malloc for 64-bit Machines

#6

Post by lotsoflumens » Mon, Dec 21, 2015, 2:38 PM UTC

DiThi wrote: post #4

lotsoflumens wrote: post #3

"dual licensed" - guaranteed to double your licensing confusion.

Can't we just use the MIT part? What's the point of having two FOSS licenses, one being non-viral?

Also, beware of submerged patent mines.

Re: SuperMalloc: A Super Fast Multithreaded Malloc for 64-bit Machines

#7

Post by listic » Mon, Dec 21, 2015, 2:40 PM UTC

wyldfire wrote: post #5

Someone wrote: post #2

Code at https://github.com/kuszmaul/SuperMalloc . Paper seems to be there, too, at https://github.com/kuszmaul/SuperMalloc/blob/master/paper/cp...

"SuperMalloc is an implementation of malloc(3) designed for x86 Hardware Transactional Memory (HTM). It turns out that the same design also makes it fast even without HTM. " Does anyone know if there's a cpuid feature flag for HTM? Do all haswell-generation SKUs support the feature?

That's Intel Transactional Synchronization Extensions, right? https://en.wikipedia.org/wiki/Transactional_Synchronization_...

Re: SuperMalloc: A Super Fast Multithreaded Malloc for 64-bit Machines

#8

Post by rayiner » Mon, Dec 21, 2015, 2:55 PM UTC

wyldfire wrote: post #5

Someone wrote: post #2

Code at https://github.com/kuszmaul/SuperMalloc . Paper seems to be there, too, at https://github.com/kuszmaul/SuperMalloc/blob/master/paper/cp...

"SuperMalloc is an implementation of malloc(3) designed for x86 Hardware Transactional Memory (HTM). It turns out that the same design also makes it fast even without HTM. " Does anyone know if there's a cpuid feature flag for HTM? Do all haswell-generation SKUs support the feature?

TSX is disabled in most (all?) Haswell SKUs due to bugs.

Re: SuperMalloc: A Super Fast Multithreaded Malloc for 64-bit Machines

#9

Post by wyldfire » Mon, Dec 21, 2015, 3:06 PM UTC

rayiner wrote: post #8

wyldfire wrote: post #5

Earlier quoted context omitted.

"SuperMalloc is an implementation of malloc(3) designed for x86 Hardware Transactional Memory (HTM). It turns out that the same design also makes it fast even without HTM. " Does anyone know if there's a cpuid feature flag for HTM? Do all haswell-generation SKUs support the feature?

TSX is disabled in most (all?) Haswell SKUs due to bugs.

Interesting. "In August 2014, Intel announced a bug in the TSX implementation on current steppings of Haswell, Haswell-E, Haswell-EP and early Broadwell CPUs, which resulted in disabling the TSX feature on affected CPUs via a microcode update." -- So only (non-early) Broadwell CPUs support it now?

Re: SuperMalloc: A Super Fast Multithreaded Malloc for 64-bit Machines

#10

Post by putterson » Mon, Dec 21, 2015, 3:15 PM UTC

wyldfire wrote: post #9

rayiner wrote: post #8

Earlier quoted context omitted.

TSX is disabled in most (all?) Haswell SKUs due to bugs.

Interesting. "In August 2014, Intel announced a bug in the TSX implementation on current steppings of Haswell, Haswell-E, Haswell-EP and early Broadwell CPUs, which resulted in disabling the TSX feature on affected CPUs via a microcode update." -- So only (non-early) Broadwell CPUs support it now?

As well as Skylake CPUs now. Check in the "TSX-NI" table entry. http://ark.intel.com/products/88191/Intel-Core-i5-6600K-Proc...