4 ms·

Sadly the T1/T2 are ancient now. The non-open SPARC is up to T7, which is 20nm process, 256 threads on 32 cores.

nickpsecurity 10y ago
Sort of. I run the bloated Web, movies, IDE's, servers, and VM's on a Core Duo with 2 cores under 2GHz done on 65nm. I'd loose some single-threaded performance probably with an OpenSPARC T2 but otherwise it should handle my modern workload. Doesn't quite feel ancient. ;) The cool thing about open-source CPU's is one can always improve on them to, say, have an extra dozen cores on more recent nodes. Like Oracle does but probably less impressive with less money.
- apaprocki 10y ago
  As a 2011 MBA user, I totally understand sticking with old tech that works. T1/T2 isn't really apples-to-apples, though. I came across this random quote from the libgmp devs that made me laugh: "SPARC chips before T4 under-perform on GMP. This is not because the GMP code is inadequately optimised for SPARC, but due to the basic v9 ISA as well as the micro-architecture of these chips. The T1 and T2 chips perform worse than any other SPARC chips; they compare to a 15 year older 486 chip." ouch!
  - nickpsecurity 10y ago
    Ouch indeed. That's pretty bad. It's believable when I think back to the purpose of these chips: handling workloads that were more I/O and concurrency bound than CPU bound. So, probably identify and eliminate these problems early on even if developer has to drop some threads or cores. Note: There's always my other recommendation of turning open-source Leon3 into a multicore. Im curious what it or Leon4 get on such a benchmark versus Intels on similar process nodes.
- pgeorgi 10y ago
  Some anecdata: Somebody tried to use a university department's T2 because it was unused and reasonably parallel for some experiment on hashing. Single core SHA1 speed on a t2 was ~300KB/s or so. Even with its 32 threads I recommended going for any ancient desktop that happened not to be used for a couple of days because the experiment would finish much faster there. Those desktops were about Core Duo class and managed 50-100MB/s SHA1 throughput (I think, it's been a couple of years).
  - nickpsecurity 10y ago
    What the hell...??? Ok, it's looking like Im going to retract that recommendation for workstations. Maybe still servers thst are I/O bound.

Sort of. I run the bloated Web, movies, IDE's, servers, and VM's on a Core Duo with 2 cores under 2GHz done on 65nm. I'd loose some single-threaded performance probably with an OpenSPARC T2 but otherwise it should handle my modern workload. Doesn't quite feel ancient. ;) The cool thing about open-source CPU's is one can always improve on them to, say, have an extra dozen cores on more recent nodes. Like Oracle does but probably less impressive with less money.

apaprocki 10y ago
As a 2011 MBA user, I totally understand sticking with old tech that works. T1/T2 isn't really apples-to-apples, though. I came across this random quote from the libgmp devs that made me laugh: "SPARC chips before T4 under-perform on GMP. This is not because the GMP code is inadequately optimised for SPARC, but due to the basic v9 ISA as well as the micro-architecture of these chips. The T1 and T2 chips perform worse than any other SPARC chips; they compare to a 15 year older 486 chip." ouch!
- nickpsecurity 10y ago
  Ouch indeed. That's pretty bad. It's believable when I think back to the purpose of these chips: handling workloads that were more I/O and concurrency bound than CPU bound. So, probably identify and eliminate these problems early on even if developer has to drop some threads or cores. Note: There's always my other recommendation of turning open-source Leon3 into a multicore. Im curious what it or Leon4 get on such a benchmark versus Intels on similar process nodes.
pgeorgi 10y ago
Some anecdata: Somebody tried to use a university department's T2 because it was unused and reasonably parallel for some experiment on hashing. Single core SHA1 speed on a t2 was ~300KB/s or so. Even with its 32 threads I recommended going for any ancient desktop that happened not to be used for a couple of days because the experiment would finish much faster there. Those desktops were about Core Duo class and managed 50-100MB/s SHA1 throughput (I think, it's been a couple of years).
- nickpsecurity 10y ago
  What the hell...??? Ok, it's looking like Im going to retract that recommendation for workstations. Maybe still servers thst are I/O bound.