New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
@FredQc benched the OVH box with E3-1270v6 with geekbench. I took that as comparison to a random picked E3-1245v2 - here you go:
http://browser.geekbench.com/v4/cpu/compare/1987631?baseline=2543940
The v6 runs at slightly higher frequency than the v5, so it gets a small speedup from that. It gets almost no speedup from the newer design. Maybe a little more power efficiency or something.
I also found the E3-1270v6 to be slightly better than the I7-6700K, single core wise or multi-cores.
That is really impressive ;-)
I've posted some UnixBench tests on WHT
http://www.webhostingtalk.com/showthread.php?t=924581&p=9875350#post9875350
http://www.webhostingtalk.com/showthread.php?t=924581&p=9878987#post9878987
Here's one I ran:
https://browser.primatelabs.com/v4/cpu/2577756
I've been happy with the CPU performance so far.
Hmm, Ubuntu seems to give a performance boost over Debian with this CPU... weird.
LLVM ... 8159 vs 45942 pts - wtf is this ? Seems to be the culprit with Debian.
Completely different compiler than gcc.
Yeah, I've just read a bit about llvm.
https://www.phoronix.com/scan.php?page=article&item=ubuntu-1604-compilers&num=1
Thanks everyone!
Got the info that i needed
yup that will be key to intel processor performance especially for Intel v4, v5 and v6 processors
What's the diff? Vectorization for avx512 or something?
Comes down to specific GCC version's support for specific Intel processor instructions as not all GCC versions are created equal. Nice example of Intel v4 processor regressions to v3 on older GCC versions discussed at https://www.webhostingtalk.com/showthread.php?t=1624698
and specific post https://www.webhostingtalk.com/showthread.php?t=1624698&p=9830111#post9830111
It's why I made sure Centmin Mod Nginx and PHP-FPM installs optionally support Clang, native CentOS GCC 4.4/4.8 (CentOS 6/7) and GCC 5.3 and GCC 6.2.1 builds (and GCC 7.x to come soon) for newer Intel processors
Well what I mean is, what do the v6 processors do that earlier ones don't, that clang is able to speed up programs with? Wider avx is the one thing I can think of. I'm not sure exactly when that showed up. I don't think it will help typical programs much, and for the kind of programs where it does help, a GPU will probably help even more.
We're truly in another time- when our video devices are now more important than the processor.
v6 uArch is basically identical to v5, and ISA is identical. Only differences are higher clockspeed and improved iGPU (for CPUs that have one). As such, compilers aren't going to perform better for one but not the other.
v2 to v3 was the last "big" uArch change. v4 is very similar to v3, v5 includes some small changes (mostly higher FP throughput), and v6 is identical to v5. Overall, there's not that much of a difference from v3 to v6, and I'd expect compilers to perform similarly across these CPUs.
Skylake-EP (E5v5; currently unavailable to general public unless you have special access to select GCP servers) introduces wider AVX.
1270v6: https://benchmark.stream/2017/04/17/benchmark-e3-1270v6-32gb-ram-2x2tb-sata/
1245v2:
https://benchmark.stream/2017/04/06/benchmark-xeon-e3-1245-32gb-ram-2x240gb-ssd/
https://benchmark.stream/2017/03/03/benchmark-xeon-e3-1245/