It's not just the process, but also the cores. Gracemont is Skylake class IPC, not power. And there's 8 of them.The i5 6400 is a 65W CPU, no way that perf/watt has been increased by a 4 factor from 14nm to 10nm
It's not just the process, but also the cores. Gracemont is Skylake class IPC, not power. And there's 8 of them.The i5 6400 is a 65W CPU, no way that perf/watt has been increased by a 4 factor from 14nm to 10nm
Right that with 8C perf/watt could be increased 4x comparatively to a 14nm based 4C/4T SKL but there s 4 cores in both the N100 and N95, 8C is for the N300 and N305.It's not just the process, but also the cores. Gracemont is Skylake class IPC, not power. And there's 8 of them.
Ah yeah, forgot about that weird split with the full die. Even so, GRT is better than just a shrunk Skylake.Right that with 8C perf/watt could be increased 4x comparatively to a 14nm based 4C/4T SKL but there s 4 cores in both the N100 and N95, 8C is for the N300 and N305.
Processor Comparison - Head 2 Head
Intel Processor N100 vs Intel Processor N95 - Benchmarks, Tests and Comparisonswww.notebookcheck.net
Ah yeah, forgot about that weird split with the full die. Even so, GRT is better than just a shrunk Skylake.
That is not what the vast majority of code looks like, however, and the empirical testing results show it. Besides, for the stated use cases, you don't need a ton of vec throughput.SKL has way more throughput if the code can make full use of the theorical max throughput without relying on SMT
I may not bother with Blue Iris on it, at least on a production scale. i may play with it though. I have a i5-6600 box provisioned for my blue iris machine.Once you receive it please let us know what your thoughts on it.
Im still thinking it will be a waking nightmare with Blue Iris and a USB Drive on USB 3.0.
As for a PFsense box, it will perform magnificently and probably have more power then my Ubiquiti UDM PRO SE even.
That is not what the vast majority of code looks like, however, and the empirical testing results show it. Besides, for the stated use cases, you don't need a ton of vec throughput.
By what logic?Your point was that a shrinked SKL wouldnt be better, wich is not the case
They're effectively even in IPC across a range of workloads, not just pure int. Have you not looked at any of the benchmarks since ADL came out?but obviously SKL is one if not two gen ahead of GRT despite Intel claiming that it is on par in integer code
Dunno, seems that you dont rememeber your own sayings..By what logic?
They're effectively even in IPC across a range of workloads, not just pure int.
Have you not looked at any of the benchmarks since ADL came out?
That has what to do with the IPC claims you're making here?Dunno, seems that you dont rememeber your own sayings..
So you've been making claims without bothering to do any research on their accuracy? Here're some starting points:Precisely Intel stated that they are equal in Integer, without saying the same for FP..
The data exists regardless of you being too lazy to look it up.Saying better, on par or worse without tests mean nothing