By: Foo_ (foo.delete@this.nomail.com), May 20, 2019 12:58 am
Room: Moderated Discussions
Adrian (a.delete@this.acm.org) on May 19, 2019 9:09 pm wrote:
>
> That is why I said that these improvements apply mainly to professional users,
> not for typical users. I am only using programs compiled specifically for the
> processors on which I run them, but few non-professional users do that.
Even among professional users, the tendency is not to use programs compiled specifically for the processor. Much professional software is off-the-shelf software that comes in precompiled binaries (even open source software you generally don't recompile yourself).
> Unfortunately like for AVX, AVX2 & FMA, Intel has chosen to not implement the BMI1, BMI2, ABM & ADX instruction
> groups on the Atom series of processors, including on the more recent Apollo Lake, Denverton and Gemini Lake,
> so many programs choose not to use these modern instructions, to support more existing computers.
And perhaps simply because they wouldn't benefit much.
>
> That is why I said that these improvements apply mainly to professional users,
> not for typical users. I am only using programs compiled specifically for the
> processors on which I run them, but few non-professional users do that.
Even among professional users, the tendency is not to use programs compiled specifically for the processor. Much professional software is off-the-shelf software that comes in precompiled binaries (even open source software you generally don't recompile yourself).
> Unfortunately like for AVX, AVX2 & FMA, Intel has chosen to not implement the BMI1, BMI2, ABM & ADX instruction
> groups on the Atom series of processors, including on the more recent Apollo Lake, Denverton and Gemini Lake,
> so many programs choose not to use these modern instructions, to support more existing computers.
And perhaps simply because they wouldn't benefit much.
| Topic | Posted By | Date |
|---|---|---|
| Intel's roadmap | Laurent | 2019/05/13 08:37 AM |
| Intel's roadmap | Alberto | 2019/05/13 09:44 AM |
| Intel's roadmap | blue | 2019/05/13 10:26 AM |
| Intel's roadmap | Maynard Handley | 2019/05/13 11:04 AM |
| Intel's roadmap | Adrian | 2019/05/13 01:15 PM |
| Actually not bad for Intel | Chester Lam | 2019/05/14 05:26 PM |
| Actually not bad for Intel | Maynard Handley | 2019/05/14 06:33 PM |
| Actually not bad for Intel | Chester Lam | 2019/05/14 08:52 PM |
| Easily, just grab the LN2... (NT) | blue | 2019/05/14 10:41 PM |
| Actually not bad for Intel | Maynard Handley | 2019/05/14 11:32 PM |
| Application matters | Chester Lam | 2019/05/15 03:15 AM |
| Intel's roadmap | Alberto | 2019/05/15 07:58 AM |
| Intel's roadmap | none | 2019/05/15 08:25 AM |
| Intel's roadmap | Chester Lam | 2019/05/15 08:32 AM |
| Sh*** is Sh** forever | Alberto | 2019/05/15 08:47 AM |
| Sh*** is Sh** forever | none | 2019/05/15 09:05 AM |
| benchmarks... | Chester Lam | 2019/05/15 09:33 AM |
| benchmarks... | none | 2019/05/15 10:09 AM |
| benchmarks... | Chester Lam | 2019/05/15 04:51 PM |
| benchmarks... | Doug S | 2019/05/16 01:10 PM |
| benchmarks... | chester lam | 2019/05/16 03:20 PM |
| benchmarks... | Doug S | 2019/05/16 03:28 PM |
| benchmarks... | chester lam | 2019/05/16 04:00 PM |
| benchmarks... | Doug S | 2019/05/17 03:39 AM |
| benchmarks... | Chester Lam | 2019/05/17 04:54 AM |
| benchmarks... | Doug S | 2019/05/17 11:52 AM |
| because per-core perf has stagnated | chester lam | 2019/05/17 01:55 PM |
| because per-core perf has stagnated | Nathan | 2019/05/17 02:54 PM |
| because per-core perf has stagnated | Adrian | 2019/05/17 10:39 PM |
| because per-core perf has stagnated | chester lam | 2019/05/19 06:32 PM |
| because per-core perf has stagnated | Adrian | 2019/05/19 09:09 PM |
| because per-core perf has stagnated | Foo_ | 2019/05/20 12:58 AM |
| because per-core perf has stagnated | Michael S | 2019/05/20 01:48 AM |
| because per-core perf has stagnated | Adrian | 2019/05/20 03:12 AM |
| because per-core perf has stagnated | Michael S | 2019/05/20 04:23 AM |
| because per-core perf has stagnated | Michael S | 2019/05/20 07:07 AM |
| because per-core perf has stagnated | Adrian | 2019/05/20 04:28 AM |
| because per-core perf has stagnated | Michael S | 2019/05/20 07:15 AM |
| because per-core perf has stagnated | none | 2019/05/20 04:41 AM |
| benchmarks... | Maynard Handley | 2019/05/16 04:35 PM |
| benchmarks... | dmcq | 2019/05/20 08:15 AM |
| benchmarks... | Maxwell | 2019/05/16 10:47 PM |
| Sh*** is Sh** forever | Anon | 2019/05/15 09:43 AM |
| Intel's roadmap | Doug S | 2019/05/13 01:24 PM |
| Intel's roadmap | wumpus | 2019/05/14 07:52 AM |
| Intel's roadmap | Alberto | 2019/05/15 08:10 AM |