jabbr Posted March 10, 2019 Share Posted March 10, 2019 Ok I'm going to post CPU usage as I upgrade my new server: Starting with Dell T420 with a single 6 core 2.4 Ghz E5-2440 V2 and 48 Gb ECC DDR3 and stock 1 Gbe Ethernet (Broadcom) HQPe -> NAA (Pro-Ject S2D) Ubuntu 18.10 DSD512 upsampling (%cpu as reported by top): ASDM7 poly-sinc-short-mp-2s: 350-360% poly-sinc-xtr-mp-2s: 370% poly-sinc-xtr-mp: 460% (stutters) Custom room treatments for headphone users. Link to comment
gdpr Posted March 10, 2019 Share Posted March 10, 2019 28 minutes ago, jabbr said: Ok I'm going to post CPU usage as I upgrade my new server: Starting with Dell T420 with a single 6 core 2.4 Ghz E5-2440 V2 and 48 Gb ECC DDR3 and stock 1 Gbe Ethernet (Broadcom) HQPe -> NAA (Pro-Ject S2D) Ubuntu 18.10 DSD512 upsampling (%cpu as reported by top): ASDM7 poly-sinc-short-mp-2s: 350-360% poly-sinc-xtr-mp-2s: 370% poly-sinc-xtr-mp: 460% (stutters) Could you please enligthen me on this measurement? How can a processor usage be higher than 100%? Dirk Link to comment
Jud Posted March 10, 2019 Share Posted March 10, 2019 2 minutes ago, ddetaey said: Could you please enligthen me on this measurement? How can a processor usage be higher than 100%? Dirk Divide by number of cores. One never knows, do one? - Fats Waller The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature. Link to comment
jabbr Posted March 10, 2019 Share Posted March 10, 2019 Just now, ddetaey said: Could you please enligthen me on this measurement? How can a processor usage be higher than 100%? It is a measure of virtual cpus, so 6 cores and 12 SMT. Rarely possible to get to all cores fully active given inefficiencies in software -- thats presumably why I am hearing stuttering at 450% usage (6 cores each 100% active would be 600%) Custom room treatments for headphone users. Link to comment
Miska Posted March 10, 2019 Share Posted March 10, 2019 "top" reports loads in two ways. At the top in the header part is total system load. There all cores (including virtual) count as 100%, so there 50% means all physical cores fully loaded for a HyperThreaded CPU. In the lower per-process listing part load is per single core. So for example 300% load means 3 cores (physical or virtual) fully loaded. Single-threaded process being 100% busy would list as 100%. Process with two threads totally busy 200%. Since HQPlayer adapts number of threads to number of cores, the load is capped by the hardware, not by the software, but fully loaded is number of cores x 100. Signalyst - Developer of HQPlayer Pulse & Fidelity - Software Defined Amplifiers Link to comment
Miska Posted March 10, 2019 Share Posted March 10, 2019 18 minutes ago, jabbr said: It is a measure of virtual cpus, so 6 cores and 12 SMT. Rarely possible to get to all cores fully active given inefficiencies in software -- thats presumably why I am hearing stuttering at 450% usage (6 cores each 100% active would be 600%) Here you can try to fine-tune a little. Either multi-core set to "enabled" vs "auto". When set to "enabled" it blasts all cores with all kinds of tasks, while with "auto" different cores are assigned different tasks in a balanced way. Signalyst - Developer of HQPlayer Pulse & Fidelity - Software Defined Amplifiers Link to comment
jabbr Posted March 10, 2019 Share Posted March 10, 2019 Now 2x E5-2470 v2 (20 cores) and 96 Gb ECC DDR3: poly-sinc-xtr-short-mp-2s: 360-400% poly-sinc-xtr-mp-2s: 420-450% poly-sinc-xtr-mp: 1400% (and still some stutters!) poly-sinc-xtr-mp: 900% (DSD256) and plays beautifully (OMG the web interface on HQPe makes this so vastly easier to do) Custom room treatments for headphone users. Link to comment
sig8 Posted March 11, 2019 Share Posted March 11, 2019 9 hours ago, jabbr said: Ok I'm going to post CPU usage as I upgrade my new server: Starting with Dell T420 with a single 6 core 2.4 Ghz E5-2440 V2 and 48 Gb ECC DDR3 and stock 1 Gbe Ethernet (Broadcom) HQPe -> NAA (Pro-Ject S2D) Ubuntu 18.10 DSD512 upsampling (%cpu as reported by top): ASDM7 poly-sinc-short-mp-2s: 350-360% poly-sinc-xtr-mp-2s: 370% poly-sinc-xtr-mp: 460% (stutters) Let us know your new server configuration. Link to comment
Miska Posted March 14, 2019 Share Posted March 14, 2019 Some new interesting hardware for running things like HQPlayer Embedded! I'll try to get one sometime soon! https://up-board.org/up-xtreme/ The CPU is 15W TDP, so it has plenty of passive cooling opportunity. asdf1000 1 Signalyst - Developer of HQPlayer Pulse & Fidelity - Software Defined Amplifiers Link to comment
asdf1000 Posted March 14, 2019 Share Posted March 14, 2019 1 hour ago, Miska said: Some new interesting hardware for running things like HQPlayer Embedded! I'll try to get one sometime soon! https://up-board.org/up-xtreme/ The CPU is 15W TDP, so it has plenty of passive cooling opportunity. Similar CPU of the 8th Gen NUC7i7DN** ? Similar performance to Intel Core i7-8650U? Link to comment
Miska Posted March 14, 2019 Share Posted March 14, 2019 1 hour ago, Em2016 said: Similar CPU of the 8th Gen NUC7i7DN** ? Similar performance to Intel Core i7-8650U? Somewhat newer version, this one: https://ark.intel.com/content/www/us/en/ark/products/149091/intel-core-i7-8565u-processor-8m-cache-up-to-4-60-ghz.html https://en.wikipedia.org/wiki/Whiskey_Lake_(microarchitecture) Includes some important features like AVX2 and FMA. asdf1000 1 Signalyst - Developer of HQPlayer Pulse & Fidelity - Software Defined Amplifiers Link to comment
asdf1000 Posted March 14, 2019 Share Posted March 14, 2019 21 minutes ago, Miska said: Includes some important features like AVX2 and FMA. Are these features helpful with HQP performance? Link to comment
Miska Posted March 14, 2019 Share Posted March 14, 2019 47 minutes ago, Em2016 said: Are these features helpful with HQP performance? Yes, minimum baseline is SSE4.2, then AVX, AVX2 and AVX-512 add more performance. FMA pretty much landed along with AVX. SSE4.2/AVX can compute two 64-bit floating point values per CPU core with single instruction, AVX doubles number of registers compared to SSE4.2. AVX2 extends computations to four 64-bit floating point values per instruction and thus the registers grow from 128-bit to 256-bit wide. AVX-512 extends this further to eight 64-bit floating point values per instruction and registers thus to 512-bit. FMA improves precision and speed of commonly used multiply-add operation on these registers. asdf1000 1 Signalyst - Developer of HQPlayer Pulse & Fidelity - Software Defined Amplifiers Link to comment
asdf1000 Posted March 14, 2019 Share Posted March 14, 2019 29 minutes ago, Miska said: Yes, minimum baseline is SSE4.2, then AVX, AVX2 and AVX-512 add more performance. FMA pretty much landed along with AVX. SSE4.2/AVX can compute two 64-bit floating point values per CPU core with single instruction, AVX doubles number of registers compared to SSE4.2. AVX2 extends computations to four 64-bit floating point values per instruction and thus the registers grow from 128-bit to 256-bit wide. AVX-512 extends this further to eight 64-bit floating point values per instruction and registers thus to 512-bit. FMA improves precision and speed of commonly used multiply-add operation on these registers. Oh that’s cool. My fanless NUC7I7DN works really well, even at DSD512 with -2s filters. So this should work well. I wonder if they’ll make a ‘gateway’ model, in fanless case. Link to comment
Miska Posted March 14, 2019 Share Posted March 14, 2019 40 minutes ago, Em2016 said: I wonder if they’ll make a ‘gateway’ model, in fanless case. Yes, that's what I'm hoping too. The two earlier ones have been really nice. So far in addition to those there are not that many of such on the market, FitPC and LogicSupply come to mind. Of course custom built audiophile ones, but not that many generic ones. asdf1000 1 Signalyst - Developer of HQPlayer Pulse & Fidelity - Software Defined Amplifiers Link to comment
jabbr Posted March 15, 2019 Share Posted March 15, 2019 Installed the NVIDIA RTX2080 Ti wow! DSD512 poly-sinc-xtr-mp (Dell T420, 2x Xeon E5-2470 V2, 96Gb -- 20 cores) -- stereo without CUDA: 1400% with CUDA: 300% Miska 1 Custom room treatments for headphone users. Link to comment
asdf1000 Posted March 15, 2019 Share Posted March 15, 2019 16 minutes ago, jabbr said: Installed the NVIDIA RTX2080 Ti wow! DSD512 poly-sinc-xtr-mp (Dell T420, 2x Xeon E5-2470 V2, 96Gb -- 20 cores) -- stereo without CUDA: 1400% with CUDA: 300% The thousand dollar question that few of us can answer.... (at DSD512..) how does non-2s sound versus -2s? Any different? Link to comment
Solstice380 Posted March 16, 2019 Share Posted March 16, 2019 3 hours ago, Em2016 said: The thousand dollar question that few of us can answer.... (at DSD512..) how does non-2s sound versus -2s? Any different? To me, there is a little difference. It’s subtle, but music is smoother while retaining excellent impact and quickness. A little less harshness, but my Lampi doesn’t tend toward harshness. @jabbr I upgraded to a 2080 Gaming with my i7-6700K and can do the non-2s. I’ll have to go through some of the filters AGAIN and note the CPU/GPU. I mostly use the sinc-shrt-mp and that runs around a reported 22% CPU (so 44%, right?) and 13% GPU. I’ll check xtr again. When the filter builds, however, it’s a different story. The CPU pegs at 100% for over 2 round trips of the progress bar, and the GPU is at 2%. asdf1000 1 https://audiophilestyle.com/profile/21384-solstice380/?tab=field_core_pfield_3 Link to comment
asdf1000 Posted March 16, 2019 Share Posted March 16, 2019 20 minutes ago, Solstice380 said: To me, there is a little difference. It’s subtle, but music is smoother while retaining excellent impact and quickness. A little less harshness, but my Lampi doesn’t tend toward harshness. Thanks. My fanless NUC7i7DNHE can do poly-sinc-short-lp at DSD512 with non-2s and comparing with -2s I don't hear any difference. Maybe with xtr the difference is there but sadly my little fanless NUC runs out of puff trying this filter at DSD512 non-2s. Link to comment
Solstice380 Posted March 16, 2019 Share Posted March 16, 2019 4 hours ago, jabbr said: Installed the NVIDIA RTX2080 Ti wow! DSD512 poly-sinc-xtr-mp (Dell T420, 2x Xeon E5-2470 V2, 96Gb -- 20 cores) -- stereo without CUDA: 1400% with CUDA: 300% The 2080 series seems very powerful. 44K > DSD512 xtr-mp CPU 29% (58) GPU 70% and the filter builds in 3 secs, again with no real GPU contribution. But playing, 70% means it’s workin’! I’ll try the closed form when I get a chance. Enjoying Aja with xtr-mp now... great bass and great spatial cues but seems the slightest bit muffled or congested compared to sinc-short-mp. https://audiophilestyle.com/profile/21384-solstice380/?tab=field_core_pfield_3 Link to comment
Miska Posted March 16, 2019 Share Posted March 16, 2019 13 hours ago, jabbr said: Installed the NVIDIA RTX2080 Ti wow! DSD512 poly-sinc-xtr-mp (Dell T420, 2x Xeon E5-2470 V2, 96Gb -- 20 cores) -- stereo without CUDA: 1400% with CUDA: 300% Just curious how much is the GPU load with Ti? Would be nice to know how much faster the Ti is in this case compared to non-Ti version. You can see the GPU info with "nvidia-smi", you can make it loop constantly for example with "nvidia-smi -l 10", then it refreshes every 10 seconds. Signalyst - Developer of HQPlayer Pulse & Fidelity - Software Defined Amplifiers Link to comment
jabbr Posted March 16, 2019 Share Posted March 16, 2019 Sat Mar 16 14:49:23 2019 +-----------------------------------------------------------------------------+ | NVIDIA-SMI 418.43 Driver Version: 418.43 CUDA Version: 10.1 | |-------------------------------+----------------------+----------------------+ | GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC | | Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. | |===============================+======================+======================| | 0 GeForce RTX 208... Off | 00000000:0A:00.0 Off | N/A | | 45% 68C P2 172W / 260W | 349MiB / 10989MiB | 47% Default | +-------------------------------+----------------------+----------------------+ +-----------------------------------------------------------------------------+ | Processes: GPU Memory | | GPU PID Type Process name Usage | |=============================================================================| | 0 28775 C /usr/bin/hqplayerd 339MiB | +-----------------------------------------------------------------------------+ Custom room treatments for headphone users. Link to comment
jabbr Posted March 16, 2019 Share Posted March 16, 2019 above is for PCM source, with DSD64 source: Sat Mar 16 14:55:18 2019 +-----------------------------------------------------------------------------+ | NVIDIA-SMI 418.43 Driver Version: 418.43 CUDA Version: 10.1 | |-------------------------------+----------------------+----------------------+ | GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC | | Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. | |===============================+======================+======================| | 0 GeForce RTX 208... Off | 00000000:0A:00.0 Off | N/A | | 40% 55C P2 73W / 260W | 333MiB / 10989MiB | 10% Default | +-------------------------------+----------------------+----------------------+ +-----------------------------------------------------------------------------+ | Processes: GPU Memory | | GPU PID Type Process name Usage | |=============================================================================| | 0 28775 C /usr/bin/hqplayerd 323MiB | +-----------------------------------------------------------------------------+ Custom room treatments for headphone users. Link to comment
jabbr Posted March 16, 2019 Share Posted March 16, 2019 18 hours ago, Em2016 said: The thousand dollar question that few of us can answer.... (at DSD512..) how does non-2s sound versus -2s? Any different? It’s at best subtle — I’m not going to give impressions until I optimize setup: The filters (at least seem to) only affect PCM -> DSD conversion not DSD->DSD upsampling, as reflected by the 10% GPU usage with DSD64->DSD512 whereas 50% usage PCM->DSD512 and I’m getting some stuttering with HD PCM sources ... I need to do some work to optimize unless @Miska is able to shed some light... (note that I’m using generic 4.18.0 kernel with Ubuntu 18.04.02 — yes I’ll be maximally compliant to reduce variables asdf1000 1 Custom room treatments for headphone users. Link to comment
Miska Posted March 16, 2019 Share Posted March 16, 2019 1 hour ago, jabbr said: above is for PCM source, with DSD64 source: Sat Mar 16 14:55:18 2019 +-----------------------------------------------------------------------------+ | NVIDIA-SMI 418.43 Driver Version: 418.43 CUDA Version: 10.1 | |-------------------------------+----------------------+----------------------+ | GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC | | Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. | |===============================+======================+======================| | 0 GeForce RTX 208... Off | 00000000:0A:00.0 Off | N/A | | 40% 55C P2 73W / 260W | 333MiB / 10989MiB | 10% Default | +-------------------------------+----------------------+----------------------+ +-----------------------------------------------------------------------------+ | Processes: GPU Memory | | GPU PID Type Process name Usage | |=============================================================================| | 0 28775 C /usr/bin/hqplayerd 323MiB | +-----------------------------------------------------------------------------+ I mean 44.1k source to 44.1k x512 with poly-sinc-xtr, so we can compare it with the other figures for 2080... Was the first one for that case? Signalyst - Developer of HQPlayer Pulse & Fidelity - Software Defined Amplifiers Link to comment
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now