Intel should add a fifth core, call it "MediaCore" technology.

[VirtualLarry]

And dedicate it to media playback.

Even on Linux, which isn't hobbled by MS's "Multimedia scheduler" technology (that retards network transfer speeds horribly), my media playback (streaming radio) isn't smooth. It glitches every few minutes. (Wired gigabit LAN).

If Intel is looking for things to add to their CPUs to take up silicon area, other than CPU cores and cache, then I suggest a dedicated media-playback core.

Edit: "Compute smarter, not faster". Intel has traditionally just thrown bigger, faster, CPU cores at problems (and do everything in software, hence things like Larrabee), rather than develop specialized, dedicated solutions, like the ARM camp has in the form of "media accelerators".

 

[ElFenix]

intel has some dedicated hardware paths for various media things, particularly video, to help with power consumption. it's where quicksync comes from.

arm had dedicated hardware decoding for media in part because their cores just weren't all that powerful.

 

[Essence_of_War]

Are you even sure that your streaming radio problems are CPU related?

 

[Arachnotronic]

And dedicate it to media playback.

Even on Linux, which isn't hobbled by MS's "Multimedia scheduler" technology (that retards network transfer speeds horribly), my media playback (streaming radio) isn't smooth. It glitches every few minutes. (Wired gigabit LAN).

If Intel is looking for things to add to their CPUs to take up silicon area, other than CPU cores and cache, then I suggest a dedicated media-playback core.

Edit: "Compute smarter, not faster". Intel has traditionally just thrown bigger, faster, CPU cores at problems (and do everything in software, hence things like Larrabee), rather than develop specialized, dedicated solutions, like the ARM camp has in the form of "media accelerators".

Um, you do realize that Intel builds in a lot of dedicated video decode/encode functionality into their iGPUs...right?

 

[VirtualLarry]

intel has some dedicated hardware paths for various media things, particularly video, to help with power consumption. it's where quicksync comes from.

arm had dedicated hardware decoding for media in part because their cores just weren't all that powerful.

Except, most of Intel's "media decoding" hardware, still requires CPU cycles, and thus is subject to the whims of the OS's scheduler. (For example, QSV using Handbrake, still gets my G3258 OCed dual-core, to nearly 100% CPU time when converting a BR MKV to an MP4 H.264 file.)

 

[TheELF]

Just raise the priority of your radio player to high and see if it helps,also maybe search for a program that will buffer a bit.

 

[TheELF]

(For example, QSV using Handbrake, still gets my G3258 OCed dual-core, to nearly 100% CPU time when converting a BR MKV to an MP4 H.264 file.)

And why shouldn't it?
That's the fastest way to do it.

 

[VirtualLarry]

Um, you do realize that Intel builds in a lot of dedicated video decode/encode functionality into their iGPUs...right?

And... does Flash Player-based internet radio playback utilize any of the hardware? Especially since it's driving the HDMI port on my AMD 7950 discrete card, and not the IGP's HDMI port audio.

I guess, I just feel that a dedicated processor core, that was programmable (so it could be used by Flash Player, or other media-playback sofware), but not subject to the OS'es software thread scheduler, would be useful. (Granted, I'm not sure how well that would work, if you wanted to play back multiple media streams at one time.)

 

[VirtualLarry]

And why shouldn't it?
That's the fastest way to do it.

If it's "dedicated hardware", it should be using nearly zero CPU cycles of the main processor cores.

 

[ShintaiDK]

If you say streaming radio makes it glitch. I dont think you got a CPU issue. And some extra random core wouldnt fix it.

 

[Essence_of_War]

If you say streaming radio makes it glitch. I dont think you got a CPU issue. And some extra random core wouldnt fix it.

This is what I was implying above. Audio is really easy to decode. It seems more likely that the problem is network related.

 

[VirtualLarry]

If you say streaming radio makes it glitch. I dont think you got a CPU issue. And some extra random core wouldnt fix it.

It seems as through, some piece of code in the whole stack, doesn't get scheduled on time, every time. So occasionally, there's a glitch.

On Windows, media playback triggers the usage of their "multimedia scheduler". Which really hurts networking performance. But media playback is better. Not perfect, but much better.

In Linux, network performance isn't hampered, but sometimes media playback suffers. I don't know if Linux Mint 17.1 uses a kernel with the preempt patches. (Did that ever reach mainline?)

I'm currently seeding Linux Torrents, with this same box. So media-playback is competing with network interrupt load.

Granted, I'm not currently running DC, so CPU usage is around 20% on each of my two cores.

The radio player, is running in Firefox, it's not a dedicated program for playback. (I assume it uses flash player? Or Firefox's native media-playback capabilities in HTML5?)

 

[VirtualLarry]

This is what I was implying above. Audio is really easy to decode. It seems more likely that the problem is network related.

I acknowledge that it could be... but why would a wired network have worse performance with streaming than wireless? I have a Gateway AIO with Kabini E1-2500 1.4Ghz dual-core, running Windows 8.1, playing the same streaming station, and it plays just fine, no glitches. Over wireless N.

Edit: And if it were my router (torrents, after all), or my internet connection (torrents, again), then why wouldn't the wireless PC be affected just the same as the wired?

I have a 75/75 connection, and I have my torrent client set to 7000KB/sec limits, up and down. I'm not currently downloading anything.

 

[ShintaiDK]

Considering how many millions that listen to web radio daily without issue. Then its quite obvious that its a relatively isolated case.

You could start dropping the torrent part to begin with when looking for a cause.

 

[VirtualLarry]

Instead of adding a dedicated core just for media, what about this instead - be able to, in both hardware (such as APIC interrupt routing connections / code) and software, be able to take an existing core, and dedicate it to media playback. Yes, setting affinity, and not sharing the core would be part-way there, but I'm thinking, of allowing media-related interrupts to be routed to that core directly too, possibly. That way, interrupts generated by media playback wouldn't have to compete with system interrupts on other cores.

Yes, this is un-SMP. The way that some OSes were designed before true SMP.

 

[dahorns]

This is what I was implying above. Audio is really easy to decode. It seems more likely that the problem is network related.

Given that this is a VirtualLarry post, I think it is safe to assume that he is doing something crazy on his end that is causing problems. Like, I picture VirtualLarry running seti@home on a "supercomputer" of 10,000 gameboys. But then complaining when that same computer struggles playing netflix while running windows in VMware.

Or something like that.

[Just kidding. . .mostly]

 

[VirtualLarry]

LOL. Perhaps this thread should instead be titled "Why does media playback suck in Linux"?

It could be the HDMI drivers for the 7950, I'm running the free drivers. I know that when I use using a DVI-to-HDMI adapter on this card, that carried audio, the audio in Linux was horrible and screechy, but worked in Windows 7. Now that I have a different case, I can connect the HDMI cable directly to the HDMI connection on the card, and it's not screechy any more.

Edit: The occasional buffer-glitch reminds me a lot of the terribad 29.97 problem with SB/IB video playback hardware. It's kind of like, the audio playback isn't clocked exactly right, so the buffer has to "adjust" every once in a while to maintain the stream.

 

[escrow4]

Flash has always been and always will be a dog in any OS. Combine that with average at best Linux drivers and there is your problem. In theory a toaster could stream radio without issues; CPU usage is irrelevant its down to bandwidth.

EDIT: Check DPC latency too.

 

[frozentundra123456]

No, it should be titled: "Why doesnt Larry just buy an i5 or i7"?

Like Dahorns, just kidding...mostly.

 

[Tsavo]

No, it should be titled: "Why doesnt Larry just buy an i5 or i7"?

Like Dahorns, just kidding...mostly.

Intel sells i5/7 branded NICs?

 

[VirtualLarry]

EDIT: Check DPC latency too.

Does Linux even use Deferred Procedure Calls? Pretty sure Linux has a different driver architecture. Not the mention, NT is fully pre-emptable, Linux is not, unless the kernel has the pre-empt patches applied.

No, it should be titled: "Why doesnt Larry just buy an i5 or i7"?

Like Dahorns, just kidding...mostly.

Gee whiz. You would think I wouldn't get these kinds of comments, now that I've moved up from a C-60 to a G3258 (Oced). Is the ultimate step a 5820K, just to listen to internet radio, without glitches? I'm not even doing DC, although I am seeding Linux distros.

 

[escrow4]

Does Linux even use Deferred Procedure Calls? Pretty sure Linux has a different driver architecture. Not the mention, NT is fully pre-emptable, Linux is not, unless the kernel has the pre-empt patches applied.

Gee whiz. You would think I wouldn't get these kinds of comments, now that I've moved up from a C-60 to a G3258 (Oced). Is the ultimate step a 5820K, just to listen to internet radio, without glitches? I'm not even doing DC, although I am seeding Linux distros.

A G3258 is only a tired dual core. I doubt this is hardware related. Something software. A new i5 never hurt nobody either. :whiste:

 

[cytg111]

And... does Flash Player-based internet radio playback utilize any of the hardware? Especially since it's driving the HDMI port on my AMD 7950 discrete card, and not the IGP's HDMI port audio.

I guess, I just feel that a dedicated processor core, that was programmable (so it could be used by Flash Player, or other media-playback sofware), but not subject to the OS'es software thread scheduler, would be useful. (Granted, I'm not sure how well that would work, if you wanted to play back multiple media streams at one time.)

No amount of hardware will fix retarded software. Flash is retarded. It is like they hanve ONE job left to do, play videos on the internet, and they fall flat on their face doing that as well.

 

[NTMBK]

Does Linux even use Deferred Procedure Calls? Pretty sure Linux has a different driver architecture. Not the mention, NT is fully pre-emptable, Linux is not, unless the kernel has the pre-empt patches applied.

Gee whiz. You would think I wouldn't get these kinds of comments, now that I've moved up from a C-60 to a G3258 (Oced). Is the ultimate step a 5820K, just to listen to internet radio, without glitches? I'm not even doing DC, although I am seeding Linux distros.

You're seeding a bunch of torrents on a two thread system, while trying to listening to streaming audio at the same time... no shit, the scheduler is going to struggle to fit all of those tasks into time slices smoothly without something losing out. Two threads is still two threads. Start messing with thread affinities, or just buy a good CPU already.

 

[biostud]

The CPU has enough power to run a media stream, but maybe the lack of HT makes it unsuitable for doing multiple processes.

I think it is OK if a modern task scheduler is optimized for running 4 threads as a minimum. A slower clocked i3 might deliver a better result.