First  Prev  1  2  3  4  Next  Last
Post Reply Crunchyroll Playback Choppy Again
78185 cr points
Send Message: Send PM GB Post
46 / M / Central USA
Online
Posted 11/12/15

RAWRwins254 wrote:

So I will address what technical "concerns" I can even though I don't work for CR and don't know their configurations. First, your "game box" does not have 12 cores. It has 6 cores with HyperThreading resulting in 2 threads per core thus Windows and basically any other OS will identify it as 12 graphs pre-Windows 8 and 6 cores 12 threads from Windows 8 on.


Hey, thanks for keeping the thread alive. Hmm, what technical concerns are you talking about? Already knew about hyperthreading and in general how it works. While there aren't twelve physically distinct cores in the processor as far as the operating system is concerned there are. And the advantages of hyperthreading over not can easily be found with a google search near you. Was it really that important to clarify this? If so I'm glad you got that off your chest. Would you have preferred I called it my gaming box?


RAWRwins254 wrote:
Second, your switch is not measured in GHz... it's measured in Gbps (Gigabits per second) which that switch listed can't support (48/24=2Gbps per port while that switch only does 1Gbps per port) for bandwidth.


Not sure why I typed ghz, higher brain to typing muscle-memory miscommunication. Sorry for the confusion for those who took umbrage. You're one of those 'Sulu fires the phasers' kinds of guys? I'm afraid it can do 1 gig full-duplex per port which is an aggregate 48 gbps. Of course actual performance probably falls short but in theory it's capable and was, back when the switch was new, a selling point of the model. The reason I mentioned this is due to your average gig-e switch, while being capable of 2 gbps per port, doesn't typically have the bandwidth internally to actually support all the ports transmitting and receiving their maximum simultaneously. This can be a pretty big issue depending on the environment.


RAWRwins254 wrote:

Third, in my section of my country (Minnesota, USA), Comcast hasn't opened up their Gigabit Pro service which is 4Gbps (2down/2up) total and that can't be used commercially. In fact 35Gbit isn't possible on a DOCSIS 3.1 modem even (which your modem is a DOCSIS 3.0 as an FYI).


Doh, no idea why I said gbit, I meant mbit believe it or not. As much as I would like otherwise I don't have the kind of money to run even a single gigabit connection let alone 35 of them. I think I had gig on the brain at the time. Thank you for catching that.


RAWRwins254 wrote:

However, I will give you the TCP crap. Even though it's being delivered through a website, the website tells Flash where the content is and then the content is streamed over UDP typically. I also will be unsubbing, but not due to your post. I'm also fed up with the stuttering and I have 130Mbps+ down typically (never drops below 100Mbps unless I'm downloading tons of small files (<1MB)). BTW, Hope you didn't sign a term agreement with Cox for Business-grade Internet. 25Mbps (which is what you more than likely actually have considering that their website doesn't show a 35Mbps plan for business) is shitty. Then again, I'm glad Cox doesn't do business in my state of residence. I pay $56/month for 105+down/10+up. I can't say I like Comcast's views on certain business ethics, but they're the best residential ISP in my area (CenturyLink needs to move to native dual-stack vs the 6rd garbage they're currently running and they need to up speeds AND get rid of their data caps.)


UDP would be much more efficient as those who've had reason to stream data have already discovered. I'm going to guess TCP is necessary due to Crunchy's completely pointless DRM.
27024 cr points
Send Message: Send PM GB Post
26 / M
Offline
Posted 11/12/15

iamsemiazas wrote:


RAWRwins254 wrote:
Second, your switch is not measured in GHz... it's measured in Gbps (Gigabits per second) which that switch listed can't support (48/24=2Gbps per port while that switch only does 1Gbps per port) for bandwidth.


Not sure why I typed ghz, higher brain to typing muscle-memory miscommunication. Sorry for the confusion for those who took umbrage. You're one of those 'Sulu fires the phasers' kinds of guys? I'm afraid it can do 1 gig full-duplex per port which is an aggregate 48 gbps. Of course actual performance probably falls short but in theory it's capable and was, back when the switch was new, a selling point of the model. The reason I mentioned this is due to your average gig-e switch, while being capable of 2 gbps per port, doesn't typically have the bandwidth internally to actually support all the ports transmitting and receiving their maximum simultaneously. This can be a pretty big issue depending on the environment.


DOH! Here i was talking about 4Gbps (2down/2up) from comcast and I neglected to remember full-duplex functionality... well i feel like a moron now. I concede my original point about max bandwidth. Although we could then argue about signaling rate versus effective data transfer and VLAN link aggregation too... not like either is really necessary in your particular use scenario. Either way, Nice to know that my original fears were misplaced. Then again, most "fake geeks" wouldn't know how to really use an LSI RAID controller with FastPath.... BTW, you said you're running 8 SSDs in RAID 0 right? why not RAID 10? There's not a performance gain from RAID 0 over RAID 10 at that level iirc. Then again, I could be being a moron again. But RAID 10 ensures redundancy while still giving RAID 0 performance. I personally would make 4 RAID 1 arrays and then RAID 0 them together to form my RAID 10 set up. But I'm kinda broke ATM so I can't really test a lot of stuff that I'd really like to.
37709 cr points
Send Message: Send PM GB Post
45 / Seattle
Offline
Posted 11/12/15

FLjerry2011 wrote:
to nuke them. It's probably simple coincidence that it helped me today (no guarantee for the long term), but I don't think it can hurt.

I domt have to woory about that detail for now since the vidoes dont load at all let alone the website / slow / disconnects! They havec some issues going on ! All the other websites I went to work only CR has a patent on it's unique problems !

Forget about producing anime get the website to work better!
Sorry about that, I forgot to post a followup that it was indeed coincidence.
78185 cr points
Send Message: Send PM GB Post
46 / M / Central USA
Online
Posted 11/12/15

RAWRwins254 wrote:


DOH! Here i was talking about 4Gbps (2down/2up) from comcast and I neglected to remember full-duplex functionality... well i feel like a moron now. I concede my original point about max bandwidth. Although we could then argue about signaling rate versus effective data transfer and VLAN link aggregation too... not like either is really necessary in your particular use scenario. Either way, Nice to know that my original fears were misplaced. Then again, most "fake geeks" wouldn't know how to really use an LSI RAID controller with FastPath.... BTW, you said you're running 8 SSDs in RAID 0 right? why not RAID 10? There's not a performance gain from RAID 0 over RAID 10 at that level iirc. Then again, I could be being a moron again. But RAID 10 ensures redundancy while still giving RAID 0 performance. I personally would make 4 RAID 1 arrays and then RAID 0 them together to form my RAID 10 set up. But I'm kinda broke ATM so I can't really test a lot of stuff that I'd really like to.


Not tagging here. I've got a mix of SOHO routers of various types (as well as a Cisco 2506 that does nothing but rate-shaping), three servers (Linux opensuse) of various formats, a handful of desktops, some TVs, a bunch of IP cameras, a couple multimedia receivers, etc. absolutely no reason for breaking the LAN into virtual pieces.

Space. I want every drop of that solid state goodness. I back it up to a 23 tera RAID 6 (Areca ARC-1170 with a bunch of WD three tera drives) nightly (incremental) and don't keep anything really important on it. Plus the drives (OCZ Vertex 4 256 gig) have a 2 million hour MTBF (yes, they're past end of life, I'm broke too).

Speaking of geek, sitting in my closet is a dual-processor vapor phase chiller custom built and shipped from Taiwan. It used to cool two 2.6 gig Xeons down to -60 celsius, both overclocked to 4.8 gigs stable (prime 95 48 hour burn-in). Unfortunately that PC died (it was inevitable, I found water damage around the sockets when I took it apart) and I didn't/don't have the money to build something similar (thank you Republican party) so the chiller languishes, waiting for a new monster to freeze. For the curious, it runs on 240 volts and when cold-starting pulls almost 5 amps. It would brown-out my house. But it was extremely loud (hence the 50' cables run from my closet through my bathroom into my bedroom) and while running almost doubled my electric bill. Some day I need to fire it back up and make sure it hasn't leaked all its coolant. I actually have two of them, the first didn't make it from overseas intact having spewed coolant and lubricant all over itself and the inside of its crate. The crazy Taiwanese that built it for me sent me a second unit free of charge. IT showed up in a crate that was reinforced with steel framing bars bolted together in a cage. BUT the compressor in the first is still good so a decent A/C guy (which I'm not) could probably get it running again.... Don't believe me ask me for pics...
27024 cr points
Send Message: Send PM GB Post
26 / M
Offline
Posted 11/13/15

iamsemiazas wrote:
Not tagging here. I've got a mix of SOHO routers of various types (as well as a Cisco 2506 that does nothing but rate-shaping), three servers (Linux opensuse) of various formats, a handful of desktops, some TVs, a bunch of IP cameras, a couple multimedia receivers, etc. absolutely no reason for breaking the LAN into virtual pieces.

Space. I want every drop of that solid state goodness. I back it up to a 23 tera RAID 6 (Areca ARC-1170 with a bunch of WD three tera drives) nightly (incremental) and don't keep anything really important on it. Plus the drives (OCZ Vertex 4 256 gig) have a 2 million hour MTBF (yes, they're past end of life, I'm broke too).

Speaking of geek, sitting in my closet is a dual-processor vapor phase chiller custom built and shipped from Taiwan. It used to cool two 2.6 gig Xeons down to -60 celsius, both overclocked to 4.8 gigs stable (prime 95 48 hour burn-in). Unfortunately that PC died (it was inevitable, I found water damage around the sockets when I took it apart) and I didn't/don't have the money to build something similar (thank you Republican party) so the chiller languishes, waiting for a new monster to freeze. For the curious, it runs on 240 volts and when cold-starting pulls almost 5 amps. It would brown-out my house. But it was extremely loud (hence the 50' cables run from my closet through my bathroom into my bedroom) and while running almost doubled my electric bill. Some day I need to fire it back up and make sure it hasn't leaked all its coolant. I actually have two of them, the first didn't make it from overseas intact having spewed coolant and lubricant all over itself and the inside of its crate. The crazy Taiwanese that built it for me sent me a second unit free of charge. IT showed up in a crate that was reinforced with steel framing bars bolted together in a cage. BUT the compressor in the first is still good so a decent A/C guy (which I'm not) could probably get it running again.... Don't believe me ask me for pics...


I was referring to VLAN'ing for the purpose of Link Aggregation (LACP stuff). How many drives can fail in a RAID 6 though and how long does it take to rebuild the array?
78185 cr points
Send Message: Send PM GB Post
46 / M / Central USA
Online
Posted 11/13/15 , edited 11/13/15

RAWRwins254 wrote:

I was referring to VLAN'ing for the purpose of Link Aggregation (LACP stuff). How many drives can fail in a RAID 6 though and how long does it take to rebuild the array?


Ah, apologies for anyone following this thread concerned about Crunchyroll's persistent stuttering problem. FYI the compression artifacts are back but for two days now i've not had any stuttering issues AND I can now freely skip back and forth without any hangs or the inexplicable 'play the next video now' problem I mentioned previously. I've made no changes to any of my systems, software, etc. I haven't even updated my video driver even though there's one available. If this keeps up I may not change my underwear either..

Hmm, VLANs and trunking (LAGs etc.) only work together when the logical port happens to be tagged/untagged with particular VLANs/VLAN. The SMC does support LACP although they call it trunking. I may experiment some day with a 2 gig lag to a file server but none of the ones I currently use have more than a single gig-e port (one has a second 100 meg port pfft). Since it's just me and I've yet to need more than any one client loading down the one of my file servers I've not really had reason to mess with it. Even if I did build a lag between the switch and a file server/etc. I still don't see reason to segregate off that port (and whatever else I'd stick with it) from the rest of the devices on my network. Just in case we're mixing up terms here, a VLAN is a virtual (virtualized shrug) local area network, it's where a switch, or group of switches, logically isolate physical members from each other as if they were physically isolated. A typical small office implementation would use a single switch and everything plugged into that switch could communicate with everything else (setting aside layer 3). If that switch were broken into two VLANs and say the first half of the ports were on VLAN 1 and the second on VLAN 2 the switch wouldn't pass packets from VLAN 2 ports to VLAN 1 ports and vice versa. Tagging comes into play where more than one switch is sharing the topology, it's how the switches communicate what VLAN a particular ethernet frame is intended for by "tagging" the frame with a little bit of extra data (802.1q is the protocol standard). I only have a single switch in my environment so I wouldn't have reason to tag, however, if I wanted to isolate say a couple of devices on ports one and two from the remaining twenty-two ports I would untag them a different VLAN than the default (which is 1). BTW, any device out there, with the correct software, can poison ethernet frames to appear on any VLAN they wish (it's just extra bits in the ethernet frame header) so relying upon VLANs for network security is a mistake.

Hmm, you can read up about all you want on RAID online but I suspect you already know the answer to your question. A RAID six can lose two drives (whereas RAID five can lose only one) and still retain integrity. Unlike RAID 5 where a single drive is set aside for parity, a RAID six stripes the parity data across all the drives. I've had drives fail twice in the past three years. It took about fourteen hours to rebuild after they were replaced. For the curious, a RAID 0 is where data is striped across all the drives in the RAID, this means fast fast read and write speeds because (assuming the RAID controller is decent) data can be read/written to every drive simultaneously. Think of each drive as if it's a sink faucet and the data being read or written as the water flowing through the faucet. A RAID 0 with eight drives is like a sink with eight faucets. RAID five and six CAN enjoy similar benefits BUT typically don't especially for write speeds because there's additional processing overhead while generating the parity data. The Areca's (I've got a ARC-1160 also) while pretty old are still pretty quick but I don't care as much about performance as I do a huge honking space to save all my stuff.

55012 cr points
Send Message: Send PM GB Post
30 / M
Offline
Posted 11/15/15
I also have the choppy playback issues. However, not all the time. I guess it is related on how many users use the site (or maybe even show related).
However, it is the only site where I have those issues and from what I can find in the forum I am not the only one for whom this the case.

I also think it would be a good idea if the video player were to show buffering status - maybe the issue is in general buffering related, e.g. to small a buffer could cause issues in case of bandwidth limitations (either own Internet or CR)
32078 cr points
Send Message: Send PM GB Post
31 / M
Offline
Posted 11/15/15
This problem has been happening off and on for years on CR. ITS PATHETIC. It's so bad this week I am torrenting my anime. I couldn't watch last night's One Piece on CR if my life depended on it.

If this keeps up, I will stop paying for Premium. So tired of it.
27024 cr points
Send Message: Send PM GB Post
26 / M
Offline
Posted 11/15/15

iamsemiazas wrote:

Hmm, VLANs and trunking (LAGs etc.) only work together when the logical port happens to be tagged/untagged with particular VLANs/VLAN. The SMC does support LACP although they call it trunking. I may experiment some day with a 2 gig lag to a file server but none of the ones I currently use have more than a single gig-e port (one has a second 100 meg port pfft). Since it's just me and I've yet to need more than any one client loading down the one of my file servers I've not really had reason to mess with it. Even if I did build a lag between the switch and a file server/etc. I still don't see reason to segregate off that port (and whatever else I'd stick with it) from the rest of the devices on my network. Just in case we're mixing up terms here, a VLAN is a virtual (virtualized shrug) local area network, it's where a switch, or group of switches, logically isolate physical members from each other as if they were physically isolated. A typical small office implementation would use a single switch and everything plugged into that switch could communicate with everything else (setting aside layer 3). If that switch were broken into two VLANs and say the first half of the ports were on VLAN 1 and the second on VLAN 2 the switch wouldn't pass packets from VLAN 2 ports to VLAN 1 ports and vice versa. Tagging comes into play where more than one switch is sharing the topology, it's how the switches communicate what VLAN a particular ethernet frame is intended for by "tagging" the frame with a little bit of extra data (802.1q is the protocol standard). I only have a single switch in my environment so I wouldn't have reason to tag, however, if I wanted to isolate say a couple of devices on ports one and two from the remaining twenty-two ports I would untag them a different VLAN than the default (which is 1). BTW, any device out there, with the correct software, can poison ethernet frames to appear on any VLAN they wish (it's just extra bits in the ethernet frame header) so relying upon VLANs for network security is a mistake.

Hmm, you can read up about all you want on RAID online but I suspect you already know the answer to your question. A RAID six can lose two drives (whereas RAID five can lose only one) and still retain integrity. Unlike RAID 5 where a single drive is set aside for parity, a RAID six stripes the parity data across all the drives. I've had drives fail twice in the past three years. It took about fourteen hours to rebuild after they were replaced. For the curious, a RAID 0 is where data is striped across all the drives in the RAID, this means fast fast read and write speeds because (assuming the RAID controller is decent) data can be read/written to every drive simultaneously. Think of each drive as if it's a sink faucet and the data being read or written as the water flowing through the faucet. A RAID 0 with eight drives is like a sink with eight faucets. RAID five and six CAN enjoy similar benefits BUT typically don't especially for write speeds because there's additional processing overhead while generating the parity data. The Areca's (I've got a ARC-1160 also) while pretty old are still pretty quick but I don't care as much about performance as I do a huge honking space to save all my stuff.



Hey sorry for the late reply. RAID 5 is distributed parity at a block level as well... I think RAID 6 allows two failures because of double parity distibution though whereas RAID 5 only has one parity section.... Either way, the rebuild is required though the extra parity should speed that up in theory. Also, VLANs being insecure is a moot point in LAGs... The data that flows outside the LAG is still visible anyways to any computer on the network (minus WiFi if you use AP Isolation). If you're using a LAG for more bandwidth to a single local node, it won't matter if VLAN technology is secure or not because you'll still have machines connected to a switched outside the LAG allocation. When the packets hit the switch, it basically goes to the rest of the network stripping the original VLAN stuff except for a reply-route. At least that's the way I understand it. Besides, if someone has physical access to your network, you should assume the risks of said access.
78185 cr points
Send Message: Send PM GB Post
46 / M / Central USA
Online
Posted 11/15/15

RAWRwins254 wrote:

Hey sorry for the late reply. RAID 5 is distributed parity at a block level as well... I think RAID 6 allows two failures because of double parity distibution though whereas RAID 5 only has one parity section.... Either way, the rebuild is required though the extra parity should speed that up in theory. Also, VLANs being insecure is a moot point in LAGs... The data that flows outside the LAG is still visible anyways to any computer on the network (minus WiFi if you use AP Isolation). If you're using a LAG for more bandwidth to a single local node, it won't matter if VLAN technology is secure or not because you'll still have machines connected to a switched outside the LAG allocation. When the packets hit the switch, it basically goes to the rest of the network stripping the original VLAN stuff except for a reply-route. At least that's the way I understand it. Besides, if someone has physical access to your network, you should assume the risks of said access.


Bah. RAID 3 is striping with a single drive set aside for parity data. RAID 5 distributes the parity data (as does RAID 6). For some reason I got the two mixed up in my head. I blame Monty Python. If nothing else this conversation has helped me refresh my knowledge. I wear many hats and it's easy to forget or mix up things I've not had reason to think about in over a decade.

Uh most switches (all of them as far as I know) treat a LAG as if it were a single physical connection. This allows the aggregation of the bandwidth of all the members of the LAG. I'm not sure how that applies to data that flows outside a LAG or what you mean by saying such. Switches don't send ethernet frames they know the destination for "outside" the port (be it a physical port or a LAG) they've learned where the destination is. That's the nature of a switch and why it's superior to a hub. My original point, which I should have made more clear, was I didn't understand what link aggregation and VLANs had to do with each other in terms of performance impact. LAGs are just logical ports made up of groups of physical ports, and VLANs are just numbers defining layer two boundaries in broadcast domains. The two aren't really related to each other even though they're often used together in big environments. In particular, getting back to the original off-topic, they really don't have anything to do with the performance or lack thereof of my little twenty-four port SMC switch. It can do VLAN, it can do LACP, but I'm not using either because to do so would be silly.

41366 cr points
Send Message: Send PM GB Post
54 / M / East Coast
Offline
Posted 11/16/15

millotalius wrote:

This problem has been happening off and on for years on CR. ITS PATHETIC. It's so bad this week I am torrenting my anime. I couldn't watch last night's One Piece on CR if my life depended on it.

If this keeps up, I will stop paying for Premium. So tired of it.


They seemed to have made changes because of Windows 10 but it affected other platforms ! Fixing one creates problems for others!
78185 cr points
Send Message: Send PM GB Post
46 / M / Central USA
Online
Posted 11/18/15
So far choppy free, incredibly going on day five. Undies starting to giving off a green vapor. Going to experiment with swapping them out for clean ones so my coworkers stop complaining about "that rancid goat cheese smell." Wish me luck.
First  Prev  1  2  3  4  Next  Last
You must be logged in to post.