These forums have been archived and are now read-only.

The new forums are live and can be found at https://forums.eveonline.com/

EVE General Discussion

 
  • Topic is locked indefinitely.
Previous page123Next page
 

time dilation problem ?

First post
Author
Previous Topic Next Topic
Vaerah Vahrokha
Vahrokh Consulting
#21 - 2013-06-16 21:09:37 UTC
CCP Explorer wrote:
Azami Nevinyrall wrote:
I'm having similar problems, earlier I was in a fleet, each jump caused TiDi to jump up to 80% in some cases.

90-100 people jumping around shouldn't cause this, with client freezing (and in some cases crashing) every jump in the new and better titled "Black warp tunnel of death."
CCP Veritas has been working on a project called Brain in a Box for a while to address this exact issue.


Please tell CCP Veritas that certain video cards that used to work pretty fine, crash quite hard ("driver has stopped working and has been reinitialized error") from about half of the new warp tunnel to its end. This regardless of installing old or new drivers, flashing the BIOS, updating DirectX.

EvE Online portal | EvE markets tutorials
CCP Explorer
C C P
C C P Alliance
#22 - 2013-06-16 22:49:12 UTC
Angmar Udate wrote:
Please take a look at Aridia region, lowsec systems of Hophib and surrounding - N3 deployed there and there is tidi permanently. Evil Perhaps you can reallocate the constellation to its own node?
There are no nodes under TiDi right now.

Erlendur S. Thorsteinsson | Senior Development Director | EVE Online // CCP Games | @CCP_Explorer
pmchem
Brutor Tribe
Minmatar Republic
#23 - 2013-06-16 22:58:59 UTC
CCP Explorer wrote:
Azami Nevinyrall wrote:
I'm having similar problems, earlier I was in a fleet, each jump caused TiDi to jump up to 80% in some cases.

90-100 people jumping around shouldn't cause this, with client freezing (and in some cases crashing) every jump in the new and better titled "Black warp tunnel of death."
CCP Veritas has been working on a project called Brain in a Box for a while to address this exact issue.


Single most important project of 2013. I hope he can be freed up to work on that exclusively; hire another guy to work on other responsibilities of his. :)

https://twitter.com/pmchem/ || http://community.eveonline.com/news/dev-blogs/community-spotlight-garpa/ || Goonswarm Economic Warfare Cabal
Ace Boogi
Republic University
Minmatar Republic
#24 - 2013-06-16 23:11:35 UTC
Sephira Galamore wrote:
Tiven loves Tansien wrote:
Don't talk like you know everything, EVE really needs extra power. Why do i have to experience TiDi if there's a fight in some other constellation away?

The main issue is, that from a business / common sense perspective, it makes little sense to have only one, or only a few systems on a node, when those are empty 90% of the time and only have a dozen people in them 9.9% of the time, just for that off chance that a huge fleet passes thru them today or tomorrow.
Hence the possibility to inform CCP ahead of time, if you have a known large operation, or a battle with a high escalation chance.

even when the nodes are supposedly 'reinforced' i notice little to no difference. the tidi in this game is a soul crushing joke of a mechanic and removes desire for me to train for large expensive ships or participate in fleet fights.

'i was there' at some of the largest battles of the last year or two, including ashakai or whatever it was spelled, and i don't know about anyone else, but that experience was far from being fun. do you guys really find enjoyment in taking 20 minutes or more to jump into system, another 20 or 30 minutes to warp, and having only fired 7 volleys from your guns in 3 hours? oh and ofc all the disconnects... yeah that really makes me want to train into and buy a supercarrier, so i can take an hour to cyno in, then blackscreen and lose my ship as i try to log back in
Paul Maken
Deep Core Mining Inc.
Caldari State
#25 - 2013-06-17 00:05:39 UTC
CCP Explorer wrote:
There is then action ongoing in another solarsystem on that same node. There are 252 location and service nodes on TQ (on 57 blades; most of them quad core, some with 8 cores) and 70 proxies (on 12 blades with 1 to 8 cores). Time Dilation is across an entire location node since the CPU on that node is then struggling due to load. A location node hosts a number of solarsystem, most often a number of constellations.


How do the nodes map onto the cores? Is it one node per core, or does Jita's node get an 8 core server all to itself? Are nodes single or multi-threaded? If they are multi-threaded, what is done to optimize for locality of reference?

Are the blades single or multiple socket? Dealing with NUMA may be a pain, but it works wonders at keeping the slower memory and interconnect busses from committing suicide under load.
Azami Nevinyrall
172.0.0.1
#26 - 2013-06-17 00:40:54 UTC
Vaerah Vahrokha wrote:
CCP Explorer wrote:
Azami Nevinyrall wrote:
I'm having similar problems, earlier I was in a fleet, each jump caused TiDi to jump up to 80% in some cases.

90-100 people jumping around shouldn't cause this, with client freezing (and in some cases crashing) every jump in the new and better titled "Black warp tunnel of death."
CCP Veritas has been working on a project called Brain in a Box for a while to address this exact issue.


Please tell CCP Veritas that certain video cards that used to work pretty fine, crash quite hard ("driver has stopped working and has been reinitialized error") from about half of the new warp tunnel to its end. This regardless of installing old or new drivers, flashing the BIOS, updating DirectX.

Essentially this, only been starting since last patch.

And cool, I've heard about the brain box project. Hopefully it fixes the issue.....Hopefully it doesn't **** up and wipe all characters.

...
Lors Dornick
Kallisti Industries
#27 - 2013-06-17 02:42:59 UTC
Vaerah Vahrokha wrote:
CCP Explorer wrote:
Azami Nevinyrall wrote:
I'm having similar problems, earlier I was in a fleet, each jump caused TiDi to jump up to 80% in some cases.

90-100 people jumping around shouldn't cause this, with client freezing (and in some cases crashing) every jump in the new and better titled "Black warp tunnel of death."
CCP Veritas has been working on a project called Brain in a Box for a while to address this exact issue.


Please tell CCP Veritas that certain video cards that used to work pretty fine, crash quite hard ("driver has stopped working and has been reinitialized error") from about half of the new warp tunnel to its end. This regardless of installing old or new drivers, flashing the BIOS, updating DirectX.


CCP Veritas (and others on that team) has been working on solving critical performance issues in the Eve back end for some time, and it appears that they still are.

But you think they should drop that issue and bother about some some random graphic drivers, in the other end of the game?

Seems like a smart move, not.


CCP Greyscale: As to starbases, we agree it's pretty terrible, but we don't want to delay the entire release just for this one factor.
NEONOVUS
Mindstar Technology
Goonswarm Federation
#28 - 2013-06-17 03:09:53 UTC
Eve's back end as with many asses is obstinate slow to work and refuses attempts to improve it.
But without she would not be such a smoking hot lady.

Also CCP Explorer you are making me very happy with this data, largely as it makes calls for monumental destruction much less likely to happen.
dark heartt
#29 - 2013-06-17 03:20:46 UTC
Warcalibre wrote:
DTson Gauur wrote:
CCP Explorer wrote:
There is then action ongoing in another solarsystem on that same node. There are 252 location and service nodes on TQ (on 57 blades; most of them quad core, some with 8 cores) and 70 proxies (on 12 blades with 1 to 8 cores). Time Dilation is across an entire location node since the CPU on that node is then struggling due to load. A location node hosts a number of solarsystem, most often a number of constellations.


Offtopic, but: Interesing facts about the size of TQs blade farm, I thought it was bigger.


EVE is dying?


And has been since 2003!

http://tetrisisunrealistic.blogspot.com.au/
Kossaw
Body Count Inc.
Mercenary Coalition
#30 - 2013-06-17 03:28:53 UTC
CCP Explorer wrote:
Azami Nevinyrall wrote:
I'm having similar problems, earlier I was in a fleet, each jump caused TiDi to jump up to 80% in some cases.

90-100 people jumping around shouldn't cause this, with client freezing (and in some cases crashing) every jump in the new and better titled "Black warp tunnel of death."
CCP Veritas has been working on a project called Brain in a Box for a while to address this exact issue.


CCP Veritas is a very smart programmer. Thank you for letting him work on projects like this outside the normal SCRUM cycle. The results of his TiDi changes were spectacular. Here's hoping he can solve the problems associated with moving characters from one system to another in an equally effective way

WTB : An image in my signature
NEONOVUS
Mindstar Technology
Goonswarm Federation
#31 - 2013-06-17 03:31:07 UTC
I am curious why is it that the data is only being considered for preservation now, and not being preserved between alterations?
Frostys Virpio
State War Academy
Caldari State
#32 - 2013-06-17 05:30:45 UTC
CCP Explorer wrote:
Azami Nevinyrall wrote:
I'm having similar problems, earlier I was in a fleet, each jump caused TiDi to jump up to 80% in some cases.

90-100 people jumping around shouldn't cause this, with client freezing (and in some cases crashing) every jump in the new and better titled "Black warp tunnel of death."
CCP Veritas has been working on a project called Brain in a Box for a while to address this exact issue.


Did CCP Veritas qualify this as "very hard"?

:D
James Amril-Kesh
Viziam
Amarr Empire
#33 - 2013-06-17 15:46:13 UTC
CCP Explorer wrote:
Azami Nevinyrall wrote:
I'm having similar problems, earlier I was in a fleet, each jump caused TiDi to jump up to 80% in some cases.

90-100 people jumping around shouldn't cause this, with client freezing (and in some cases crashing) every jump in the new and better titled "Black warp tunnel of death."
CCP Veritas has been working on a project called Brain in a Box for a while to address this exact issue.

I don't think you understand his point.
It's worse than before.
A jump of 100 people used to only bring tidi to maybe 50% for 30 seconds max. Nowadays it gets closer to 20% (what the poster you replied to meant, not 80% which is 8/10 speed) and slowly recovers over the course of at least two minutes. It's the same with docking and undocking. The server seems to be much more aggressive about tidi than it was before and it seems to recover much more slowly than before.

Enjoying the rain today? ;)
Ranger 1
Ranger Corp
Vae. Victis.
#34 - 2013-06-17 18:00:20 UTC
James Amril-Kesh wrote:
CCP Explorer wrote:
Azami Nevinyrall wrote:
I'm having similar problems, earlier I was in a fleet, each jump caused TiDi to jump up to 80% in some cases.

90-100 people jumping around shouldn't cause this, with client freezing (and in some cases crashing) every jump in the new and better titled "Black warp tunnel of death."
CCP Veritas has been working on a project called Brain in a Box for a while to address this exact issue.

I don't think you understand his point.
It's worse than before.
A jump of 100 people used to only bring tidi to maybe 50% for 30 seconds max. Nowadays it gets closer to 20% (what the poster you replied to meant, not 80% which is 8/10 speed) and slowly recovers over the course of at least two minutes. It's the same with docking and undocking. The server seems to be much more aggressive about tidi than it was before and it seems to recover much more slowly than before.

Makes me wonder if they needed to tweak it to be more aggressive on purpose.

View the latest EVE Online developments and other game related news and gameplay by visiting Ranger 1 Presents: Virtual Realms.
CCP Explorer
C C P
C C P Alliance
#35 - 2013-06-17 19:24:19 UTC
Paul Maken wrote:
CCP Explorer wrote:
There is then action ongoing in another solarsystem on that same node. There are 252 location and service nodes on TQ (on 57 blades; most of them quad core, some with 8 cores) and 70 proxies (on 12 blades with 1 to 8 cores). Time Dilation is across an entire location node since the CPU on that node is then struggling due to load. A location node hosts a number of solarsystem, most often a number of constellations.
How do the nodes map onto the cores? Is it one node per core, or does Jita's node get an 8 core server all to itself? Are nodes single or multi-threaded? If they are multi-threaded, what is done to optimize for locality of reference?

Are the blades single or multiple socket? Dealing with NUMA may be a pain, but it works wonders at keeping the slower memory and interconnect busses from committing suicide under load.
It's one node per core. Jita is on a high-end quad core machine, the other 3 cores are used for 3 nodes for fleet fights. Each node is single-threaded, cooperative multitasking (Python Global Interpreter Lock).

Erlendur S. Thorsteinsson | Senior Development Director | EVE Online // CCP Games | @CCP_Explorer
CCP Explorer
C C P
C C P Alliance
#36 - 2013-06-17 19:26:34 UTC
Ranger 1 wrote:
James Amril-Kesh wrote:
CCP Explorer wrote:
Azami Nevinyrall wrote:
I'm having similar problems, earlier I was in a fleet, each jump caused TiDi to jump up to 80% in some cases.

90-100 people jumping around shouldn't cause this, with client freezing (and in some cases crashing) every jump in the new and better titled "Black warp tunnel of death."
CCP Veritas has been working on a project called Brain in a Box for a while to address this exact issue.

I don't think you understand his point.
It's worse than before.
A jump of 100 people used to only bring tidi to maybe 50% for 30 seconds max. Nowadays it gets closer to 20% (what the poster you replied to meant, not 80% which is 8/10 speed) and slowly recovers over the course of at least two minutes. It's the same with docking and undocking. The server seems to be much more aggressive about tidi than it was before and it seems to recover much more slowly than before.

Makes me wonder if they needed to tweak it to be more aggressive on purpose.
No changes have been made to TiDi; you are probably experiencing another issue, logging we added just before Odyssey that CCP Veritas and CCP QC are actively working on optimizing.

Erlendur S. Thorsteinsson | Senior Development Director | EVE Online // CCP Games | @CCP_Explorer
Tinkerbel Ducttape
Sebiestor Tribe
Minmatar Republic
#37 - 2013-06-17 19:42:01 UTC  |  Edited by: Tinkerbel Ducttape
CCP Explorer wrote:
No changes have been made to TiDi; you are probably experiencing another issue, logging we added just before Odyssey that CCP Veritas and CCP QC are actively working on optimizing.
So, can't CCP just remove that and get it to work properly before readding it?
Azami Nevinyrall
172.0.0.1
#38 - 2013-06-17 20:06:42 UTC
Tinkerbel Ducttape wrote:
CCP Explorer wrote:
No changes have been maSadde to TiDi; you are probably experiencing another issue, logging we added just before Odyssey that CCP Veritas and CCP QC are actively working on optimizing.
So, can't CCP just remove that and get it to work properly before readding it?

I think it would be better if they actually wrote the game from scratch and fix all the known issues while doing so.

Eventually this HAS to be done, CCP knows it too. It's just a matter of time, and I hope it sooner then waiting until the game is no longer playable.

Yes, I know what a large task that is. Just stop being lazy already!

...
Ranger 1
Ranger Corp
Vae. Victis.
#39 - 2013-06-17 20:41:14 UTC
CCP Explorer wrote:
Ranger 1 wrote:
James Amril-Kesh wrote:
CCP Explorer wrote:
Azami Nevinyrall wrote:
I'm having similar problems, earlier I was in a fleet, each jump caused TiDi to jump up to 80% in some cases.

90-100 people jumping around shouldn't cause this, with client freezing (and in some cases crashing) every jump in the new and better titled "Black warp tunnel of death."
CCP Veritas has been working on a project called Brain in a Box for a while to address this exact issue.

I don't think you understand his point.
It's worse than before.
A jump of 100 people used to only bring tidi to maybe 50% for 30 seconds max. Nowadays it gets closer to 20% (what the poster you replied to meant, not 80% which is 8/10 speed) and slowly recovers over the course of at least two minutes. It's the same with docking and undocking. The server seems to be much more aggressive about tidi than it was before and it seems to recover much more slowly than before.

Makes me wonder if they needed to tweak it to be more aggressive on purpose.
No changes have been made to TiDi; you are probably experiencing another issue, logging we added just before Odyssey that CCP Veritas and CCP QC are actively working on optimizing.

That would explain it. Smile

View the latest EVE Online developments and other game related news and gameplay by visiting Ranger 1 Presents: Virtual Realms.
Ranger 1
Ranger Corp
Vae. Victis.
#40 - 2013-06-17 20:42:55 UTC
Azami Nevinyrall wrote:
Tinkerbel Ducttape wrote:
CCP Explorer wrote:
No changes have been maSadde to TiDi; you are probably experiencing another issue, logging we added just before Odyssey that CCP Veritas and CCP QC are actively working on optimizing.
So, can't CCP just remove that and get it to work properly before readding it?

I think it would be better if they actually wrote the game from scratch and fix all the known issues while doing so.

Eventually this HAS to be done, CCP knows it too. It's just a matter of time, and I hope it sooner then waiting until the game is no longer playable.

Yes, I know what a large task that is. Just stop being lazy already!

I think we'll let you explain to the rest of the player base why there won't be any updates for the next few years while everyone at CCP is busy reinventing the wheel. Smile

View the latest EVE Online developments and other game related news and gameplay by visiting Ranger 1 Presents: Virtual Realms.
Previous page123Next page