Jump to content


Founders [premium]
  • Content Count

  • Joined

  • Last visited

Posts posted by Dave

  1. On 20/04/2018 at 5:50 PM, ZachariasX said:

    maybe your wingman was in fact inside the cloud, you jus thought that the vloud bitmap looked farer back.


    We don't have clouds as volumes in the sim. There is no way to get near them, they are just scaled cardboard in front of you. I guess in VR this spacial artifact is even more prominent than on a monitor. Still, it's working as intended.

    I have tested this to death. I am always in constant voice comms with my wingman. This issue occurs when he is between me and a cloud. This is a bug. Unless bugs are intended it is not working as intended. If this doesn't happen to you I'm happy for you. It does happen to me - all the time - and I'm growing tired of other people proclaiming there is no problem just because it doesn't affect them. Regardless of whether the clouds are volumetric or billboarded, the game clearly tracks whether or not a player is "inside" cloud, and it is possible to transition from outside cloud to inside cloud consistently when approaching a cloud from the same side. Therefore the game seems to consistently model a cloud being on one side or another of a player. Similarly another aircraft when viewed from your own is either in front of the cloud, or obscured by it. If a cloud and a second player are both to one side of player one and player two sees the same cloud to the same side of their aircraft as the first, then player two is between the cloud and player one. If this relation does not hold (away from boundary conditions ) we have either a bug or a major design flaw.

    • Upvote 3

  2. On 02/05/2018 at 2:56 AM, =BMAD=kirumovka said:

    And yes about using VPN as you said, it can reduce latency if it is located in a place like the middle east. I'm not that dumb to accept logic like connecting VPN located in antarctica and expect it to reduce my latency for connecting to sydney from perth. However it is apparently possible to redirect routing through vpn that isn't at a deliberate middle of the intended route location like india, but rather having one at the destination such as in Paris. I'm not sure how that works however, so maybe VPNs do have some control over chosen routing.

    You aren't understanding - your VPN terminator simply being in the Middle East doesn't reduce latency - it just doesn't make it impossible like a terminator in Antarctica would. Any VPN termination point off the shortest network path is only going to add latency in addition to the additional latency inherent in encapsulating traffic - which is what a VPN does. People use VPNs, and pay the attendant cost of latency, for secrecy or to alter their apparent location. VPNs do not - except in some very exceptional circumstances - reduce latency, simply because the packets still have to traverse all the same network segments - just encapsulated.

    I am sure - designing and maintaining these networks has been my day job for several years. 


    10 hours ago, 216th_Jordan said:


    Wifi will in almost all cases produce packet loss and sometimes high jitter or connection interferences lasting multiple hundreds of milliseconds, thats why with wifi you get a lot of disconnects in Il-2 MP.


    But as you habe ethernet its rather bad routing by the ISPs.

    The WiFi advice is correct. Do not use it for timing-sensitive applications.


    However, with respect to his particular problem, its not bad routing - its physics.

    You can't beat the speed of light, and the speed of light in glass is about 200,000km/s. Either way you go from Australia, WOL is about half a great circle away. But fibre paths don't follow great circles so the fibre path is actually about 1.5 times as long. Thats more than 30,000km - 60,000km for a round trip (which is what ping reports). Thats approximately 300ms - and that is with a single fibre and no switch gear, routers, firewalls or shitty copper tails or radio links over last-mile connections. Then you have inevitable packet loss due to collisions, congestion, wave cancellation, state table overflows, etc and the consequent retransmission that is just part of how the Internet works. That can add between 50 and 100ms as your packets transit tens or hundreds of intermediary networks and their associated switching and routing equipment. Packet handling in routers adds latency - and there are at least 20 between you and WOL (only some are detected by ping - most are "bump-in-the-wire" devices invisible to basic topology mapping tools).


    Then there is quantisation. To maximise aggregate throughput, networking equipment typically waits to fill a buffer (for the next transport layer network frame) before forwarding. If the MAC frames for a given link are large (to increase total bandwidth) and your packets are tiny then the delay to fill frames can become significant. You can naturally configure this behaviour for the links you manage but it isn't automatic and takes into account the general usage pattern of the link.


    Last but not least (actually not even last but I don't have time for more) there is processing overhead at the source and destination. You may have noticed your ping increases - sometimes dramatically - when the server has more players. This is because the server is not able to process network traffic at anywhere near line-rate and the added load of maintaining state for the additional players makes it slower to produce, consume and otherwise handle network traffic.


    The only way we antipodeans are going to have pings below 350 would be for a popular server to be hosted in the US. I even deployed one - making a personal compromise of 150ms over an AU hosted server - but noone would leave WOL.


    3 hours ago, 307_Tomcat said:

    BTW long ago I was using rerouting proxies to reduce my ping while playing at foreign servers - sometimes it was reduced by 10-20 ms :-)

    This can help, but you need to find a location for your proxy that is close to the network shortest path and which lies on a route which bypasses a problematic link or device (ie has a destination IP that will not be routed over the same slow link anyway). The proxy will at least avoid encapsulation overhead. The dynamic nature of routing rules outside your control will often defeat this over time and the gains are minimal anyway - WOL ping fluctuates by more than 30 ms from one minute to the next.

    • Like 1
    • Upvote 2

  3. On 29/04/2018 at 11:22 PM, =BMAD=kirumovka said:

    It will if it re-routes through the middle east to access the european hub rather than well, going around the earth.

    The route traversed is not determined by a VPN. It is decided by backbone providers' border gateway and autonomous system routing rules. These take into account peering relationships between providers, cost, expected and measured latency, link state, link utilisation, congestion, cost and capacity between peering networks. There isn't even a single consistent path taken by packets traversing the Internet, due to the factors above and the underlying design of the Internet to be fault tolerant.

    Unless your VPN provider has built a shadow internet or owns significantly (in terms of Internet scale) large AS's they have little control over the flow of traffic beyond their own directly controlled networks in a way that could shorten the communication path relative to that taken by unencapsulated IP traffic.

    VPNs almost never reduce latency (it isn't impossible though) - and they always add overhead that increases it. They may also encapsulate traffic in ways that defeat a game's mechanisms to reduce latency - eg by tunnelling UDP inside SSTP. 


    One way the path could be shortened physically is if your VPN tunnel connected from Perth to say Dubai or somewhere in India. But AS routing rules are designed in the main to minimise the transit time between origin and destination addresses. There is a reason the Perth to Indonesia and Middle East links are avoided - they are the oldest cables out of Australia with very low capacities and the other ends are in countries with shithouse network connectivity.

  4. On 19/04/2018 at 12:43 AM, Space_Ghost said:

    No, it pretty obvious that wasn't what he was saying. They're two separate non-issues.

    Not on board with all the complaining but I do think marginalising an issue you don't personally have a non-existent is kinda selfish. For me in VR, when my wingman flies between me and a cloud he completely disappears. This is a problem.

    • Upvote 2

  5. You're welcome.

    And the author did mispell ensure. Even if you aren't a grammar nazi comparison with the stencilled text on the real aircraft should confirm this. The difference between ensure and insure is the same today as it was in 1939.

    Having made the effort to excel in English at school I am often irritated by modern corruption of the language and the ubiquitous acceptance of poor spelling and grammar today, even among those who write for a living. So, as if affected by  some literary form of OCD, I am often compelled to highlight examples as they appear. My apologies. ;)

  6. On 02/04/2018 at 11:20 PM, Pict said:


    Nice skin, I always had a preference for VVS Kobras that'd had started out in the RAF temperate scheme...


    just a small detail about the RAF roundels, they were not all the same size, with the upper wing roundel being significantly larger, if you look at the standard "grey" skin in game or in the viewer you can see this, and it is historically correct.


    The tail flash would also need to be blanked out as they were also put on at the Bell factory and any operational markings if the aircraft had been one of the few that flew operationally.


    Here's some detail of what I'm on about




    Most of that looks correct as I skim over it - but it makes me wonder how many other details are wrong when the author couldn't spell "ensure".

    Some potentially useful info here http://www.rafweb.org/Squadrons/Sqn Markings/sqn_codes.htm

  7. On 09/04/2018 at 10:42 AM, Gambit21 said:


    GPU is far more the limiting factor, especially at higher resolutions.

    My CPU hardly notices IL2, bit at 4K my 1080 is running at 80 - 100%.


    Your 1080 can easily be brought to it's knees.

    Not for me. I run out of CPU long before GPU. Specs in my sig. Basically multicore optimisation in BoX is pretty poor to non-existant and hardware innovation moved to massively parallel a few years ago.

    13 hours ago, ShamrockOneFive said:



    Seconded that GPU is going to be the bigger issue.


    I've been watching my performance levels carefully since 3.001 came out to figure out what's going on when my frame rate or perceived game smoothness drops off.


    I've got a Core i5 6600, 8GB DDR4 RAM (with fairly aggressive timings can't remember off the top of my head), GTX 960 2GB, etc. etc. Of the issues that I'm having I see the GPU being worked to about 80-90% most of the time but what really causes issues is the VRAM. That's when it gets choppy. That and my system RAM I'm finding is frequently at 7.9 GB.


    So RAM upgrades are simple and easy and I'll be doing that the next time there's a sale and while GPU upgrades are simple... The price is not so easy.


    Even RAM prices are high. When I bought my system 3 years ago the DDR4 memory sticks were down around the $90 CDN mark. Now its a $150 for the same pair.

    I think the biggest issue for you is that 2GB is not enough VRAM to hold all the required textures on the card, so bandwidth is being wasted pushing textures from RAM to your GPU. I think 8GB is almost a minimum for fluid rendering now with many titles, BoX included. A 1080Ti will be plenty good enough and if you hold off about 3 months the price will drop by about 50%.

  8. Latest info on the next gen cards from Nvidia is:


    - name will be GTX 11x0 with the top end being the 1180 at launch;

    - 12nm process Turing architecture for gaming cards and Ampere for HPC (Nvidia architecture is diverging and specialising);

    - both will use GDDR6 - probably in 8GB and 16GB configurations (based on leaked info from Hynix);

    - should be announced at the conference in Taipei in June;

    - should start shipping in late July;

    - due to the higher production cost of GDDR6 expect about a 20% price hike over the same relative level 10 series cards.

    (I expect higher prices due to all the intolerable bandwagon-chasing crypto-currency noobs)


    I would have bought a Titan V already if I thought it would yield any improvement with BoX, but TBH my 1080Ti isn't being fully utilised as is. The CPU bottleneck is what needs to be resolved.

  9. Is there a plan to make the Multiplayer (Dogfight) browser actually work at all?

    For me the Ping column has never worked - usually displays 0 for most servers. And now sorting by anything other than server name doesn't work either. Given that there are dozens of empty servers these flaws render the multiplayer browser less than useless.

    Given also that the only populated server (during the hours I am typically able to play) is in some distant corner of Faknoeswheristan and kicks players with pings above their threshold it would be nice to know if joining was a waste of time before commencing the usual 5 minute join-fail-join cycle.

    While we're at it could you maybe address the really poor server rejoin experience.

    I understand that hotfixes to 3.001 should and will take priority, but this has recently reached an all time low for user experience and we've been waiting for most of these issues to be addressed since 2014. So it would be nice if improvement of the presently dismal multiplayer browse/join experience was prioritised as soon as 3.00X has stabilised.




    • Thanks 1

  10. La5FN seems a bit fast at altitude, but I love it'll finally have a mirror as well.


    So, the G6 will be a slowly flying brick of 20 Dollar rip-off disappointment. It lacks the interesting modifications it had in reality, various boosts, rockets, 30mm gunpowder, etc.


    Fail. Before engaging in a s***storm on reddit and other forums over it, does anyone know where to write to in order to get a refund on the G6?



    • Upvote 5

  11. Same day as everyone else a new Mита :rofl:   -  I was in Sacramento bought the Calendar for 2018 there and there stands in black and white 20 March is the "First day of Spring".  :happy:

    Meteorological speeking the Frst day of Spring is 1st March :big_boss:


    Having just had to build this into an application I can tell you that there is no global standardisation of seasonal dates - even in Met. 


  12. You have missed out a possibility: that the anecdotes do not match reality.  For instance, P-40 pilots who tried to turn with German planes and failed are far less likely to survive to tell anecdotes about their comparative turning performance.


    This is called survivor bias. So back in the German pilots' mess, there is a successful pilot talking about how he turned inside a P-40.


    Hence the developers correct emphasis on measurable performance.


    I fear that you are going to be even more unhappy if the developers manage to fix the measurable and acknowledged issue with 109 turn times being too low  [edit: too high! Sunday.] 


    All first hand accounts I have read, which include that of a German pilot flying F4s in North Africa against P40Es, are consistent in this regard. The P40 easily out-turned 109s. The German pilots almost always stuck to diving through the circling P40s (RAF tactics of the day called for a defensive circle). According to all accounts, 109s that abandoned their advantage and turned with the P40s were quickly outturned. 

  13. I think that given the fact collector planes are not necessarily tied to a "battle" release and that most planes are available across a wide range of scenarios and time periods, the Erla canopy and maybe even MW50 might make sense as modifications on the G-6, only locked in BoK scenarios. Then the plane would be very useful when BoBp comes out and could be used in that without being sub-par.


    But the development and release of each of the collector planes is tied to a particular battle. This is clear from both the Dev Diary posts and the sales material on the website.

    People shouldn't get too uptight about it though. I expect all the mods that have been mentioned in this thread will ship in BoBp, either by default or as DLC. Why wouldn't they - its easy development and guaranteed money.

  14. November 1943 is close enough in my book. I think it should be there.


    It should be remembered that what we have bought is Battle of Kuban not Battle of 1943.


    This isn't directed at you specifically - your post simply mentions dates so it prompted this thought.

    I don't think it is bad to want certain mods and options and I agree that mission makers have ample freedom to constrain types and field modifications ot be historically accurate as they see fit, but people need to stop short of demanding that things be included in BOK because they existed somewhere in the world during the time of the Kuban campaign.


    They may not be anachronistic from a manufacturing or even availability standpoint, but many of the options people have said emphatically must be included simply weren't used on the Eastern Front at this time. There are plenty of 109 mods we shouldn't get in this particular title in the series in the same way you shouldn't see Mustangs, Mk.IX Spitfires, Corsairs etc. They will undoubtedly come in titles that target the appropriate theatres.

    • Upvote 1

  15. But if you have a few data points and have worked a least-squares fit then I guess that would answer it - for the game implementation at least. It would be hard, experimentally to account for ram air effect at various airspeeds without quite a lot of messing about. Might be an interesting exercise.

  16. Done a few checks and I think this is what is exactly happening with P40: 100% throttle is 70" MP, not 56" as I thought. After 10" on MP gauge (@ 30% throttle), to gain extra 5" you have to open throttle by 5%, too. At 50" MP throttle is around 80% - extrapolating to 100% MP should rise to... 70" MP. Max emergency power as stated in specification (at 56" MP) is approx 85% throttle then.


    Looks like things are alright after all. We just have an option to over-boost the over-boost, blowing up the engine shortly, as it probably should.


    I think your extrapolation, if linear, is suspect.

    I don't think mass flow is linear with throttle butterfly angle, either with or without forced induction.

  • Create New...