devtrackers.gg
about 5 years
ago -
Hunt: Showdown
-
Direct link
Transcript (by Youtube)
| 3s | [Music] |
|---|---|
| 22s | [Music] |
| 29s | [Music] |
| 37s | [Music] |
| 57s | [Music] |
| 65s | [Music] |
| 81s | [Music] |
| 97s | [Music] |
| 109s | [Music] |
| 139s | [Music] |
| 156s | [Music] |
| 186s | [Music] |
| 203s | [Music] |
| 222s | [Music] |
| 229s | [Music] |
| 238s | [Music] |
| 257s | [Music] |
| 265s | [Music] |
| 281s | [Music] |
| 297s | [Music] |
| 309s | [Music] |
| 339s | [Music] |
| 357s | [Music] |
| 386s | [Music] |
| 404s | [Music] |
| 423s | [Music] |
| 430s | [Music] |
| 438s | [Music] |
| 458s | [Music] |
| 465s | [Music] |
| 471s | yes it's great |
| 473s | [Music] |
| 475s | hello everyone |
| 477s | wow I hope you guys can clearly hear us |
| 480s | oh yeah it's fine so no you don't have |
| 488s | to yell we have a mic capture everything |
| 491s | so today we're gonna talk about update |
| 494s | 4.2 that literatures on life some of you |
| 498s | already played it on a desk server and |
| 499s | first we're gonna have our two guests |
| 502s | I mean introduce yourself I am Chris you |
| 506s | may remember me from previous dreams I'm |
| 508s | the design director at some of you might |
| 510s | know me as Barney in the channels I |
| 512s | think we've revealed that before right |
| 514s | that's good yeah for a while hey my name |
| 519s | is Paolo and then principle game |
| 520s | programmer it's a hunt I don't have any |
| 524s | or I'm in this course but I'm too shy to |
| 526s | actually get Flair so so we're gonna |
| 533s | first have them talk a little bit about |
| 535s | what they've done for the badge and what |
| 537s | they're still doing at cetera and then |
| 540s | we're gonna go over some survey results |
| 541s | so some of you if they did thank you |
| 544s | because that's very helpful for us fill |
| 546s | them to survey that we're adding during |
| 547s | the DES server there will be another |
| 549s | survey later and then we're gonna go |
| 553s | through some questions so yeah so 4.2 so |
| 559s | basically we should be going live today |
| 561s | if it's not already live I think yeah ok |
| 567s | that's good so what you guys are gonna |
| 569s | be getting in the patch I don't know how |
| 571s | many of you have been using the test |
| 573s | server I guess we got quite a lot of |
| 575s | responses this time on the surveys which |
| 577s | means a lot of people were on the test |
| 578s | server which is really encouraging |
| 579s | because that helps us helps us hippo so |
| 583s | 4.2 is going to contain mostly |
| 585s | two areas that we focus on the first |
| 587s | area it's going to be smoothness of |
| 589s | gameplay so we spent a lot of time |
| 592s | working on reducing stalls reducing |
| 594s | stutters frame drops making sure that |
| 597s | your you know |
| 598s | overall experience was nice and clean |
| 599s | and nice yeah a lot of people required |
| 601s | like said they had a waste mother yeah I |
| 604s | know we've mentioned that in the past I |
| 606s | think we've done a couple of performance |
| 608s | updates and stuff where we've kind of |
| 609s | said we're gonna focus on stores first |
| 611s | and then bring the overall frame rate up |
| 613s | and I think that we needed another round |
| 616s | on that it was clear that we hadn't done |
| 618s | enough we had a lot of feedback from |
| 620s | people that you know they wanted to see |
| 622s | much more improvements to these |
| 624s | framedrops and we had so many sort of |
| 628s | even one-on-one discussions with the |
| 629s | community about yeah yeah we've had a |
| 632s | special channel to this Creek but in |
| 634s | general also the fact that the stutters |
| 637s | are very diverse like the first reasons |
| 638s | we have them for different reasons |
| 640s | yeah my boss is putting it very mildly I |
| 642s | said we found and by we I actually mean |
| 646s | I found this guy for your smooth frame |
| 649s | rates by the way content of the patch |
| 656s | and we found quite a lot of stuff and |
| 658s | it's all been bundled into 5.2 which |
| 659s | should be available for you today place |
| 662s | for people they say yes there's a lot of |
| 672s | stuff that went into the patch that was |
| 674s | related to started stores framedrops |
| 675s | other things people on lower end |
| 677s | hardware should see some improvements to |
| 678s | their performance just generally as well |
| 681s | you should see a bit of an improvement |
| 683s | in memory usage there's some more |
| 685s | general CPU optimizations and things a |
| 689s | lot of people on the test servers |
| 690s | actually mentioned that they were able |
| 691s | to sort of bump this back up a notch I |
| 693s | mean prove the detail up and stuff like |
| 695s | that so hopefully a good patch we wait |
| 698s | on feedback the test server feedback was |
| 700s | so far pretty good yeah so hopefully you |
| 704s | guys experience yeah nice so yeah what |
| 708s | have you guys been doing so far in |
| 712s | understandable terms as much as possible |
| 716s | [Laughter] |
| 718s | yes the card is part of everything |
| 722s | yeah do you work in performance right |
| 723s | easy yeah yeah actually |
| 728s | let me check our schedule of things what |
| 730s | we I mean there's this question there's |
| 732s | many different improvements like mostly |
| 735s | during this patch we were concentrating |
| 736s | on CPU performance as cpu performance |
| 740s | was like one of the main bottlenecks we |
| 744s | were seeing on the low spec on the |
| 748s | command expects high specs are not that |
| 750s | effective because usually they can pump |
| 751s | up like people have much better CPUs |
| 753s | more cores more memory for them as usual |
| 756s | okay so these people potentially will be |
| 759s | not experience occurrs better like as |
| 762s | much of improvement and but even even |
| 764s | for them the experience hopefully should |
| 768s | get smoother like let's start to me like |
| 771s | more I think we'll just go with the two |
| 773s | words performance and smooth just |
| 782s | mention goes I know a lot of people |
| 784s | sometimes get confused so they're like |
| 786s | why are they working working on the CPU |
| 788s | when it's not using the full GPO GPU and |
| 793s | the reason that is because it's good at |
| 796s | the moment still heavier on the CPU |
| 797s | because it's not using the full rather I |
| 800s | mean it's it's PC so everyone has |
| 803s | completely different spec like pcs like |
| 806s | we have to have telemetry and scene just |
| 808s | the whole matrix of all the possible |
| 810s | configurations it's mind-boggling people |
| 812s | play the super finance computers |
| 814s | crippled by some way one bad part all |
| 817s | right other other people play on like |
| 820s | laptops which are not even like gaming |
| 822s | laptops and and still expect very good |
| 825s | experience and then some people actually |
| 826s | have just decent pcs maybe you like just |
| 829s | from four years ago so what greatest |
| 831s | thing yeah but I think all of the ones |
| 835s | that we see in what we do is we notice |
| 836s | that there's the vast majority of people |
| 839s | have hardware that should very easily be |
| 841s | able to present the kind of performance |
| 842s | that they would expect so you know |
| 843s | getting 60 frames per second at 1080p |
| 846s | could not be a big deal yeah in no way |
| 848s | I'm saying that oh it's user |
| 850s | it's user problem that they not getting |
| 851s | that better pcs not at all actually but |
| 854s | my point is that even such vast |
| 857s | differences in the in the cardboard |
| 859s | people use we try to at least find the |
| 862s | like the biggest point of impact where |
| 864s | they impact will be the like but where |
| 867s | more people will benefit from it |
| 868s | yes so we see that the CPU profile like |
| 871s | the CPU cost of the game on is one of |
| 874s | the worst offenders for most of the |
| 876s | people besides we try to just increase |
| 878s | the overall destruction of people how |
| 880s | can they receive like and most people |
| 882s | have like average too low computers so |
| 887s | if we can give them better user |
| 889s | experience so they can actually play |
| 891s | well hit their shots without start |
| 893s | suffering from this better yeah and I |
| 895s | think it's also important to remember as |
| 896s | well that if you are limited by your |
| 898s | processor or you limited by you know |
| 900s | things like the clock speed and other |
| 902s | stuff on there is that it will limit |
| 904s | everything it's it means that your CV or |
| 906s | GPU can't work as hard and you can't get |
| 908s | the full the attention of the GPU so we |
| 910s | get asked a lot is you know I have a |
| 911s | 1080 Ti and maybe an i5 4790 Intel or |
| 916s | something like that and why am I not |
| 918s | getting 60 frames or 70 friends or |
| 919s | smooth experience in general and the |
| 921s | reason is because the CPU is yeah if I |
| 924s | do that or if it might be like if |
| 925s | everybody would CPU and GPU but then I'm |
| 927s | good like a gigabytes of memory right |
| 929s | now the game is barley 15 cannon |
| 931s | gigabytes so it means you've costly |
| 933s | gonna wanna go in the swap file so you |
| 935s | kinda limited by your hard drive through |
| 938s | which is kind of surprising for many |
| 940s | people hmm |
| 941s | it's really hard to tell what is the |
| 943s | case |
| 944s | I mean what we find is that we get asked |
| 949s | by a lot of time a lot of times is why |
| 950s | does performance take so long to to |
| 952s | address and why is this been going for |
| 953s | you know for you know several months |
| 955s | eight nine months now since Hunt has |
| 956s | been released and the truth of the |
| 958s | matter is is that you have to plan for |
| 961s | different specs different combinations |
| 964s | of Hardware different driver types |
| 966s | different platforms different windows |
| 967s | types different settings and all of |
| 970s | those can combine into you know we're |
| 971s | talking hundreds of thousands of |
| 973s | combinations of things that could happen |
| 975s | and we have to go in and we have to test |
| 977s | all that stuff and it takes a lot of |
| 978s | time |
| 979s | you're jumping back quickly to the |
| 980s | original |
| 981s | question about the performance side of |
| 983s | things and what was done we found a |
| 986s | couple of very interesting things yeah |
| 987s | so didn't mean we can go like more |
| 990s | specifically so like one of the |
| 995s | offenders which we found and it was |
| 996s | actually quite curious one that we found |
| 999s | that so there was one piece of like work |
| 1003s | which was done for the rendering and it |
| 1005s | was it's actually a very CPU heavy job |
| 1008s | and we actually some time ago already |
| 1012s | identified that this is beef heavy and |
| 1013s | we don't really need it to happen on |
| 1015s | client it's like car engine is ready we |
| 1018s | change it from a thing that happened in |
| 1020s | runtime to something that was pre |
| 1021s | calculated yeah |
| 1022s | engine is generally like ready like |
| 1024s | everything has to be the runtime |
| 1026s | especially for the editor experience for |
| 1028s | like our normal development cycle for |
| 1030s | the end user not everything has to be |
| 1032s | runtime we can recalculate some things |
| 1033s | what element is condensing the main so |
| 1036s | we implemented this it was great we |
| 1038s | measure the performance improvements was |
| 1040s | quite a while ago and then we kind of |
| 1042s | forgot about this issue and during this |
| 1045s | cycle profiling I find out |
| 1047s | hey this is wrong why he never saw this |
| 1051s | so I mean yeah the fix was just yeah |
| 1054s | make sure that it's actually properly |
| 1055s | streamed and not happening in the wrong |
| 1058s | time the more important in this case is |
| 1062s | so the reason was that we have multiple |
| 1064s | configurations of the game there's one |
| 1066s | configuration which is a release which |
| 1067s | is affected about to ship to players it |
| 1069s | doesn't have almost no like checks no in |
| 1073s | no introspection is what is happening is |
| 1075s | that again and then there is a profile |
| 1076s | builds which has much more debug |
| 1078s | functionality introspection call |
| 1079s | performance works like we can have |
| 1081s | markers to see what exactly is happening |
| 1082s | at which point we'll frame program so |
| 1092s | basically we have a couple of different |
| 1094s | modes that we run the game in one is |
| 1096s | called profile profile mode is as we're |
| 1098s | working on the game there's debug stuff |
| 1101s | in and then slows the game down yes then |
| 1104s | we have something called release mode |
| 1105s | and release mode is where you compile |
| 1107s | the game and you take all of the crap |
| 1109s | out that you don't need that you don't |
| 1112s | need to debug and have all this extra |
| 1113s | information |
| 1114s | but your game gets sort of less a 15% |
| 1117s | faster yes and that's what you guys get |
| 1119s | to play so we have a debug version and |
| 1121s | you guys have a live version and the bug |
| 1123s | we found with the stools was on the |
| 1126s | light part of it yeah yeah yeah so this |
| 1133s | bus was written just in this version |
| 1135s | just in this configuration technically |
| 1137s | yeah because this one is actually should |
| 1139s | be faster and the reason why we were not |
| 1142s | seeing this is specifically because it |
| 1143s | was at least because the release doesn't |
| 1145s | have this additional markers additional |
| 1147s | introspection is what's happening can |
| 1149s | again so we found it just because this |
| 1153s | time we started using some additional |
| 1155s | new tools which are not inherent to |
| 1157s | CryEngine but rather external tools |
| 1159s | developer Microsoft like Microsoft peeks |
| 1162s | if anyone interested and yeah |
| 1165s | so this external profiling tools allowed |
| 1168s | us to support something which we were |
| 1169s | previously was not able to find out what |
| 1171s | was also quite interesting is these |
| 1173s | these two issues may or may not there |
| 1176s | was two different cases some would cause |
| 1178s | your framerate to drop and you would |
| 1181s | stutter so you'd be moving around like a |
| 1183s | lot of you have probably come across the |
| 1185s | experience where you're moving around |
| 1186s | the level and you're trying to track |
| 1188s | something down or go to the objective or |
| 1189s | whatever and you feel this pause or like |
| 1192s | this stall for a few seconds or whatever |
| 1194s | as you're moving around and it's super |
| 1196s | annoying or you know you're raising your |
| 1198s | gun up to shoot an AI or another player |
| 1200s | whatever and he's so annoying and that's |
| 1206s | one of the ones that we fixed and then |
| 1208s | and that basically will cause you're |
| 1209s | actually will cause your framerate to |
| 1211s | peak downwards so you'll get a drop and |
| 1213s | there was another issue that we found |
| 1215s | which meant that you're you're you have |
| 1218s | those stutters installs and you would |
| 1220s | feel this kind of movement that felt |
| 1222s | very very strange and a bit you know |
| 1224s | disconnected somehow but your framerate |
| 1226s | didn't change people with high-end |
| 1229s | hardware like people with like eight |
| 1230s | core machines and really good machines |
| 1232s | would feel these weird like movements |
| 1235s | and these weird kind of glitches as they |
| 1236s | were moving around but they had nothing |
| 1238s | to do with their performance they would |
| 1239s | be getting nice frame rates but the |
| 1241s | movement felt like shit well in this |
| 1242s | performance issue it's just it was |
| 1244s | affecting the like main threat |
| 1247s | performance it was a physics test it was |
| 1249s | in issue is the physics system and what |
| 1251s | does that mean so we split our like game |
| 1255s | simulation and multiple threats of |
| 1258s | execution because nowadays the computers |
| 1261s | all have a face for course so we can |
| 1263s | theoretically in the perfect world do |
| 1265s | work four times |
| 1267s | like yeah not faster but at least what |
| 1269s | types of work in the single unit of time |
| 1271s | yeah yeah so the good news is but so our |
| 1281s | physics system is running in parallel |
| 1282s | with the normal game simulation and |
| 1285s | there were issues when the physics |
| 1288s | system was overloaded and it was not |
| 1291s | properly synchronized with the game so |
| 1292s | then for the game it was looking like I |
| 1295s | requested a movement like move me like |
| 1297s | with this velocity next frame it expects |
| 1299s | to have result because the thesis thread |
| 1301s | was not catching up with the main threat |
| 1303s | so maybe optimized the main thread and |
| 1306s | it begins shorter than the physics it |
| 1308s | the physics was not ready to give |
| 1311s | results in result the entity was not |
| 1313s | moving so you kind of your frame rate is |
| 1315s | great your entity is not moving and this |
| 1318s | kind of starter is like the good |
| 1319s | telltale kind of stutter is you can |
| 1321s | update yourself like the moment you |
| 1327s | start moving you start stuttering or |
| 1329s | what was the other entities like you |
| 1332s | kind of like it to be and I think that's |
| 1334s | actually something I don't think we put |
| 1336s | on the patch notes is it's very likely |
| 1338s | that that has also been fixed with the |
| 1340s | fix for this thing so if you guys had |
| 1342s | experienced where you were you know |
| 1344s | playing the game and the AI was sort of |
| 1345s | being a bit twitchy and a bit weird that |
| 1347s | wasn't a new direction in the animation |
| 1355s | just to add to the problems it was just |
| 1360s | one case of a starter of course there is |
| 1362s | another case of starter and generally |
| 1364s | the game is overloaded and yeah your |
| 1367s | friend is actually like the whole frame |
| 1368s | is missing but this is B so going on |
| 1372s | like we saw instigating yeah |
| 1374s | which is but this one with the physics |
| 1376s | hopefully we fixed it we still gonna |
| 1379s | work more on the physics system because |
| 1381s | we do see that it's doing more work than |
| 1383s | supposed to do yeah and there's multiple |
| 1385s | ways like there's some bugs there's some |
| 1387s | like improvement general there's just |
| 1390s | like asset improvements there's too many |
| 1392s | things on the level which are |
| 1393s | potentially too expensive or don't |
| 1395s | activate themselves well what are the |
| 1397s | strategies we normally do like when you |
| 1398s | normally build a game you build a lot of |
| 1400s | content and at the end you optimize the |
| 1402s | hell out of it and then it gets better |
| 1403s | but with a game I can't because we're |
| 1406s | doing early access and we're doing so |
| 1407s | causes we add content and then we have |
| 1409s | to go back and optimize and content |
| 1411s | optimizing one strategy that we do is |
| 1414s | that I mean this is again why you'll get |
| 1416s | some more optimizations in a later date |
| 1419s | hopefully in 5.0 or whenever we actually |
| 1423s | do just blanket optimizations where we |
| 1425s | don't necessarily target specific things |
| 1427s | we just say you're using too much let's |
| 1430s | try to back it off let's try to reduce |
| 1431s | stuff so that you reduce the pressure on |
| 1434s | the physic system or the animation |
| 1435s | system or the CPU or whatever we're |
| 1437s | doing you know it doesn't hurt just to |
| 1440s | try to optimize things where you can |
| 1442s | yeah then I'm actually curious because |
| 1445s | that is also something that some people |
| 1447s | asked like if we are working on |
| 1450s | performance what does the rest of the |
| 1452s | team is doing you know are they also you |
| 1454s | know and they make new bosses but |
| 1458s | figuring have to increase the poly count |
| 1461s | yes so there are more things I mean you |
| 1464s | can probably explain better than me like |
| 1467s | for example what are some of the teams |
| 1468s | doing also for performance so what we |
| 1473s | mentioned earlier again we will be |
| 1475s | putting out more optimizations as we go |
| 1477s | so this is not the only one we're gonna |
| 1479s | be shipping what we tended to do is |
| 1483s | because people had performance issues |
| 1485s | over Christmas and the 300 to 400 issue |
| 1487s | that we had we wanted to make sure that |
| 1489s | we fix that as soon as we could and the |
| 1491s | community was very vocal and asking for |
| 1493s | it and they absolutely should be |
| 1503s | you |
| 1545s | can you hear us now again yeah I think |
| 1548s | it's maybe the cables I'm good |
| 1550s | long all right safe to say apples in |
| 1552s | chat if you can hear us again |
| 1554s | heckles no no the lane didn't say apples |
| 1562s | like ended apples okay there's no many |
| 1565s | apples well we were saying what the |
| 1570s | others are doing okay so yes so |
| 1573s | basically what we wanted to do is there |
| 1576s | was a performance drop between 3.0 and |
| 1578s | 4.0 which was not good that was on our |
| 1581s | side we messed that up and we needed to |
| 1584s | fix it quickly which is why we wanted to |
| 1586s | make sure that you guys had patched |
| 1588s | sooner rather than later so this is why |
| 1589s | you're getting patch 4.2 which hopefully |
| 1592s | fixes those issues and makes things |
| 1594s | smoother as well and we will be |
| 1597s | continuing to develop more patches |
| 1598s | you'll actually get another one that |
| 1600s | will update content for 5.0 and the rest |
| 1603s | of the team have been working a lot on |
| 1605s | sside optimizations and you know things |
| 1607s | like improving the level reducing the |
| 1610s | amount of poly counts entities and |
| 1612s | making sure all the objects have the |
| 1614s | right lofts and stuff like that so yeah |
| 1616s | another thing which |
| 1625s | you |
| 1633s | you |
| 1669s | it's okay I don't know what's wrong |
| 1672s | thank you is it back again yeah oh good |
| 1677s | now say another word bananas bananas |
| 1680s | bananas you were saying I'm saying you |
| 1695s | were saying something I was saying that |
| 1697s | the memory consumption may improve the |
| 1699s | performance for some users especially if |
| 1701s | they have limited amount of actual |
| 1703s | physical memory like remember yeah of |
| 1705s | course for rich people who have 30 |
| 1707s | gigabytes of memory maybe not for them |
| 1710s | but yeah this is what the team is |
| 1714s | working on there was perfect question |
| 1715s | why the optimized assets right away yeah |
| 1720s | yeah it's a good question the reason |
| 1724s | being is it would take us a lot longer |
| 1725s | to give you a patch if we put all the |
| 1728s | assets in we have to do a lot of |
| 1730s | different testing it takes a lot longer |
| 1732s | and you would get a delay of several |
| 1734s | weeks before you got your patch and we |
| 1737s | already know that people are very sad |
| 1738s | that they don't have the boss yet before |
| 1740s | missus going first right now so it's |
| 1742s | been already a little bit late I'm |
| 1746s | actually curious so what do we do to |
| 1749s | prepare for like a batch with the DES |
| 1750s | server and the live server right now we |
| 1754s | have to test over a lot of people played |
| 1757s | it on something that's quite important |
| 1759s | to keep in mind and it's actually |
| 1761s | something we kind of wanted to remind |
| 1762s | people about when it comes to the test |
| 1763s | server and performance which is the test |
| 1767s | server performance is only as good as |
| 1769s | the number of players that are in the |
| 1770s | session if you play on the test server |
| 1773s | and it's a low population this can |
| 1775s | affect performance because the more |
| 1777s | players that are involved with this it |
| 1780s | will activate more entities what AI will |
| 1782s | be active and your frame rate can change |
| 1784s | depending on how many players are in the |
| 1787s | session with you so if you go onto the |
| 1789s | test server and you immediately find |
| 1791s | your frame rate sort of doubled I would |
| 1794s | say try it a few more times see if you |
| 1795s | can get into sessions if you have more |
| 1797s | players it would always be good and we |
| 1800s | need to make sure that you actually |
| 1801s | with eight to ten other players that's |
| 1804s | really important |
| 1805s | typically we see at least internally |
| 1807s | maybe upwards of 8 to 10 frames |
| 1809s | difference between a populated server |
| 1811s | and a non populated server on a medium |
| 1814s | spec PC so yeah I mean to the point we |
| 1816s | like we do performance cultures and we |
| 1819s | don't even do them this is one player |
| 1821s | our QA team when I'm doing performance |
| 1823s | cultures they all play 10 / 10 people |
| 1826s | not all of them capturing the data for |
| 1829s | our analysis but it's just a few usually |
| 1832s | pieces like the target hardware but |
| 1834s | everyone else just creating the traffic |
| 1835s | creating action yeah I mean if some of |
| 1838s | you guys might have seen a post that we |
| 1839s | hide up on or social media where you saw |
| 1841s | that we were testing who have done the |
| 1843s | test rooms and we all have different |
| 1845s | specs so we can see how they perform on |
| 1847s | each of those computers and then all |
| 1849s | playing together yeah another thing I |
| 1852s | think it's also quite good to mention is |
| 1854s | we had two questions before about what |
| 1856s | kind of NASA based hard way to develop |
| 1860s | the games on and I can tell you right |
| 1861s | now is a hugely mixed variety of |
| 1864s | hardware that we actually develop on my |
| 1867s | machine I use this is the machine I use |
| 1874s | is I think it's a 1060 and I use a Xeon |
| 1879s | e3 because if we have to generate stuff |
| 1883s | for level design we have to use a lot of |
| 1884s | memory and the CPU is quite good |
| 1887s | I don't remember specifically something |
| 1890s | with an eight-course yeah yeah listen I |
| 1906s | you see smug she's not even programming |
| 1909s | the operators rap I don't have a great |
| 1915s | PC and then I just respect myself but in |
| 1920s | the play test rooms and the other |
| 1922s | development machines so next to my PC I |
| 1924s | also have a 660ti with a semi I actually |
| 1928s | have 60 Ti so my my CPU is better and I |
| 1931s | have more |
| 1932s | because I need to compile the game and |
| 1944s | we do have like a really big variety in |
| 1947s | our test test areas that we do when we |
| 1949s | check for performance and it ranges like |
| 1952s | I think one of the machines is like i7 |
| 1954s | 2600k hi-fi's in there we have AMD's in |
| 1958s | there we have different rising pipes |
| 1960s | we've got six 60s all the way up to 10 |
| 1961s | 80s I don't think we have anything more |
| 1964s | powerful entities in there right now we |
| 1966s | don't have 2016 or 2017 or anything like |
| 1968s | this yes yeah and I think the most |
| 1971s | powerful CPU we have in there is an 8700 |
| 1974s | K and that goes all the way down to you |
| 1976s | know so we have a huge array of hardware |
| 1981s | and not all of the Deb machines are |
| 1984s | super NASA mega machines just the ones |
| 1987s | that I actually just realized that I |
| 1991s | brought the other piece of paper over |
| 1993s | there so do you have to run over there |
| 1995s | no you can just say what do you want to |
| 2004s | talk about later what we plan to do for |
| 2005s | a 5.0 or your well yeah and no fuck I'm |
| 2011s | always going to be optimizations we |
| 2013s | can't be too more specific at the moment |
| 2014s | just because there's so many there's so |
| 2015s | many that of asset side related there |
| 2019s | may be some stuff in there that will |
| 2020s | help for CPU the things that we've done |
| 2022s | in there but I mean it's work in |
| 2024s | progress there's like some work is just |
| 2027s | general optimizing game or yeah are we |
| 2030s | looking into putting more the systems or |
| 2033s | notice the UI systems maybe into like |
| 2037s | parallel processing one so using |
| 2039s | different threads like Java file or |
| 2042s | update just to release the main turret |
| 2045s | and putting everything more tightly in |
| 2048s | the multiple course yeah but it depends |
| 2051s | what I mean this things take time and |
| 2054s | it's not sure that it will fit into like |
| 2056s | next performance by should thank you so |
| 2058s | it really depends which one |
| 2060s | and I'm I do think also what will make a |
| 2064s | difference is what we see happening on |
| 2066s | life when the patch comes out so if we |
| 2068s | see it on live and we still see people |
| 2070s | with status and that's going to be |
| 2073s | something we will drop everything for |
| 2074s | and we will start looking at those again |
| 2075s | because those are the killer when it |
| 2077s | comes to game play again like if you |
| 2080s | have 60 frames your fragment is just |
| 2082s | going up and down and installing it |
| 2083s | doesn't matter what matters is that you |
| 2085s | have a smooth consistent experience yeah |
| 2087s | first and then we can bring the frame |
| 2089s | rate up to something better |
| 2090s | yeah so then that's the plan okay add |
| 2094s | that I'm gonna disappear for a little |
| 2096s | bit well they keep talking about the |
| 2098s | server results cuz everyone like |
| 2102s | statistics in what we thought we would |
| 2105s | do actually was what we've never done is |
| 2108s | we've done a lot of surveys for |
| 2109s | performance and other things and we |
| 2110s | actually I don't think before remember |
| 2112s | we've ever shared any of the results |
| 2114s | with you guys and I think that would be |
| 2116s | quite nice just to look at the kind of |
| 2119s | things that people were getting and the |
| 2120s | kind of feedback that they had and all |
| 2123s | of the test results you're going to be |
| 2124s | seeing are from the test server yeah |
| 2126s | this is only the results from the test |
| 2128s | so we will run another set of surveys |
| 2134s | for life so when the patch is live which |
| 2137s | it is right now we will run an |
| 2139s | additional survey with this as well all |
| 2151s | of them yeah |
| 2152s | just pretty fast I didn't you did a |
| 2153s | nineties data burst sure yeah yeah I |
| 2166s | mean we don't need to go into too much |
| 2168s | detail with these but we just thought |
| 2169s | we'd share the results for for the |
| 2172s | community since you guys have been |
| 2173s | giving us feedback on this I mean this |
| 2175s | is data that has come from you so I |
| 2177s | think it would be cool if you have a |
| 2178s | look at it with us we're not going to |
| 2180s | show you the results for every single |
| 2181s | question just because it will take you |
| 2183s | know a thousand years and we'd rather |
| 2185s | just get down to the you know the Q&A |
| 2187s | and stuff like |
| 2188s | but yeah so the first question we wanted |
| 2190s | to show you was like this is basically |
| 2192s | from their survey results of what kind |
| 2194s | of resolution you go to running out and |
| 2196s | you can see that not only from the side |
| 2198s | does it look like a middle finger but |
| 2202s | you know most people obviously running |
| 2203s | in native 1080p very few people are |
| 2207s | running in 720 or you know 900 P but a |
| 2210s | lot of people in 10 18 a lot of people |
| 2212s | in 1440 I think the total uses in this |
| 2216s | poll is about 1100 I think so I just |
| 2227s | think it's very interesting to see so |
| 2228s | you know when people ask us what we're |
| 2230s | targeting in terms of performance like |
| 2233s | one of the questions that were asked of |
| 2234s | us was like what would we expect from |
| 2236s | you know a recommend a spec PC for |
| 2239s | performances we're aiming for 60 frames |
| 2241s | at 1080p for multiplayer 60 frames is a |
| 2245s | pretty nice target and most players are |
| 2247s | running in 1080 if you have higher |
| 2250s | resolutions then you know you've better |
| 2252s | hardware and/or to change your system |
| 2254s | specs to compensate for that so the next |
| 2260s | question was the quality settings that |
| 2262s | most players have been using especially |
| 2263s | multi so and no surprise that most |
| 2265s | people are using low what was surprising |
| 2268s | to us is that it's either low or high |
| 2271s | now the data is a bit interesting here |
| 2273s | because of the custom settings so if |
| 2275s | people are making a lot of small changes |
| 2277s | to their settings that would basically |
| 2278s | be flagged as a custom which you know it |
| 2282s | doesn't really fit in any of those |
| 2283s | categories but it's still interesting to |
| 2285s | see that the majority of people 40% of |
| 2287s | the players on forcing everything down |
| 2289s | to the low spec and it's to be expected |
| 2291s | if you want to get your performance on I |
| 2293s | think you want some point try to |
| 2295s | acrylate is this actual telemetry you |
| 2297s | get you come to life |
| 2298s | yeah that would be really interesting to |
| 2303s | see especially after the catch is life |
| 2306s | it would be nice to see whether that |
| 2307s | changes and whether more people can you |
| 2309s | know use the sort of higher higher specs |
| 2313s | in one month so yeah ok next one next |
| 2317s | one vsync |
| 2320s | again not a huge surprise that most |
| 2323s | people have vsync off I guess if you're |
| 2325s | not hitting 60 frames per second it |
| 2328s | doesn't make much sense and probably a |
| 2330s | lot of people's target is 60 yeah yeah |
| 2333s | not too much to say about this but yeah |
| 2336s | it's interesting to see the ratio here |
| 2342s | this is a question we asked people |
| 2343s | whether you will put happy with the |
| 2345s | performance generally in hunt and |
| 2347s | obviously you know the feedback that we |
| 2350s | get is that it's a problem and a lot of |
| 2352s | people have feedback that it's you know |
| 2354s | not where it needs to be but this result |
| 2357s | was somewhat unexpected we had quite a |
| 2359s | lot of people you know 2/3 of people |
| 2360s | said that they were happy with it and |
| 2362s | third said that they would not now the |
| 2365s | tricky thing with this particular part |
| 2367s | of the survey is we need to understand |
| 2369s | whether this is relative to the test |
| 2371s | over or whether this is relative to the |
| 2373s | game as a whole so we will probably |
| 2376s | adjust this in the future just to make |
| 2378s | sure that that's a bit clearer but yeah |
| 2380s | interesting nonetheless then we wanted |
| 2385s | to ask people a little bit about what |
| 2387s | areas of the game that they think that |
| 2389s | we should focus more on when it comes to |
| 2391s | performance in general and I can't see |
| 2395s | the image anymore yeah ok so basically |
| 2403s | people were mentioning that the two |
| 2405s | things that were obviously too high on |
| 2406s | my list was the framerate he's more |
| 2408s | stable so it's not going up and down and |
| 2410s | it's not sort of dipping and spiking and |
| 2412s | you know dropping when you're aiming and |
| 2414s | the stuff that we talked about earlier |
| 2416s | and the overall frame rate was too low |
| 2419s | you know coming behind that was that |
| 2422s | performance changes too much between |
| 2424s | patches so if you scroll down a little |
| 2426s | bit please so performance was changing a |
| 2429s | lot between patches was quite high on |
| 2431s | the list a lot of these other ones we |
| 2433s | have some data from you guys actually |
| 2435s | people typed in stuff which is really |
| 2436s | cool to see and we're not going to show |
| 2438s | that here because there's just too many |
| 2439s | entries but really interesting to see |
| 2441s | that stuff so this is the data for that |
| 2443s | question and the next one so this is |
| 2448s | related specifically to what we wanted |
| 2451s | to check between sort of 4.0 and 4.2 |
| 2455s | this question basically is asking you |
| 2458s | know did you guys have these stalls and |
| 2459s | starters in earlier patches and that |
| 2461s | could be anywhere from you know from 2.5 |
| 2463s | onwards we want to know about this and |
| 2464s | what was interesting is a lot of people |
| 2466s | were having the stalls I think that was |
| 2468s | very so again it's one of the reasons we |
| 2472s | were focusing so much on that and so |
| 2474s | on the smooth experience was because so |
| 2476s | many people were having issues with this |
| 2478s | next one please and this essentially is |
| 2483s | the kind of follow-up to this question |
| 2485s | so what we see in here is after people |
| 2488s | tried 4.2 on the tests over after they |
| 2491s | played it they said that they had a much |
| 2494s | lower degree of scores and much lower |
| 2496s | you know issues that were happening |
| 2498s | after 4.2 when the experience of |
| 2500s | generally smoother which is positive and |
| 2502s | it's a you know it's a good thing from |
| 2504s | our sight but again we need to see how |
| 2506s | that behaves on the below servers yeah |
| 2509s | okay yeah that gives multitasking yes |
| 2515s | it's not worth installing sometimes yeah |
| 2518s | and stuttering so we asked you about |
| 2522s | also a little bit about how you describe |
| 2524s | these stalls status or something that |
| 2526s | was kind of difficult for us in the past |
| 2527s | was that and people describe you know I |
| 2530s | have frame drops or I'm lagging or yeah |
| 2533s | this pictures come in so much variety |
| 2536s | like and for some people yeah it's fine |
| 2539s | and others say oh it's horrible and they |
| 2541s | describe exactly the same thing yes I |
| 2545s | mean we try to kind of break that out |
| 2546s | and and come up with some descriptions |
| 2548s | that kind of make sense for you know |
| 2550s | performance stalls and you know a lot of |
| 2553s | you described it as basically being kind |
| 2555s | of prolonged periods where your frame |
| 2557s | rate had dropped down but had never kind |
| 2559s | of come to a stop which is what the the |
| 2562s | big blue bar here is here you know that |
| 2565s | you're not fully freezing but you do see |
| 2567s | like a moment where your frame rates |
| 2568s | like kind of collapsing down to you know |
| 2570s | a lower number and then maybe kind of |
| 2572s | recovering again so it's interesting to |
| 2575s | see this it's interesting to see you |
| 2576s | know what kind of issues people have and |
| 2579s | to what degree they have it and again we |
| 2581s | need to follow up on that on the test |
| 2583s | service one with your line yeah then we |
| 2588s | have some just general questions about |
| 2589s | satisfaction so this is again related to |
| 2591s | the patch on the test server so this is |
| 2593s | only on the test so hopefully it means |
| 2597s | that a lot of the issues have been |
| 2599s | resolved we'd like to think so yes our |
| 2601s | job is done yeah we can go home yeah so |
| 2605s | that's |
| 2606s | we want we're not going to just expect |
| 2609s | that to be the case we're not going to |
| 2611s | take the results and just say yep |
| 2612s | everything's perfect what we need is we |
| 2614s | need you guys to keep telling us your |
| 2617s | your thoughts fill in the surveys |
| 2619s | feedback us tell us if it got better for |
| 2621s | you got worse for you tell us what kind |
| 2624s | of issues you have and be vocal about a |
| 2626s | really tell us s it's super important |
| 2628s | because we need the with that |
| 2629s | information we can you know plan in |
| 2632s | favor of this yeah exactly |
| 2634s | Thanks 1 and when it comes to |
| 2638s | prioritizing this was one of the final |
| 2641s | questions we had in the survey which was |
| 2642s | actually one of the most useful for us |
| 2644s | was we can see what it is that you guys |
| 2646s | are more interested in when it comes to |
| 2649s | performance as in do you want us to |
| 2651s | focus on smoothing the experience or |
| 2653s | improving the frame rate overall or |
| 2656s | changing the graphical quality or giving |
| 2658s | more options and you know things like |
| 2660s | this and I don't want to talk too much |
| 2662s | about this but you know have a look at |
| 2664s | the graph and you know getting tell us |
| 2666s | more about it if you guys feel like we |
| 2669s | should be prioritizing something more |
| 2671s | for the other then you know communicate |
| 2673s | with us and let us know and we'll react |
| 2676s | on my own yeah and then the last |
| 2682s | question so what we did is we wanted to |
| 2686s | know from you guys are we doing enough |
| 2688s | to make performance better and the only |
| 2690s | people that can tell us this is you |
| 2692s | internally we can look at performance |
| 2694s | analysis we can look at the data we have |
| 2696s | we can look at the performance we get in |
| 2697s | the game but at the end of the day we |
| 2699s | need you guys in terms of the |
| 2702s | cross-section of hardware that you have |
| 2703s | the time that you've put into it the |
| 2705s | experience that you have and you need to |
| 2708s | tell us are we doing enough to make the |
| 2711s | experience good for you so I leave you |
| 2714s | with this one and you know when the |
| 2717s | patch is live fill in the survey again |
| 2719s | and let us know that would be really |
| 2720s | great yeah I hope that wasn't too tough |
| 2725s | I mean graphs and stuff it's kind of |
| 2727s | boring lots of words cool numbers |
| 2735s | all right lift it run back here in a |
| 2738s | system |
| 2745s | [Music] |
| 2757s | [Music] |
| 2759s | [Applause] |
| 2763s | no it's not it's good I was gonna say |
| 2769s | something really this attested it must |
| 2776s | have been like when you got up it must |
| 2778s | have like this had a tiny tiny tremor |
| 2783s | you say like Casagrande tremor yeah it |
| 2787s | must go in the chair clearly there were |
| 2789s | just a question of update trust me the |
| 2802s | street it's it's part of it you know |
| 2803s | it's like the traditional feeling where |
| 2805s | you go back to the stream the three |
| 2807s | spinning while actually since we've seen |
| 2811s | you |
| 2812s | yeah but we have it's it's there's |
| 2815s | always something audio is like the bane |
| 2816s | of my existence it will never work |
| 2818s | should we just buy it you Mike it's not |
| 2820s | them I think it's a cable Pazuzu money |
| 2822s | buy a new cable yeah that would be nice |
| 2824s | definitely if anyone wants to support us |
| 2827s | and to send some cables it would be |
| 2830s | really cool |
| 2832s | so we got some questions I'll just go |
| 2836s | through them and I also saw one that I |
| 2839s | already wrote down that I saw during |
| 2840s | history and then we'll catch some from |
| 2844s | the stream which is a little bit harder |
| 2847s | right now so just answer these first so |
| 2849s | if you have a good performance question |
| 2851s | I know you guys are very eager for |
| 2852s | content but performance questions that |
| 2854s | they can answer hold on to decide and |
| 2858s | then let us know in a second so one of |
| 2861s | the first questions that I got was |
| 2862s | skiing to give us any info about Vulkan |
| 2864s | API renderer it's supposed to give |
| 2866s | better performance and I wrote eggs and |
| 2868s | it's on the roadmap so it depends |
| 2873s | Okin and DirectX 12 effects in the cpu |
| 2876s | performance so if you do feel about |
| 2878s | bottlenecked not really for cpu |
| 2882s | performance yes probably but we are not |
| 2885s | exhausted |
| 2886s | options what we can do is their kicks of |
| 2887s | 11 and B yeah it will come in the future |
| 2891s | the improvements there so we first want |
| 2894s | to see how far we can push that and only |
| 2896s | then another thing is implementing new |
| 2899s | VI is very major undertaking not only |
| 2903s | from just getting it to work but getting |
| 2905s | it to work stable reliable and on the |
| 2908s | variety of hardware like going back to |
| 2910s | the topic pc is such a big field so yeah |
| 2916s | it's a lot of work to do and right now |
| 2918s | at least yearbooks of statistic of |
| 2921s | experience call it runs under for people |
| 2925s | coming again we always never say never |
| 2927s | and I can't really comment too much |
| 2929s | about what we're doing internally with |
| 2931s | stuff like this but given the Crytek's |
| 2934s | history has been very technically |
| 2936s | booming and very kind of graphically |
| 2937s | driven other stuff it's probably if a |
| 2939s | fairly safe thing to assume that it's |
| 2941s | somewhere in the pipeline but we know we |
| 2943s | can't we don't have anything new to tell |
| 2950s | you about it at the moment because right |
| 2952s | now our priority is just make sure that |
| 2954s | performance is where it needs to be at |
| 2956s | present dx12 is not going to help with |
| 2958s | that at present so we've been focusing |
| 2961s | on the things that we can make it |
| 2962s | difference with which is in this cases |
| 2964s | human source what is the end goal for |
| 2968s | performance what will be the minimum |
| 2970s | specifications that people could run at |
| 2972s | 1080p 60fps I think we touched on that |
| 2975s | one earlier I think we said so what our |
| 2978s | goal is always gonna be on PC for |
| 2981s | performance shooter is we want to get as |
| 2982s | many people as we can to 60 frames per |
| 2985s | second or tank yeah now that depends |
| 2988s | enormously on the hardware that you have |
| 2990s | and the settings that you use but with |
| 2993s | the recommended specs hardware that we |
| 2995s | assigned to the game 60 frames at 1080 |
| 2998s | we think is something that I would my |
| 3000s | high or it should be on ideally medium I |
| 3003s | mean medium and recommended you should |
| 3004s | go hand-in-hand |
| 3005s | what we do is we tend to give you enough |
| 3007s | options hopefully to make those |
| 3008s | adjustments yourself but you know that |
| 3011s | will take a little bit of time probably |
| 3012s | to get to that exact sort of frame rate |
| 3015s | we're not I don't think we're too far |
| 3016s | off that at the moment but |
| 3018s | people will probably be a bit more |
| 3020s | willing to accept slightly lower frame |
| 3022s | rate as long as it yeah but you know our |
| 3025s | goal is always going to be 60 frames |
| 3027s | 1080p recommend aspect we're not we're |
| 3031s | not decreasing it it's more for the |
| 3033s | recommended spec which is probably not |
| 3035s | you computer to make it that everyone |
| 3039s | even with like potato Dorito wasn't the |
| 3042s | reto computers can play the game yeah if |
| 3044s | you have a really high end machine it's |
| 3046s | not like we want to reduce it was oh |
| 3064s | yeah I know that we were planning to |
| 3066s | have more graphical settings |
| 3068s | yeah at the moment it's not we're not |
| 3073s | going to be adding new graphical |
| 3074s | settings yet the specs what we want to |
| 3078s | do it actually goes back to one the |
| 3081s | questions we're going to talk about a |
| 3082s | little bit later but we can cover it now |
| 3083s | something that I personally would like |
| 3085s | to improve in 5.0 since I'm saying on a |
| 3089s | stream that means that we're obviously |
| 3090s | committing to it so no brightness might |
| 3094s | i I would take full responsibility for |
| 3096s | it we wanted to make sure that the level |
| 3100s | of vegetation detail would be similar on |
| 3102s | low spec medium spectrum high specs so |
| 3104s | right now low spec you have a lot of |
| 3107s | detail vegetation and grass that's |
| 3108s | missing |
| 3109s | yeah and that shouldn't really be in |
| 3111s | 2019 we don't feel we turned off some |
| 3114s | stuff because we wanted to just make |
| 3115s | sure that people would lower in Hardware |
| 3117s | had a better chance at getting good |
| 3118s | framing now that we've done some |
| 3120s | optimization we kind of feel like it |
| 3122s | should be possible for everyone to have |
| 3123s | that and everyone should have that level |
| 3125s | of quality |
| 3126s | nice-looking blended yeah I mean it's |
| 3128s | it's mostly gonna be a GPU so thing so |
| 3131s | we kind of feel that it's it's good if |
| 3133s | you'll expect you should have a much |
| 3134s | more comparable graphical quality to |
| 3136s | other specs as well there was I think we |
| 3141s | have a question here about graphical |
| 3142s | downgrade yeah yeah I'm pretty sure we'd |
| 3144s | get to that as well I cover that quickly |
| 3146s | now okay so let me see |
| 3150s | I think that was your |
| 3161s | you |
| 3182s | you |
| 3196s | okay just in one question |
| 3199s | okay okay I'll just quickly repeat a |
| 3203s | question for tomorrow stream K explained |
| 3205s | the differences in a state of graphics |
| 3206s | of the launch bill than what we have now |
| 3208s | what changes were made what change has |
| 3211s | made the biggest visible difference in |
| 3212s | quality and what change has made the |
| 3214s | biggest difference in frame rate without |
| 3215s | a visible change I know that many people |
| 3218s | have complained by graphic downgrades |
| 3219s | and the developers might like to answer |
| 3221s | them directly in the changes this is not |
| 3224s | a question this is an essay it's pretty |
| 3226s | long I just really got what they were |
| 3229s | condensed it down so what happens to the |
| 3232s | graphics over time what have we done |
| 3233s | graphics white to the game's launch and |
| 3236s | what makes the biggest something like |
| 3239s | that so the biggest thing that we did |
| 3243s | was I believe it was in 2.5 we |
| 3245s | introduced the graphical options to |
| 3246s | player so initially when I launched you |
| 3248s | had one setting and that was it which is |
| 3251s | not good 2.5 we actually added the |
| 3254s | ability to set high medium a lot of |
| 3256s | different things and individual spec |
| 3257s | settings but for each object group or |
| 3260s | vegetation things like this now I think |
| 3263s | what this is probably mostly referring |
| 3265s | to hope that the person who's listening |
| 3267s | who wrote the second correct as before |
| 3269s | on there was an actual issue I think it |
| 3272s | was perhaps 2.5 to 3 oh yeah there was |
| 3279s | there was a kind of a criticism that was |
| 3281s | that we removed a lot of stuff for |
| 3284s | performance and I'm so far I can tell |
| 3286s | you now that we have never deliberately |
| 3289s | reduced the amount of stuff we render |
| 3291s | for performance except for when we |
| 3294s | changed you know high medium and low |
| 3295s | when we reduce those because you have to |
| 3297s | separate the Spector there was a part of |
| 3300s | a patch that actually removed a lot of |
| 3302s | the vegetation and that was kind of |
| 3305s | interpreted as us optimizing and it |
| 3307s | wasn't it was a bug there going in and |
| 3309s | we fixed it and bought the vegetation |
| 3311s | back and 45.0 we would actually like to |
| 3314s | take that another much upwards and |
| 3316s | actually gave low spec people |
| 3318s | more grass yeah more detail and stuff in |
| 3320s | there as well and even then like |
| 3322s | technically the throw distance doesn't |
| 3324s | change much between aspects so you |
| 3327s | shouldn't have a competitive difference |
| 3329s | nothing like you see something to risk |
| 3331s | op you should see it on all suspects |
| 3333s | yeah yeah apart from the box oh yeah |
| 3339s | yeah because I know a lot of people |
| 3341s | always like I went the lighting back |
| 3343s | from like the launch build but like |
| 3345s | change doesn't like artistic change |
| 3348s | right yeah yeah in many cases yeah I |
| 3353s | mean like we said like our goal is never |
| 3355s | to make things look worse for |
| 3356s | performance I mean there's there's lots |
| 3358s | of things that we can do to maintain |
| 3359s | performance without degrading with your |
| 3361s | quality and if visual quality has ever |
| 3362s | degraded it's not something that's our |
| 3364s | goal our goal is to give the best of |
| 3367s | yeah we're making a game because yeah |
| 3371s | exactly |
| 3372s | okay um let's see okay the current |
| 3375s | temporal anti-aliasing used |
| 3380s | some senses are not written very |
| 3382s | correctly but use the current temporal |
| 3389s | anti-icing use causes oh yeah so what |
| 3394s | this is asking is that some of our antia |
| 3395s | lighting can cause ghosting around I |
| 3398s | think specifically weapons so scopes so |
| 3404s | when you're moving you get this kind of |
| 3405s | ghosting effect yeah you should just |
| 3409s | read the whole question it was more but |
| 3410s | I was like hmm I didn't see them so I |
| 3412s | made sure the question was that there is |
| 3414s | a ghost in Carter part rendering craft |
| 3416s | apart and will you fix this and they so |
| 3422s | we know about this artifact we fear it's |
| 3425s | not necessarily caused by te |
| 3427s | implementation it's probably the mix of |
| 3429s | TA maybe the or mr. Specter it might be |
| 3433s | the global illumination actually |
| 3434s | affecting it a lot we have to see how to |
| 3437s | improve that okay but it's not as simple |
| 3440s | according to performance issues so |
| 3441s | where's the constellation say it is |
| 3444s | something that I know a few people have |
| 3445s | mentioned before it's it's not the |
| 3447s | easiest issue to fix unfortunately it's |
| 3449s | actually |
| 3449s | conflict between two kind of rendering |
| 3451s | things that we're trying to do at the |
| 3452s | same time and it's basically if we spend |
| 3455s | time fixing that we spend less time |
| 3457s | optimizing or less time fixing more |
| 3459s | critical stuff so it's not that we don't |
| 3461s | want to fix it it's not that we don't |
| 3462s | care about the issue it's more just we |
| 3464s | have to assign people in the right it's |
| 3472s | the new performance update going to fix |
| 3473s | us sometimes metal sheets or walls don't |
| 3475s | render and give people one way wall so |
| 3480s | it's a buck it's not the performance |
| 3482s | like not necessarily a performance like |
| 3485s | consideration we specifically were |
| 3488s | addressing vixen in proper |
| 3491s | implementation of rendering 3d engine mm |
| 3494s | depends there's not sure because a |
| 3497s | particular fix is no I mean so the thing |
| 3500s | with that bug is it keeps coming back |
| 3501s | and going away and coming back and going |
| 3503s | away and it's we've investigated it |
| 3506s | quite a lot and it's it's quite a very |
| 3508s | low-level problem it's quite a difficult |
| 3510s | one to solve and if with all of these |
| 3513s | kinds of issues like if it were easy to |
| 3515s | fix with this specific one it is it is |
| 3520s | important to get it fixed it has to be |
| 3522s | fixed because it's giving advantage of |
| 3523s | this advantage other players and |
| 3525s | something that we can't accept it's not |
| 3526s | good but it takes quite a lot of effort |
| 3530s | to look into that one it is quite low |
| 3532s | level it's quite difficult it's possibly |
| 3533s | countdown and the driver level of the of |
| 3537s | the rendering and other stuff like that |
| 3538s | but it's something we will I can tell |
| 3546s | you now I've done it's important it's |
| 3548s | important to us because it's important |
| 3550s | to the balance of the game and we can't |
| 3552s | have walls missing and everybody's |
| 3553s | looking for it's just the performance |
| 3558s | difference between system assessment' |
| 3560s | single channel memory and those were |
| 3562s | dual channel is massive there's 30 FPS |
| 3565s | or now write more in load one percents |
| 3569s | and n FPS more on average when going |
| 3571s | from single to dual channel memory gone |
| 3572s | big the difference is huge most of the |
| 3574s | games get no benefit from more memory |
| 3576s | channels there are very few exceptions |
| 3578s | like that of the five could you please |
| 3580s | explain why |
| 3581s | my hunt put such a huge demand all |
| 3583s | around bandwidth that's a big question |
| 3586s | it's a lot of those big questions so in |
| 3591s | general the game is generally using lots |
| 3593s | of memory actually more than we expect |
| 3598s | more than be said for us as they like |
| 3600s | memory budget so I remember a budget was |
| 3602s | less than 8 gigabytes closer to |
| 3604s | something like 5 6 on PC and then of |
| 3607s | course it may increase when you set your |
| 3608s | talk I like support what some textures |
| 3611s | need more memory |
| 3613s | yeah be working on this so like next two |
| 3617s | days we already internal worked and a |
| 3619s | significant improvement some how much |
| 3622s | may be consumed and yeah apart from that |
| 3625s | we were not getting much into specific |
| 3628s | differences between like this card about |
| 3630s | configurations yeah we probably have to |
| 3633s | spend a bit of time as well seeing the |
| 3636s | spectrum of people that actually are |
| 3638s | affected by this issue is something we |
| 3640s | have to but we can tell people go and |
| 3647s | set your computer in this mode or go buy |
| 3649s | better rom it's always just download |
| 3654s | there's some websites to be proposed |
| 3656s | that yeah so be with the roster |
| 3660s | disclaimer don't actually do that yeah |
| 3662s | no don't but we would rather spend our |
| 3666s | effort on changing what we can change |
| 3669s | yeah hard part because really change we |
| 3672s | can adapt to that we need to make sure |
| 3673s | that the game runs well no matter |
| 3675s | whether you have you know dual channel |
| 3677s | Ram it's it's just strange to kind of |
| 3681s | exactly what I said is we can't expect |
| 3682s | people to adjust their hardware to our I |
| 3684s | don't enjoy the software works for |
| 3686s | everyone |
| 3687s | yep um let's see ok this is a nice short |
| 3692s | start the question hey please explain |
| 3694s | the difference between borderless and |
| 3696s | fullscreen windows in technical terms |
| 3698s | one has a border well that's something I |
| 3700s | set that to as entering nailing no I see |
| 3703s | the both of that border so basically one |
| 3705s | of them allows and I have to be probably |
| 3708s | technically checked against this but I |
| 3710s | think one of them allows you but |
| 3712s | the OS can actually like put stuff on |
| 3715s | top of the other one and give us some |
| 3717s | notifications and overlays and the other |
| 3719s | one is like the priority is the game the |
| 3722s | game has all the priority I don't know |
| 3725s | if that helps because I'm not |
| 3728s | particularly like an engineer or rather |
| 3730s | like if you know windows API that well |
| 3733s | to have a specific differences really |
| 3735s | sure how our implementation is done so I |
| 3738s | wouldn't really be able to tell |
| 3740s | technical in technical terms but in |
| 3743s | general would suggest that this is |
| 3744s | correct like in portals window it's the |
| 3746s | window |
| 3747s | so windows compositors still have to do |
| 3749s | all the work like render windows behind |
| 3751s | being rendered windows in front that |
| 3753s | composite them together on a single |
| 3754s | picture when you're going to promote |
| 3756s | windows can say oh you you definitely |
| 3759s | will never be seen yes Ben I'm not even |
| 3761s | bother to throw you I guess it's like |
| 3762s | the between having an office with the |
| 3764s | door open or door closed doors open |
| 3767s | people can bug you and sound can get in |
| 3769s | and annoy you but whether the door is |
| 3770s | closed you're exclusive and you're on |
| 3772s | your own without being interrupted shit |
| 3778s | I see what kind of office you describe |
| 3783s | [Laughter] |
| 3786s | where you guys be implementing and |
| 3789s | videos ELSS or any or end or ray tracing |
| 3793s | to Gabriel it's really good for the |
| 3795s | experts could really benefit from the |
| 3797s | u.s. we will focus on making sure that |
| 3802s | is as good a game as it can be for |
| 3804s | everyone first before we start looking |
| 3806s | into additional features and and |
| 3808s | supporting things that may be not |
| 3811s | supporting the riches we take from rich |
| 3813s | and gifts facing stuff is really awesome |
| 3817s | it's really cool but at the moment our |
| 3820s | priority has been we need to get |
| 3821s | performance going I need to make sure |
| 3823s | that yeah it's it's good for it so right |
| 3825s | now I can't say yes or no I'm just gonna |
| 3828s | say that we're focusing program |
| 3830s | performance primary and the lights |
| 3831s | turning off again yeah yeah it's like |
| 3835s | maybe we should just throw something at |
| 3837s | it out and others |
| 3842s | so while I was trying okay this is the |
| 3845s | longest while I was trying to reduce the |
| 3848s | stutters thing I noticed that produced |
| 3850s | the starburst of most was to say Blake |
| 3851s | the control flow guard in the exploit |
| 3853s | protection of the security section of |
| 3856s | Windows settings later noticed that this |
| 3858s | particular issue has been affecting most |
| 3860s | computers since the major windows update |
| 3862s | almost a year ago and was much worse |
| 3864s | than this |
| 3865s | all that time ago my question is this |
| 3869s | are you aware of this particular issue |
| 3871s | and if you are what can I do to achieve |
| 3873s | the same effect religion stutter while |
| 3875s | keeping control flow guard on since its |
| 3878s | since its off it reduces my computer so |
| 3880s | I guess we need to ask or need to answer |
| 3883s | what is control flow guard it's a |
| 3890s | Microsoft Windows feature to protect |
| 3895s | programs from being tampered with |
| 3898s | malware this viruses this kind of like |
| 3901s | malicious software so first no we were |
| 3906s | not aware that it might be a case or |
| 3908s | rather because it's a big news feature |
| 3910s | it's not necessarily a game feature we |
| 3913s | were not actually looking into that from |
| 3917s | what I know the control guard uses again |
| 3920s | quite a big chunk of memory so my hope |
| 3923s | is that reducing memory consumption may |
| 3925s | affect the effectiveness of control card |
| 3928s | as well or rather you not become |
| 3930s | constantly page swapped from these can |
| 3932s | to memory to these to memory so that may |
| 3935s | improve the stutters apart from that ya |
| 3938s | know we didn't really look to this issue |
| 3939s | yet I literally do today is the first |
| 3942s | time I talked about it might be |
| 3943s | improvement but I wouldn't recommend |
| 3947s | given that this is a security feature of |
| 3949s | the operating system I wouldn't |
| 3951s | recommend to anyone just to go and |
| 3952s | disable it sounds dangerous I think it |
| 3956s | comes back to the question we answered |
| 3957s | earlier which is if you have to kind of |
| 3958s | do weird workarounds to make hunt work |
| 3961s | that's a bad thing and it shouldn't be |
| 3963s | the case |
| 3965s | making sure that you don't need to do |
| 3967s | that and it it just works out of the box |
| 3969s | and that's really what the goal is for |
| 3970s | us |
| 3971s | there was another quick that was the |
| 3973s | last question well yeah I wonder the |
| 3974s | jeopardy if you have one I saw one just |
| 3977s | actually from killer grease who asked |
| 3979s | about my model quality and they saw |
| 3982s | something we've talked about I think |
| 3983s | last which was we did say at some point |
| 3989s | we wanted to do a very high spec we've |
| 3992s | not given up on the idea of doing a very |
| 3994s | high spec but it was important for us to |
| 3996s | make sure that the performance had got |
| 3998s | more focus before we started doing yeah |
| 4001s | super high-end graphics yeah I would |
| 4007s | love to have a high-end spec and just |
| 4010s | you know really turn on everything and |
| 4011s | have people you know the coolest like |
| 4013s | super awesome 4k everything it will |
| 4018s | hopefully come I'm not gonna guarantee |
| 4020s | it's something we want to do first |
| 4023s | performance for the whole demo think |
| 4027s | about I see someone say really high spec |
| 4029s | on 429 yes okay I saw a question early |
| 4034s | image at happy thought of lowering the |
| 4037s | AI to boost performance I mean in |
| 4041s | general here's the entities and did some |
| 4045s | amount of identities and regulators in |
| 4047s | the world affects the performance a lot |
| 4049s | not performance network traffic all that |
| 4052s | this is the reason why we actually test |
| 4054s | visited and ten players in the game |
| 4055s | because the having ten players actually |
| 4058s | makes the game to populate the world |
| 4060s | with the I like when area doesn't player |
| 4062s | in there there's no yen they're |
| 4067s | decreasing the perform or decreasing the |
| 4069s | accounts would be a way to find it but |
| 4071s | at the same time if one player to be |
| 4073s | entertained by amount of AI characters |
| 4075s | around them this was one of the |
| 4076s | considerations why we were cautious |
| 4079s | about increasing the amount of players |
| 4080s | because yeah technically game was |
| 4082s | supported but the performance will |
| 4085s | suffer from be increased so either |
| 4088s | perform suffers or the experience |
| 4089s | suffers doesn't like player go in |
| 4091s | compound and it's empty |
| 4092s | yeah basically that we will our mantra |
| 4096s | internally has always been that we will |
| 4097s | never |
| 4097s | want to adjust gameplay for the sake of |
| 4100s | performance yeah we would not want to |
| 4103s | start with cutting features or reduce |
| 4105s | quality or make things from a gameplay |
| 4108s | perspective worse the one exception that |
| 4111s | we've done recently for that actually in |
| 4113s | this patch was the fusee time so this is |
| 4116s | something that is in the patch notice is |
| 4117s | something that we had to adjust which we |
| 4119s | can discuss quickly so we found that |
| 4121s | there were some performance issues with |
| 4122s | that because they have a very long time |
| 4124s | that they they were quite heavy on GPU |
| 4128s | especially even for very high-end |
| 4130s | machines they have some some cost to |
| 4132s | them so for this patch we reduce those |
| 4134s | to a five-minute timer and this is one |
| 4137s | at one example we've done where we have |
| 4139s | made a gameplay adjustment for the sake |
| 4141s | of performance but it's not something we |
| 4143s | want to do for long-term so long term we |
| 4145s | want to optimize them and make them how |
| 4147s | they were but just run faster and we |
| 4150s | just didn't have the bandwidth to do |
| 4151s | that for this patch so for this there |
| 4153s | will be a little bit less time but you |
| 4156s | know we will go in and optimize already |
| 4162s | changed I think it was always a 10 we |
| 4167s | can check if someone then is he will |
| 4175s | answered it but ok so some more of |
| 4182s | performance questions that they could |
| 4184s | answer let me know to check the lights |
| 4190s | gone off again |
| 4191s | this room is haunted this is her isn't |
| 4193s | it I think it's just me I think I wants |
| 4196s | it he did not pay attention what was the |
| 4200s | question that we do it is actually |
| 4202s | Dennis it was about the fuses are you |
| 4209s | planning to put CryEngine at its best |
| 4212s | in-game coming back to that yes the |
| 4216s | question was are we gonna be putting a |
| 4218s | nice pic awesome yeah okay we just |
| 4220s | answered that yeah I wouldn't see hunt |
| 4222s | it yeah it's hunting all I can say is |
| 4225s | just in summary |
| 4227s | the questions that people had is please |
| 4230s | check it out long German it's what we |
| 4232s | always had sorry yeah yeah okay well |
| 4234s | that's good to know |
| 4235s | okay then I just maybe just never went |
| 4240s | like I'm high - Archie by the way yes |
| 4244s | getting special treatment it's getting |
| 4245s | special treatment it's getting cold out |
| 4250s | well okay well the performance be up to |
| 4253s | par in 5.0 we'll dig up keep updated |
| 4255s | after the goal is to always make it |
| 4258s | better yeah so I know that you say I |
| 4261s | want content but no worried content is |
| 4263s | coming I want to quickly say something |
| 4266s | about that we are going to update the |
| 4268s | roadmap very soon but we start with the |
| 4271s | short term so what is coming very soon |
| 4275s | so what we're gonna do is we're gonna |
| 4278s | work on the roadmap that will be up to |
| 4280s | 1.0 and some post 1.0 so like before |
| 4284s | release and what's going after but we |
| 4287s | will give you more information when we |
| 4289s | update it hopefully this week so I |
| 4292s | tonight which it out so probably |
| 4295s | tomorrow so you guys can keep an eye on |
| 4298s | that which has content and obviously as |
| 4301s | you might know the boss unfortunately is |
| 4303s | a little bit delayed but we have gotten |
| 4306s | way better performance now which is more |
| 4308s | important for us but wasn't our fault we |
| 4311s | didn't delay the boss |
| 4312s | I mean the reason why we delayed it is |
| 4314s | because we wanted to have to prefer mr. |
| 4316s | people can actually play the boss yeah |
| 4318s | the boss was the thing that broke the |
| 4320s | performance right now it's it wasn't I |
| 4322s | didn't break first |
| 4324s | but it's we want to make sure that |
| 4325s | everyone can actually play the game and |
| 4327s | enjoy the boss because when we had the |
| 4330s | update after 3.0 the performance went |
| 4333s | down and people actually said like I |
| 4335s | can't play the game anymore which we |
| 4336s | obviously don't want to so I know some |
| 4340s | people are like I have this high end PC |
| 4341s | I never have issues but guys think about |
| 4344s | everyone it's not I think we should get |
| 4347s | more werewolf memes going yeah we get we |
| 4350s | keep getting a lot of werewolf in art |
| 4353s | let me see why his hair works |
| 4358s | can't ask questions I can't answer |
| 4361s | questions about the bosses just |
| 4362s | community of just performance questions |
| 4365s | right now so if you have a performance |
| 4367s | question still left we will do our best |
| 4369s | to finish it we already asked it was |
| 4371s | like that we never changed anything |
| 4373s | specific for the games do not look like |
| 4375s | III video obviously which is better |
| 4381s | you mean the video no yeah they said |
| 4385s | they want the performers are how it |
| 4387s | looked back in the e3 video they think |
| 4389s | that the graphical quality had gone down |
| 4391s | between these three years yeah it's |
| 4394s | possible that there was some more high |
| 4397s | quality effects purely because we had a |
| 4399s | single speck before so we had only high |
| 4403s | spec perform and in took 1/5 that was |
| 4405s | split into low medium and high so a lot |
| 4407s | of stuff got turned off for media my |
| 4408s | lover so maybe that is one of the |
| 4410s | reasons for that but if you guys have |
| 4412s | some specific kind of concerns about |
| 4414s | that specific question zone yeah I want |
| 4416s | me to send them our way yeah you can in |
| 4419s | general also messages in the discord so |
| 4421s | we can get specific things now we know |
| 4424s | who serve attention like Pope yeah no |
| 4426s | it's like those labs that turn off |
| 4428s | automatically after nobody's moving and |
| 4430s | one of them is kind of far away so let |
| 4434s | me see netcode is something we can't |
| 4438s | really answer we are still looking into |
| 4439s | the servers we know that there are some |
| 4441s | people that have experience hyping we |
| 4443s | also if you haven't already I just |
| 4449s | realized what I said it was personal so |
| 4454s | we have to just also released a blog |
| 4457s | last week this week this week about our |
| 4460s | hit registration and that's dude also |
| 4463s | with the server reaction stuff I do |
| 4465s | think there might be a chance for else |
| 4467s | to make another blog that explains more |
| 4468s | because I know what people want more |
| 4469s | details also about travel bullet speed |
| 4471s | and such of that and I'll make sure I |
| 4474s | get on to that too you guys can get |
| 4476s | another vlog hopefully sometime soon |
| 4479s | with more information about that |
| 4482s | merchandise is something we're still |
| 4484s | working on |
| 4486s | so yeah there is a question about this |
| 4490s | teaming performance of MIT much so I |
| 4492s | guess textures |
| 4493s | I guess it's blurriness my rom just |
| 4497s | can't handle dependence is too damn high |
| 4500s | yeah we're working on improving the |
| 4502s | streaming system so it's like besides |
| 4505s | better quality streaming works from the |
| 4513s | disk all right now the only like advice |
| 4517s | I can give is if you have SSD try to put |
| 4519s | them game on SSD it improves the stream |
| 4521s | like this access time yeah so it may |
| 4525s | improve the Hulk was the world case |
| 4528s | yourself yeah and there was another one |
| 4531s | I know that this not really related to |
| 4533s | performance directly but some people are |
| 4535s | asking about the reflection of the |
| 4536s | weapon in the water in the water I know |
| 4539s | this one has been a bug that's been |
| 4540s | around for a long time it's another case |
| 4543s | of if it was easy to fix we would have |
| 4545s | fixed it already yeah I can tell you |
| 4554s | anything like the player cans the player |
| 4567s | mode of the weapons they are not normal |
| 4569s | they're not going through the normal |
| 4571s | rendering function else they're like |
| 4573s | first-person mode rendering so therefore |
| 4575s | they not necessarily work well with some |
| 4578s | other rendering features it's definitely |
| 4581s | on the list of things that when we free |
| 4584s | up some resources after we fixed the |
| 4586s | performance side of things a bit more |
| 4588s | you know solving some of those weird |
| 4591s | graphical issues that everyone gets the |
| 4592s | rotated bias is something that we would |
| 4594s | like to do but I'm with you guys a |
| 4597s | hundred percent I find it is irritating |
| 4598s | if you do that it is a question of the |
| 4602s | resources that we have to spend on that |
| 4603s | and the priorities and stuff like this |
| 4605s | and it has been worked on but it's not |
| 4607s | ready for use |
| 4609s | yeah there's also a question about the |
| 4611s | shed of the gun on the ground it looks |
| 4613s | white I mean I think that's how much a |
| 4614s | similar case |
| 4615s | it's actually very similar in the reason |
| 4617s | because so the gun hands and everything |
| 4619s | they're special german special enigma |
| 4623s | but the shadow caster is normal |
| 4625s | third-person model and they probably |
| 4627s | mentioned together properly therefore it |
| 4629s | looks like a but I remember we had a |
| 4631s | buck about that and we need to I mean |
| 4634s | it's a buck we need to fix it and yeah I |
| 4636s | just don't remember specifically details |
| 4638s | what exactly it was broken there but |
| 4640s | something about this atop first person |
| 4642s | misses shadow casting third person let |
| 4646s | me see if there are no questions anymore |
| 4651s | for it now obviously people can still |
| 4655s | answer questions on the discord or on |
| 4658s | social media just send this message on |
| 4660s | either of our platforms and yeah I think |
| 4665s | that was pretty much it for today I |
| 4667s | think we have gotten a good explanation |
| 4668s | about what's going on with a performance |
| 4672s | currently what they've been doing for |
| 4673s | four point two which is live right now |
| 4675s | so definitely play it will have another |
| 4678s | survey up probably tomorrow or otherwise |
| 4681s | beginning next week let us know what you |
| 4686s | guys think please let us know how your |
| 4687s | performance is specifically for that |
| 4690s | patch so we can tell you know how it is |
| 4692s | compared to previous patches and and |
| 4694s | then hopefully you guys will have a good |
| 4697s | day so thank you guys so much for |
| 4699s | watching thank you our guests for being |
| 4702s | here and answering the question speed |
| 4705s | recap a speed recap or they wondered I |
| 4710s | mean okay we talked about before it's |
| 4712s | really hard cause it's like so audio |
| 4714s | kept breaking off we are haunted and |
| 4718s | that's about it we asked you a question |
| 4719s | as before minutes has gone better we |
| 4721s | still work on performance the content |
| 4722s | comes after it got to just so come and |
| 4724s | don't worry the boss along the way also |
| 4725s | another a is on the way and we have some |
| 4727s | questions that were very really 5.0 |
| 4730s | walking team will contain more |
| 4732s | optimization as well as content as well |
| 4734s | they will be a survey for the life |
| 4737s | service that we would really appreciate |
| 4739s | you guys filling out and tell us about |
| 4740s | your performance is happening |
| 4742s | apples apples bananas that's christosp |
| 4745s | as long your mechanics |
| 4748s | right yeah survey survey yeah |
| 4754s | CJ's broken |
| 4769s | [Music] |
| 4780s | [Music] |
| 4787s | [Music] |
| 4798s | [Music] |